Report No. 87913 Understanding the Impact of Climate Change on Hydropower: the case of Cameroon April 27, 2014 AFTEG Africa Energy Practice i Understanding the Impact of Climate Change on Hydropower: the case of Cameroon Climate Risk Assessment for hydropower generation in Cameroon by Johan Grijsen With ii Contents Executive Summary 1. Introduction and objective 1.1. Objective of the Climate Risk Assessment 1.2. Previous studies on climate risk to Cameroon’s water resources 1.3. Outline of the report 2. Decision scaling methodology for a risk-based assessment of climate change impacts on WR 2.1. Top-down approach 2.2. Decision - scaling methodology 2.3. Identification of climate hazards and thresholds 2.4. Vulnerability assessment: Assessment of system response to changes in runoff 2.5. Climate Informed Risks: Estimating likelihood of climate conditions and hazards 2.6. Summary of the adopted methodology for Climate Risk Assessment 3. Hydrometeorological data for the five main river basins in Cameroon 3.1. Cameroon’s river basins 3.2. Runoff data 3.3. Precipitation, temperature and potential evapotranspiration data 3.4. Trends and abrupt changes in rainfall across Cameroon 4. Runoff response to climate change 4.1. Assessment of climate elasticity of streamflow through regression analysis 4.2. Use of the aridity index to assess climate change impacts on annual runoff 4.3. Climate and hydrological modeling 4.4. Regression analysis of basin runoff, rainfall and temperature 5. Vulnerability analysis: impacts of future runoff changes on WR system performance indicators 5.1. Seasonal water management and hydro-energy generation model 5.2. Runoff elasticity of hydro-energy generation in the Sanaga basin 5.3. Runoff elasticity of the economic performance of Lom Pangar Hydropower Project 5.4. Target flow Nachtigal and additional seasonal storage capacity in Djerem Basin 5.5. Runoff elasticity of hydro-energy generation in other basins i 6. Climate change projections and impacts on runoff for the main river basins in Cameroon 6.1. Climate projections for the main river basins of Cameroon 6.2. Seasonal climate projections for Cameroon 6.3. Runoff projections from the Climate Portal 6.4. Runoff projections based on climate projections from the Climate Wizard 6.5. Runoff scenarios for the economic analysis of water and hydropower projects in Cameroon 6.6. CMIP-5 climate projections 7. Climate risks for hydropower generation in Cameroon 7.1. Hydro-energy Sanaga basin and EIRR Lom Pangar project 7.2. Lagdo dam in Niger Basin 7.3. Nyong and Ntem River Basins 8. Conclusions and recommendations References Annexes 1 Terms of Reference 2 Principles of climate change projections 3 Climate Risk Assessment for water resources development in the Niger Basin 4 Monthly runoff data series for key hydrometric stations 5 Annual and monthly data series of rainfall and temperature (CRU-TS 3.10) and runoff 6 Turc-Pike model for assessment of climate change impacts on annual runoff 7 Climate change projections for 2050 and 2080 from the Climatewizard ii Acronyms and Abbreviations AFTEG Africa Energy Unit of the World Bank BADC British Atmospheric Data Centre CC Climate Change CCKP Climate Change Knowledge Portal (WBG) CDF cumulative density function CGIAR Consultative Group of International Agricultural Research CRA Climate Risk Assessment εP Precipitation elasticity of runoff εQ Runoff elasticity of a performance indicator εT Temperature elasticity of runoff E Actual evapotranspiration E0 Potential Evapotranspiration (PET) EDC Electricity Development Corporation EIRR Economic Internal Rate of Return FAO Food and Agriculture Organization of the United Nations GCM Global Circulation Model (a.k.a. General Climate Model) GHG Green House Gas GRDC Global Runoff Data Centre, Koblenz, Germany GWh Giga Watt hour (energy) GWP Global Water Partnership H Head of power station (m) HP Hydropower IC Installed Capacity (MW) IPCC Intergovernmental Panel on Climate Change IWRM Integrated Water Resources Management MINADER Ministry of Agriculture MINEE Ministry of Energy and Water MINEP Ministry of Environment and Protection of Nature MW Mega Watt (power) iii NRB Niger River Basin P Precipitation (mm) PANGIRE Plan d’Action Nationale de Gestion Intégrée des Ressources en Eau Q Runoff, streamflow (m3/s or mm/year) RCM Regional Climate Model RDM Robust Decision Making SDAP Sustainable Development Action Plan (for the Niger Basin) SRES Special Report on (GHG) Emission Scenarios T Temperature (0C) TFESSD Trust Fund for Environmentally and Socially Sustainable Development WatBal Water Balance (hydrological) model WBG World Bank Group iv Acknowledgments The Africa Energy Unit (AFTEG) of the World Bank has obtained financing from the Trust Fund for Environmentally and Socially Sustainable Development (TFESSD) – funded by Finland and Norway - for work towards “Understanding the Impact of Climate Change on Hydropower Generation in Cameroonâ€?. This scoping study serves as a technical assistance to the Lom Pangar Hydropower Project to support the operation of the dam by the Electricity Development Corporation (EDC) of Cameroon. Historical climatological data as well as climate projections for the 21st century were obtained from internet sources, such as the CRU TS3.10 data set, the Climate Wizard, and WBG’s Climate Portal. Historical runoff data for about 20 selected sub-catchments in Cameroon were obtained from the Global Runoff Data Centre in Koblenz, Germany (GRDC 1 ), as well as from the Electricity Development Corporation of Cameroon (EDC). The study was carried out by Johan Grijsen (Consultant Water Resources Management) and Hrishi Patel (Consultant GIS). It was managed by Daniel Murphy (TTL) with assistance of Gabriella Puz and later of Farah Mohammadzadeh of the World Bank. The report was presented to a validation workshop organized jointly by the World Bank and EDC in Yaoundé, Cameroon on December 4, 2013. Many valuable comments and suggestions were made by the participants, which have been incorporated in this final report. 1 The Global Runoff Data Centre, GRDC in the Bundesanstalt für Gewässerkunde, 56068 Koblenz, Germany v Executive summary Main conclusion: the Lom Pangar and Nachtigal storage and hydropower projects in the Sanaga Basin are economically robust and climate resilient projects, but hydro-energy generated at Lagdo dam in the Benue basin may suffer a significant decrease due to climate change. Based on the presently available climate projections for the 21st century this study concludes that by 2050 the total long-term average hydro-energy generation by the Edea, Song Loulou, Lom Pangar and Nachtigal power plants in the Sanaga Basin could vary between -15% and +5% of the base case value (present hydrology); results for 2080 are similar. It is highly unlikely that hydro-energy generation would decrease more than 20% due to climate change. Most importantly, since the direct and indirect contribution of Lom Pangar to hydro-energy generation in the basin is overall fairly constant for runoff variations in the range of -20% to +20%, the EIRR is similarly not sensitive to runoff changes, unless decreases in runoff exceed 20%. In the worst case scenario the EIRR of the Lom Pangar project would be reduced with less than 10%, e.g. from about 14.5% in the base case (baseline post-1971 hydrology) to 13% under severe climate change impacts. Potential climate change impacts on hydro-energy generation at the Njock, Mouila and Memve-Ele stations in the Nyong and Ntem Basins are expected to be equally minor to moderate. The probability that the annual hydro-energy generation at anyone of these three plants would reduce with 20% or more is negligible. On the contrary, under the 2050 climate conditions there is – according to GCM projections - nearly 20% probability that the annual hydro-energy generation at Lagdo dam in the Benue Basin would reduce with 20% or more2 (30% probability by 2080). Thus, hydro-energy generation at Lagdo dam may suffer a significant decrease due to climate change, and is less climate resilient than Lom Pangar and Nachtigal. Objective of Climate Risk Assessment: This report presents a Climate Risk Assessment (CRA) for the five main river basins of Cameroon, focusing on the potential climate change impacts on water resources availability for hydro-energy generation, particularly in the Sanaga, Benue, Nyong and Ntem River Basins. Central and West Africa have experienced a marked climate variability in the 20th century and more distant past. Most recently, around 1970 an abrupt downward shift in precipitation is believed to have occurred, which reduced for example runoff from the Sanaga basin with about 16%. Lack of recent discharge records prohibits such assessments for other basins in Cameroon. However, even larger abrupt shifts were observed in the western part of the neighboring Niger Basin. Understanding the possible impacts of the present hydrological variability and the potential future changes in climate on existing and planned water resources and hydropower infrastructure is crucial for development planning. Therefore, the objective of this study is to carry out a CRA for the five main river basins in Cameroon and to assess the potential future impacts of climate change on water resources availability and hydro-energy potential in Cameroon. An effort is made to quantify the climate risks to future hydro- energy generation in the Sanaga and other basins and to the operation of the Lom Pangar storage reservoir for the 2050 and 2080 investment horizons. 2 One of five (20%) climate models indicate changes in P and T yielding at least 20% reduction in Lagdo hydro- energy output. vi Flexible and robust methodology for CRA: Given the substantial uncertainty in climate projections from Global Circulation Models (GCMs), a flexible and robust bottom-up, risk-based methodology was used to assess future risks of climate change to water resources availability and hydro-energy generation in Cameroon. This methodology seeks to identify the response of important performance metrics to parametrically varied changes in climate, and then uses available climate change projections to assess which of these responses are more or less likely. The reason for doing so is to first put the emphasis on understanding the current water resources systems, planned investments and their sensitivity to climate variation. Thus, the methodology focuses primarily on identifying potential climate hazards to infrastructure through the analysis of the sensitivity to changes in river runoff of performance indicators significant for hydro-energy generation. Seasonal water resources and hydro-energy system models (developed in Excel) were used to investigate the impacts of future changes in runoff on the performance of existing and planned hydropower facilities in the Sanaga and other basins in Cameroon. This analysis was based on monthly runoff data for the hydrological base line period 1971 – 2003. Subsequently, we used climate projections to assess the plausibility of such climate hazards. An ensemble of 15 climate projections for the 21st century, generated by 15 GCMs for the A1B emission scenario (provided by the climate wizard website3), was used to capture the possible future annual average climate across the five river basins of Cameroon in terms of probability distributions of precipitation and temperature changes. Analytic hydrological techniques were subsequently used to estimate the response of runoff to changes in precipitation and temperature, in terms of the climate elasticity4 of runoff. This climate elasticity was subsequently used to translate the projected relative changes in annual temperature and precipitation into projected relative changes in annual runoff, and derive a probability distribution of future runoff changes. Based on the earlier assessed sensitivity of system performance indicators to changes in runoff, we then translated for each basin the probability distribution of future runoff changes into the probability distribution of future changes in the selected performance indicators, and estimated the probabilities of exceedance of identified risk levels. Hydro-energy generation performance indicators: For this study we focused our analysis on performance indicators linked to hydro-energy production, e.g. guaranteed production (in MW) during the dry season and total annual energy production (in GWh/yr). Water use for irrigation/agriculture and domestic/industrial water supply will remain small in the foreseeable future, and the economic importance of river navigation is negligible. Environmental flows would need to be guaranteed under any future development and/or climate change scenario, and can thus be treated as a constant. 3 Climate Wizard: http://climateknowledgeportal.climatewizard.org and http://climatewizard.org. The Climate Wizard offers results for only 15 GCMs (a new version for 26 GCMs under CMIP5 is under preparation), but the climate sensitivity analysis of HP infrastructure analyses a wider range of potential future climates. However, similar results were obtained from Climate Change Knowledge Portal (http://climateknowledgeportal.org) of the World Bank for an ensemble of 22 projections, based on 22 GCMs. 4 The concept of elasticity is used to define the response of runoff to climate changes or the response of hydro- energy generation or a project’s EIRR to changes in runoff; e.g. a precipitation elasticity of runoff of 2.2 and runoff elasticity of hydro-energy generation of 0.8 implies that 10% change in annual precipitation would cause 10% * 2.2 = 22% change in annual runoff and 22% * 0.8 = 17.6% change in annual hydro-energy generation. vii Seasonal water resources and hydro-energy generation models (developed in Excel, using monthly flow data as input) were used to seek an understanding of how the present and future hydro-energy system of Cameroon will respond to changes in runoff caused by climate changes. This process identifies runoff conditions that cause unacceptable performance levels, and determines the runoff elasticity of hydro- energy generation, as well as the runoff elasticity of the economic performance of the new Lom Pangar reservoir and hydropower station (under construction). Runoff elasticity of hydro-energy generation in the Sanaga Basin: Under the present climate conditions (base case hydrology) the integrated operation of the new Lom Pangar reservoir – with a live storage of 6,000 MCM and an installed capacity of 30 MW – and existing Mapé and Bamendjing reservoirs is shown to satisfy a guaranteed flow of 600 m3/s (10% percentile) for the future Nachtigal Run-of-the-River power station. Lom Pangar reservoir significantly improves the dry season hydro-energy generation at Nachtigal, increasing the guaranteed capacity (10% percentile) with 104 MW to 263 MW and the overall annual generation at Nachtigal with 464 GWh/yr (21%). Similarly, Lom Pangar improves the guaranteed flow (10% percentile) at Song Loulou and Edea with 236 m3/s to 930m3/s. The total guaranteed capacity (10% percentile) of both power plants in the dry season will increase with 137 MW (34%) and the overall annual generation will increase with 553 GWh/yr (12.5%). Altogether, including Nachtigal, Edea, Song Loulou and the energy generation at Lom Pangar itself, the new reservoir enables an increase of the dry season guaranteed capacity with 46% from 563 to 824 MW, an increase of the average dry season load factor from 60% to 80%, and an increase of the overall annual generation with 18% from 6,613 to 7,807 GWh/yr. Lom Pangar reservoir enables a more uniform distribution of hydro-energy generation over the year; average generation for the four power stations ranges from 620 GWh/month during the dry season (7 months) to 693 GWh/month during the rainy season (5 months). The impacts of Lom Pangar reservoir on the (possible) future Song Ndong, Song Mbengue and Kikot power plants were also assessed, indicating again a significant increase in dry season load factors (about 15%). To determine the runoff elasticity of hydro-energy generation in the Sanaga Basin, climate change impacts on runoff were introduced by varying the monthly (naturalized) flow data for the period 1971 – 2003 parametrically, i.e. by adopting uniform changes ranging from +20% to -30% in steps of 10%. The direct and indirect contribution of the Lom Pangar reservoir to hydro-energy generation in the Sanaga Basin is shown to be significant and nearly constant for runoff variations in the range of -20% to +20%, contributing in this range nearly 20% to the average dry season load factor for Edea, Song Loulou, Nachtigal and Lom Pangar power stations5. Therefore, the annual variability of dry season hydro-energy generation under the present flow regime (without Lom Pangar reservoir) is more or less preserved in absolute terms and the runoff elasticity of dry season hydro-energy generation is shown to improve only marginally from about 0.7 for the present conditions (base case hydrology) to 0.5 for the situation with Lom Pangar reservoir. Thus, while the Lom Pangar reservoir enables a significant increase in dry season 5 Note that Lom Pangar’s contribution to dry season hydro -energy generation would be constant in absolute terms in the hypothetical case that the reservoir would each rainy season be filled to its maximum capacity (like the Mbakaou reservoir). viii hydro-energy generation, it contributes only modestly to the climate resilience of dry season hydro- energy6. The runoff elasticity of the guaranteed dry season generation capacity was assessed at 1.0. The impact of Lom Pangar reservoir on rainy season hydropower generation is negligible under all considered climate induced runoff changes due to the relative abundance of runoff in this period. Runoff elasticity of the EIRR of the Lom Pangar and Nachtigal projects in the Sanaga Basin: Three development scenarios were used to assess the runoff elasticity of the Economic Internal Rate of Return (EIRR) of the Lom Pangar project, i.e. 1: Lom Pangar only (including incremental energy generated at Edea and Song Loulou), 2: same as 1 with addition of the Nachtigal project (including total cost and total generated energy), and 3: same as 2 with addition of Song Mbengue (including total cost and total generated energy). Without climate change induced runoff changes, the EIRR varies between 14.1 and 15.4%, depending on the scenario. Most importantly, since the direct and indirect contribution of Lom Pangar to hydro-energy generation in the basin is overall fairly constant for runoff variations in the range of -20% to +20%, the EIRR is similarly not sensitive to runoff changes, unless decreases in runoff exceed 20%. The EIRR tends to reduce slightly for increasing runoff when there is relatively less need for additional storage in the Lom Pangar reservoir. The runoff elasticity of the EIRR is about 0.5 at 30% flow reduction for scenario 1 (0.3 for scenario 2 and 0.2 for scenario 3), i.e. the EIRR is reduced with about 15% for such large runoff reduction due to climate change. Overall, the EIRR of the Lom Pangar project is robust (well above the 12% threshold) and insensitive to moderate runoff changes (<20%) due to climate change. Even for a runoff reduction of 30% it stays at 12% or higher under all three scenarios. Runoff elasticity of hydro-energy generation in other basins: We have assessed the runoff elasticity of hydro-energy generation in other river basins for Lagdo dam in the Benue basin (located near Riao), Njock and Mouila stations on the Nyong River (located near Eseka), and Memve-Ele on the Ntem river (downstream of Ngoazik). No recent flow data were available for these basins and we have, therefore, based the CRA for these power stations on the available monthly flow data for the period 1950 – 1980. The guaranteed monthly output of Lagdo power station is sensitive to runoff changes, at a runoff elasticity of about 1.5. The runoff elasticity of Lagdo’s annual hydro-energy output is about 0.8 to 1.0. The runoff elasticity of guaranteed monthly power for Njock, Mouila and Memve-Ele is 1.0, as expected due to the absence of upstream storage other than for daily flow modulation. The ratio between maximum turbine discharge and average annual runoff (1951-1979) is 0.67 for Njock, 0.60 for Mouila and 1.14 for Memve-Ele. Accordingly, the runoff elasticity of annual hydro-energy output is the lowest for Mouila (about 0.25) and the highest for Memve-Ele (about 0.5); similarly, the ratio of guaranteed power and installed capacity is the lowest for Memve-Ele (0.26) and the highest for Mouila (0.60). 6 A further reduction of the annual variability of dry season hydro-energy generation would only be possible by prioritizing over-annual storage, focusing mainly on supplementing dry season flows after ‘dry’ rainy seasons; thereby reducing Lom Pangar’s potential to contribute each year significantly to dry -season hydro-energy generation. ix Climate projections: Globally, a warmer world is projected to bring more precipitation, but the consensus projection indicates that this will happen only marginally in Cameroon. Multiple Global Circulation Models (GCM) project on average no significant changes in annual precipitation across Cameroon by 2050 (A1B emission scenario); the standard deviation of the projected long-term average precipitation changes is about 7% for 2050. For 2080 a small increase in precipitation (4%) is projected, with a standard deviation of 11%. Projections for 2050 vary mostly between -10% and +15%, on average +1%. Projected increases in precipitation are distributed homogeneously across the country. It is recommended to adopt for impact analysis a worst case scenario of 15% reduction in precipitation by 2050 and 20% reduction by 2080. Given that multiple studies have shown that the available GCMs posses little skill for climate projections for West and Central Africa, minor changes are not significant. However, we cannot rule out even larger future changes in climate given the significant model uncertainty for the region and the prevailing variability of the West and Central African climate. Whereas climate scientists not nearly understand the causes of the Sahelian droughts during the 1970s and 1980s, our ability to forecast future hydro-climate for the region is equally fraught with uncertainty. All GCMs project significant increases in temperature, on average 2.00C by 2050, varying mostly between 1.30C and 2.70C (on average 3.00C by 2080, varying between 2.00C to 40C). Projected increases in temperature are distributed homogeneously across the country. The temperature sensitivity of the potential evapotranspiration varies between about 2% per 0C in the North of Cameroon and 2.4% per 0C in Central and South Cameroon. Thus, a 20C increase in temperature by 2050 will increase potential evapotranspiration, and thus crop water requirements, by nearly 4 to 5%. Runoff projections: The response of runoff to climate changes depends primarily on the runoff coefficient (or alternatively on the degree of aridity) of a basin. Available hydrological data suggest that basin runoff coefficients decline from >0.5 in the Northern Coastal Basins with heavy rainfall, to 0.25 to 0.3 in the Sanaga, Southern Coastal (Nyong and Ntem rivers) and Congo Basins, to about 0.18 for the Benue Basin and <0.05 for the extreme North of Cameroon. This reflects the increasing aridity from South to North due to decreasing rainfall and increasing temperature and potential evapotranspiration. The response of runoff to climate changes varies accordingly: the estimated precipitation elasticity of runoff varies between 1.9 for the Coastal Basins, 2.2 for the Sanaga and Congo basins, 2.6 for the Benue basin and nearly 3 for the extreme North of Cameroon; similarly, the temperature sensitivity of runoff varies between -2.5% per 0C for the Coastal basins, -3% per 0C for the Sanaga and Congo basins, -3.5% per 0C for the Benue basin and -4% per 0C for the extreme North. Thus, runoff is the least sensitive to climate changes in the cooler and high rainfall regions and most vulnerable in the hotter and arid extreme North of Cameroon. Based on the available climate projections and assessed climate elasticities of runoff, we project a small decrease in average annual runoff by 2050 (-4%), and no change by 2080. However, the spread between individual projections is significant, with a standard deviation of nearly 17% by 2050 and 24% by 2080. Projections of average runoff vary mostly between -35% and +30% by 2050 and between -40% and +40% by 2080. x Climate risks for hydro-energy generation in Cameroon: The projected runoff changes have been translated into changes in hydro-energy generation and changes in the EIRR of the Lom Pangar and Nachtigal projects, based on the assessed runoff elasticities of hydro-energy. By 2050 the total long- term average hydro-energy generation by the Edea, Song Loulou, Lom Pangar and Nachtigal power plants could vary between -15% and +5% of the base case value (present post-1971 hydrology). Results for 2080 are similar. Given the present climate projections, it is highly unlikely that hydro-energy generation would decrease more than 20% due to climate change. Changes in the EIRR for Nachtigal and Lom Pangar due to climate change are projected to be very limited. It is not likely that by 2050 the EIRR would decrease with more than 5% of its base case value, while in the worst case the EIRR would be reduced with less than 10%, e.g. from about 14.5% in the base case to 13% under severe climate change impacts. Hence, the Lom Pangar and Nachtigal projects are economically robust and resilient to long- term climate changes (how reservoir operation will deal with short-term climate variability is a separate issue to be considered). Moreover, one may assume that climate change impacts occur gradually over time, and the largest impacts on runoff and hydro-energy generation would occur towards the end of the economic analysis period (2012 - 2064). On the contrary, by 2050 total long-term average hydro-energy generation at Lagdo dam in the Benue Basin could vary between -35% and +15% of the base case value (present hydrology). Results for 2080 show even a larger potential deviation. Under the 2050 climate conditions there is nearly 20% probability7 that the annual hydro-energy generation would reduce with 20% or more (30% probability by 2080). Thus, Lagdo’s hydro-energy generation may suffer a significant decrease due to climate change, and is less climate resilient than the Lom Pangar and Nachtigal projects. Finally, by 2050 and 2080 total long-term average hydro-energy generation at Njock and Mouila stations on Nyong River could vary between -10% and +5% of the base case value (present hydrology), and between -15% and +10% at Memve-Ele station on Ntem River. The probability that the annual hydro- energy generation at any of these three plants would reduce with 20% or more is negligible. Hence, climate change impacts on hydro-energy generation at these stations will only be minor to moderate. Learning lessons from the management of historical climate variability: Significantly, the historical variability of precipitation and runoff is of the same magnitude as climate trends projected for the 21 st century. Therefore, learning lessons from managing the present impacts of intra-seasonal and inter- annual variability of Cameroon’s climate has the potential to better prepare water managers for dealing with long-term climate change impacts. Successfully managing the historical climate variability and droughts is likely the best adaptation strategy that can be recommended. Robust decision analysis: Climate change confronts decision makers with deep uncertainties, requiring robust decision analysis to inform good decisions by identifying system vulnerabilities, assessing 7 One of five (20%) climate models indicate changes in P and T yielding at least 20% reduction in Lagdo hydro- energy output xi alternatives for ameliorating those vulnerabilities, and assessing opportunities which may arise under certain future climate scenarios, such as e.g. due to increased runoff. From this perspective, it may be worthwhile to re-evaluate the proposed installed capacity for the future Nachtigal power station (presently designed at 360 MW), enabling it to benefit from the present high flow conditions during the rainy season and from future increased flow conditions which may occur with nearly 50% probability8. Moreover, the economic performance of water and energy infrastructure projects must also be tested for worst case scenarios. It is recommended to use the following climate change induced runoff scenarios for analysis of the robustness of water and energy infrastructure projects in Cameroon, for the 2050 investment horizon: ï‚· Average future (2050) scenario under climate change: no significant changes in annual runoff ï‚· Medium (2050) dry scenario of climate change: decrease of annual runoff by 15% (25% probability) ï‚· Worst case/Dry scenario by 2050: decrease of annual runoff by 35% (3% probability) ï‚· Medium wet scenario of climate change: increase of annual runoff with 10% (20% exceedance probability) ï‚· Extreme wet scenario: increase of annual runoff with 30% (2% probability). Robust methodology: The results of all available climate projections shown in this report demonstrate the large uncertainty in individual climate projections and the importance of taking all GCM projections into consideration, rather than building a CRA on the projections of only a few selected GCMs. Also, climate models which performed well for the present climate conditions may not necessarily give the same good performance for future conditions. Multi-model ensembling is found to be the appropriate approach for assessing the impacts of climate change on the water resources in Cameroon. This helps reducing the effects of model errors in one particular model and effects of the natural variability in any particular run. Anticipatory adaptation is most important for investments or decisions that are inflexible or irreversible, and have long lifetimes or lead times. Lom Pangar dam with multiple future downstream hydropower developments is an exemple of long-lived, climate-sensitive infrastructure investments. The methodology and modeling tools prepared under this CRA study provide powerful tools to identify system vulnerabilities and future climate change scenarios where the proposed developments might fail to meet their goals. These scenarios can be used to identify potential actions to address vulnerabilities and evaluate tradeoffs among them. Urgent need for a nation-wide comprehensive Hydrological or Water Information System: This study has shown significant gaps in hydrometeorological data available for Cameroon since the 1980s. High priority should thus be given to the monitoring of the present status of the country’s water resources, current runoff trends, minimum flows and similar performance metrics. Moreover, consistent records of 8 The projections of one of two climate models suggest increased runoff in the future xii long-term recorded rainfall could not be easily located for this study, even though they are reported to exist to some degree. Actual precipitation (and runoff) data are required for more enhanced and more accurate future analyses of rainfall – runoff relationships and hydrological modeling. Therefore, it is recommended to revive, upgrade and possibly expand the previously existing hydrometeorological networks in Cameroon, with particular emphasis on the collection of runoff data. Equal priority should be given to the preparation of a comprehensive and nation-wide data base of already available hydrometeorological data. Thus, it is recommended to urgently develop and operationalize a comprehensive Hydrological or Water Information System (HIS/WIS) for the country. Caveats It is important to emphasize that in this report climate risks are calculated based on the long term (i.e. 30 years) shifts in mean precipitation and temperature; they do not account for changes in the inter- annual or decadal variability. We do not consider this a serious limitation since the hydrological baseline period used for our Excel model simulations and the design of the Lom Pangar dam (1971-2003) comprises arguably a variable period in the historical water resources data for Cameroon, with lower runoff than observed prior to 1970. The worst droughts in living memory occurred in the region between 1970 and 1990. In fact, the Sanaga basin has experienced in this period runoff shortages of the same order of magnitude as the maximum projected decreases in runoff. xiii 1. Introduction and objective 1.1 Objective of the Climate Risk Assessment The Africa Energy Unit (AFTEG) of the World Bank has obtained financing from the Trust Fund for Environmentally and Socially Sustainable Development (TFESSD) – funded by Finland and Norway - for work towards “Understanding the Impact of Climate Change on Hydropower: the Case of Cameroonâ€?. The development objective of this activity is (i) to develop tools for assessing climate change impacts on the operation of hydraulic infrastructure such as regulating dams and hydropower plants in the Sanaga River basin, and (ii) to take steps towards an institutional framework for climate resilient water resources management in Cameroon. The aim of this initiative is to build resilience to climate risks into water management in general and hydropower development in Cameroon in particular. The study includes three components: i. Develop suitable climate change scenarios for the Sanaga basin, support the Electricity Development Corporation (EDC) of Cameroon to develop a reliable hydrological model for the Sanaga River basin, and derive climate change impacts on the potential generation capacity in the Sanaga basin in the context of changing hydrology; ii. Assess the impact of climate change on the future operation of Lom Pangar dam and three other regulating dams in the Sanaga basin and support the establishment of an operational regime of hydraulic infrastructures in the Sanaga River basin, in a consultative manner with water users and taking into account equitable sharing of resources between users and environmental flows. iii. Assess future impacts of climate change on water resources availability and management in Cameroon. The objective of the present study is to carry out a preliminary Climate Risk Assessment (scoping level) for the five river basins in Cameroon and to assess future impacts of climate change on water resources availability and management in Cameroon (item iii above), with special attention to potential climate change impacts on future hydropower generation in Cameroon and the operation of the Lom Pangar storage reservoir under conditions of climate change (parts of items i and ii); see Annex 1 for Terms of Reference. This assessment also aims to provide an analytical base for increased dialogue on climate variability and change and on integrated management of water resources in Cameroon. The principal audiences for this report are policy makers at the national level in Cameroon, particularly in the Ministry of Water and Energy (MINEE), the Ministry of Environment, Nature Protection and Sustainable Development (MINEPDED), and the Ministry in charge of Agriculture (MINADER). The assessment will identify information and knowledge gaps and priorities for future studies/activities. 1 1.2 Previous studies on climate risks to Cameroon’s water resources There are multiple studies regarding the abrupt changes observed around 1970 in precipitation and runoff regimes across Cameroon and across West Africa in general. The general consensus is that around 1970 an abrupt change in precipitation occurred, as discussed in Chapter 3 (Tarhule et al, 2013; Tarhule and Grijsen, 2013; ISL et al, 2005b; Kpoumie, 2010; Liénou et al, 2008; Dzana et all, 2011). However, only a few studies have been published regarding the potential future changes in river regime for Cameroon, particularly in such quantitative terms as required for this report and relevant for the planning of future hydropower development in Cameroon. MINEE and GWP (Volume 2, 2009) briefly discuss potential future climate change impacts on Sanaga Basin runoff, albeit only by summarizing the work of Sighomnou et al (2007), discussed further below. Similarly, MINEP and UNDP (2012) provide only limited quantitative results regarding future changes in water resources relevant to hydro-energy generation. This study analyzed the output of only one GCM run (MPI-ECHAM5 for emission scenario A1B) and used a regional climate model REGCM for dynamical downscaling of model output to the scale of river basins in Cameroon. The main risks to climate change were assessed by agro-ecological zone, with a focus on extreme climate risks such as droughts, erosion, high winds and flooding. Potential impacts on overall water resources availability, as relevant to hydro- energy generation, were not discussed. Data analysis showed (for the MPI-ECHAM5 run) a slight increase in precipitation until 2035, followed by a sharp and remarkable decrease by 2100 in nearly all agro-ecological zones except. The annual average temperatures would rise by the end of the 21st century from 0.7 0C to 4.6 0C in the northern part of Cameroon and from 0.5 0C to 3.5 0C in the forest zone in the South of the country. However, the selected GCM simulation run is not representative for the wide spectrum of GCM projections readily available for Cameroon, which each represent a possible future climate. This report shows that a single run style study has little validity and utility. Sighomnou et al (2007) analyzed the potential impacts of climate changes during the 21st century on the runoff regime of the Sanaga River, again only based on the output of one GCM (HADCM3) for the A2 emission scenario. The GR2M hydrological model was used to translate projected future changes in precipitation and potential evapotranspiration (PET) into runoff changes. Compared to the period 1971- 2000 an increase of 3.4% in average annual precipitation is projected, along with a gradual increase in PET, in the order of 30% increase by 2100. As a result of such large increase in PET with only a modest increase in precipitation, runoff in the Sanaga basin could by 2100 decrease by as much as 20%. However, several issues arise regarding the accuracy and representativeness of these results, mainly: a) The main issue is that an increase of 30% in PET due to an increase of about 4 0C would appear to be unrealistic. As shown in Annex 6, the temperature sensitivity of PET is about 2.5% per 1 0C according to the Hargreaves method (Hargreaves et al, 1982, 1985, 1994, 2003; Allen et al, 1998; Droogers and Allen, 2002) and Penman-Monteith method (Monteith, 1965). These methods are generally considered to fall in the category of best methods (Xu and Singh, 2001) and would yield an increase 2 of PET with only about 10% by 2100 instead of 30%. The newest version of Climate Wizard (included in the WBG’s Climate Change Knowledge Portal: http://climateknowledgeportal.climatewizard.org) uses also the Hargreaves method and its results point in the same direction. Consequently, Sighomnou et al (2007) appear to have largely overestimated the decrease in future runoff. b) Results derived from the Climate Wizard (Chapter 6) indicate that the HADCM3.1 GCM is in fact the driest climate model for the Sanaga Basin (projecting decreases of precipitation rather than increases) and is as such not representative for the wide spectrum of GCM projections readily available for Cameroon. As shown in this report, it is important to consider the total spectrum of possible future climates rather than only one possible future scenario. c) Studies with the application of multiple hydrological models on the same watershed have shown that the response of runoff to changed climate input parameters (particular to changes in precipitation, temperature and thus potential evapotranspiration) largely depends on the selected model formulations (Vano et al, 2012; Section 4.3). In other words, different hydrological models can simulate quite different runoff responses to similarly changed climatic conditions. Basing the analysis of runoff response to changes in climatic conditions primarily on observed runoff and climatic data is thus preferred (Chiew, 2006). d) The various results presented by Sighomnou et al (2007) in their Table 1 suggest a very high precipitation elasticity9 of runoff, in the order of +3.8 compared to our finding of about +2.2 for the Sanaga basin (based on observed runoff and climatic data and studies for similar river basins in Africa; see Chapter 4). The same results suggest a PET elasticity of runoff in the order of -1, which is realistic in view of our results (the theoretical value would be -1.2; see Chapter 4). Consequently, a precipitation increase by 3.4% and PET increase by 10% would yield a runoff increase by about 2.2 * 3.4% – 1 * 10% = - 2.5%, i.e. no significant change in average annual runoff by 2100 instead of a decrease with 20%. Munang et al (2010; http://www.eecore.org/climate-change/) studied climate change impacts on the hydro-energy potentials of the Sanaga basin by using trend analysis to ascertain the past (and future) change of temperature, precipitation and river runoff. Regression analyses were used to establish the relationship between temperature, precipitation and river runoff. Forecasts were generated by extending the observed trends, taking into consideration the importance of each climate variable to river runoff. Results showed from 1960 – 2007 an increase in temperature by 0.8 0C, a decline in precipitation by 112 mm/yr or 6.5% and a decline in river runoff by 142 m3/s or 7.5% for the Sanaga Basin. These results suggest an unrealistic low precipitation elasticity of runoff of only 1.15 (instead of a value in the range of 2.0 to 2.5; see Chapter 4). Projections of these trends to 2037 indicated a decrease in Sanaga river runoff by 355 m3/s (or 19% compared to the 1960s). The study thus concluded that climate change leads to a decrease in precipitation and river runoff over time, which negatively affects the hydro-energy potential of the Sanaga River. However, it appears that the authors have ignored the 9 A precipitation elasticity of runoff of 3.8 suggests that runoff would increase with 13% due to an increase in precipitation with 3.4%. Similarly, a PET elasticity of runoff of -1.0 suggests that runoff would decrease with 30% due to an increase of PET with 30%, in total a runoff reduction with 17%. 3 abrupt change in runoff/precipitation, which occurred around 1970 in West and Central Africa, and has been assessed independently by many researchers (as indicated at the beginning of this section). When such abrupt change in runoff around 1970 (16% for the Sanaga Basin) is accepted, there is no negative trend in runoff beyond 1970. Instead, the authors have incorporated this abrupt negative change in a long-term declining trend and extrapolated this trend to 2037, obviously projecting a significant reduction in runoff. As shown in this report, a consistent reduction of 20% in Sanaga Basin runoff is highly unlikely to occur by 2050. 1.3 Outline of the report Chapter 1 introduces the objectives of this Climate Risk Assessment (CRA) and Chapter 2 describes the CRA methodology adopted for the five main river basins of Cameroon. Chapter 3 analyses the available hydro-meteorological data for these Basins, to the extent as deemed useful for this CRA. Chapter 4 discusses the response of basin runoff to climate changes and presents the results of regression analyses of available runoff and rainfall data. Based on simulations with a seasonal water management and hydro-energy generation model in Excel based on monthly flow data, Chapter 5 then analyzes the sensitivity of the water resources system and hydropower generating infrastructure in selected basins to long-term changes in runoff due to climate change. Subsequently, Chapter 6 reviews available climate change projections for 2050 and 2080, and translates these into projections of changes in runoff which can be expected in the distant future. Finally, Chapter 7 quantifies in probabilistic terms the risks and impacts of climate change on hydro-energy and economic performance indicators, while conclusions and recommendations are presented in Chapter 8. 4 2. Decision scaling methodology for a risk-based assessment of climate change impacts on WR Given the substantial uncertainty in climate projections from the current Global Circulation Models (GCM), it is difficult to estimate what the future climate is likely to be. Generally, a Climate Risk Assessment (CRA) aims at better understanding the dynamics of future climate over the subject river basin, and assessing its potential impacts on water resources, hydro-energy production, navigation, agriculture, dependable minimum river flows, the environment, etc., as well as possible impacts on existing and planned infrastructure. This is essential for assisting decision makers and stakeholders to better manage their water resources, prepare for extreme hydrological hazards, and enhance development planning in the subject Basin. Therefore, generally the objective of a CRA is to assess the risks of climate change to the water resources and associated development sectors of a Basin in the near (e.g. 2030), mid (e.g. 2050) and distant future (e.g. 2070); 2050 and 2070 being significant for investments in hydropower development with a typical 50 years investment horizon. 2.1 Top-down approach Conventional CRA studies typically adopt a top-down approach whereby subsequently: i. the output of a limited number of selected Global Circulation Models (CMs) for one or more emission scenarios is statistically downscaled (bias removal by e.g. quantile mapping; Wood et al., 2004) or dynamically downscaled (using a Regional Climate Model or RCM) for the subject basin; ii. downscaled climate projections are used to drive a hydrological model of the subject basin to assess potential climate change impacts on the basin’s runoff; iii. output of the hydrological model is used to drive a water resources system model of the basin, to assess potential climate change impacts on selected performance indicators, representing the basin’s socio-economic and environmental conditions associated with the present and future water resources development and of interest to stakeholders; and iv. the water resources system model is finally applied to analyze and design adaptation options. Precipitation projections for West and Central Africa vary widely, and on average GCMs project no significant change in precipitation across Cameroon, with slightly more than half of the GCMs projecting an increase in rainfall (Ref. Annex 2). According to Kundzewicz et al (2007), no single model can be regarded as the “bestâ€? for any region or parameter, and it is thus preferred to use the output of a large ensemble of climate projections for multiple GCMs and emission scenarios. Climate projections typically lack credibility at the spatial and temporal scales that are relevant to water resources planning. A variety of statistical downscaling approaches and dynamical downscaling through regional climate models (RCM) are available, which change the resolution of projections. However, downscaling does as such little to improve the credibility of the underlying large scale climate projections. Different RCMs also provide different projections and thus add not only additional (sub-) regional information but also additional uncertainties and errors. Even past performance in the reproduction of observed climate is no 5 guarantee for future performance (Annex 2). The top-down approach thus requires the statistical or dynamical downscaling of a large number of climate projections for a basin, which may provide a false sense of accuracy and is time consuming. However, several on-line tools are now available providing access to statistically downscaled climate projections, which allows this part of climate risk assessments to be conducted faster and conveniently. Subsequently, a hydrological model needs to be developed and all downscaled climate projections need to be processed with such model, to provide multiple input data series (e.g. for 20 GCMs and 2 RCMs, or 40 dynamically downscaled climate projections, each for two emission scenarios) to the chosen water resources system model. In turn, this model has to be run for all simulated future runoff scenarios. Analyzing the output of all model runs and synthesizing the results into meaningful results for decision makers constitutes in itself a challenge. Moreover, studies with the application of multiple hydrological models on the same watershed have shown that the response of runoff to changed climatic input (particular to changes in precipitation, temperature and potential evapotranspiration) largely depends on the selected model formulations (Vano et al, 2012). In other words, different hydrological models simulate different runoff responses to similarly changed climatic conditions. Basing the analysis of runoff response to changes in climatic conditions primarily on observed runoff and climatic data is thus preferred (Chiew, 2006). The response of runoff to climate changes is then determined through regression analysis and classic hydrological techniques applied to observed runoff and climate data, based on linear relationships between relative changes (%) in runoff, rainfall and temperature. Historical gridded precipitation, temperature and potential evapotranspiration data such as the CRU TS3.10 hydrometeorological data sets are readily available on the internet for a 0.50 global grid. The top-down approach is generally time consuming, reason that the analysis is often limited to a few GCMs and emission scenarios, and one RCM or statistical downscaling. A major shortcoming of the approach based on a few climate projections is, that it does not provide enough output for a probabilistic analysis of future runoff conditions, as required for assessing the risks to the water resources system under consideration. Therefore, this study adopts an alternative, less time consuming and bottom-up risk-based methodology for the analysis of climate change impacts on existing or planned infrastructure investments in water resources systems, which makes use of a large number of climate projections readily available on the internet (Brown.et al., 2012; Ghile et al., 2013; Grijsen et.al., 2013). It does not require the time consuming processing and downscaling of a large number of climate projections, nor does it necessarily require the development of hydrological models to simulate the impacts of climate changes on (sub-) basin runoff. Time consuming hydrological modeling efforts may often not be justified in view of the large uncertainties embedded in the presently available climate projections for the 21st century. While much effort may still be needed to develop a water resources system model of the subject basin, this model need not be run for a multitude of input scenarios derived from climate and hydrological models. Instead, the baseline runoff conditions (current hydrology) can be parametrically varied for each of the future investment and water demand scenarios, to span a wide range of plausible changes in runoff conditions. 6 This methodology, as described in more detail in Section 2.2, enables estimates of the plausibility of climate risks and helps to develop a conceptual framework for adaptation strategies that increases the resilience and robustness of investment planning in the subject basin, without the need for excessive climate change, hydrological and water resources modeling. The methodology places climate projections in the context of risks to investments rather than as credible predictions of the future. The methodology was applied successfully for a Climate Risk Assessment of infrastructure investment in the Niger River Basin (Ghile et al., 2013; Grijsen et al., 2013); see also HydroPredict Vienna and Annex 3. It focuses on identifying the response of a water resources system to climate change and subsequently on assessing the likelihood of risks by using climate information from a multi-GCM ensemble of climate projections and historical climate conditions, readily available for multiple emission scenarios from reliable internet based tools, such as the WBG’s Climate Change Knowledge Portal (CCKP; Strzepek et al., 2011; Climate Portal) and the Climate Wizard 10 (climatewizard). The water resources system vulnerability would generally be analyzed in terms of performance metrics of hydroelectricity production, navigation, irrigated agriculture, flooding, maintaining environmental flows and other water resources related interest of stakeholders. The described methodology can be rapidly applied for all river basins in Cameroon, particularly to obtain a first impression (scoping) of what potential future climate change impacts on the country’s water resources may look like. 2.2. Decision - scaling methodology As an alternative to a GCM-driven analysis for climate change impact assessment, researchers have proposed bottom-up approaches that focus on understanding the climate response of the water system of interest through vulnerability analysis, bringing in GCM projections at a later stage of analysis to inform risks (Prudhomme et al., 2010; Brown et al., 2012; Ghile et al., 2013); see Figure 2.1. The process is described as ‘decision-scaling’, which identifies risk in an exploration of climate sensitivity and then uses climate information such as GCM projections to quantify the relative probabilities of risky climate conditions. Ideally, the approach begins with the water resources system of concern (bottom-up) and identifies risk thresholds for relevant system performance indicators through engagement with stakeholders 11 representing various water dependent sectors (including agriculture, hydro-energy, navigation, environment, fisheries, etc.); see also Section 2.3. The runoff changes that cause threshold- crossing risks are then identified through a process of parametrically varying the runoff inputs to the applied water resources model. Subsequently, the response of runoff to climate changes is determined, yielding ultimately the response of system performance metrics to specific climate changes. An 10 University of Washington and the Nature Conservancy (2009); Data source: Global Climate Model (GCM) output, from the World Climate Research Program's (WCRP) Coupled Model Inter-comparison Project phase 3 (CMIP3) multi-model dataset (Meehl et al., 2007), were downscaled (as per Maurer et al., 2009), using the bias- correction/spatial downscaling method of Wood et al. (2004) to a 0.5 degree grid, based on the 1950-1999 gridded observations of Adam and Lettenmaier (2003). 11 Due to time constraints this first step could not be timely implemented for this CRA; risk thresholds were thus assumed arbitrarily, based on experience with CRA in the Niger Basin. These thresholds should be discussed with stakeholders during a workshop to be held for validation of this report. 7 ensemble of multiple GCM projections is then used to estimate the probability of specific climate changes, runoff changes and thus of the commensurate changes in relevant performance indicators. I. Identify hazards & thresholds Figure 2.1: Conceptual flow chart for Stakeholder climate risk assessment (source: Ghile et  Assign risk levels for system performance Inputs al., 2013)  Define performance metrics The use of climate projections at a later stage of the analysis reduces the II. Assess system response to climate propagation of uncertainties earlier in the analysis, and allows assessment of  Performance metrics response to run-off the impacts of a wider range of climate  Run-off response to climate drivers changes, which are not restricted to  Climate response functions for metrics GCM projections. Additionally, it provides a transparent framework by first specifying the climate response of III. Estimate likelihood of climate hazards the system performance, allowing Climate uncertain climate information to be  Probabilities of future climate states Information framed in terms of performance metrics  Exceedance probabilities for risk levels. and thresholds of concern to stakeholders. It also facilitates the incorporation of subjective judgment of climate experts and stakeholders into Manage Climate Risks the assessment of risks. The decision-scaling process involves thus four steps: (i) identification of water resources system performance indicators, hazards and threshold risk levels, (ii) assessment of system responses to climate changes (vulnerability assessment), (iii) estimation of the likelihood of climate hazards (tailored climate information), and (iv) economic impact assessment and the development of adaptation strategies that address the climate risks (climate risk management), as shown in Figure 2.1. The fourth step is not further elaborated in this report. 2.3 Identification of climate hazards and thresholds The first stage of the decision-scaling process is to identify the potential hazards to the system that result from changes in climate, where hazard implies the impact of a climate change but not its probability. Risk is defined as the product of impact and probability, an expected value of the loss. The process begins with stakeholder discussions to identify the relevant performance indicators of the water resources system under consideration, and also, if appropriate, thresholds of acceptable decreases in performance levels. Beyond these levels, the performance for a particular measure would be deemed unacceptable. A critical threshold of system performance could be for example a 20% or greater 8 reduction from baseline performance (i.e. the average performance based on 30-year mean historical climate) of the present system or infrastructure investment plan, for major sectors of interest. Examples of historical climate events provide useful guidance for facilitating these discussions. Stakeholders should ideally be consulted at an early stage of the climate risk assessment work. Figure 2.2 shows an example of the definition of risk levels based on percentage interval changes in performance indicators, derived from the Climate Risk Assessment for the Niger Basin (Grijsen et al, 2013; Annex 3). Fig. 2.2: Risk levels based on changes in Performance Indicators. Note: The red arrow indicates the 20% reduction in performance relative to base level beyond which – in this example - impacts are considered significant. For this study we have adopted Figure 2.2 for the definition of risks levels, but these risk levels need to be further discussed with relevant stakeholders. We will focus the climate risk analysis on performance indicators linked to hydro-energy production, e.g. guaranteed production (in MW) during the dry season and total annual energy production (in GWh/yr). Till date water use for irrigation/agriculture and domestic/industrial water supply is still small and is not expected to increase substantially in the foreseeable future (MINEE and GWP, 2009, Volume 1). The economic significance of river navigation is equally small, while the maintenance of minimum flows downstream of storage reservoirs is a pre- requisite under all foreseeable hydro-meteorological conditions, with or without climate change impacts on runoff. Hence, minimum flows should not be affected by climate change. It is likely that the people living in rural areas would be potentially most affected by climate change impacts on rainfed agriculture. However, this important impact domain is outside the scope of this study. 9 2.4 Vulnerability assessment: Assessment of system response to changes in runoff Following the definition of performance metrics and thresholds, the next step is to seek an understanding of how the system responds to changes in runoff caused by climate changes. This process identifies runoff conditions that cause unacceptable performance levels, such as the above mentioned 20% reduction from baseline performance. Through the exploration of the impacts of plausible runoff changes on the system (including impacts of increased evapotranspiration caused by increased temperatures on reservoir evaporation and irrigation water demands), conditions that cause risks – i.e. unacceptable system performance - are identified. Estimation of a basin’s response to changing climate conditions is typically conducted by: i. Modeling the response of performance indicators for the sectors of interest to relative (%) changes in annual basin runoff, by the application of a water resources system model, yielding the runoff elasticity12 (εQ) of performance indicators; and ii. Linking the response of annual basin runoff to changes in the mean annual precipitation and temperature by regression and/or other classic hydrological analyses, yielding the precipitation and temperature elasticities13 (εP and εT) of runoff, which synthesize the climate sensitivity of basin runoff. 2.4.1 Water resources modeling A water resources system model (such as HEC-ResSim) simulates inter alia streamflow, multi-sectoral water allocation, reservoir and hydropower operations, performance of irrigated agriculture (crop yields, crop failures), newly planned infrastructure, adaptation measures, etc. Based on multiple model runs with parametrically varied runoff boundary conditions (e.g. runoff changes varying between +20% and -30%), the runoff elasticity (εQ) of performance indicators can then be assessed for various infrastructure scenarios, to synthesize the runoff vulnerability of the water resources system performance. This work needs to be conducted at an early stage, and forms the key part of any CRA study. Assessing the hydrological boundary conditions (based on historical runoff and precipitation data) for the water resources model is an important part of this work, as well as assessing the impacts of conceived adaptation measures. Typically the water resources modeling study uses multiple infrastructure and water demand scenarios, including a baseline scenario with existing dams and irrigation systems in the basin, and several 12 The runoff elasticity of a performance indicator defines the response of an indicator to changes in runoff. For example, for hydro-energy the runoff elasticity varies typically between 0.75 and 1.25. A runoff elasticity of 1.0 indicates that a 10% decrease in runoff causes a 10% decrease in generated (annual) hydro-energy. 13 The precipitation elasticity of runoff defines the response of runoff to changes in precipitation and the temperature elasticity defines runoff response to changes in temperature (due to changes in evapotranspiration). A precipitation elasticity of εP = 2.5 indicates that a 10% decrease in rainfall causes a 25% decrease in runoff. A temperature elasticity of εT = -0.7 indicates that a 10% increase in temperature causes a 7% decrease in runoff. 10 investment scenarios with planned dams, hydro-energy and irrigation development, adaptation measures, etc. The water demand scenarios are a baseline scenario with agricultural, domestic and other water demands projected for a future period/scenario (e.g. 2030, 2050 and 2070) without climate change (based on the historic climate), and a scenario with an additional water demand increase reflecting warmer temperatures (typically for Cameroon by 2050 a 5% increase of potential evapotranspiration and crop water requirements due to a temperature increase of 20C; see Annex 6 and Chapter 4). Runoff conditions are parametrically varied for each of the investment and demand scenarios to span a range of plausible changes in runoff conditions and assess the system response, i.e. the runoff elasticities of performance indicators. Runoff scenarios are defined as baseline runoff (i.e. no change over the historical conditions) and selected changes to mean annual runoff, e.g. +20%, +10%, -10%, -20% and -30% respectively. This range of runoff changes is chosen to go beyond the range of runoff projections derived from the available GCM projections (as shown in Chapter 6) and includes more severe climate change scenarios that cannot be ruled out due to the high uncertainty in climate change projections for Central and West Africa. This analysis enables the establishment of quasi-linear relationships between relative (percentage) changes in runoff and relative changes in system performance, defined by the runoff elasticity εQ of performance indicators. Expressing climate change impacts in terms of relative (percentage) changes has both the advantage of simplifying the presentation of results to stakeholders, as well as eliminating most of the systematic errors and biases in absolute model output values which would occur in the simulations for the present conditions as well as in the simulations for future conditions under climate change. Linearization of the response functions - i.e. runoff response to climate changes and system performance response to runoff changes – is justifiable in view of the large uncertainties embedded in the available climate projections, greatly reduces the efforts and time required for a Climate Risk Assessment, and facilitates the presentation of results to stakeholders. Preferably a simulation period should be selected which includes observed droughts as well as above average wet periods, over at least a 30 years period. Due to an abrupt negative shift in precipitation around 1970, which was observed across the entire Niger Basin as well as across Cameroon, we have chosen the slightly drier period 1972 – 2003 for our analyses. Generally, this type of analysis reveals significant differences among the climate sensitivity of performance metrics. For example, when high priority is given to water supply for irrigation at the cost of hydro-energy generation, agriculture may show a limited sensitivity to changes in climate (εQ< 0.5). In such case, hydro-energy would tend to be much more sensitive, with at least proportionate changes in response to given changes in runoff (εQ >= 1). The highest sensitivities to changes in annual runoff may be observed for dry season environmental flows. The runoff elasticity of performance indicators is a concise way to synthesize results for stakeholders. Regular interaction with stakeholders during this phase remains important, particularly during the study of adaptation measures. 11 2.4.2 Runoff response to climate change The next step after the identification of problematic runoff conditions links changes in mean basin runoff to changes in mean climate conditions (precipitation, temperature and potential evapotranspiration). The concept of elasticity for evaluating the sensitivity of streamflow to changes in climate was introduced by Schaake (1990), who defined climate elasticity of streamflow by the proportional change (%) in streamflow Q divided by the proportional change (%) in a climate variable. No hydrological modeling is required when sufficient observed hydrometeorological data are available (including precipitation, temperature and potential evapotranspiration data readily available from internet sources, such as the CRU TS3.10 hydrometeorological data sets), and one is mainly interested in seasonal/annual runoff volumes in relation to storage capacity available in a basin. Time consuming hydrological modeling, with high data demands, may be necessary in case one is interested in daily to monthly runoff changes due to climate change. Arora (2002) provides a theoretical underpinning for the estimation of climate elasticities of runoff based on the aridity index φ = E0/P, i.e. the ratio of annual potential evapotranspiration (E0) to annual precipitation (P); see also Chapter 4. The aridity index has been shown to adequately describe the actual evaporation ratio E/P and the runoff coefficient Q/P (= 1 - E/P) of catchments for a range of climatic regimes. Based on Arora (2002) and Chiew et al (2006), Grijsen (2013) has shown for the Niger River Basin that the observed runoff coefficient Q/P provides a powerful initial estimator for the climate elasticities of runoff, as follows: ï‚· εP = 3 – 3 Q/P + (Q/P)2 ; range 3 to 1 for Q/P ranging from 0 to 1 (2.2 for Q/P = 0.3) ï‚· εT = {-2 + 3 Q/P – (Q/P)2}*{T/(T+17.8)}; range -1.15 to 0 for Q/P ranging from 0 to 1 (for most basins in Cameroon with average temperature T = 24 0C: εT = -0.7 for Q/P = 0.3). Chiew (2006) and Chiew et al (2006) analyzed concurrent annual precipitation, temperature and runoff time series for 521 (by and large) unregulated catchments around the world, varying between 100 and 76,000 km2. The precipitation elasticities of 80% of the basins varied between 1.0 and 3.0 (Figure 2.3). 12 Source: Chiew, 2006 and Chiew et al, 2006 Figure 2.3: Estimates of εP for Africa (left panel) and Australia (right panel) Higher εP values (>2) were found in Australia and in southern and western Africa, while lower values (<2) were found in the mid and high latitudes of the Northern Hemisphere. The lowest ε P values are found for Nordic conditions with only snowmelt runoff and a runoff coefficient near 1, while the highest climate elasticities of runoff are found in highly arid regions with a low runoff coefficient. The εP results shown for Australia are well described by the above equation for εP. The climate elasticities of runoff shown in Figure 2.3 (left panel) for Cameroon are inconclusive. The precipitation elasticity of runoff would likely be in the range of 2 to 2.5 for the main runoff generating basins (Congo, Coastal basins, Sanaga and Niger Basin), which will be further elaborated in Chapter 4. The empirical (regression) analysis for a specific basin is based on observed precipitation, temperature and streamflow data at multiple monitoring stations, representing particularly the sub-catchments where most of the basin runoff is generated (in case of Cameroon the entire basins). GIS based gridded historic climate data sets available from internet based sources can be explored for better representation of catchment average climate conditions. Runoff data can be obtained from the Global Runoff Data Centre in Koblenz, Germany (GRDC). Preferably a historic period should be selected which includes observed droughts as well as above average wet periods. In case runoff is impacted by man- made infrastructure, water abstractions, over-annual flow regulation and storage in reservoirs, flows need to be corrected in order to derive the naturalized runoff, which is also required as boundary condition for the water resources system model. Based on this analysis, a best fit (log-linear) empirical model can - for example - be defined as: â?„ ( â?„ ) ( â?„ ) After linearization for small relative changes (say < 25%) one obtains: dQ/Q0 = εP * dP/P0 + εT * dT/T0 where, Q, P and T denote respectively the estimated mean annual basin-runoff (mm/yr), annual precipitation (mm/yr) and annual mean temperature (°C); Q0, P0 and T0 denote the respective reference baseline values; and εP and εT denote respectively the precipitation and temperature elasticity of runoff. The impact of observed annual temperature variations on runoff is generally small, and can often not reliably be determined from available hydrometeorological records, nor from hydrological models (Vano et al, 2012). Theoretical considerations point to temperature elasticities εT in the order of -0.7 for Central and South Cameroon (see Chapter 4), which corresponds to a temperature sensitivity of runoff of about -3% per 0C. Thus, a temperature increase of about 20C (+8%) alone by 2050 could result in a 13 decrease in annual mean runoff of the order of 6%. Similarly, a 10% decrease in precipitation alone could cause a decrease in mean runoff in the order of 20% to 25%. 2.5 Climate Informed Risks: Estimating likelihood of climate conditions and hazards The next stage in the decision-scaling process is the determination of relative likelihoods of the climate and runoff conditions identified as critical in previous steps. It is where the best available climate information is incorporated into the risk assessment process. The risk of violating the initially assessed thresholds regarding system performance (Figure 2.2) can be assessed by using available climate projections and historical climate conditions from tools such as the Climate Wizard (climatewizard) and the WBG’s Climate Change Knowledge Portal (Climate Portal). The Climate Wizard provides bias- corrected climate projections for user defined areas (at a 50 km grid resolution), based on 15 GCMs, for the A2, A1B and B1 emission scenarios (IPCC’s AR4 assessment). Typically it provides probability distributions of future changes (in absolute and relative terms) in mean annual precipitation (P), temperature (T) and potential evapotranspiration (E0) for multiple time horizons, and annual areal average values for the current/past climatology (P, T, E0). The Climate Wizard also provides projections on a monthly or seasonal basis. The Climate Wizard estimates E0 based on the Hamon (1961) method, which - as shown by Xu and Singh (2001) – significantly underestimates potential evapotranspiration and severely overestimates the temperature sensitivity of E0 at a fixed 6.2% per 0C. The more accurate Hargreaves (Hargreaves and Samni, 1982) and Penman-Monteith (Monteith, 1965) methods yield instead for the basins in Cameroon a temperature sensitivity of E0 of about 2.5% per 0C (see Chapter 4). However, Climate Wizard has identified this issue and will shortly change to the use of the Hargreaves method. Adequate estimates of E0 can also be derived from the CRU TS3.10 data sets. The World Bank’s Climate Change Knowledge (CCK) Portal provides similar results (changes in relative terms) for 8,413 river basins across the world, for 22 GCMs and 3 emission scenarios (A2, A1B and B1). It also estimates for each GCM run impacts on mean annual runoff, annual base flow, flood and drought indicators, potential evapotranspiration and similar parameters of interest, in terms of relative changes from the historical baseline (1961 to 1999) to the 2030s and 2050s. Strzepek et al (2011; WB Water Paper CCK Portal) provides a detailed account of the studies and hydrological modeling underpinning the results displayed in the Climate Portal. Projected changes in runoff and E0 were simulated with the CLIRUN-II hydrologic model II (Strzepek and Fant, 2010), an extension of the WatBal water balance model (Yates, 1996). The results provide an understanding of the range of potential consequences of climate change on water resources at the country and basin scale, and are as such suitable for use as inputs to screening-level analyses of the impacts of climate change on water resources dependent investments, for the 2050 investment horizon. The results of these portals are thus considered to be adequate for the identification and scoping of the potential impacts of future climate changes on water resources availability and management in the five river basins in Cameroon. 14 Detailed climate projections for multiple GCMs and emission scenarios can also be obtained from the Data Library of the International Research Institute for Climate and Society (IRI-Columbia) and similar sources, but these projections would require extensive and time consuming data processing, statistical downscaling and bias correction. The results of the above climate change portals can be used directly, without any further processing or bias correction of GCM runs, to assess the probability distribution of projected changes in P and T for the selected emission scenarios and time horizons. The probability distributions are based on the assumption that each run is equally plausible (which is as good as any other method for assessing probabilities). Since the analysis focuses on changes in the average annual runoff, precipitation and temperature (over a period of e.g. 30 year), projections of relative changes (%) in these average parameters adhere well to the normal distribution; for example, the period 2035 – 2065 is taken as being representative for 2050. It is stressed that the probabilities presented in this report are GCM- based and thus necessarily of limited credibility due to the large uncertainties in GCM projections. But it is the best information we have at this point of time (note that in Chapter 6 also results are presented for the most recent generation of CMIP5 climate model projections). By focusing on mean annual values, we use the most credible aspect of the climate projections. Since water resources management in, for example, the Sanaga basin depends heavily on large reservoirs with possibly over- annual storage in the basin, changes in mean annual runoff are the most important runoff characteristic with respect to the annual filling of these reservoirs, as well as for total annual hydro-energy generation. Since the projected long-term changes in average annual rainfall, temperature and runoff are all approximately normally distributed, we can estimate the probability distribution function (pdf) of projected changes in runoff from the expected average changes in rainfall and temperature, and the variances of these changes, as follows: ï‚· E{dQ/Q0} = εP E{dP/P0} + εT E{dT/T0}; E{...} denotes the projected average shift in means of P and T ï‚· Cv2(dQ/Q0) = εP2 Cv2(dP/P0) + εT2 Cv2(dT/T0); Cv(...) = coefficient of variation of P and T projections The contribution of temperature variations to the variance of runoff is negligible due to its small coefficient of variation (Cv) compared to the Cv of changes in precipitation; hence: ï‚· Cv(dQ/Q0) = εP Cv(dP/P0). This also explains why the temperature elasticity cannot be estimated adequately through regression analysis from limited historical hydrometeorological data sets. Once the mean and Cv of the projected runoff changes have been established, one can proceed to assess the mean and standard deviation for the normal probability distributions of the selected Performance Indicators (PI), based on the runoff elasticity of PI (εQ), as follows: ï‚· E{dPI/PI0} = εQ E{dQ/Q0} ï‚· Cv(dPI/PI0) = εQ Cv(dQ/Q0) 15 Finally, probabilities of non-exceedance are estimated for specified risk levels, which are of particular interest to stakeholders, e.g. the risk or probability that the average hydro-energy generation in a basin may reduce with more than 20%, or the risk that pre-defined thresholds for acceptable climate change impacts are exceeded. 2.6 Summary of the adopted methodology for Climate Risk Assessment The Climate Risk Assessment (CRA) for the Niger Basin, briefly described as an example in Annex 3, has shown that – at least at basin scale – the relation between changes in runoff and changes in precipitation and temperature, and between changes in most performance indicators and changes in runoff is quasi-linear. These quasi-linear relations are characterized by the climate elasticities of runoff and the runoff elasticities of performance indicators, including the Economic Internal Rate of Return (EIRR) of a specific project or investment scenario. The pdfs of projected changes in precipitation, temperature, runoff and performance indicators are well described by the normal distribution. Keeping in view the large uncertainties in the presently available climate change projections, errors due to linearization are insignificant. The presented methodology thus allows for a relatively rapid and straightforward assessment of climate change impacts. The main components of the presented decision-scaling methodology are summarized in Figure 2.4, as follows: Figure 2.4: Schematic overview of the decision - scaling CRA methodology  Use water resources system modeling and historical hydrological conditions to assess the impact of changes in runoff on water resources system performance indicators, and determine the runoff elasticity of selected performance indicators, for the present conditions and/or for future investment scenarios14. Use the water resources system model to test alternative measures for adaptation to climate change.  Estimate the probability distributions of changes in precipitation and temperature, using a bias- corrected ensemble of GCM projections of future climate for the 21st century (as available from 14 In the present study this is mainly limited to an analysis of the impacts of runoff changes on hydro-energy generation at existing and planned dams in Cameroon (Sanaga basin dams, Lagdo dam on the Benue River and planned dams on the Nyong and Ntem Rivers). 16 internet portals), and determine the mean and standard deviation of projected changes in precipitation and temperature.  Use hydrologic analysis methods to determine the response of runoff to climate changes (climate elasticity of runoff) from historical records and theoretical considerations, and use the climate elasticities of runoff to translate the probability distributions of future changes in precipitation and temperature into probability distributions of future mean runoff.  Translate the probability distributions of future runoff changes into probability distributions of changes in performance indicators, based on the runoff elasticities of performance indicators, and estimate the probabilities (risks) of specific changes in system.  Ensure stakeholder involvement in an iterative manner in defining the thresholds and performance indicators, as well as for reviewing analytical results. 17 3. Hydrometeorological data for the five main river basins in Cameroon 3.1 Cameroon’s river basins There are five river basins in Cameroon, viz. the Lake Chad, Niger, Sanaga, Congo and Coastal Rivers Basins (Figure 3.1), three of which (Lake Chad, Niger and Congo Basins) are shared with neighboring countries. The Sanaga Basin (Figure 3.2) is the largest national basin, while the Coastal Rivers flow directly to the Atlantic Ocean. The Nyong and Ntem Rivers are also national rivers and part of the Southern Coastal Basins. The country has two main climatic regions: a humid equatorial climate in the south (between 20 and 60 N) and a tropical climate in the centre and North (between 60 and 130 N); the latter borders the Sahelian climate zone. In 2007 the total population of Cameroon amounted to about 25 million. Fig. 3.1: Main river basins and climate zones of Cameroon (Source: MINEE and GWP, 2009 (Volume 1) and Suchel, 1987) 18 A baseline study of the available water resources and water uses in Cameroon (MINEE and GWP, 2009)15 was carried out by the Global Water Partnership (GWP), under the guidance of the Ministry of Water and Energy (MINEE), as a first step in the development of a National Integrated Water Resources Management Plan. Precipitation varies significantly across Cameroon. In the Benue basin upstream of Garoua annual precipitation varies between 900 and 1,400 mm/yr (on average 1,100 mm/yr), while further North in the Lake Chad Basin it ranges between 600 to 800 mm/yr. In the Sanaga basin, annual rainfall averages between 1,350 mm/yr in the East and 2,400 mm/yr in the West at Edea, on average about 1,700 mm/yr. The spatial distribution of rainfall across the Congo Basin is more uniform, varying between 1,450 and 1,750 mm/yr (average about 1,600 mm/yr), but exhibiting a bi-modal distribution in time with peak rainfall in May and October. Annual precipitation in the Northern Coastal Basins generally exceeds 2,200 mm/yr, but exhibits a huge spatial variability, from 1,600 mm/yr at Muyuka to 9,800 mm/yr at Debundscha. Finally, precipitation across the Southern Coastal Basins varies between 1,500 mm/yr in the East and 2,900 mm/yr in the West, like the Congo Basin with a bi-modal distribution in time. The available precipitation and runoff data are discussed in more detail in subsequent Sections. Figure 3.2: The water resources system of the Sanaga basin (Source: Olivry, 1986 and MINEE and GWP, 2009) 15 It is noted that this as such comprehensive baseline study only briefly discusses potential future climate change impacts on Sanaga Basin runoff by summarizing the work of Sighomnou et al (2007). 19 3.2 Runoff data In 1980 Cameroon counted 74 hydrometric stations, but the measurement network severely declined after 1980 (Sighomnou et al, 2007); only 32 stations were operational in 2008 (MINEE and GWP, 2009). Available monthly stream flow data till 1980 for hydrometric stations in 4 of the 5 main river basins of Cameroon were obtained from the Global Runoff Data Centre in Koblenz, Germany (GRDC16). No further data were uploaded at GRDC after 1980. Supplemental flow data till 2003 were obtained from the EDC and previous studies conducted for the Lom Pangar reservoir project (e.g. ISL et al, 2005 and 2007). No discharge data became available for the Lake Chad basin in Cameroon. Table 3.1 lists the stations in the various basins for which monthly flow data were made available for this study, along with some characteristic data such as catchment area, location details, long-term average flow and the time period for which data are available. Station locations are shown in Figure 3.3. The monthly flow data used in Chapter 5 for the analysis of climate change impacts on hydropower generation in the Sanaga Basin (post 1971), and in the Niger, Nyong and Ntem basins (prior to 1980) are tabulated in Annex 4 and briefly discussed in the following sections. Basin GRDC # River Station Lat. Long. Area (km2)Flow (m3/s)MCM/yr mm/yr Alt. (m)d/s_station Data period Source 1335014 BENUE RIAO 9.05 13.68 30,650 249 7,840 256 186 1335500 1950 1980 GRDC 1335181 KEBI COSSI 9.62 13.87 25,000 97 3,064 123 195 1335500 1955 1979 GRDC Niger 1335500 BENUE GAROUA 9.3 13.38 64,000 357 11,262 176 174 1835800 1945 1980 GRDC 1335451 METCHUM GOURI 6.28 10.03 2,240 107 3,361 1,500 550 1964 1980 GRDC 1338600 DJEREM MBAKAOU 6.33 12.82 20,200 369 11,635 576 826 1338400 1959 2003 GRDC/EDC 1338201 LOM BETARE-OYA 5.92 14.13 11,100 176 5,544 499 662 1338400 1951 1980 GRDC LOM LOM PANGAR 19,700 263 8,298 421 1951 2003 EDC SANAGA GOYOUM 50,500 585 18,443 365 1972 2003 EDC 1338400 SANAGA NACHTIGAL 4.35 11.63 76,000 991 31,252 411 426 1338050 1951 2003 GRDC/EDC Sanaga MAPE MAGDA 4,020 96 3,013 749 1965 2003 EDC NOUN BAMENDJING 2,190 52 1,633 746 1965 2003 EDC 1338252 MBAM MANTOUM 5.62 11.18 14,700 323 10,196 694 660 1338300 1965 1980 GRDC 1338300 MBAM GOURA 4.57 11.37 42,300 632 19,937 471 396 1338050 1951 2003 GRDC/EDC 1338050 SANAGA EDEA 3.77 10.07 131,500 1,877 59,191 450 12 1944 2003 GRDC/EDC 1326301 MUNGO MUNDAME 4.57 9.53 2,420 167 5,270 2,178 17 1952 1977 GRDC Coast- 1326500 NKAM MELONG 5.12 10 2,280 71 2,226 976 700 1951 1977 GRDC North 1326300 WOURI YABASSI 4.45 9.97 8,026 310 9,765 1,217 10 1951 1977 GRDC 1336500 CROSS MAMFE 5.75 9.32 6,810 529 16,685 2,450 44 1967 1979 GRDC 1340500 NTEM NGOAZIK 2.3 11.3 18,100 278 8,779 485 500 1953 1979 GRDC Coast- 1339500 NYONG MBALMAYO 3.52 11.5 13,555 150 4,725 349 634 1339017 1951 1979 GRDC South 1339014 NYONG ESEKA 3.68 10.7 21,600 276 8,701 403 146 1339100 1951 1977 GRDC 1339100 NYONG DEHANE 3.57 10.12 26,400 442 13,945 528 15 1951 1977 GRDC 1448050 DJA NGBALA 2.02 14.9 38,600 421 13,271 344 335 1448100 1954 1978 GRDC Congo 1348152 KADEI PANA 4.2 14.68 20,370 247 7,783 382 570 1748500 1965 1980 GRDC Table 3.1: Overview of hydrometric stations with monthly flow data 16 The Global Runoff Data Centre, GRDC, in the Bundesanstalt für Gewässerkunde, 56068 Koblenz, Germany 20 Only the data presented for Edea near the outlet of the Sanaga basin are naturalized flow data, i.e. corrected for the regulating impact of three upstream reservoirs (Bamendjing since 1974, Mapé since 1987 and Mbakaou since 1969). The available monthly and annual flow data for Goyoum, Nachtigal and Goura in the Sanaga Basin are affected by the seasonal flow regulation of the upstream reservoirs. Fig. 3.3: Hydrometric stations for the main river basins of Cameroon 21 3.2.1 Niger – Benue Basin Figure 3.4 summarizes hydrological characteristics of the Benue Basin in Cameroon, for the gauging stations Garoua, Cossi and Riao near the present Lagdo dam. Lagdo dam was put into service in 1982, but no flow data are available after 1980. Runoff data for Edea at the outlet of the Sanaga basin are included for comparison in the upper left panel of Figure 3.4. The correlation of annual and seasonal runoff at Cossi and Riao with the runoff at Garoua is good. The coefficient of variation (CV) of annual runoff for all three stations is identical (about 0.28), but nearly twice as large as the CV of annual runoff from the Sanaga Basin (about 0.17; Section 3.2.2), reflecting the more erratic rainfall across the Benue Basin in Cameroon. The correlation with the Sanaga basin runoff is weak. The runoff data for Garoua and Riao are used to determine the climate elasticity of runoff for the Benue basin and the runoff data for Riao are used to analyze potential climate change impacts on Lagdo’s hydro-energy generation. The normal probability distribution function applies well to the Basin’s annual runoff data. Niger Basin Correlation coefficients Area (km 2) Stations Normal distribution runoff Benue basin CV Garoua Cossi Riao Edea 0.29 64,000 Garoua 1 0.84 0.95 0.44 500 0.29 25,000 Cossi 1 0.68 0.41 Garoua 0.28 30,650 Riao 1 0.66 Riao 400 Annual runoff (mm/yr) Annual runoff Niger Basin Garoua 700 Cossi 300 Riao Annual runoff (mm/yr) 600 y = 75.623x + 257.31 Edea 500 200 400 y = 51.023x + 175.96 300 100 200 100 0 -2.5 -1.5 -0.5 0.5 1.5 2.5 0 1945 1950 1955 1960 1965 1970 1975 1980 Reduced Normal variate Annual flow at Riao - Cossi vs Garoua Monthly flows June-December Cossi + Riao vs Garoua 600 Riao y = 0.9709x 3,000 Cossi R² = 0.9562 Flow Riao - Cossi (m 3/s) 500 y = 0.9491x R² = 0.9808 (m 3/s) Riao+Cossi 400 2,000 Flow Cossi + Riao 300 y = 0.7075x R² = 0.899 200 1,000 100 y = 0.267x R² = 0.6878 0 0 100 200 300 400 500 600 0 1,000 2,000 3,000 Flow Garoua (m 3/s) Flow Garoua (m 3/s) Fig. 3.4: Hydrological characteristics of the Benue basin in Cameroon 22 3.2.2 Sanaga Basin Fair to good correlations (>0.6) and consistent CV values (average 0.17) are found among most discharge stations in the Sanaga basin (Fig. 3.5), showing spatial uniformity in runoff characteristics. Goyoum on the Sanaga River is an exception; it has a larger CV value (0.25) than the other stations, correlates poorly with the downstream Nachtigal station (Fig. 3.6 - right panel), and has thus been excluded from further analysis. It is concluded that the hydrological variability of the Sanaga basin is sufficiently homogenous for the purpose of our climate risk assessment. The flow data for Lom Pangar (partially derived from flow data for the upstream Betare-Oya station; Fig. 3.6 - left panel) are essential for the analysis of the contribution of the new Lom Pangar reservoir to the future hydro-energy generation in the Sanaga basin under conditions of climate change, even though its correlation with flows at neighboring stations (Nachtigal, Mbakaou and Edea) is less than optimal. Few actually observed data are available for the location of the Lom Pangar dam. The correlation of monthly flows at Lom Pangar and Betare-Oya for the period 1972-1980, shown in Figure 3.6, was used to fill-in missing data for either station. Sanaga Basin Correlation coefficients 2 CV Area (km ) Stations Edea Nachtigal Mbakaou Bet-Oya Goura Mantoum Lom P. Goyoum Mape Bamendjing 0.14 131,500 Edea 1 0.92 0.89 0.72 0.84 0.86 0.72 0.39 0.76 0.73 0.14 76,000 Nachtigal 1 0.84 0.66 0.80 0.72 0.66 0.50 0.74 0.56 0.17 20,200 Mbakaou 1 0.56 0.75 0.84 0.56 0.27 0.79 0.71 0.18 11,100 Betare-Oya 1 0.58 0.71 1.00 0.22 0.30 0.48 0.16 42,300 Goura 1 0.94 0.58 0.54 0.41 0.73 0.17 14,700 Mantoum 1 0.71 -0.36 0.96 0.82 0.18 19,700 LomPangar 1 0.22 0.31 0.49 0.25 50,500 Goyoum 1 0.27 0.20 0.20 4,020 Mape 1 0.60 0.17 2,190 Bamendjing 1 Annual runoff Sanaga Basin 1,200 Edea Nachtigal Goura Mbakaou Lom Pangar Mape Annual runoff (mm/yr) 1,000 Bamendjing 800 600 400 200 1944 1952 1960 1964 1972 1980 1984 1992 2000 1948 1956 1968 1976 1988 1996 Fig. 3.5: Hydrological characteristics of the Sanaga basin 23 Lom Pangar vs Betare-Oya Monthly flows Goyoum vs Nachtigal (monthly flows; 1972 - 1980) 3,000 1000 y = 0.7233x y = 1.6916x R² = 0.6063 Flow Goyoum (m3/s) R² = 0.9167 Lom Pangar (m3/s) 750 2,000 500 1,000 250 0 0 0 200 400 600 0 1,000 2,000 3,000 Betare-Oya (m3/s) 3 Flow Nachtigal (m /s) Fig. 3.6: Correlation between monthly flows at Lom Pangar and Betare-Oya on the Lom River (left) and Goyoum and Nachtigal on the Sanaga River (right) Figure 3.7 shows a good agreement between the flow data measured for Edea (at Song Loulou) near the outlet of the Sanaga River, at Goura on the Mbam River and at Nachtigal further upstream on the Sanaga River. Hence, these flow data form a good basis for the analysis of climate change impacts on runoff. Figure 3.8 shows hydrographs of daily discharges observed at Edea, Nachtigal and Goura. Other than for the flood seasons of 1980 and 1982 there is a good agreement between the flows measured at these stations. During the dry season differences occur since 1970 after the impoundment of Mbakaou (July 1969) and Bamendjing (July 1974) reservoirs, since only the available flow data for Edea are naturalized flow data. Mapé dam was completed in July 1987. Annual flows Edea vs Nachtigal + Goura Monthly flows Edea vs Nachtigal + Goura 3,000 (1951-1980) 7,000 6,000 2,500 y = 1.1204x y = 1.1388x Flow Edea (m 3/s) 5,000 Flow Edea (m 3/s) R² = 0.8944 R² = 0.9909 4,000 2,000 3,000 2,000 1,500 1,000 1,000 0 900 1,400 1,900 2,400 0 1,000 2,000 3,000 4,000 5,000 6,000 Flow Nachtigal + Goura (m3/s) Flow Nachtigal + Goura (m 3/s) Fig. 3.7: Correlation of annual and monthly flow data for Edea, Nachtigal and Goura 24 8,000 Edea 1.12*(Nachtigal + Goura) Nachtigal Goura 6,000 Discharge (m 3/s) 4,000 2,000 0 1/1/52 12/31/52 12/31/53 12/31/54 12/31/55 12/30/56 12/30/57 12/30/58 12/30/59 12/29/60 12/29/61 12/29/62 8,000 Edea (naturalized) 1.12*(Nachtigal + Goura) Nachtigal Goura 6,000 Discharge (m3/s) 4,000 2,000 0 1/1/63 1/1/64 12/31/64 12/31/65 12/31/66 12/31/67 12/30/68 12/30/69 12/30/70 12/30/71 12/29/72 12/29/73 8,000 Edea (naturalized) 1.12*(Nachtigal + Goura) Nachtigal 6,000 Goura Discharge (m3/s) 4,000 2,000 0 1/1/74 1/1/75 1/1/76 12/31/76 12/31/77 12/31/78 12/31/79 12/30/80 12/30/81 12/30/82 12/30/83 Fig.3.8: Hydrographs for Edea, Nachtigal and Goura (1952 – 1983) – Note: The sum of the flows at Nachtigal and Goura is multiplied with 1.12 according 2 to the regression shown in Figure 3.7; this multiplier is approximately the same as the ratio between the respective catchment areas (Edea: 131,500 km , 2 2 Nachtigal: 76,000 km and Goura: 42,300 km ) 25 Edea naturalized monthly flows Edea naturalized annual flows 8,000 3,000 Annual flow Average Monthly flow (m3 /s) 6,000 2,500 y = -7.7896x + 2114.5 Flow (m3 /s) 4,000 2,000 2,000 1,500 0 average <=1971: 2,057 m3/yr 1 2 3 4 5 6 7 8 9 10 11 12 average >=1972: 1,719 m3/yr 1,000 Months 1944 1954 1964 1974 1984 1994 1944-71 1972-2003 Min Max Lom Pangar monthly flows Lom Pangar annual flows 1,250 400 Annual flow Average Monthly flow (m3/s) 1,000 350 750 y = -1.4387x + 301.99 300 Flow (m3/s) 500 250 250 200 0 average <=1971: 288 m3 /yr average >=1972: 247 m3 /yr 1 2 3 4 5 6 7 8 9 10 11 12 150 Months 1951 1955 1959 1963 1967 1971 1975 1979 1983 1987 1991 1995 1999 2003 Min Max 1951-71 1972-2003 Mbakaou monthly flows Mbakaou annual flows 2,000 550 Annual flow 500 Average Monthly flow (m3/s) 1,500 450 y = -1.5818x + 405.38 1,000 400 Flow (m 3 /s) 350 500 300 0 250 average <=1971: 416 m3 /yr 1 2 3 4 5 6 7 8 9 10 11 12 average >=1972: 347 m3 /yr 200 Months 1959 1963 1967 1971 1975 1979 1983 1987 1991 1995 1999 2003 2007 Min Max Average Fig. 3.9: Abrupt changes in Sanaga basin runoff in 1971 at Edea, Lom Pangar and Mbakaou Numerous studies have been devoted to the severe droughts which plagued particularly West Africa and to a lesser extent Central Africa during the 1970s and 1980s. The general consensus is that around 1970/1971 an abrupt change in precipitation occurred (Tarhule et al, 2013; Tarhule and Grijsen, 2013; 26 ISL et al, 2005b; Kpoumie, 2010; Liénou et al, Normal distribution runoff Sanaga basin 2008; Dzana et all, 2011). Accordingly, a drop of 900 Edea Nachtigal Mbakaou about 16% occurred in the Sanaga flow at Edea, a drop of 14% in the combined Lom and Pangar River flows and a drop of 17% in the Djerem Annual runoff (mm/yr) 700 y = 99.521x + 569.9 River flow at Mbakaou, as shown in Figure 3.9 above. Sighomnou et al (2007) mention a 15% 500 runoff decline for the Sanaga basin since 1970, and overall a 14% decline in runoff for the South y = 75.262x + 450.13 300 of Cameroon. Monthly flows reduced y = 75.959x + 411.21 throughout the year, albeit percentage wise more during the dry season than during the 100 rainy season. The normal probability distribution -2.5 -1.5 -0.5 0.5 1.5 2.5 function applies well to the annual runoff data Reduced Normal variate for the Sanaga basin (Fig. 3.10). Fig. 3.10: Normal distribution of annual runoff for the Sanaga Basin. 3.2.3 Coastal basins Figures 3.11 and 3.12 summarize hydrological characteristics of the Nyong and Ntem Basins, which are considered to be representative for the southern Coastal Basins. Runoff data for Edea at the outlet of the Sanaga basin are included in Figure 3.11 for comparison. The correlation between annual runoff at the various stations is generally good. The coefficients of variation (CV) at all three stations on the Nyong River (Dehane, Eseka and Mbalmayo) are identical, and similar to the CV of annual runoff for the Sanaga basin (average 0.17). The correlation with the Sanaga basin runoff is weak. The runoff data for Dehane are suspiciously high, as can be seen from the catchment area and runoff ratios shown in Table 3.2. Therefore, we will use the data for Eseka on the Nyong River and Ngoazik on the Ntem River to determine the climate elasticity of runoff for the Southern coastal basins. The monthly flow data for Eseka and Ngoazik will also be used to analyze potential climate change impacts on hydro-energy generation at the planned Njock, Mouila and Memve-Ele power plants. The normal probability distribution function applies well to the annual runoff data for the basins. Catchment area Annual runoff 2 Catchment km ratio Eseka m3/s ratio Eseka Dehane 26,400 1.222 447 1.622 Eseka 21,600 1.000 276 1.000 Mbalmayo 13,555 0.628 150 0.543 Table 3.2: Comparison of catchment area and Ngoazik 18,100 0.838 278 1.009 runoff ratios for southern coastal basins Similarly, Figures 3.13 and 3.14 summarize hydrological characteristics for the northern coastal basins, with good correlations between the stations, consistent CV values, a good correlation with the runoff at 27 Edea and normal distribution of annual runoff. Runoff data for Melong, Mundame and Yabassi were selected for further analysis of the climate elasticity of runoff. Nyong and Ntem Basins Correlation coefficients CV Area (km 2) Stations Dehane Eseka Mbalmayo Ngoazik Edea 0.18 26,400 Dehane 1 0.90 0.73 0.70 0.49 0.16 21,600 Eseka 1 0.92 0.78 0.54 0.17 13,555 Mbalmayo 1 0.81 0.43 0.21 18,100 Ngoazik 1 0.43 Annual Runoff Nyong - Ntem Basins 800 Dehane 700 Eseka Annual runoff (mm/yr) Mbalmayo 600 Ngoazik Edea 500 400 300 200 1951 1955 1959 1963 1967 1975 1971 1979 Fig. 3.11: Hydrological characteristics of the Nyong and Ntem river basins Nyong and Ntem Rivers vs Eseka NDF Nyong and Ntem basin runoff 800 800 Dehane y = 99.052x + 528.23 Dehane Ngoazik y = 1.6115x Ngoazik 700 R² = 0.8054 Eseka Mbalmayo Annual flow (m3 /s) Annual runoff (mm/yr) 600 600 y = 102.62x + 485.03 400 500 y = 1.0222x R² = 0.5786 400 200 y = 70.277x + 393.56 300 y = 0.5542x R² = 0.7973 0 200 200 250 300 350 400 -2.5 -1.5 -0.5 0.5 1.5 2.5 Annual flow Eseka (m3/s) Reduced Normal variate Fig. 3.12: Hydrological characteristics of the Nyong and Ntem river basins 28 Coastal Basins - North Correlation coefficients CV Area (km 2) Stations Gouri Mamfe Melong Mundame Yabassi Edea 0.14 2,240 Gouri 1 0.75 0.87 0.95 0.78 0.75 0.20 6,810 Mamfe 1 0.81 0.82 0.84 0.59 0.16 2,280 Melong 1 0.80 0.75 0.76 0.14 2,420 Mundame 1 0.64 0.71 0.15 8,026 Yabassi 1 0.63 Annual runoff Northern coastal areas 3,500 Mundame Mamfe Gouri Yabassi 3,000 Melong Edea Annual runoff (mm/yr) 2,500 2,000 1,500 1,000 500 0 1955 1963 1967 1971 1975 1951 1959 1979 Fig. 3.13: Hydrological characteristics of the northern coastal basins (note: Gouri is located in the neighboring south-eastern portion of the Niger Basin) Annual flows northern coastal areas NDF basin runoff northern coastal areas 3,000 3,500 Mundame y = 343.75x + 2186 Mamfe Mundame y = 2.8464x Yabassi 2,500 3,000 Gouri Melong Annual flow (mm/yr) Annual runoff (mm/yr) Yabassi y = 2.2255x 2,500 2,000 y = 196.01x + 1216.6 2,000 1,500 y = 1.6012x 1,500 1,000 y = 1.2376x y = 165.26x + 976.2 1,000 500 500 0 500 700 900 1,100 1,300 -2.5 -1.5 -0.5 0.5 1.5 2.5 Annual flow Melong (mm/yr) Reduced Normal variate Fig. 3.14: Hydrological characteristics of the northern coastal basins (note: Gouri is located in the neighboring south-eastern portion of the Niger Basin) 29 3.2.4 Congo Basin Due to paucity of data, a runoff data series of useable length was only available for the gauging station Ngbala on Dja River in south-east Cameroon. Missing data for 6 years could be estimated based on the few data available for the Pana station on Kadei River. As expected, the correlation with Sanaga basin runoff is weak, but the CV of annual runoff is similar to other basins and the annual runoff is normally distributed. Congo Basin Correlation coefficients 2 NDF Congo basin runoff CV Area (km ) Stations Pana Ngbala Edea 600 0.18 20,370 Pana 1 0.90 0.42 0.21 38,600 Ngbala 1 0.57 Ngbala 500 Annual runoff Congo Basin Annual runoff (mm/yr) 700 400 Pana Annual runoff (mm/yr) Ngbala 600 300 Edea 500 200 400 y = 75.089x + 350.91 100 300 0 200 -2.5 -1.5 -0.5 0.5 1.5 2.5 1956 1960 1968 1972 1980 1964 1976 Reduced Normal variate Fig. 3.15: Hydrological characteristics of the Congo Basin in Cameroon 3.2.5 Lake Chad Basin As previously mentioned, no runoff data became available for the Lake Chad Basin. However, MINEE and GWP (2009) provide some insight in the inter-annual flow variability of the Logone River at Bongor, located on the border with Chad at about 10.30 N latitude. The station is representative for the southern part of the Lake Chad Basin. Annual flows are plotted in Figure 3.16, which shows a severe reduction in flows during the 1970s and particularly 1980s. This flow reduction was much more severe than that observed for the Sanaga and other basins in Cameroon, as shown in the previous sections. This may indicate that the drier Northern basins of Cameroon are more sensitive to the impacts of climate change and variability that the humid central and southern basins, as will also been shown in Chapters 4 and 6. Nonetheless, flows of the Logone River at Bongor have substantially recovered from the droughts during the late 1990s and early 2000s, and were at 80% of the pre-1970 flow level during the first decade of the 21st century. 30 Fig. 3.16: Inter-annual flow variability for the Logone River at Bongor in the Lake Chad Basin (source: MINEE and GWP, 2009) 3.2.6 Summary of runoff characteristics for the main river basins in Cameroon Overall we can observe a good correlation between the annual runoff observed at multiple stations in each river basin in Cameroon. Coefficients of variation of annual runoff are uniform across the central and southern basins, in the range of 0.16 to 0.20, and about 0.3 for the Benue basin. The latter reflects this part of the Benue basin being in the transition zone between the tropical/equatorial climate in the centre and South of Cameroon and the Sahelian climate in the extreme North of Cameroon (Lake Chad Basin). The variation of annual runoff is well described by the normal probability distribution. These results enable us to focus the analysis of the climate elasticity of runoff in Chapter 4 on a limited number of gauging stations and commensurate sub-catchments considered representative for each river basin, i.e.: ï‚· Garoua and Riao for the Benue Basin ï‚· Edea, Goura, Nachtigal, Lom Pangar and Mbakaou for the Sanaga Basin ï‚· Melong, Mundame and Yabassi for the Northern Coastal Basins ï‚· Eseka (Nyong River) and Ngoazik (Ntem River) for the Southern Coastal Basins, and ï‚· Ngbala for the Congo Basin 31 3.3 Precipitation, temperature and potential evapotranspiration data Despite the existence of 408 meteorological stations in Cameroon, of which only 10% provide currently data on regular basis (MINEE and GWP, 2009), actual observations of precipitation and temperature did not become available to support a basin by basin analysis of the climate elasticity of runoff in Cameroon. Therefore, we used the updated CRU TS 3.10.01 historical climate database (see also BADC-CRU) of the Climatic Research Unit (CRU) at the University of East Anglia (UK) – published by the British Atmospheric Data Centre (BADC) - as best available estimates of monthly and annual precipitation and temperature data for the period 1901 - 2009 (Harris et al, 2013). The CRU-TS (time-series) data sets present month- by-month variations in climate since 1901. These are calculated on high-resolution (0.5x0.5 degree) grids, which are based on an archive of monthly mean climatic parameters provided by more than 4,000 weather stations distributed around the world. The time series include climate variables such as cloud cover, diurnal temperature range, daily mean temperature, monthly average daily maximum and minimum temperatures, precipitation, vapor pressure, wet day frequency and derived variables such as frost day frequency and potential evapotranspiration (PET). However, the PET data provided in the CRU data for Cameroon were judged to be unrealistically low, and we therefore used PET estimates based on the modified Hargreaves method provided by the latest climate wizard site (CKPCW) and CGIAR’s Global Aridity and PET database (CGIAR-PET); see Hargreaves et al (1982, 1985, 1994, 2003) and Annex 6. Annual and average monthly values of precipitation, temperature and potential evapotranspiration are listed in Annex 5 for the sub-catchments selected for the analysis of the climate elasticity of runoff (represented by gauging stations), for the period 1944 – 2009 for which runoff data are also available. Annual and average monthly runoff data for these gauging stations are also included in Annex 5. Annual precipitation and temperature data are graphically presented in Figures 3.16 and 3.17. Average rainfall Cameroon Basins 3,500 3,000 2,500 Rainfall (mm/yr) 2,000 1,500 1,000 500 0 Edea Nachtigal Goura Garoua Eseka Ngoazik Mundame LC-North Fig. 3.16: Long-term average annual rainfall for sub-basins in Cameroon (1901-2009) 32 Average temperature Cameroon River Basins 29 28 27 Rainfall (mm/yr) y = 0.0024x + 27.016 26 y = 0.0022x + 24.673 25 24 23 y = 0.002x + 22.993 22 Edea Nachtigal Goura Garoua Eseka Ngoazik Mundame LC-North Fig. 3.17: Long-term average annual temperature for sub-basins in Cameroon (1901-2009) Gridded long-term annual average precipitation and temperature for the entire country are depicted in Figures 3.18 and 3.19. Rainfall generally varies across the Sanaga, Congo and Southern Coastal Basins between 1,500 and 2,250 mm/yr, with very high rainfall (>3,000 mm/yr) in parts of the northern coastal basins and sharply decreasing rainfall north of the Sanaga Basin, about 1,100 mm/yr as an average for the Benue Basin upstream of Garoua and less than 500 mm/yr in the extreme North of Cameroon close to Lake Chad. Annual temperatures (Figure 3.19) range from 23 0C to 25 0C in most of Central and South Cameroon to 27 0C in the Benue Basin and >28 0C in the Northern part of the Lake Chad Basin. Annual potential evapotranspiration (PET) is shown in Figure 3.20. This map was prepared from CGIAR’s Global Aridity and PET database (CGIAR-PET; Zomer et al, 20006, 2008) and is consistent with PET values obtained from the Climate Wizard. Both systems use the modified Hargreaves method (Hargreaves et al, 1982, 1985, 1994 and 2003; Allen et al, 1998; Droogers and Allen, 2002) to estimate monthly values of PET. Xu and Singh (2001) analyzed, compared and generalized the various popular evaporation equations that belong to the category of temperature-based methods for the estimation of E0. For monthly evaporation values, they found that the modified Blaney–Criddle (1950) and modified Hargreaves methods produced the least errors. The modified Hargreaves method provides good estimates of the reference crop evaporation compared to estimates obtained with the standard Penman-Monteith method (Monteith, 1965). PET varies across Cameroon, from > 2,000 mm/yr in the North to 1,500 mm/yr along the coast. Probability distributions of annual rainfall and temperature are shown in Figures 3.21 and 3.22. The normal probability distribution function applies well, similar to what was observed in Section 3.2 for the annual runoff data. The coefficient of variation of annual precipitation varies uniformly between 0.08 and 0.10 across most of Cameroon, except in the northern part of the Lake Chad Basin (0.17). 33 Fig. 3.18: Spatial distribution of annual precipitation and temperature across Cameroon (1901 – 2009); Source: CRU TS 3.10.01 34 Fig. 3.19: Spatial distribution of annual precipitation and temperature across Cameroon (1901 – 2009); Source: CRU TS 3.10.01 35 Fig. 3.20: Spatial distribution of annual potential evapotranspiration across Cameroon (Source: CGIAR) 36 Normal distribution of annual rainfall 1944 - 2009 3,500 y = 304.77x + 2660.4 3,000 2,500 y = 190x + 2223.6 Rainfall (mm/yr) 2,000 1,500 y = 128.5x + 1559.6 1,000 y = 107.33x + 1076.6 y = 117.77x + 682.42 500 0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 Reduced normal variate Edea Nachtigal Garoua Eseka Ngbala Mundame Yabassi LC-North Fig.3.21: Normal probability distributions of annual rainfall for sub-basins in Cameroon Normal distribution annual temperature (1944 - 2009) 30 29 y = 0.4726x + 27.959 28 Temperature (0C) 27 y = 0.3835x + 27.137 26 y = 0.353x + 24.778 25 24 23 y = 0.3443x + 23.678 22 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 Reduced normal variate Edea Nachtigal Garoua Eseka Ngbala Mundame Yabassi LC-North Fig.3.22: Normal probability distributions of annual temperature for sub-basins in Cameroon 37 Monthly climatic parameters (annual P, T and PET) are shown in Figure 3.23 and the comparable annual runoff is shown in Figure 3.24. The duration of the rainy season across the Benue Basin (upstream of Garoua) is shorter than further South. South of Ngaoundare rains generally start in March and cease in October/November. Rainfall over southern Cameroon (Coastal Rivers and Congo Basin) shows two rainy seasons, with relatively low rainfall during June to August, but this pattern is not distinctly discernible in the rainfall distribution over the Sanaga basin and the northern coastal basins. The spatial and temporal distribution of monthly temperatures is fairly uniform across the Sanaga, Congo and southern coastal basins, with the lowest temperatures (22 0C) occurring during the peak of the rainy season (July, August), and the highest temperatures (25 to 26 0C) occurring during March to April, at the onset of the rainy season. Temperatures in the northern coastal basins tend to be two degrees lower, while temperatures increase significantly while moving north from the Sanaga basin. Monthly potential evapotranspiration (PET) reflects the same pattern as temperature, with maximum values prior to the onset of the rainy season. The temporal and spatial distribution of monthly runoff (Figure 3.24) is commensurate with the distribution patterns for monthly rainfall (Figure 3.23). Only few recorded rainfall data were available to verify the representativeness of the CRU-TS 3.10 precipitation data for Cameroon, i.e. annual precipitation data for 4 stations in the Sanaga Basin with each continuous data available for the period 1952 - 2002, namely: Edea (south-west of Basin), Bertoua (south-east), Banyo (north-west) and Bafia (centre-west). Figure 3.25 (left panel) shows the regression between the average rainfall for these 4 stations and the CRU-TS 3.10 gridded precipitation data for the Sanaga Basin. Since the station Edea is located on the far western end of the basin, its rainfall is more representative for the higher rainfall across the western coastal basins than for the Sanaga Basin proper. Therefore, the right panel of Figure 3.25 also shows the correlation between CRU precipitation data and the average annual rainfall at Bertoua, Banyo and Bafia. Overall, the average long-term rainfall appears to be well captured by the CRU data set, even though the correlation shown in Fig. 3.25 is rather poor. The long-term coefficients of variance (CV) are similar for both data sources, i.e. 0.09. Hence, we can assume that the CRU-TS 3.10 precipitation data set reflects the overall rainfall characteristics of the Sanaga basin, and its neighboring basins. 38 500 500 400 400 Rainfall (mm/month) Rainfall (mm/month) 300 300 200 200 100 100 Month 0 Month 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 Edea Goura Nachtigal Betare-O Yabassi Mund. Melong Gouri Mbakaou Riao Garoua LC-North Ngoazik Eseka Ngbala 32 32 30 Temperature (0C) 30 Temperature (0C) 28 28 26 26 24 24 22 22 Month Month 20 20 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Edea Goura Nachtigal Betare-O Yabassi Mund. Melong Gouri Mbakaou Riao Garoua LC-North Ngoazik Eseka Ngbala 200 200 PET (mm/month) PET (mm/month) 150 150 Month Month 100 100 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Edea Goura Nachtigal Betare-O Yabassi Mund. Melong Gouri Mbakaou Riao Garoua LC-North Ngoazik Eseka Ngbala Fig. 3.23: Monthly variations of rainfall (upper panels), temperature (middle panels) and PET (lower panels) for selected sub-basins (Sanaga and Niger basins in left panels and coastal basins and Congo basin in right panels) 39 150 400 Runoff (mm/month) Runoff (mm/month) 300 100 200 50 100 Month 0 0 Month 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Edea Goura Nachtigal Betare-O Yabassi Mund. Melong Gouri Mbakaou Riao Garoua Ngoazik Eseka Ngbala Fig. 3.24: Monthly variations of runoff for selected sub-basins (Sanaga and Niger basins in left panel and coastal basins and Congo basin in right panes) Sanaga Basin rainfall CRU-TS 3.10 data vs Sanaga Basin rainfall CRU-TS 3.10 data vs 2,200 areal precipitation 4 stations 2,200 areal precipitation 3 stations Sanaga Basin CRU rainfall Sanaga Basin CRU rainfall y = 0.9411x y = 1.072x 2,000 2,000 (mm/yr) (mm/yr) 1,800 1,800 1,600 1,600 1,400 1,400 Pre-1972 Pre-1972 1,200 Post-1971 1,200 Post-1971 1,200 1,400 1,600 1,800 2,000 2,200 1,200 1,400 1,600 1,800 2,000 2,200 Rainfall 4 stations (mm/yr) Rainfall 3 stations (mm/yr) Fig. 3.25: Correlation between CRU-TS 3.10 data and actual precipitation data Sanaga Basin 3.4 Trends and abrupt changes in rainfall across Cameroon A full discussion of hydroclimatic variability in Cameroon is beyond the scope of the present report. It is briefly discussed here to indicate that hydrological data (where available) for the period 1972 – 2003 are most suitable for the analysis of the climate sensitivity of the present and future water resources and hydropower infrastructure in the Sanaga and other river basins in Cameroon. Hydroclimatic variability manifests in several ways, including the occurrence of abrupt shifts in the mean and trends in precipitation. An abrupt shift is a point in time at which the parameters of the underlying distribution (e.g. mean, variance, or trend) or the parameters of the model used to describe the time series abruptly change (Beaulieu et al., 2012, p. 1229). A common type of change point is a regime shift, which is a rapid reorganization from one relatively stable state to another (Rodionov, 2004). The regimes may last for several decades and then abruptly change to other regimes. Figure 3.26 shows examples of various change points. 40 Fig. 3.26: Examples of time series with change point in (a) the mean, (b) the variance, (c) both the mean and variance (adapted from Beaulieu et al., 2012, p. 1231) The occurrence of an abrupt change point suggests that the statistical properties, primarily the mean and standard deviation, of the time series before and after the change point are significantly different and therefore should be treated as two separate time series. This finding has important implications for water resources planning because the results of analyses which take the occurrence of the change point into account can be significantly different from those which ignore the change point. For example, if one were to ignore in Figure 3.9 the occurrence around 1970 of an abrupt negative shift of 16% in the available runoff time series for the Sanaga Basin, one would conclude that there is a negative trend in runoff in the order of 0.4% to 0.5% per year. Extrapolation of such trend would project a further reduction of runoff with 20% between 2005 and 2050. However, by taking the change point into account, one may conclude that likely no trends, whether positive or negative, exist in the hydro- meteorological time series. Acceptance of the existence of a change point implies that we accept the time series pattern since the change point as the new normal hydrological regime. Thus, the assessment of climate risks and water resources planning should be based on this new normal. Tarhule et al (2013) extensively analyzed hydro-meteorological time series for 14 sub-basins of the Niger Basin. Results showed that an abrupt change point occurred around 1970 in the rainfall and thus in the streamflow (but not in the temperature) time series in all sub-watersheds of the Niger Basin. For the entire Benue Basin an abrupt change of -7% was assessed, against -16% for the Upper and Middle Niger Basins. Examples of abrupt changes or trends in precipitation based on recorded rainfall in Cameroon are shown in Figure 3.27 for the average of 3 rainfall stations in the Sanaga basin and for the Edea rainfall station, representative for the coastal basins. Without considering an abrupt negative change of about 10% by 1971, one would conclude that there is a negative trend of about 0.3% per year in precipitation. 41 Average precipitation for Bertoua, Banyo and Annual precipitation Edea rainfall station Bafia stations in the Sanaga Basin 2,200 3,600 Annual rainfall Annual rainfall Average y = -9.4824x + 2831.2 Average Precipitation (mm/yr) Precipitation (mm/yr) 2,000 3,200 y = -4.591x + 1724.7 1,800 2,800 1,600 2,400 1,400 2,000 average <=1971: 2,742 average <=1971: 1,682 mm/yr mm/yr average >=1972: 1,556mm/yr 1,200 1,600 1944 1949 1954 1959 1964 1969 1974 1979 1984 1989 1994 1944 1949 1954 1959 1964 1969 1974 1979 1984 1989 1994 1999 Fig. 3.27: Trends and abrupt changes in recorded precipitation data for the Sanaga Basin Precipitation Nyong Basin at Eseka (CRU Precipitation Sanaga Basin at Edea (CRU data) data) 2,200 2,200 Annual… Annual rainfall y = -0.9684x + 1752.6 y = -1.5466x + 1793.4 Average Average Precipitation (mm/yr) Precipitation (mm/yr) 2,000 2,000 1,800 1,800 1,600 1,600 1,400 1,400 average <=1971: 1,756 mm/yr average <=1971: 1,792 mm/yr average >=1972: 1,694mm/yr average >=1972: 1,706mm/yr 1,200 1,200 1999 1954 1944 1949 1954 1959 1964 1969 1974 1979 1984 1989 1994 1944 1949 1959 1964 1969 1974 1979 1984 1989 1994 Fig. 3.28: Trends and abrupt changes in CRU-TS 3.10 precipitation data for the Sanaga and Nyong Basin 1999 Similar plots derived from the CRU data sets are shown in Figure 3.28 above for the Sanaga and Nyong basins. It is observed that the abrupt changes, if they were to exist around 1971, are much less pronounced (only about -4%) than the changes observed in the actual precipitation and runoff data, while trend lines indicate likely insignificant negative trends (<0.1% per year). Thus, the CRU precipitation data series appear to underestimate the regime change, which may have taken place around 1970. However, the CRU time series do present to an extent the droughts of the early 1970s and the 1980s. In view of these results, and the reduced runoff after 1970 shown in Figure 3.9, we will base our analysis for the Sanaga basin on the available data for the period 1972 – 2003. For other basins no runoff data became available beyond 1980, and available previous runoff data will be used. No change points or significant trends occur in the CRU temperature time series shown in Fig. 3.17. 42 4. Runoff response to climate change For this study, the primary goal is to determine the relative changes (in %) in annual runoff and system performance criteria such as annual and seasonal hydro-energy generation due to relative changes (in %) in annual climate parameters (notably precipitation P and temperature T) caused by projected future climate changes. For this purpose Schaake (1990) introduced the concept of elasticity for evaluating the sensitivity of streamflow to changes in climate, and defined climate elasticity of streamflow by the proportional change in streamflow Q divided by the proportional change in a climate variable, such as precipitation. Accordingly, the precipitation and temperature elasticities of runoff are defined as: εP = [dQ/Q]/[dP/P] and εT = [dQ/Q]/[dT/T] The first task, then, is to determine the climate elasticities of runoff; Chapter 5 elaborates the response of performance indicators to changes in runoff. Unfortunately, the response of runoff to climate change is complicated by a number of confounding factors, including land-use and land-cover characteristics and complex vegetation feedback processes. “Different catchments respond differently to the same change in climate drivers, depending largely on catchment physio-geographical and hydro-geological characteristics and the amount of lake or groundwater storage in the catchment.â€? (Kundzewicz et al, 2007). We have extensively reviewed available literature on the climate elasticity of basin runoff, inter alia Wigley and Jones (1985), Gedney et al (2006), Gleick (1986, 1987), Karl and Riebsame (1989), Risbey and Entekhabi (1996), Vogel et al (1999), Labat etal (2004), Legates et al (2005), Sankarasubramanian et al (2001), Chiew (2006), Chiew et al (2006), Teng et al (2012) and Grijsen and Brown (2013). Arora (2001) provided a useful theoretical underpinning for the estimation of climate elasticities of runoff on the basis of an aridity index, or alternatively the runoff coefficient. Reviewed studies also included the estimation of climate change impacts on runoff through hydrological modeling studies for similar river basins in Africa, inter alia Deksyos and Abebe (2006), Strzepek and McCluskey (2006) and SNC Lavalin (2007). We also applied regression models to the available rainfall and runoff data for various sub-basins in Cameroon. The data used were spatially aggregated (gridded) annual precipitation data for approximately the period 1950-2009 for major sub-basins and annual stream flow data obtained from GRDC and EDC, as discussed in Chapter 3. Regression between temperature and runoff data was not attempted, for reasons discussed below. Hydrological modeling was not attempted due to lack of data for model calibration and verification, and previous rather unsuccessful modeling attempts. 4.1 Assessment of climate elasticity of streamflow through regression analysis The classic approach consists of the analysis of climate-geomorphology-runoff relationships, which involves the “correlation between measured climate and hydrological parameters such as precipitation, runoff or temperatureâ€? (Labat et al, 2004). This approach dates back to the 1940s, when Langbein (1949) related mean annual runoff from 22 drainage basins in the United States to the mean annual total precipitation and the weighted temperature. According to Karl and Riebsame (1989), the results of runoff estimates due to temperature changes based on Langbein’s curves are not consistent with the 43 findings of other studies, mainly because they overestimated the temperature impacts on evapotranspiration. In their empirical analysis and testing, they showed that temperature fluctuations in the United States are not as great a factor in runoff changes as suggested by previous studies. They noted, ‘‘our version of Langbein’s nomogram ..., based on temporal fluctuations of climate and runoff in 82 basins with minimum human impact, indicates that precipitation changes may be amplified one to six times relative to runoff changes. However, even 1 to 2°C average temperature changes often have little effect on annual runoffâ€?. Wigley and Jones (1985) also show that the influences of precipitation changes dominate the impact of evapotranspiration changes (caused by temperature changes) on runoff. Subsequently, Gleick (1986, 1987) concluded from using several scenarios of future climate as input to a water-balance model of California's Sacramento Basin that annual runoff is affected primarily by precipitation changes, not by temperature changes, while the seasonal distribution of runoff is affected by changes in mean monthly temperature. Risbey and Entekhabi (1996) also showed that multivariate (at-site) models can be misleading because they tend to show greater sensitivity to temperature than either observations or modeling approaches suggest. The Third IPCC Assessment Report concludes “where data are available, changes in annual streamflow usually relate well to changes in total precipitationâ€?. Ultimately, they concluded, “..., precipitation trends can easily explain the observed changes in continental-scale runoffâ€?. Vogel et al (1999) carried out a study for 1,553 undeveloped basins across the USA, where runoff was not significantly impacted by human development activities, such as storage in reservoirs and abstractions for irrigation. The authors sought to develop regional relationships between the first two moments of annual streamflow and readily measured basin and climate characteristics, for 18 major U.S. water resource regions. Using a log-linear model, the authors related the average and standard deviation of annual streamflow to a variety of parameters, including climate and geomorphic basin characteristics, while keeping the number of basin characteristics used to a minimum. The adopted log- linear regression model is of the form: Q = ea Xb Yc Zd.... or ln(Q) = a + b ln(X ) + c ln(Y ) + d ln(Z) + ….. The authors showed significantly improved results in the estimation of annual stream flow when climate information (P = annual precipitation and T = annual temperature) is added to the catchment area (A) as variables in the regional regression equations. The regression model for the annual stream flow is then: Q = α Ad Pb Tc or ln (Q) = a + b ln (P) + c ln (T) with scale factor a (=α Ad) and shape factors b and c. A good estimate of b and c can be obtained from the linearized version, based on ordinary least squares: dQ/Q = b dP/P + c dT/T, or (Q-μQ)/μQ = b (P-μP)/μP + c (T-μT)/μT b = εP = (CvQ/CvP) [(Ï?QP – Ï?PT.Ï?QT)/(1-Ï?PT2)] 44 c = εT = (CvQ/CvT) [(Ï?QT – Ï?PT.Ï?QP)/(1-Ï?PT2)] Due to the low correlation between Q and T, as well as between P and T, b equals approximately b = εP = Ï?QP . CvQ / CvP (similar to linear and non-linear Q – P models discussed below). For this log-linear model the precipitation elasticity εP equals the coefficient b, and the coefficient c represents the temperature elasticity εT of stream flow. The precipitation elasticity is positive and the temperature elasticity is generally negative, indicating that increases in temperature tend to increase evapotranspiration and decrease stream flow. Unfortunately, the authors’ treatment of temperature in the process was not explicit. As a result, it is impossible to isolate in their paper the contribution of temperature to the overall improvement of the regression from the contribution of precipitation. The variables P and T were added jointly in the analyses, and it is not clear how much of the stream flow variance was explained by temperature. Generally, variations in annual average temperatures are very small compared to the annual variations in rainfall and the random noise contained in the runoff data series. Based on the CRU precipitation and temperature data sets we found that for various sub-basins in Cameroon, the coefficient of variation of annual rainfall is 7 times the coefficient of variation of annual average temperatures. This makes it difficult, if not impossible, to reliably detect the impacts of temperature on runoff from the available flow, precipitation and temperature data series, even though temperature affects the E0 and thus the actual evaporation losses. Grijsen and Brown (2013) also concluded that available runoff and temperature data for the Niger Basin could not be used for determining the temperature elasticity of Niger Basin runoff. Instead, they found across the Niger Basin consistent values in the range of 2.0 to 2.5 for the precipitation elasticity. Krakauer and Fung (2008) used essentially the same USGS data base as Vogel et. al (1999), i.e. a set of over 1,000 stream gauges, primarily from small and minimally disturbed watersheds. Using multiple regressions in order to isolate the impacts of global warming and increasing CO2 levels on streamflow, they found that “changing precipitation ...... explains most of the inter-annual and longer term variability in streamflowâ€?, and that “multiple regression of streamflow against precipitation, temperature and CO2 suggests that higher CO2 levels may increase streamflow, presumably from lower transpiration due to the physiological plant response to CO2, but that this positive response is offset by concomitant increasing evaporation due to global warming.â€? It was also found that the positive CO2 impact (lower transpiration) on streamflow is strongest compared to the negative temperature impact (higher E0) in parts of the USA where summer precipitation dominates (Mid-West and South-West). The impact of increased plant water use efficiency on streamflow is thus relatively largest where precipitation is concentrated during the growing season and therefore is used by plants rather than running off and contributing to streamflow. By contrast, abiotic evaporation increases with temperature regardless of the season or CO2 level. Thus, the study by Krakauer and Fung (2008) suggests that the future impacts of rising temperatures and CO2 levels on streamflow could be relatively small compared to the impact of future rainfall changes, due to the observed offset between lower transpiration due to higher CO2 levels and higher abiotic evaporation due to higher temperatures. 45 Sankarasubramanian et al (2001) studied the precipitation elasticity of runoff for 1,291 basins across the USA, and produced the elasticity contour map shown in Figure 4.1; for most of continental USA precipitation elasticities vary between 1.5 and 2.5 . Watershed model-based estimates of εP were shown to be highly sensitive to model structure and calibration errors. Through Monte Carlo experiments the authors showed that the non-parametric estimator: εP = median {[dQ/μQ]/[dP/μP]}, applied to annual runoff and precipitation data series, provides a robust and low-bias estimator for εP, where dQ and dP denote annual deviations from the long-term average runoff μQ and precipitation μP. It was found that εP tends to be low for basins with significant snow accumulation (values 1.0 to 1.5), as well as for basins whose moisture and energy inputs are seasonally in phase which each other. These conditions do not prevail in Cameroon, thus higher precipitation elasticities (2 to 2.5) can be expected. Figure 4.1: Contour map of precipitation elasticity of stream flow for the continental USA (Source: Sankarasubramanian et al, 2001) Sankarasubramanian et al (2001) also show that similar robust and non-biased estimates of εP can be obtained on the basis of simple linear and non-linear runoff-precipitation models, as follows: Linear runoff – precipitation model (Q - P): Q = a (P - P0), with a = Ï?QP . σQ / σP (scale factor) and P0 = μP [1– CvP / (CvQ. Ï?QP)], where: Ï?QP = correlation coefficient for Q and P; σQ and σP are standard deviations of runoff (Q) and precipitation (P), μP = mean P; μQ = mean Q, and Cv = coefficient of variation The precipitation elasticity of runoff is then obtained from (Q-μQ)/μQ = εP (P-μP)/μP, as: εP = μP/(μP – P0) = Ï?QP . CvQ / CvP 46 Non - linear runoff – precipitation model (Q - P): Q = a P b, with a = μQ /[Σ Pb/n] (scale factor) and b = shape factor and precipitation elasticity of Q An estimate of the shape factor b is obtained from the linearized version: dQ/Q = b dP/P, or (Q-μQ)/μQ = b (P-μP)/μP εP = b = Ï?QP . CvQ / CvP (same as for the linear Q – P model) Here we used the precipitation elasticity estimators εP = Ï?QP . CvQ / CvP and - due to the low correlation assessed between CRU precipitation and observed runoff data – we used as well εP = CvQ / CvP. We also used findings from literature and a theoretical relationship between long-term precipitation, and actual and potential evapotranspiration (Section 4.2) to derive theoretical values of the climate elasticities of runoff. Sankarasubramanian’s (2001) estimator εP = median {[dQ/μQ]/[dP/μP]} produced erratic results. 4.2 Use of the aridity index to assess climate change impacts on annual runoff Precipitation and available energy (expressed in terms of potential evapotranspiration E0) largely determine the actual annual evapotranspiration (E) and runoff (Q) rates in a region. The aridity index φ = E0/P, i.e. the ratio of annual potential evapotranspiration (E0) to annual precipitation (P), has been shown to describe the actual evapotranspiration ratio E/P and the runoff coefficient Q/P (= 1-E/P) of catchments for a range of climatic regimes. Based solely on the aridity index of a basin Arora (2002) derived simple analytic expressions to estimate changes in runoff due to changes in precipitation and (temperature driven) changes in potential evapotranspiration, as a first order estimate of the effect of climate change on annual runoff, i.e. the precipitation elasticity εP and the evaporation elasticity εE0 = [dQ/Q]/[dE0/E0] of runoff. The latter is then used to derive the temperature elasticity and temperature sensitivity of runoff, εT and ST. Arora’s analysis and the temperature elasticity/sensitivity of E0 are discussed in detail in Annex 6. The author discussed five functional forms, which describe the actual evaporation ratio E/P as a function of the aridity index φ, and assessed how the results from the Canadian Centre for Climate Modeling and Analysis (CCCma) third-generation atmospheric GCM (AGCM3) compared with those estimated by these five functional forms. We used the functional form introduced by Turc (1954) and Pike (1964): E/P = [1+ φ-2]-0.5 (evaporation ratio) and Q/P = 1 – E/P (runoff coefficient), which ultimately yields (see Annex 6). εP = 1 + β = 3 – 3 Q/P + (Q/P)2; precipitation elasticity of runoff εE0 = -β; potential evapotranspiration elasticity of runoff εT = -β T/(T+17.8) = -0.57 β; temperature elasticity of runoff for T=240C ST = - β/(T+17.8); temperature sensitivity of runoff (change in runoff per 1oC change in temp.) β = [1 + φ2]-1 / {[1 + φ-2]0.5 – 1} = (1 + E/P) E/P = 2 – 3 Q/P + (Q/P)2 47 φ 0.0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.5 3 4 5 E/P 0.0 0.243 0.447 0.600 0.707 0.781 0.832 0.868 0.894 0.928 0.949 0.970 0.981 Q/P 1.0 0.757 0.553 0.400 0.293 0.219 0.168 0.132 0.106 0.072 0.051 0.030 0.019 σE/σP 0.0 0.014 0.089 0.216 0.354 0.476 0.576 0.655 0.716 0.800 0.854 0.913 0.943 σQ/σP 1.0 0.986 0.911 0.784 0.646 0.524 0.424 0.345 0.284 0.200 0.146 0.087 0.057 β 0.0 0.301 0.647 0.960 1.207 1.391 1.524 1.622 1.694 1.791 1.849 1.911 1.942 εP 1.0 1.301 1.647 1.960 2.207 2.391 2.524 2.622 2.694 2.791 2.849 2.911 2.942 εEo 0.0 -0.301 -0.647 -0.960 -1.207 -1.391 -1.524 -1.622 -1.694 -1.791 -1.849 -1.911 -1.942 εT (T=24 0C) 0.0 -0.173 -0.372 -0.551 -0.693 -0.798 -0.875 -0.931 -0.973 -1.028 -1.061 -1.097 -1.115 ST (T=24 0C) 0.0 -0.7% -1.5% -2.3% -2.9% -3.3% -3.6% -3.9% -4.1% -4.3% -4.4% -4.6% -4.6% Table 4.1: Theoretical values of actual evapotranspiration, runoff coefficient and climate elasticities Results obtained with the above equations are shown in Table 4.1. Generally, higher rainfall decreases the aridity index φ as well as the climate elasticities of runoff, and increases the runoff coefficient Q/P. The reverse is true for increasing potential evapotranspiration E0 or decreasing rainfall, which increases the aridity index and the climate elasticities of runoff and decreases the runoff coefficient. For example, for large φ (infinite) we obtain: E/P = 1, Q/P = 0, β = 2, εP = 3 and εT = -1.15 (for T = 24 0C). However, for large values of the aridity index, such as for Northern Cameroon (Lake Chad Basin), where the aridity index is about 3 (E0 = 2,000 mm and P = 700 mm), runoff is only a few percents of precipitation and a relatively large climate elasticity has only minor impacts in absolute terms of runoff. Instead, for the Sanaga and southern basins, φ varies between 0.7 and 0.9, and we obtain, E/P = 0.62, Q/P = 0.38, β = 1.0, εP = 2.0, εT = -0.6 (for T = 24 0C) and ST = -2.4%/0C. Finally, for φ = 0 we obtain: E/P = 0, Q/P = 1, β = 0, εP = 1 and εT = ST = 0. Under cold Nordic conditions with mainly snowmelt as runoff and minimal evaporation the precipitation elasticity of runoff is thus about 1. For Cameroon, the temperature-based Hargreaves (1982) method for the estimation of E0 yields a temperature elasticity of E0 equal to 24/(24+17.8)=0.57 and a temperature sensitivity of E0 = (24+17.8)-1 = 2.4% (see for details Annex 6). Thus, a 20C increase in temperature by 2050 will increase potential evapotranspiration by nearly 5%. Chiew et al (2006) and Chiew (2006) analyzed concurrent annual precipitation, temperature and runoff time series for 521 (by and large) unregulated catchments around the world, varying between 100 and 76,000 km2. The precipitation elasticities of 80% of the basins varied between 1.0 and 3.0, as predicted by the Turc-Pike model. Higher εP values (>2) were found in Australia and in southern and western Africa (Figure 4.2), while lower values (<2) were found in the mid and high latitudes of the Northern Hemisphere, with the lowest εP values in cold as well as very wet tropical climates. The results shown for Australia are well described by the equation εP = 3 – 3 Q/P + (Q/P)2, with Q/P = runoff coefficient. 48 Fig. 4.2: Estimates of εP for Africa (left panel) and Australia (right panel)(source: Chew et al, 2006b) The application of these theoretical formulas will be shown in Section 4.4. Even though the above equations do not accurately represent the hydrological cycle, particularly due to the seasonality of the precipitation, they indicate the order of magnitude of the climate elasticity of runoff. A similar conclusion was reached by Teng et al (2012), who studied the climate change impact on mean annual runoff across continental Australia. Impacts were estimated using the Budyko-Fu equations, initially formulated by Budyko (1948) as a functional form in principle similar to the Turc-Pike model used in this report. Estimates were informed by projections from 15 global climate models and compared with the results from extensive hydrological modeling. Averaged across large regions, the estimates from the Budyko-Fu equations were reasonably similar to those from the hydrological models. They concluded that in view of the simplicity of the Budyko equation, the similarity in the results, and the large uncertainty in global climate model projections of future precipitation, these types of equations (including the Turc-Pike equation used in this report) are suitable for estimating climate change impact on mean annual runoff across large regions. They found such equations particularly useful for data- limited regions, for studies where only estimates of climate change impact on long-term water availability are needed, and for investigative assessments prior to a detailed hydrological modeling study. Arora (2001) reached similar conclusions. The equations are, however, limited to estimating the change in mean annual runoff for a given change in mean annual precipitation and potential evapotranspiration. Hydrological models, on the other hand, are required to also take into account potential changes in sub-annual and other climate characteristics, as well as provide a continuous simulation of daily and monthly runoff, which can be important for water availability studies. For the purpose of our study, we are mainly interested in future changes in mean annual runoff 4.3 Climate and hydrological modeling The deterministic approach to determining runoff response to climate change is through GCM modeling, “often coupled to distributed models of surface and subsurface hydrologic processesâ€? (Labat et al, 2004, 49 p.631). The key element of these methods is that various IPCC (4th assessment) emission scenarios are used to drive hydrological models to produce an estimate of expected runoff. Associated with this approach is the use of downscaling from the climate model scale to the catchment scale. Unfortunately, numerous studies reviewed in Kundzewicz et al (2007) tend to focus primarily on Europe, North America, and Australasia and were rather inconclusive regarding West-Africa. Based on a review of the various studies and approaches, Kundzewicz et al (2007) concluded, “In general, these studies have shown that different ways of creating scenarios from the same source (a global-scale climate model) can lead to substantial differences in the estimated effect of climate change, but the hydrological model uncertainty may be smaller than errors in the modeling procedure or differences in climate scenarios. However, the largest contribution to uncertainty in future river flows comes from the variations between the GCMs used to derive the scenarios.â€? Moreover, “In regions with little or no snowfall, changes in runoff are dependent much more on changes in rainfall than on changes in temperature. A general conclusion from studies in many rain dominated catchments is that flow seasonality increases, with higher flows in the peak flow season and either lower flow during the low flow season or extended dry periods.â€? Deksyos and Abebe (2006) assessed the potential impacts of climate change on the water resources of the Lake Tana catchment in Ethiopia, constituting the headwaters of the Blue Nile, with an average runoff of 230 mm/yr. The regional monthly water balance cum rainfall-runoff model WatBal (Yates, 1996) was used to assess the sensitivity of runoff to climate change for synthetic climate change scenarios. An overall increase in temperature of 20C and unchanged rainfall conditions yielded a decrease of 11.3% in the annual runoff. Temperature in the basin shows small seasonal variations with an annual average of 200C, which sets the temperature elasticity at about -1.1. The precipitation elasticity was assessed at 1.8 for decreasing rainfall and 2.2 for increasing rainfall, on average 2.0. Strzepek and McCluskey (2006) applied the conceptual WatBal model to gridded data to simulate changes in soil moisture and runoff across Africa. The model inputs were the climate variables of the 1961–1990 climatology and physiological parameters (e.g. soil properties and land use) derived from global datasets for each of the 0.5o latitude/longitude cells across the continent. To ascertain the possible impacts of climate change this study used GCM-based climate change scenarios as input to the WatBal model. A subset of the 20 scenarios - produced by the Climate Research Unit (CRU), University of East Anglia, Norwich, UK - for which data are available at 0.5º x 0.5º for the globe, was employed to represent a range of equally plausible future climates, with differences attributable to the different climate models used and to different emission scenarios (A2 and B2) that the world may follow. The WatBal model simulated the impacts of these different scenarios on runoff and actual evaporation. The possible variation of Africa-wide climate change impacts on runoff ranged for 2050 from a decrease of 15% to an increase of 5% compared to the 1961–1990 base case hydrology. Runoff variations due to climate change for West Africa, represented by Ghana, Burkina Faso, Cameroon and Niger, ranged 50 between a decrease of 20% and an increase of 10%, with an average decrease of 4% (B2 scenario) to 8% (high A2 scenario). For the Upper Nile Basin, including Lake Victoria, SNC Lavalin (2007) provides an assessment of the potential impacts of climate change on hydro-energy generation in the region. The WatBal model, developed for a ½ degree by ½ degree grid of Africa, was used to calculate potential evapotranspiration, actual evapotranspiration, runoff and relative soil moisture, and test the sensitivity of runoff to climate change. The model was calibrated at the ½ degree level against the GRDC11 Global Gridded Runoff Database. The A1B and A1F1 emission scenarios were used for assessing climate and corresponding runoff changes, representing a relatively high economic growth worldwide and a relatively low growth in population. The output of 7 GCM models, which best simulated the climate of East Africa, were used to project changes in temperature and precipitation for 2050 and 2100 relative to the 20th century. The pertinent hydro-climatic conditions for 3 test basins are shown in Table 4.2, along with theoretical values for climate elasticities, derived from the Pike-Turc equations (Section 4.2). The Kyoga region is located North of Lake Victoria in Uganda, the Tanganyika region represents the Kagera river Basin in Rwanda, Burundi and West Tanzania, and the Nyasa region represents southern Tanzania. Theoretical and actual values of E/T and Q/T agree very well for the Tanganyika and Nyasa basins, and less so for the Kyoga basin. The aridity indices are in the order of 1.3, the theoretical precipitation elasticity is about 2.4 and the temperature elasticity is about -0.8. Parameter Kyoga Tanganyika Nyasa Theory Watbal Theory Watbal Theory Watbal E0 (mm/yr) 1,770 1,653 1,715 P (mm/yr) 1,388 1,249 1,291 Q (mm/yr) 183 272 254 0 22.9 20.2 21.7 Temp ( C) Aridity Index (φ) 1.275 1.323 1.328 E/P 0.787 0.868 0.798 0.782 0.799 0.803 σE/σP 0.487 0.604 0.508 0.539 0.510 0.612 β 1.406 1.434 1.437 Precip. Elasticity 2.406 2.434 2.437 E0 elasticity -1.406 -1.434 -1.437 Temp. elasticity -0.791 -0.763 -0.790 Q/P 0.213 0.132 0.202 0.218 0.201 0.197 Table 4.2: Climate elasticities for the Nile Equatorial Lakes Basin Overall, the GCM models predicted for 2050 a temperature increase of 1.80C and an increase in precipitation of 4 to 14% for the A1B scenario. All three regions exhibit an approximately linear relationship between changes in temperature and precipitation and changes in runoff, as simulated by the WatBal model. Table 4.3 provides a summary of changes in annual runoff as a function of (i) changes 51 in annual precipitation uniformly imposed on each month, and (ii) an increase in average annual temperature also uniformly imposed on each month. The precipitation elasticity is about 2.5 for an increase in rainfall and 2.2 for a decrease in rainfall. The response of streamflow to temperature increases for the Tanganyika and Nyasa basins is 8% decrease in runoff due to a 2 0C increase in temperature, corresponding to a temperature elasticity of runoff -0.84 (average temperature T = 210C). The temperature elasticity of runoff for the Kyoga basin is only half of this value, about -0.4. Theoretical elasticity values – as per Table 4.1 - and the values derived from model simulations agree very well. Change in Change in runoff Precip. ε Change in Change in runoff Temp. ε Precip. Kyoga Tanganyika Nyasa average Temp. Kyoga Tanganyika Nyasa Tan/Nyasa -25% -45% -48% -53% 1.95 0 -2% -4% -4% -0.84 +1 C -10% -20% -22% -24% 2.20 +2 0 C -4% -8% -8% -0.84 -5% -10% -11% -13% 2.27 0 -8% -15% -16% -0.81 +4 C 5% 11% 12% 13% 2.40 0 -12% -21% -22% -0.75 +6 C 10% 23% 24% 28% 2.50 25% 61% 65% 77% 2.71 Table 4.3: Sensitivity of runoff for variations in precipitation and temperature (NEL region) Research suggests that hydrological models cannot produce reliable estimates of the impacts of temperature increases on runoff (Sankarasubramanian, 2001). This is also shown in Vano et al (2012), who studied changes in runoff with respect to precipitation and temperature elasticities for the Colorado River in the southwestern USA through comparisons of multi-decadal simulations from 5 commonly used Land Surface Models (LSM); see Figure 4.3. Annual runoff in this catchment is dominated by snowmelt, and as such this basin is not representative for the river basins in Cameroon. Fig. 4.3: Precipitation elasticities and temperature sensitivities for the Colorado River (Vano et al, 2012) 52 However, their main finding that precipitation and temperature elasticities/sensitivities derived from hydrological models vary largely across the applied LSMs is relevant. Depending on the LSM used, they found for the present hydrology precipitation elasticities varying mostly in the range of +2.5 to +4, and temperature sensitivities varying mostly in the range of -3% to -7% (decline in runoff) per 10C temperature increase, well outside the range expected based on other research and the Turc-Pike equation discussed in Section 4.2. 4.4 Regression analysis of basin runoff, rainfall and temperature As indicated above, the log-linear Q-P-T model is not very suitable for estimating the temperature elasticity of runoff, due to the relatively small impact of temperature on runoff and the usually rather large random noise in the precipitation and runoff signals. Hence, we have limited our analysis to deriving theoretical climate elasticity values and linear regression for the estimation of the precipitation elasticity of runoff, i.e.: εP = Ï?QP . CvQ/CvP. As mentioned before, Sankarasubramanian’s median estimator εP = median {[dQ/μQ]/[dP/μP]} yielded erratic results. Fig. 4.4 shows graphical results of the runoff-precipitation (Q-P) regression analyses for selected sub- basins. These analyses were performed on the standardized data series, respectively q = [Q- μQ]/μQ = dQ/Q and p = [P- μP]/μP = dP/P, and the precipitation elasticity thus equals (in principle) the slope of the trend line in the graphs. However, the correlation between CRU precipitation and observed runoff data is generally poor (R2 < 0.5; Ï?QP < 0.7), which causes possibly unrealistic low values of the precipitation elasticity defined as: εP = Ï?QP . CvQ/CvP. Intuitively, looking at the graphs in Figure 4.4, one would expect the slope of a trend line with high correlation to be in the order of 2 for the Sanaga, Congo and Southern Coastal Basins (Nyong and Ntem rivers), 2.5 for the Benue basin and <2 for the Northern Coastal Basins, commensurate with the ratios between the coefficients of variation, as an alternative definition for the precipitation elasticity: εP = CvQ/CvP (for Ï?QP = 1). Various statistics of the gridded CRU precipitation and temperature data and available runoff data for selected sub-basins have been summarized in Table 4.4, as well as theoretical climate elasticities of runoff on the basis of the aridity index, results of the linear regression analyses of the available P-Q data sets, and the alternative estimator εP = CvQ/CvP. It is observed that the runoff coefficient Q/P is about 0.18 for the Benue Basin, 0.25 to 0.3 for the Sanaga Basin and the Southern Coastal and Congo Basins, and 0.5 for the Northern Coastal Basins. Fig. 4.5 compares for the Turc-Pike model (Section 4.2) theoretical and actual values of the parameters Q/P, εP and σQ/σP (inter-annual variability of Q and P) against the aridity index Ï• , and εP, φ and Cv(Q)/Cv(P) against the runoff coefficient Q/P. Overall, the theoretical relationships for Q/P (φ) and σQ/σP (φ) describe the actual observed values very well. However, the Q - P regression analysis appears to significantly underestimate the precipitation elasticity of runoff across Cameroon. This is caused by the overall poor correlation found between gridded CRU precipitation data and observed runoff data as indicated above (R2 < 0.5). 53 Edea Nachtigal Goura 0.50 0.50 0.50 0.25 0.25 0.25 q = dQ/Q dq = Q/Q q = dQ/Q 0.00 0.00 0.00 -0.25 -0.25 -0.25 y = 1.3424x y = 1.2714x y = 1.5239x R² = 0.4193 R² = 0.3382 R² = 0.3873 -0.50 -0.50 -0.50 -0.25 -0.13 0.00 0.12 0.25 -0.25 -0.13 0.00 0.12 0.25 -0.25 -0.13 0.00 0.12 0.25 p = dP/P p = dP/P p = dP/P Mbakaou Betare-Oya Garoua 0.50 0.50 0.75 0.50 0.25 0.25 0.25 q = dQ/Q q = dQ/Q q = dQ/Q 0.00 0.00 0.00 -0.25 -0.25 -0.25 y = 1.2097x y = 0.5432x -0.50 y = 2.0646x R² = 0.3946 R² = 0.0852 R² = 0.3333 -0.50 -0.50 -0.75 -0.25 -0.13 0.00 0.12 0.25 -0.25 -0.13 0.00 0.12 0.25 -0.25 -0.13 0.00 0.12 0.25 p = dP/P p = dP/P p = dP/P Riao Eseka Ngoazik 0.75 0.50 0.50 0.50 0.25 0.25 0.25 q = dQ/Q q = dQ/Q q = dQ/Q 0.00 0.00 0.00 -0.25 -0.25 -0.25 -0.50 y = 1.7857x y = 1.7228x y = 1.475x R² = 0.5161 R² = 0.2933 R² = 0.559 -0.75 -0.50 -0.50 -0.25 -0.13 0.00 0.12 0.25 -0.25 -0.13 0.00 0.12 0.25 -0.25 -0.13 0.00 0.12 0.25 p = dP/P p = dP/P p = dP/P Ngbala Mundame Yabassi 0.50 0.50 0.50 0.25 0.25 0.25 q = dQ/Q q = dQ/Q q = dQ/Q 0.00 0.00 0.00 -0.25 -0.25 -0.25 y = 1.7037x y = 1.0761x y = 1.4784x R² = 0.6096 R² = 0.7015 R² = 0.5603 -0.50 -0.50 -0.50 -0.25 -0.13 0.00 0.12 0.25 -0.25 -0.13 0.00 0.12 0.25 -0.25 -0.13 0.00 0.12 0.25 p = dP/P p = dP/P p = dP/P Fig 4.5: Correlation of relative changes in rainfall and runoff for selected sub-catchments 54 Betare- Lake Hydrometric station Edea Nachtigal Goura Mbakaou Garoua Riao Eseka Ngoazik Ngbala Melong Mundame Yabassi Gouri Oya Chad Benue-Niger Basin Nyong Ntem Congo Niger - Basin - Sub-basin Sanaga Basin North-Coastal Basins North in Cameroon Basin Basin Basin South Surf. Area (km2) 131,500 76,000 42,300 20,200 11,100 64,000 30,650 21,600 18,100 38,600 2,280 2,420 8,026 2,240 27,470 Precipitation (mm) Average 1,715 1,560 1,838 1,556 1,498 1,077 1,171 1,746 1,736 1,623 2,186 2,660 2,224 2,052 682 St. dev 133 126 158 139 155 107 111 164 181 154 209 304 187 203 116 Cv 0.08 0.08 0.09 0.09 0.10 0.10 0.09 0.09 0.10 0.09 0.10 0.11 0.08 0.10 0.17 E0 1,631 1,658 1,609 1,710 1,675 1,934 1,901 1,548 1,572 1,632 1,472 1,499 1,526 1,541 2,069 Theoretical values climate elasticities (Turc - Pike) Aridity index φ 0.95 1.06 0.88 1.10 1.12 1.80 1.62 0.89 0.91 1.01 0.67 0.56 0.69 0.75 3.03 Q/P 0.31 0.27 0.34 0.26 0.25 0.13 0.15 0.34 0.33 0.29 0.44 0.51 0.43 0.40 0.05 E/P 0.69 0.73 0.66 0.74 0.75 0.87 0.85 0.66 0.67 0.71 0.56 0.49 0.57 0.60 0.95 σQ/σP 0.67 0.61 0.71 0.60 0.59 0.33 0.38 0.71 0.70 0.64 0.83 0.88 0.82 0.78 0.14 β 1.16 1.26 1.09 1.29 1.30 1.64 1.58 1.10 1.12 1.21 0.87 0.73 0.89 0.96 1.85 εP 2.16 2.26 2.09 2.29 2.30 2.64 2.58 2.10 2.12 2.21 1.87 1.73 1.89 1.96 2.85 εT -0.67 -0.72 -0.62 -0.73 -0.74 -0.99 -0.95 -0.64 -0.65 -0.70 -0.47 -0.43 -0.51 -0.53 -1.13 Actual values Turc - Pike parameters Q/P 0.26 0.26 0.26 0.37 0.31 0.16 0.22 0.23 0.28 0.22 0.45 0.82 0.55 0.70 E/P 0.74 0.74 0.74 0.63 0.69 0.84 0.78 0.77 0.72 0.78 0.55 0.18 0.45 0.30 σQ/σP 0.56 0.59 0.61 0.72 0.53 0.47 0.65 0.40 0.55 0.50 0.77 1.09 1.04 1.02 Temperature (0C) Average 23.9 23.7 23.1 23.4 23.4 27.1 26.7 24.3 24.1 24.1 21.4 24.8 23.7 22.3 28.0 St. dev 0.30 0.31 0.32 0.32 0.36 0.38 0.37 0.29 0.30 0.31 0.34 0.35 0.34 0.34 0.47 Cv 0.013 0.013 0.014 0.014 0.015 0.014 0.014 0.012 0.013 0.013 0.016 0.014 0.014 0.015 0.017 Table 4.4: Climate and hydrological data for selected sub-basins in Cameroon (Note: time periods vary depending on available runoff data ) 55 Betare- Lake Hydrometric station Edea Nachtigal Goura Mbakaou Garoua Riao Eseka Ngoazik Ngbala Melong Mundame Yabassi Gouri Oya Chad Nyong Ntem Congo Niger - Basin - Sub-basin Sanaga Basin Niger Basin North-Coastal Basins Basin Basin Basin South North Surf. Area (km2) 131,500 76,000 42,300 20,200 11,100 64,000 30,650 21,600 18,100 38,600 2,280 2,420 8,026 2,240 27,470 3 Runoff (m /s) Average 1,877 991 632 369 165 357 249 276 278 433 71 168 310 102 St. dev 308 179 130 64 29 102 71 45 57 94 12 26 50 15 Cv 0.16 0.18 0.21 0.17 0.17 0.29 0.28 0.16 0.21 0.22 0.16 0.15 0.16 0.14 Runoff (mm/yr) Average 450 411 471 576 469 176 256 403 485 354 976 2,186 1,217 1,439 St. dev 74 74 97 100 82 50 73 65 100 77 161 332 195 207 Cv 0.16 0.18 0.21 0.17 0.17 0.29 0.28 0.16 0.21 0.22 0.16 0.15 0.16 0.14 Correlation coefficients Ï?QP 0.647 0.582 0.622 0.631 0.292 0.577 0.542 0.748 0.718 0.783 0.630 0.838 0.749 0.786 Ï?QT -0.207 0.015 -0.401 -0.091 0.043 0.155 0.039 0.364 0.383 0.351 -0.468 -0.478 -0.258 -0.226 Ï?PT -0.221 -0.136 -0.248 -0.116 0.142 0.023 0.014 -0.194 0.089 0.030 -0.220 -0.280 -0.358 -0.188 Regression models Q - P -T Linear Q-P model a 0.360 0.343 0.381 0.455 0.154 0.272 0.354 0.298 0.397 0.391 0.483 0.917 0.781 0.801 P0 464 359 602 291 -1,552 430 450 394 514 717 166 276 665 256 εP 1.37 1.30 1.49 1.23 0.49 1.67 1.62 1.29 1.42 1.79 1.08 1.12 1.43 1.14 Non-linear Q-P-T model approximation: b = εP 1.34 1.33 1.33 1.23 0.49 1.66 1.62 1.47 1.36 1.77 0.95 1.02 1.43 1.12 approximation: c = εT -0.87 1.33 -3.90 -0.24 0.01 2.91 0.66 7.15 5.28 5.54 -3.57 -2.88 0.12 -0.77 Theoretical values Turc - Pike εP 2.16 2.26 2.09 2.29 2.30 2.64 2.58 2.10 2.12 2.21 1.87 1.73 1.89 1.96 2.85 εT -0.67 -0.72 -0.62 -0.73 -0.74 -0.99 -0.95 -0.64 -0.65 -0.70 -0.47 -0.43 -0.51 -0.53 -1.13 ST -2.8% -3.0% -2.7% -3.1% -3.2% -3.6% -3.5% -2.6% -2.7% -2.9% -2.2% -1.7% -2.1% -2.4% -4.0% εP = Cv(Q)/Cv(P) 2.12 2.23 2.39 1.95 1.68 2.89 3.00 1.73 1.98 2.29 1.72 1.33 1.91 1.45 1944- 1951- 1951- 1959- 1951- 1945- 1950- 1951- 1954- 1956- 1951- 1953- 1951- 1964- Period with flow data 2003 2003 2003 2008 2003 1980 1980 1976 1979 1976 1976 1976 1976 1980 Table 4.4 (continued): Climate and hydrological data for selected sub-basins in Cameroon (Note: time periods vary depending on available runoff data ) 56 Instead, the modified estimator εP = Cv(Q)/Cv(P), which assumes perfect correlation between annual precipitation and runoff, shows an excellent agreement with the theoretical values of εP (right panel of Figure 4.5). Therefore, we have used this modified estimator for εP, as well as the theoretical values of the runoff climate elasticities according to the Turc-Pike model in guiding us in selecting appropriate (read ‘conservative’) climate elasticities for the river basins in Cameroon, as summarized in Table 4.5. Overall, the precipitation elasticity of runoff is seen to decrease from 2.8 in the arid north to about 2 in the tropical centre and south of the country. These values compare well to the climate elasticities found in studies for other river basins in Africa. 3.0 3.0 2.5 2.5 2.0 2.0 Values 1.5 1.5 Values 1.0 1.0 0.5 0.5 0.0 0.0 0.0 1.0 2.0 3.0 4.0 5.0 0.0 0.2 0.4 0.6 0.8 1.0 Aridity index Runoff coefficient (Q/P) Q/P Q/P-act εP εP εP-act Cv(Q)/Cv(P) εP-act σQ/σP σQ/σP-act φ φ - act Fig. 4.5: Theoretical and actual hydrological parameter based on aridity index and runoff coefficient Lake Northern Southern Basin Benue Sanaga Congo Chad Coast Coast εP 2.8 2.6 2.2 1.9 2.1 2.2 εT -1.1 -1.0 -0.7 -0.5 -0.65 -0.7 ST -4.0% -3.5% -3.0% -2.2% -2.6% -2.9% Table 4.5: Recommended climate elasticities of runoff for Cameroon river basins (Note: The values for the 0 northern part of the Lake Chad Basin are determined for an aridity index φ = 4 and temperature T = 30 C) It is seen that the observed runoff coefficient Q/P provides a powerful estimator for the climate elasticities of runoff, i.e.: ï‚· εP = 3 – 3 Q/P + (Q/P)2 ï‚· εT = [-2 + 3 Q/P - (Q/P)2].[T/(T + 17.8)] ï‚· ST = εT /T The runoff coefficient Q/P can also be estimated from the aridity index φ = E0/P (see Annex 6): ï‚· Q/P = 1 - [1+ φ-2]-0.5 57 5. Vulnerability analysis: response of hydro-energy generation to changes in runoff For this study we focus our analysis on performance indicators linked to hydro-energy production, e.g. guaranteed production (in MW) during the dry season and total annual energy production (in GWh/yr). Water use for irrigated agriculture and domestic/industrial water supply is relatively small and is not expected to increase significantly in the foreseeable future (GWP, 2009; Volume 117). The economic importance of river navigation is equally negligible, while the maintenance of minimum flows downstream of storage reservoirs is a pre-requisite under all foreseeable hydro-meteorological conditions, with or without climate change impacts on runoff. Hence, minimum flows should not be affected by climate change18. The next step is thus to seek an understanding of how the present and future hydro-energy system of Cameroon will respond to changes in runoff caused by climate changes. This process identifies runoff conditions that cause unacceptable performance levels, and determines the runoff elasticity of hydro-energy generation. Table 5.1 summarizes the main characteristics of present and planned future hydropower plants and storage reservoirs in Cameroon19. Max. Target Max. DS evap. Max. Max. Installed Basin Life Annual Period Reservoir - HP Year flow flow Reserv. + irrig. net DS River Head Capacity area storage flow flow data station built turbine turbine area losses release (m) (MW) (km 2 ) (MCM) (MCM) (years) (m 3 /s) (m 3 /s) (km 2 ) (MCM) (MCM) Edea Sanaga 1954 24 1,250 264 1,100 131,500 54,200 1972-2003 Song Loulou Sanaga 1981-88 42 1,100 406 1,100 129,800 54,200 1972-2003 Mbakaou Djerem 1970 20,200 348 2,600 220 2,380 11,000 1972-2003 Bamendjing Noun 1975 2,190 251 1,675 220 1,455 1,646 1972-2003 Mape (Magba) Mape 1988 4,020 530 3,100 275 2,825 2,990 1972-2003 Lagdo Benue 1982 28 436 72 30,650 697 4,550 550 4,000 7,840 1959-1980 Existing dams and HP stations 742 1,826 11,925 1,265 10,660 Lom Pangar Lom 2016 36.7 93 30 25 1,970 570 6,000 220 5,780 7,780 1972-2003 Nachtigal Sanaga 50 820 360 700 76,000 28,200 1972-2003 Song Mbengue Sanaga 81 1,250 890 1,100 129,000 54,200 1972-2003 Song Ndong Sanaga 25 1,250 275 1,100 129,000 54,200 1972-2003 Kikot Sanaga 45 1,575 623 1,100 124,400 54,200 1972-2003 Njock Nyong 73 185 119 21,600 8,700 1951-1976 Mouila Nyong 140 160 197 20,800 8,380 1951-1976 Memve-Ele Ntem ongoing 51 450 202 26,350 12,500 1953-1979 Planned dams and HP stations 2,671 570 6,000 220 5,780 Total 3,413 2,396 17,925 1,485 16,440 Table 5.1: Main characteristics of present and planned hydropower plants and storage reservoirs 17 The only significant abstractions for irrigation occur from the Lagdo reservoir in the Benue basin (Section 5.5.1). 18 Farmers living in rural areas could potentially be most affected by climate change impacts on rainfed agriculture. However, this impact domain is outside the scope of this study. 19 Based on the Joint Development Agreement signed between IFC, EDF, RTC and the Government of Cameroon, the design data for Nachtigal have been adjusted to the commissioning of 360 MW for a head of 50 m by 2020. 58 Figure 5.1 presents an overview of potential hydropower stations on the Sanaga River between Edea and Nachtigal. It is noted that the indicated generation potential is outdated, but the main purpose of this graph is to show locations of future sites and the enormous hydro-energy potential of the Sanaga basin. Most reports consulted quote indeed slightly different installed capacities (IC), maximum turbine discharges, minimum turbine discharge targets for the dry season and maximum heads for various existing and potential future power stations. Hence, we have used ‘consensus’ estimates, which – even if not exactly correct - suffice for the purpose of this Climate Risk Assessment. Generated power E (in MW) is calculated based on: E = Ï? η g Q H/1000 (maximum equal to IC) with: IC = installed capacity (MW), Ï? = water density (0.998 kg/l), η = plant efficiency (90%), g = gravity constant (9.80 m/sec2), Q = turbine discharge (m3/s) and H = Head (m) Plant availability has been set at 96% for all HP stations, with the exception of the plant availability of Edea which was set at 80% based on World Bank (2012a). Most of the hydropower stations will be of the Run-of-the-River (R-o-R) type, which will as such be directly exposed to runoff changes due to climate change. However, the runoff elasticity of hydro-energy generation for such plants depends foremost on the ratio between maximum possible turbine discharges and the magnitude of rainy season river discharges; the lower the ratio the less its sensitivity to runoff changes. Fig. 5.1: Potential future hydropower stations on the Sanaga River (Source: MINEE and GWP, 2009) The R-o-R power plants in the Sanaga Basin also benefit from storage of rainy season runoff in the existing Bamendjing, Mapé and Mbakaou reservoirs, and in the future Lom Pangar reservoir (see Figure 5.2). Storage in these reservoirs will be sensitive to significant negative runoff changes when the ratio between storage capacity and annual runoff is close to one (such as is the case for the existing 59 Bamendjing and Mapé reservoirs and to a lesser extent for the new Lom Pangar reservoir). Storage in the Mbakaou reservoir is insensitive to runoff changes since annual runoff exceeds its storage capacity more than four times. Lagdo dam in the Benue Basin (near the Riao gauging station) is at an intermediate position. Mapé 3,2Md m3 Bamendjin 3.1 BCM Figure 5.2: Existing and planned 1,8Md m3 1.7 BCM Mbakaou 2,5Md m3 reservoirs in the Sanaga Basin 2.6 BCM As shown in Chapter 4 potential evapotranspiration would increase Bassin Versant Intermédiaire by 2050 with about 5% due to a 20C Bassin Versant BV02 increase of temperature. However, Intermédiaire BV01 since Lom evaporation losses from Pangar 6Md m3 reservoirs and water volumes used Lom Pangar for irrigated agriculture are generally 6 BCM 30 MW small compared to annual runoff volumes, such small increase (5%) of Edéa Song Loulou Nachtigal (projet) 264MW 384MW 280MW an already small amount has not 406 MW 360 MW been taken into account. MWMW MW 60 5.1 Seasonal water management and hydro-energy generation model for the Sanaga Basin Using the available monthly flow data, we developed a seasonal water management and hydro-energy generation model in excel for the Sanaga Basin, to determine the runoff elasticity of hydro-energy for the Sanaga basin, as well as the runoff elasticity of the economic performance of the new Lom Pangar reservoir and hydropower station. This section describes the assumptions and monthly flow data adopted for building (in Excel 2007) a seasonal water management and hydro-energy generation model for the Sanaga basin. Monthly flow data: As shown in Chapter 4, it is apparent that an abrupt change in precipitation and runoff occurred around 1970. Therefore, we have based our analysis on the available monthly flow data for the period 1971 - 2003, as listed in Annex 4. The data for Edea are naturalized flow data provided by EDC, i.e. actual flow data corrected for the monthly runoff changes caused by seasonal storage in the Mbakaou, Bamendjing and Mapé reservoirs since the start of their operations in respectively 1970, 1975 and 1988 (see Fig. 5.2 for reservoir locations). The flow data in Annex 4 for these reservoirs represent the inflows of the reservoirs. Inflow and outflow data are shown in Figure 5.3 for the period July 1994- June 1995 (the 1994 rainy season presented relatively high inflows). In principle all inflow of Bamendjing and Mapé reservoirs is captured, except in above average wet years; only a fraction of Mbakaou’s inflow is stored during the rainy season due to Mbakaou’s limited storage capacity, even in dry years. Inflow and outflow hydrographs reservoirs (July 1994 - June 1995) 600 Mape-inflow 1,800 Mape-outflow Bamendjing - inflow Flow Bamendjing & Mape (m3/s) 500 1,500 Bamendjing-outflow Mbakaou-inflow Flow Mbakaou (m3/s) 400 Mbakaou-outflow 1,200 300 900 200 600 100 300 0 0 1 31 61 91 121 151 181 211 241 271 301 331 361 Fig. 5.3: Inflow and outflow hydrographs for Mbakaou, Bamendjing and Mapé reservoirs The available monthly flow data for Nachtigal are not yet naturalized and are thus affected by the flow regulation provided since 1970 by the Mbakaou reservoir. For the period 1971-1979 the available EDC 61 data represent the inflow of Mbakaou and the GRDC data provided the reservoir outflow. Similarly, inflow and outflow data were available from EDC for the period 1990-2003. The differences between these inflow and outflow data were used to naturalize the flow data for Nachtigal for the same periods, while average monthly corrections were applied for the period 1980-1989. Definition of rainy and dry seasons: During the months of June and December there are regular shortages in the naturalized river flows at Edea, Song Loulou and Nachtigal compared to the target flows set for these power stations, respectively 1,100 m3/s for Edea and Song Loulou and 700 m3/s for Nachtigal (see Section 5.4). Therefore, for the purpose of this CRA analysis we define the dry season as the period December to June (212 days) and the rainy season – during which period water is stored in all reservoirs - as the period July to November (153 days). An exception is made for the Bamendjing and Mapé reservoirs, for which we assume water is also stored during the month of June. Losses from reservoirs and maximum net contributions to dry season flows: On average the available reservoir inflow and outflow data for the period 1990-2003 show a net evaporation loss (corrected for rainfall directly on the reservoir) of 12 m3/s from Mbakaou, 12 m3/s for Bamendjing and 15 m3/s for Mapé. Based on ISL et al (2005b) losses for Lom Pangar for the dry season are also set at 12 m 3/s. Total dry season losses from these four reservoirs in the Sanaga basin total to 935 MCM. Thus, the net dry season contribution of each reservoir to maintaining minimum flows at Nachtigal, Edea and Song Loulou is less than its maximum storage capacity (Table 5.1). Reservoir management efficiency: The model adopted for this study assumes perfect advance knowledge and foresight of the upcoming dry season flows, to determine which minimum flow can be maintained at various locations. While the recession flows during December to March can to some extent be estimated from the flows on December 1st, this is not the case for the basin runoff during April – June, when occasional rain storms occur. Therefore, we have introduced – analogous to previous studies (inter alia ISL, 2005) – a reservoir management efficiency of 90%. This implies that 10% of reservoir storage is assumed to be lost due to inadequate runoff prediction and reservoir management, without contributing to maintaining the minimum turbine discharges at the various power stations. Based on model runs, commensurate losses in average annual hydro-energy generation are estimated at only 1.5%, but the loss of guaranteed power during the dry season is 9 to 10%. Lom Pangar reservoir management and Nachtigal operation: In our model the Lom Pangar dam is primarily operated to satisfy the minimum turbine discharge requirement for Nachtigal (which has been shown to also satisfy the minimum requirement for Song Loulou and Edea; see Figure 5.2), as follows: ï‚· Determine the surplus in rainy season flow at Nachtigal above the target flow threshold (700 m3/s). ï‚· Deduct the maximum storage in Mbakaou reservoir (2,600 MCM) as well as its rainy season evaporation losses (160 MCM); the balance is the maximum surplus which can be stored in Lom Pangar reservoir (including rainy season evaporation losses) without reducing the rainy season flow at Nachtigal below the above minimum threshold turbine discharge. 62 ï‚· Determine the total rainy season flow volume at Lom Pangar, corrected for a minimum flow of 25 m3/s to be maintained downstream of the dam at all times; the minimum of this surplus at Lom Pangar and the above surplus at Nachtigal, corrected for evaporation losses from the reservoir, can be stored, keeping also into account the initial storage on July 1st (in case of over-annual storage) and the maximum storage capacity of the reservoir (6,000 MCM). ï‚· Calculate the dry season flow shortage at Nachtigal below the minimum turbine flow target (700 m3/s), and compare this volume with the storage on December 1st in the Mbakaou and Lom Pangar reservoirs (corrected for dry season evaporation losses and the adopted reservoir management efficiency of 90%). In case of a surplus, the balance will be stored in Lom Pangar reservoir as over- annual storage and the turbine flow target (700 m3/s) will be met; in case of a shortage, the average dry season turbine flow at Nachtigal is calculated. ï‚· Calculate the total reduction of rainy season flow at Nachtigal, including storage in Mbakaou and Lom Pangar reservoirs and evaporation losses from both reservoirs, and reduce the August to November flows at Nachtigal accordingly, respectively with 10%, 40%, 40% and 10% of the total storage and losses. Subsequently, determine the average rainy season turbine flow at Nachtigal, keeping the maximum turbine flow (818 m3/s; see Table 5.1) into account. ï‚· Finally, calculate the seasonal and annual hydro-energy generation at Nachtigal for the period 1972 – 2003, as well as percentiles, long-term average and guaranteed capacities (MW) and turbine flows, based on a plant availability of 96%. Guaranteed capacities and turbine flows are based on the 10% percentiles of seasonal energy generation and flows. The calculation of hydro-energy generation at Lom Pangar is based on the data provided in Figure 5.4. For this analysis it is assumed that the filling and depletion of the reservoir takes place in a linear fashion between the reservoir volumes determined for July 1st and December 1st, which enables the estimation of average head during the seasons. Turbine flows are estimated based on the water balance of the reservoir with a minimum flow of 25 m3/s. Max. turbine flow (m 3 /s) 93 Lom Pangar reservoir - Volume - Stage-volume-area Lom Pangar reservoir Head curve 6,000 Tail race (m) 636 Efficiency 90% Level Life storage Head Power H-13 5,000 Volume (MCM) (m NGC) MCM (m) MW (m) y = 9.769x2 + 19.656x 4,000 649 0 13.0 10.6 0.0 650 60 14.0 11.5 1.0 3,000 655 540 19.0 15.6 6.0 2,000 660 1,410 24.0 19.6 11.0 665 2,750 29.0 23.7 16.0 1,000 670 4,690 34.0 27.8 21.0 0 672.7 6,000 36.7 30.0 23.7 0 5 10 15 20 25 Head - 13 (m) Fig. 5.4: Lom Pangar reservoir characteristics 63 Results for Nachtigal hydro-energy generation (IC = 360 MW), with and without Lom Pangar reservoir are summarized in Table 5.2. The guaranteed flow (10% percentile) is nearly 600m 3/s, a value which is often quoted as a requirement for Nachtigal. However, by adopting a target flow of 700 m3/s, the overall generation was improved. Due to the storage of rainy season runoff in Lom Pangar reservoir, hydro-energy generation during this season will be slightly reduced with 4%, but dry season generation will be significantly improved with 56%, increasing the guaranteed capacity (10% percentile) with 104 MW (+65%) and the overall annual generation with 464 GWh/yr (21%). Hydro-energy generation results for Lom Pangar are also included in Table 5.2. Average generation for the two HP stations ranges from 220 GWh/month during the dry season to 254 GWh/month during the rainy season. The results agree well with results of other more detailed modeling studies (ISL, 2005a and 2007; World Bank, 2012a). Nachtigal Lom Pangar Results with Lom Pangar Dec-June July-Nov Total Dec-June July-Nov Total Guaranteed capacity (MW) 263 305 21 8 3 Guaranteed flow (m /s) 597 694 85 32 Average load factor (%) 78% 91% 83% 74% 56% 67% Average production (GWh/yr) 1,423 1,208 2,631 114 62 175 Results without Lom Pangar Guaranteed capacity (MW) 159 345 3 Guaranteed flow (m /s) 363 785 Average load factor (%) 50% 95% 69% Average production (GWh/yr) 914 1,253 2,167 Table 5.2: Hydro-energy generation results for Nachtigal, with and without Lom Pangar Bamendjing and Mapé reservoir management and operation of Edea and Song Loulou HP stations: The sub-model for Edea and Song Loulou HP stations takes the results of the Nachtigal-Mbakaou sub- model as input, and the Bamendjing and Mapé reservoirs (Fig. 5.2) are subsequently operated to satisfy the minimum turbine flow requirements for Edea and Song Loulou as much as possible, as follows: ï‚· Determine the total rainy season flow volume at Bamendjing and Mapé reservoirs; these surpluses, adjusted for evaporation losses from the reservoir are stored for release during the next dry season, also keeping into account the initial storage on July 1st (in case of over-annual storage) and the maximum storage capacity of the reservoirs (Table 5.1). ï‚· Calculate the dry season flow shortage (December – June) at Edea and Song Loulou below the turbine target flow (1,100 m3/s), and deduct the flow volumes already released from Mbakaou and Lom Pangar. Compare the remaining shortage with the storage on December 1st in the Bamendjing and Mapé reservoirs (adjusted for dry season evaporation losses and the adopted reservoir management efficiency of 90%). In case of a surplus, the balance will be stored in both reservoirs as over-annual storage and the turbine target flow (1,100 m3/s) will be met; in case of a shortage, the average dry season turbine flow at Edea/Sang Loulou is calculated. 64 ï‚· Determine the reductions of monthly flows at Edea and Sang Loulou for the rainy season based on the flow reductions estimated above for Nachtigal, and the reservoir inflows for Bamendjing and Mapé reservoirs. Subsequently, determine the average rainy season turbine flows at Edea and Song Loulou, keeping the maximum turbine flows (respectively 1,250 and 1,100 m3/s; see Table 5.1) into account. ï‚· Finally, calculate the seasonal and annual hydro-energy generation at Edea and Song Loulou for the period 1972 – 2003, as well as percentiles, long-term average and guaranteed capacities (MW) and turbine flows, based on a plant availability of 80% for Edea and 96% for Song Loulou. Guaranteed capacities and turbine flows are based on the 10% percentiles of seasonal energy generation and flows. Results for Edea and Song Loulou hydro-energy generation, with and without the Lom Pangar reservoir are summarized in Table 5.3. Total hydro-energy generation results for Edea, Song Loulou, Nachtigal and Lom Pangar are also included. The guaranteed flow (10% percentile) at Song Loulou and Edea is 931m3/s, an increase with 236 m3/s compared to the ‘without Lom Pangar’ case. By using the limited storage capacity upstream of Song Loulou (4 MCM) for daily flow modulation, peak hour turbine discharges can be increased to 1,100 m3/s during at least 6 hours, while reducing the guaranteed flow during the remainder of time to about 870 m3/s. Due to Lom Pangar reservoir, the guaranteed capacity (10% percentile) of both stations in the dry season will increase with 137 MW (34%) and the overall annual generation will increase with 553 GWh/yr (12.5%). Hydro-energy generation results for Lom Pangar and Nachtigal are also included in Table 5.3. Average generation for the four HP stations ranges from 620 GWh/month during the dry season to 693 GWh/month during the rainy season. Overall the difference between average generation per season and the 10% percentile is less than 5%. Lom Pangar reservoir enables a uniform distribution of hydro-energy generation over the year. The results agree well with results of other detailed modeling studies (ISL, 2005a and 2007; World Bank, 2012a). Song Loulou Edea Edea+Song LL+Nachtigal+Lom P Results with Lom Pangar Dec-June July-Nov Total Dec-June July-Nov Total Dec-June July-Nov Total Guaranteed capacity (MW) 344 402 197 252 824 967 Guaranteed flow (m3 /s) 931 1,089 931 1,194 Average load factor (%) 92% 96% 93% 67% 79% 72% 80% 89% 84% Average production (GWh/yr) 1,900 1,429 3,328 905 768 1,672 4,341 3,466 7,807 Results without Lom Pangar Guaranteed capacity (MW) 257 402 147 255 563 1,002 3 Guaranteed flow (m /s) 695 1,089 695 1,209 Average load factor (%) 74% 96% 83% 54% 79% 65% 60% 91% 73% Average production (GWh/yr) 1,523 1,429 2,952 725 770 1,495 3,162 3,451 6,613 Table 5.3: Hydro-energy generation results for Song Loulou and Edea, with and without Lom Pangar Performance of other planned hydropower stations: We have included three potentially viable Run-of- the-River power stations, which are presently under consideration and are all located on the Sanaga 65 River between Edea and the junction with Mbam River, i.e. Song Ndong, Song Mbengue and Kikot (ref. Table 5.1 and Figure 5.1). Thus, river flow conditions are assumed to be identical to the flow conditions for Edea and Song Loulou. Hydro-energy generation results - with and without the Lom Pangar reservoir - are summarized in Table 5.4. Total hydro-energy generation results for all 7 HP stations are also included. Due to Lom Pangar reservoir, the guaranteed capacity (10% percentile) of all seven HP stations in the dry season will increase with 575 MW (39%) and the overall annual generation will increase with 2,526 GWh/yr (14%). Average generation for the seven HP stations ranges from 1,596 GWh/month during the dry season to 1,932 GWh/month during the rainy season. Overall, the difference between average dry season generation and the 10% percentile is only 6%, against 12% for the ‘without case’. Thus, Lom Pangar reservoir enables a more uniform distribution of hydro-energy generation over the year; it has little to no impact on hydro-energy generation during the rainy season. Song Ndong Song Mbengue Kikot Total energy 7 HP stations Results with Lom Pangar Dec-June July-Nov Total Dec-June July-Nov Total Dec-June July-Nov Total Dec-June July-Nov Total Guaranteed capacity (MW) 205 263 663 851 369 560 2,061 2,640 Guaranteed flow (m3/s) 931 1,194 931 1,194 931 1,414 Average load factor (%) 81% 95% 87% 81% 95% 87% 64% 93% 76% 77% 92% 83% Average production (GWh/yr) 1,131 960 2,090 3,663 3,109 6,773 2,035 2,123 4,158 11,170 9,658 20,829 Results without Lom Pangar Guaranteed capacity (MW) 153 266 495 861 275 571 1,486 2,700 3 Guaranteed flow (m /s) 695 1,209 695 1,209 695 1,442 Average load factor (%) 65% 95% 78% 65% 95% 78% 51% 93% 69% 60% 93% 74% Average production (GWh/yr) 907 962 1,869 2,937 3,118 6,055 1,632 2,135 3,767 8,637 9,666 18,303 Table 5.4: Hydro-energy generation results for 7 HP stations, with and without Lom Pangar The results of the seasonal model in Excel agree well with results of other more detailed modeling studies (ISL, 2005a and 2007; World Bank, 2012a). Thus, the seasonal model can be used for the assessment of the impacts of climate change induced runoff changes on hydro-energy generation in the Sanaga basin. Since changes in runoff and performance indicators are expressed in percentages of present values, remaining systematic model errors will cancel out for the most part. 5.2 Runoff elasticity of hydro-energy generation in the Sanaga basin Climate change impacts on runoff are introduced by varying the available monthly (naturalized) flow data parametrically, i.e. by adopting uniform changes ranging from +20% to -30% in steps of 10%. This range of change was adopted based on the analysis of available climate change projections for Cameroon, as discussed in Chapter 6. For best results the target turbine flow for Nachtigal was modified with 5 m3/s for each percentage change in long-term average runoff, for example 600 m3/s for 20% decrease in runoff and 800 m3/s for 20% increase in runoff (see Section 5.4). Results for the situation with Lom Pangar dam are shown in Table 5.5, for Edea, Song Loulou, Nachtigal, Lom Pangar, the sum of the previous four HP stations and the sum of seven HP stations, including Song Ndong, Song Mbengue and Kikot. Results for the situation without Lom Pangar are shown in Table 5.6. 66 With Lom Pangar (GWh) With Lom Pangar (GWh) Edea Dec-June 20% 10% 0% -10% -20% -30% Song LL Dec-June 20% 10% 0% -10% -20% -30% Min 822 718 615 537 464 393 Min 1,727 1,508 1,290 1,127 975 825 Max 946 946 946 946 916 841 Max 1,986 1,986 1,986 1,986 1,925 1,766 0.10 946 896 800 719 645 570 0.10 1,986 1,882 1,680 1,510 1,354 1,196 0.20 946 946 887 809 725 619 0.20 1,986 1,986 1,862 1,700 1,522 1,300 Dec. - June 0.50 946 946 946 887 797 709 Dec. - June 0.50 1,986 1,986 1,986 1,862 1,674 1,490 average 941 929 905 856 775 688 average 1,976 1,951 1,900 1,799 1,628 1,445 load factor 70.0% 69.1% 67.3% 63.7% 57.7% 51.2% load factor 95.5% 94.3% 91.8% 86.9% 78.7% 69.8% 10% m 3 /s 1,101 1,043 931 837 751 663 10% m 3 /s 1,101 1,043 931 837 751 663 guar. MW 232 220 197 177 158 140 guar. MW 407 385 344 309 277 245 July-Nov. average 774 772 768 761 750 729 July-Nov. average 1,433 1,432 1,429 1,421 1,408 1,384 Year average 1,715 1,701 1,672 1,618 1,525 1,418 Year average 3,410 3,383 3,328 3,219 3,036 2,830 Year load factor 74.1% 73.5% 72.3% 69.9% 65.9% 61.3% Year load factor 95.7% 95.0% 93.4% 90.4% 85.2% 79.4% Elasticity Dec - June 0.20 0.27 0.40 0.53 0.71 0.80 Elasticity Dec - June 0.20 0.27 0.40 0.53 0.71 0.80 load Year 0.13 0.17 0.25 0.33 0.44 0.51 load Year 0.12 0.16 0.25 0.33 0.44 0.50 Nachtigal Dec-June 20% 10% 0% -10% -20% -30% Lom P. Dec-June 20% 10% 0% -10% -20% -30% Min 1,227 1,101 976 850 734 619 Min 95 90 84 75 41 14 Max 1,720 1,612 1,505 1,397 1,290 1,182 Max 129 131 132 133 134 131 0.10 1,436 1,356 1,283 1,139 1,033 917 0.10 109 106 104 100 96 93 0.20 1,513 1,412 1,334 1,228 1,112 975 0.20 112 109 106 102 99 94 Dec. - June 0.50 1,640 1,565 1,505 1,397 1,290 1,174 Dec. - June 0.50 112 112 113 113 111 103 average 1,607 1,517 1,423 1,319 1,207 1,088 average 114 114 114 113 109 105 load factor 87.7% 82.8% 77.7% 72.0% 65.9% 59.4% load factor 74.3% 74.4% 74.3% 73.6% 71.5% 68.4% 10% m 3 /s 668 631 597 530 481 427 10% m 3 /s 90 87 85 83 81 82 guar. MW 294 278 263 233 212 188 guar. MW 22 22 21 21 20 19 July-Nov. average 1,244 1,231 1,208 1,174 1,116 1,027 July-Nov. average 69 66 62 53 47 43 Year average 2,850 2,749 2,631 2,493 2,322 2,116 Year average 182 180 175 165 156 148 Year load factor 90.4% 87.2% 83.4% 79.0% 73.6% 67.1% Year load factor 69.3% 68.3% 66.6% 62.8% 59.4% 56.1% Elasticity Dec - June 0.64 0.66 0.70 0.73 0.76 0.78 Elasticity Dec - June 0.00 0.01 0.05 0.09 0.18 0.26 load Year 0.42 0.45 0.49 0.53 0.59 0.65 load Year 0.20 0.25 0.41 0.57 0.55 0.53 Sum 4 HP Dec-June 20% 10% 0% -10% -20% -30% Sum 7 HP Dec-June 20% 10% 0% -10% -20% -30% Min 3,872 3,418 2,964 2,589 2,214 1,851 Min 10,080 8,840 7,604 6,642 5,721 4,816 Max 4,781 4,675 4,569 4,462 4,256 3,906 Max 11,923 11,817 11,710 11,604 11,176 10,256 0.10 4,477 4,255 3,869 3,465 3,130 2,793 0.10 11,619 11,028 9,902 8,894 7,999 7,094 0.20 4,558 4,433 4,191 3,828 3,429 2,951 0.20 11,699 11,575 10,925 9,940 8,900 7,647 Dec. - June 0.50 4,684 4,610 4,497 4,222 3,866 3,472 Dec. - June 0.50 11,825 11,751 11,626 10,897 9,861 8,819 average 4,638 4,511 4,341 4,086 3,720 3,326 average 11,744 11,525 11,170 10,553 9,575 8,522 load factor 85.9% 83.6% 80.4% 75.7% 68.9% 61.6% load factor 81.0% 79.5% 77.0% 72.7% 66.0% 58.8% 10% m 3 /s 10% m 3 /s guar. MW 955 905 824 740 667 592 guar. MW 2,417 2,290 2,061 1,851 1,664 1,472 July-Nov. average 3,520 3,501 3,466 3,409 3,320 3,184 July-Nov. average 9,791 9,745 9,658 9,524 9,310 8,964 Year average 8,157 8,012 7,807 7,495 7,040 6,510 Year average 21,535 21,271 20,829 20,077 18,884 17,486 Year load factor 87.8% 86.2% 84.0% 80.7% 75.8% 70.1% Year load factor 86.2% 85.2% 83.4% 80.4% 75.6% 70.0% Elasticity Dec - June 0.34 0.39 0.49 0.59 0.72 0.78 Elasticity Dec - June 0.26 0.32 0.44 0.55 0.71 0.79 load Year 0.22 0.26 0.33 0.40 0.49 0.55 load Year 0.17 0.21 0.29 0.36 0.47 0.53 Table 5.5: Impacts of selected runoff changes on hydro-energy generation with Lom Pangar dam The impacts of runoff changes on hydro-energy generation with and without Lom Pangar dam are graphically shown in Figures 5.5 and 5.6, while Figure 5.7 depicts the runoff elasticity of the dry season and annual generated hydro-energy. The following is observed: ï‚· The guaranteed dry season generation capacity (10% percentile) varies for all HP stations nearly linearly in a ratio 1:1 with the adopted change in average annual runoff, i.e. the runoff elasticity for this indicator is about 1.0. The same is also true for the contribution of Lom Pangar reservoir to the guaranteed dry season generation capacity. 67 Without Lom Pangar (GWh) Without Lom Pangar (GWh) Edea Dec-June 20% 10% 0% -10% -20% -30% Song LL Dec-June 20% 10% 0% -10% -20% -30% Min 633 586 538 491 444 393 Min 1,329 1,230 1,130 1,031 931 825 Max 946 937 883 828 766 706 Max 1,986 1,968 1,854 1,739 1,609 1,483 0.10 698 650 597 544 491 437 0.10 1,465 1,366 1,254 1,142 1,030 919 0.20 740 695 642 589 530 472 0.20 1,554 1,459 1,349 1,236 1,114 991 Dec. - June 0.50 810 768 720 659 597 530 Dec. - June 0.50 1,702 1,613 1,511 1,384 1,253 1,113 average 817 775 725 669 608 542 average 1,715 1,628 1,523 1,406 1,276 1,138 load factor 61% 57.7% 54.0% 49.8% 45.2% 40.3% load factor 82.9% 78.7% 73.6% 67.9% 61.7% 55.0% 10% m 3 /s 812 757 695 633 571 509 10% m 3 /s 812 757 695 633 571 509 guar. MW 171 160 147 134 121 107 guar. MW 300 280 257 234 211 188 July-Nov. average 774 772 770 764 753 736 July-Nov. average 1,433 1,432 1,429 1,424 1,413 1,390 Year average 1,591 1,548 1,495 1,434 1,361 1,278 Year average 3,148 3,060 2,952 2,830 2,689 2,528 Year load factor 68.8% 66.9% 64.6% 62.0% 58.8% 55.3% Year load factor 88.4% 85.9% 82.9% 79.4% 75.5% 71.0% Elasticity Dec - June 0.63 0.69 0.73 0.77 0.81 0.84 Elasticity Dec - June 0.63 0.69 0.73 0.77 0.81 0.84 load Year 0.32 0.35 0.38 0.41 0.45 0.48 load Year 0.33 0.37 0.39 0.41 0.44 0.48 Nachtigal Dec-June 20% 10% 0% -10% -20% -30% Lom P. Dec-June 20% 10% 0% -10% -20% -30% Min 783 738 694 650 606 561 Min Max 1,355 1,274 1,187 1,099 1,012 923 Max 0.10 885 832 779 726 674 619 0.10 0.20 908 855 800 745 690 635 0.20 Dec. - June 0.50 1,050 989 927 862 797 730 Dec. - June 0.50 average 1,038 976 914 851 788 723 average load factor 56.7% 53.3% 49.9% 46.4% 43.0% 39.5% load factor 10% m 3 /s 412 387 363 338 314 288 10% m 3 /s guar. MW 181 170 159 149 138 127 guar. MW July-Nov. average 1,264 1,259 1,253 1,243 1,224 1,187 July-Nov. average Year average 2,302 2,235 2,167 2,094 2,011 1,910 Year average Year load factor 73.0% 70.9% 68.7% 66.4% 63.8% 60.6% Year load factor Elasticity Dec - June 0.68 0.68 0.69 0.69 0.69 0.70 Elasticity Dec - June load Year 0.31 0.32 0.33 0.34 0.36 0.40 load Year Sum 4 HP Dec-June 20% 10% 0% -10% -20% -30% Sum 7 HP Dec-June 20% 10% 0% -10% -20% -30% Min 2,815 2,618 2,421 2,224 2,027 1,830 Min 7,638 7,080 6,522 5,964 5,405 4,802 Max 4,288 4,179 3,923 3,667 3,387 3,058 Max 11,429 11,257 10,588 9,919 9,170 8,368 0.10 2,956 2,764 2,554 2,344 2,133 1,923 0.10 8,224 7,675 7,063 6,451 5,838 5,226 0.20 3,209 3,022 2,820 2,587 2,351 2,114 0.20 8,804 8,272 7,670 7,057 6,381 5,701 Dec. - June 0.50 3,579 3,377 3,163 2,939 2,690 2,418 Dec. - June 0.50 9,673 9,145 8,534 7,899 7,195 6,421 average 3,569 3,379 3,162 2,926 2,672 2,403 average 9,735 9,233 8,637 7,980 7,261 6,496 load factor 68.1% 64.4% 60.3% 55.8% 50.9% 45.8% load factor 67.8% 64.3% 60.2% 55.6% 50.6% 45.3% 10% m 3 /s 10% m 3 /s guar. MW 653 610 563 516 469 422 guar. MW 1,731 1,615 1,486 1,357 1,228 1,098 July-Nov. average 3,471 3,463 3,451 3,431 3,390 3,313 July-Nov. average 9,751 9,719 9,666 9,578 9,431 9,191 Year average 7,041 6,843 6,613 6,357 6,062 5,716 Year average 19,485 18,952 18,303 17,558 16,692 15,686 Year load factor 78.0% 75.8% 73.2% 70.4% 67.1% 63.3% Year load factor 78.9% 76.7% 74.1% 71.1% 67.5% 63.5% Elasticity Dec - June 0.64 0.69 0.72 0.75 0.78 0.80 Elasticity Dec - June 0.64 0.69 0.73 0.76 0.80 0.83 load Year 0.32 0.35 0.37 0.39 0.42 0.45 load Year 0.32 0.35 0.38 0.41 0.44 0.48 Table 5.6: Impacts of selected runoff changes on hydro-energy generation without Lom Pangar dam ï‚· The direct and indirect contribution of the Lom Pangar reservoir to hydro-energy generation in the Sanaga Basin is shown to be significant and nearly constant for runoff variations in the range of -20% to +20%, contributing in this range 20% to the average load factor for Edea, Song Loulou, Nachtigal and Lom Pangar power stations during the dry season20 (11% or 1,200 GWh annually). 20 Note that Lom Pangar’s contribution to dry season hydro -energy generation would be constant in absolute terms in the hypothetical case that the reservoir would each rainy season be filled to its maximum capacity (like the Mbakaou reservoir). 68 Edea - generated energy with and w/out Song LL - generated energy with and w/out LP (dashed lines) LP (dashed lines) 2,000 250 4,000 450 Guaranteed Power (MW) Guaranteed Power (MW) 3,500 400 Energy prod. (GWh) Energy prod. (GWh) 1,500 200 3,000 350 2,500 300 1,000 150 2,000 250 500 100 1,500 200 1,000 150 0 50 500 100 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Runoff change (%) Runoff change (%) Year Dec-June Dec-June (10%) 10% - MW Year Dec-June Dec-June (10%) 10% - MW Nachtigal - generated energy with and Lom Pangar - generated energy w/out LP (dashed lines) 3,000 300 Guaranteed Power (MW) 200 25.0 Guaranteed Power (MW) 2,500 250 175 22.5 Energy prod. (GWh) Energy prod. (GWh) 2,000 200 150 20.0 1,500 150 125 17.5 1,000 100 100 15.0 500 50 75 12.5 0 0 50 10.0 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Runoff change (%) Runoff change (%) Year Dec-June Dec-June (10%) 10% - MW Year Dec-June Dec-June (10%) 10% - MW Edea + SLL + Nachtigal + LP - generated 7 HP stations - generated energy with and energy with and w/out LP (dashed lines) w/out LP (dashed lines) 9,000 1,000 25,000 2,500 Guaranteed Power (MW) Guaranteed Power (MW) Energy prod. (GWh) Energy prod. (GWh) 7,000 800 20,000 2,000 5,000 600 15,000 1,500 3,000 400 10,000 1,000 1,000 200 5,000 500 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Runoff change (%) Runoff change (%) Year Dec-June Dec-June (10%) 10% - MW Year Dec-June Dec-June (10%) 10% - MW Fig. 5.5: Impacts of runoff changes on hydro-energy generation with and without Lom Pangar dam (note that green lines ‘10%-MW’ refer to guaranteed power on the right axis) 69 Impact LP on guaranteed power Dec - June Impact LP on average annual energy (dashed lines refer to right axis) (dashed lines refer to right axis) 120 700 600 3,000 Guaranteed power (MW) Guaranteed Power (MW) Energy (GWh/yr) 100 600 500 2,500 Energy (GWh/yr) 80 500 400 2,000 60 400 300 1,500 40 300 200 1,000 20 200 100 500 0 100 0 0 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Runoff change (%) Runoff change (%) Edea Song LL Nachtigal Edea Song LL Nachtigal Lom Pangar 4 HP 7 HP Lom Pangar 4 HP 7 HP Impact LP on average load factor Dec - June (Lom Pangar results refer to right axis) 35% 75% Average load factor DS (%) Average load factor DS (%) 30% 70% 25% 65% 20% 60% 15% 55% 10% 50% 20% 10% 0% -10% -20% -30% Fig. 5.6: Impact of Lom Pangar reservoir on dry Runoff change (%) season guaranteed power and load factor, and Edea Song LL Nachtigal average annual energy 4 HP 7 HP Lom Pangar Increased runoff is not fully effective in terms of producing more energy due to the prevailing maximum turbine capacities, while reduced runoff leads initially to less runoff being lost to the ocean during the rainy season. Therefore, the annual variability of dry season hydro-energy generation under the present flow regime (without Lom Pangar reservoir) is more or less preserved in absolute terms and the runoff elasticity of dry season hydro-energy generation is shown to improve only marginally from 0.6 to 0.8 for the present conditions (base case hydrology) to 0.3 to 0.8 for the situation with Lom Pangar reservoir. Thus, while the Lom Pangar reservoir enables a significant increase in dry season hydro-energy generation, it contributes only modestly to the climate resilience of dry season hydro-energy. A further reduction of the annual variability of dry season hydro-energy generation would only be possible by prioritizing over-annual storage, focusing mainly on supplementing dry season flows after ‘dry’ rainy seasons; thereby reducing Lom Pangar’s potential to contribute each year significantly to dry-season hydro-energy generation. 70 ï‚· The runoff elasticity of rainy season hydro-energy generation is overall small – less than 0.2 - due to the relative abundance of runoff in this period. Thus, a change of 20% in runoff yields a change in generated hydro-energy of 4% or less. The impact of Lom Pangar reservoir on rainy season hydropower generation is negligible under all considered climate induced runoff changes. ï‚· Because of the upper limit of turbine capacities, annual runoff increases produce proportionally less additional hydro-energy than lost due to similar decreases in annual runoff. Runoff elasticity generated energy (GWh) Runoff elasticity generated energy (GWh) December - June with Lom Pangar December - June without Lom Pangar 1.0 1.0 0.8 0.8 Elasticity (%) Elasticity (%) 0.6 0.6 0.4 0.4 0.2 0.0 0.2 20% 10% 0% -10% -20% -30% -0.2 0.0 Runoff change (%) 20% 10% 0% -10% -20% -30% Runoff change (%) Edea Song LL Nachtigal Edea Song LL Nachtigal LP 4 HP 7 HP 4 HP 7 HP Runoff elasticity annual generated energy Runoff elasticity annual generated energy (GWh/yr) with Lom Pangar (GWh/yr) without Lom Pangar 1.0 1.0 0.8 0.8 Elasticity (%) Elasticity (%) 0.6 0.6 0.4 0.4 0.2 0.2 0.0 0.0 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Runoff change (%) Runoff change (%) Edea Song LL Nachtigal Edea Song LL Nachtigal LP 4 HP 7 HP 4 HP 7 HP Fig. 5.7: Runoff elasticities dry season and annual generated hydro-energy, with and without Lom Pangar 5.3 Runoff elasticity of the economic performance of Lom Pangar Hydropower Project The impacts on Lom Pangar reservoir and hydropower station on hydro-energy generation in the Sanaga Basin have also been used to test the runoff elasticity of the Economic Internal Rate of Return (EIRR) of 71 this project. We have used for this purpose the spreadsheet applied by the World Bank for the economic analysis of the project (World Bank, 2012a). Certain corrections had to be made to the hydrological baseline data and results included in this spreadsheet, particularly regarding the increase in average dry season flows in the Sanaga River enabled by the Lom Pangar reservoir. Based on our more detailed hydrological analysis it was found that this long-term average increase in dry season flows was overestimated in the original analysis, thus yielding about 3% higher EIRR values in the original economic analysis of the project. Three development scenarios are considered, similar to the original analysis, i.e. Scenario 1: Lom Pangar only (including incremental energy generated at Edea and Song Loulou), Scenario 2: same as scenario 1 with commissioning of the Nachtigal project (330 MW) by 201721 (including total cost and generated energy), and Scenario 3: same as scenario 2 with addition of Song Mbengue (including total cost and total generated energy). Results are shown in Table 5.7 and Figure 5.8. Without climate change induced runoff changes, the EIRR varies between 14.1 and 15.4%, depending on the scenario. Runoff changes due to climate change HP station 20% 10% 0% -10% -20% -30% Scenario 1: Edea + Song Loulou + Lom Pangar Edea (GWh/yr) 125 153 177 184 164 140 Song LL (GWh/yr) 262 323 376 389 347 301 LP (GWh/yr) 182 180 175 165 156 148 Sub-total 569 656 729 739 667 588 Elasticity -1.10 -1.00 -0.57 -0.13 0.42 0.64 EIRR (%) 11.9% 13.2% 14.1% 14.1% 13.0% 11.8% Elasticity -0.78 -0.68 -0.34 0.00 0.40 0.55 Scenario 2: Scenario 1 + Nachtigal Nachtigal (GWh/yr) 2,850 2,749 2,631 2,493 2,322 2,116 Sub-total 3,419 3,405 3,360 3,232 2,990 2,704 Elasticity 0.09 0.13 0.26 0.38 0.55 0.65 EIRR (%) 14.4% 14.4% 14.5% 14.4% 13.9% 13.2% Elasticity -0.06 -0.06 0.01 0.08 0.22 0.31 Scenario 3: Scenario 2 + Song Mbengue Song Mb. (GWh/yr) 6,940 6,883 6,766 6,545 6,171 5,736 Sub-total 10,358 10,287 10,126 9,777 9,160 8,439 Elasticity 0.11 0.16 0.25 0.34 0.48 0.56 EIRR (%) 15.4% 15.4% 15.4% 15.2% 14.9% 14.6% Elasticity 0.03 0.03 0.07 0.10 0.15 0.18 Table 5.7: EIRR for 3 development scenarios: Lom Pangar only, with Nachtigal and with Nachtigal and Song Mbengue (Note: table shows incremental energy generated at Edea and Song Loulou and total energy generated at Nachtigal and Song Mbengue; scenario 2 shows total energy of scenario 1 + energy Nachtigal, etc. ) Most importantly, since the direct and indirect contribution of Lom Pangar to hydro-energy generation in the basin is overall fairly constant for runoff variations in the range of -20% to +20%, the EIRR is 21 The commissioning data for the Nachtigal HP plant is now set for 2020, at an installed capacity of 360 MW (as used in the present analysis), but these differences are not expected to impact the EIRR of scenario 2 significantly. 72 similarly not very climate sensitive, unless decreases in runoff exceed 20%. The EIRR is optimal for the historical hydrology and tends to be constant or reduce slightly for increasing runoff when there is less need for additional storage in Lom Pangar reservoir to supplement dry season flows at Nachtigal, Song Loulou and Edea (under constant target turbine discharges). The EIRR decreases with decreasing flows due to the reservoir not always being fully utilized under such reduced flow conditions. The runoff elasticity of the EIRR is about 0.5 at 30% flow reduction for scenario 1, i.e. the EIRR is reduced with about 15% of its base case value for such large runoff reduction due to climate change. Overall, the EIRR of the Lom Pangar project is robust and insensitive to moderate runoff changes (<20%) due to climate change. Even for a runoff reduction of 30% it stays near to or above the 12% threshold, depending on the development scenario. Runoff elasticity of EIRR Lom Pangar, EIRR for Lom Pangar, Nachtigal and Song Nachtigal and Song Mbengue Mbengue 0.6 16% Runoff elasticity EIRR (%) 0.3 15% 0.0 14% EIRR (%) -0.3 13% -0.6 12% -0.9 11% 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Runoff change (%) Runoff change (%) LP only (1) LP + Na (2) LP + Na + SMb (3) LP only (1) LP+ Na (2) LP + Na + SMb (3) Fig. 5.8: Runoff elasticity of the EIRR and EIRR of Lom Pangar Project under various development scenarios. 5.4 Target flow Nachtigal and additional seasonal storage capacity in Djerem Basin In our analysis we have set the target turbine flow for the Nachtigal power station at 700 m 3/s, for calculating the maximum surplus which can be stored in Lom Pangar and Mbakaou reservoirs (including rainy season evaporation losses) without reducing the rainy season flow at Nachtigal below the threshold turbine discharge. Reducing this target flow does not affect the guaranteed (10% percentile) dry season flow at Nachtigal (about 600 m3/s), but slightly decreases the dry season and annual generation (in GWh/yr) at the 4 power stations Edea, Song Loulou, Nachtigal and Lom Pangar (denominated 4 HP; see Table 5.8). Instead, increasing the target flow above 700 m3/s decreases the guaranteed dry season flow at Nachtigal – and thus its guaranteed capacity (10% percentile) – without significantly increasing the dry season and annual hydro-energy generation at the 4 power stations. Therefore, the selected target turbine flow of 700 m3/s is considered to be optimal. For best results 73 regarding the impacts of climate change on hydro-energy generation (see also Table 5.8), the target flow for Nachtigal was modified with 5 m3/s for each percentage change in long-term average runoff, for example 600 m3/s for 20% decrease in runoff and 800 m3/s for 20% increase in runoff. Overall, results are not very sensitive to the selection of the target turbine discharge for Nachtigal. We have also tested the potential for further storage development upstream of Nachtigal on Djerem river by hypothetically (as a proxy) increasing the storage capacity of Mbakaou reservoir from 2,600 MCM to 4,600 MCM and increasing the target turbine discharge for Nachtigal to 750 m 3/s. As shown previously, there is a large surplus of water on Djerem River beyond the present storage in Mbakaou reservoir. As expected, the guaranteed and average dry season generation capacities would increase, but at the cost of a slight reduction in rainy season generation capacity. Overall, the positive impacts of such increase in storage on hydro-energy generation in the basin are limited, increasing the guaranteed (10% percentile) dry season generation capacity with only about 10% or 90 MW across Edea, Song Loulou, Nachtigal and Lom Pangar. The average annual production at these four stations would be increased with 175 GWh/yr, equivalent to an increase of the average load factor with 2% from 84% to 86%. Thus, the potential positive impacts of increased storage (beyond Lom Pangar) upstream of Nachtigal are limited and may economically not be very interesting, unless such storage could be combined with significant additional energy generation on the site of the additional storage reservoir. The reason for these modest impacts is that the annual storage in the Lom Pangar reservoir already achieves a rather uniform distribution of energy generation throughout the year, as shown in Table 5.3, i.e. the average generation for the four power stations ranges from 620 GWh/month during the dry season to 693 GWh/month during the rainy season. The additional storage of 2,000 MCM would increase average dry season generation to 650 GWh/month and reduce rainy season generation to 686 GWh/month. Present hydrology -20% change in runoff +20% change in runoff 3 Target Nachtigal (m /s) 600 650 700 750 800 600 650 700 700 750 800 3 Guaranteed flow N. (m /s) 601 598 597 562 562 481 470 461 672 668 668 4 HP -10% GWh dry season 3,932 3,972 3,869 3,822 3,822 3,130 3,130 3,101 4,488 4,477 4,477 4 HP – average GWh dry s. 4,147 4,280 4,341 4,355 4,356 3,720 3,756 3,741 4,534 4,593 4,638 4 HP –average GWh rainy s. 3,498 3,477 3,466 3,459 3,460 3,320 3,309 3,312 3,537 3,529 3,520 4 HP – average total GWh/yr 7,645 7,757 7,807 7,813 7,815 7,040 7,065 7,053 8,071 8,122 8,157 Table 5.8: Impact of target turbine flow for Nachtigal on hydro-energy generation 74 5.5 Runoff elasticity of hydro-energy generation in other basins 5.5.1 Lagdo dam in the Benue Basin No recent flow data were available for Lagdo reservoir and we, therefore, had to base this CRA on the available monthly flow data for the period 1950 – 1980. Lagdo reservoir characteristics are summarized in Fig. 5.9 and the location of Lagdo dam within the Benue Basin is shown in Figure 5.10. Surface area Reservoir Lagdo reservoir- Level-volume-height curves Reservoir 2 Volume level (m) (km ) 10,000 1000 (MCM) Volume 180 0 0 8,000 Area 800 Volume (km3) Area (km²) 206 232 1,450 Dead storage 6,000 600 216 697 6,000 Full Supply 4,000 400 218.18 840 7,680 2,000 200 0 0 180 185 190 195 200 205 210 215 220 Figure 5.9: Lagdo reservoir characteristics Level (m) Fig. 5.10: Location of Lagdo dam in the Benue Basin (Source: MINEE and GWP, 2009) For the purpose of this analysis we have simplified the operation of Lagdo reservoir as follows: ï‚· As for most other plants we have assumed a plant efficiency of 90% and a plant availability of 96%; see Table 5.1 for other plant and reservoir characteristics. Abstractions for irrigation were set at 725 MCM/yr (23 m3/s 22 ) and net losses for evaporation (corrected for net rainfall on the reservoir) at 550 MCM/yr or 17 m3/s, in total 40 m3/s or 1,275 MCM/yr (source BRLi, 2007). 22 3 3 BRLi (2007) indicates that abstractions for irrigation equal 9 m /s for the left bank and 14 m /s for the right bank; MINEE and GWP (2009) mentions instead only abstractions for about 50% of these amounts. 75 ï‚· Storage of rainy season runoff is limited to the months August to October. To achieve a rather uniform flow distribution over the year, we have assumed that during these 3 months 75% of the net runoff (reservoir inflow minus losses for reservoir evaporation and irrigation abstractions) will be stored up to a maximum of 4,550 MCM, while the balance is released uniformly during August – October. The storage volume on November 1st is then released uniformly over time over the period November till July, and added to the natural inflow during this period (after deduction of losses for reservoir evaporation and irrigation abstractions). Tests have shown that nearly identical results are obtained for a storage fraction of 70% and 80%, and that 75% is the optimal storage fraction. ï‚· Reservoir levels are subsequently calculated from: H (m) = 206 + (Volume – 1,450)/455 (see Fig. 5.10), and the head for the power station is calculated as the difference of reservoir level H and tail race level: h = 188 + (Q/52.7)0.667, with Q = outflow of Lagdo reservoir (m3/s). Finally, the monthly power output and annual hydro-energy generation is calculated, keeping in view the maximum turbine flow and output restrictions, as listed in Table 5.1. Guaranteed monthly power output (in MW) is calculated as the 10% percentile of all monthly outputs over the period 1950 – 1980. ï‚· Monthly runoff was parametrically varied with changes ranging between -30% and +20%. Results are shown in Table 5.9 and Figure 5.11. The guaranteed monthly output of Lagdo HP station is sensitive to runoff changes, at a runoff elasticity of about 1.5. The runoff elasticity of Lagdo’s annual hydro-energy output is about 0.8 to 1.0. Similar results are shown in Figure 5.12, based on annual variations in runoff and hydro-energy generation, simulated with the Mike Basin model for the period 1966 – 1989 (BRLi, 2007 and Grijsen et al, 2013). Figure 5.11 shows an average runoff elasticity of about 0.9, with lower values for runoff increases and higher values for runoff decreases (similar to the results in Table 5.9). Climate Change impact on Lagdo hydro-energy (GWh/yr) Runoff change 20% 10% 0% -10% -20% -30% Min 183 160 139 117 97 76 Max 389 386 382 364 351 301 0.10 263 227 195 182 152 124 0.20 287 250 233 208 176 144 0.50 341 313 286 262 227 191 average 324 307 286 260 230 195 Load factor 51.4% 48.7% 45.3% 41.3% 36.5% 31.0% Guaranteed MW 25 23 20 17 14 11 Elasticities: Average (GWh/yr) 0.67 0.75 0.82 0.89 0.97 1.05 Guaranteed MW 1.43 1.53 1.58 1.63 1.55 1.51 Table 5.9: Impact of runoff changes on generated energy for Lagdo power station 76 Energy output Lagdo dam Runoff elasticity Lagdo dam output 400 30 2.00 Energy generation (GWh/yr) Guaranteed power (MW) 1.50 Runoff elasticity 300 20 1.00 200 10 0.50 Annual energy Annual energy Guaranteed MW Guaranteed MW 100 0 0.00 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Runoff change (%) Runoff change (%) Fig. 5.11: Impact of runoff changes on generated energy for Lagdo power station Lagdo dam - 2005 hydrology 150% Annual hydro-energy change y = 0.9235x R² = 0.9194 100% 50% (GWh) 0% -50% Fig. 5.12: Runoff elasticity of Lagdo hydro- -100% energy generation (Source: Mike Basin -100% -50% 0% 50% 100% 150% Annual flow change (m3/s) model simulations; 1966 – 1989) 5.5.2 Njock and Mouila on Nyong River and Memve-Ele on Ntem River No recent flow data were available for Nyong and Ntem Rivers and we have, therefore, based this CRA on the available monthly flow data for the period 1951 – 1979. All three HP stations are R-o-R plants, with no storage other than for daily flow modulation. Njock will be located at Eseka (catchment area 21,600 km2), for which flow data are available from 1951-1956. Mouila is located upstream of Eseka, with a catchment area of 20,800 km2; hence the flow at Mouila was estimated at 96% of the flow at Eseka. Memve-Ele is located downstream of Ngoazik (catchment area 18,100 km2), for which flow data are available for the period 1953-1979. The catchment area of Memve-Ele is 26,350 km2 and the flow at this station was thus estimated at 146% of the flow at Ngoazik. Plant availability and efficiencies were set respectively at 96% and 90%. Monthly power output and annual hydro-energy generation were calculated from the monthly flows and plant characteristics shown in Table 5.1. Monthly runoff was parametrically varied with changes ranging between -30% and +20%. Results are shown in Table 5.10 and Figure 5.13. The runoff elasticity of guaranteed monthly power is for all 3 stations 1.0, as is expected 77 due to the absence of upstream storage, other than for daily flow modulation. The ratio between maximum turbine discharge and average annual runoff (1951-1979) is 0.67 for Njock, 0.60 for Mouila and 1.14 for Memve-Ele. Accordingly, the runoff elasticity of annual hydro-energy output is the lowest for Mouila (about 0.25) and the highest for Memve-Ele (about 0.5); similarly, the ratio of guaranteed power and installed capacity is the lowest for Memve-Ele (0.26) and the highest for Mouila (0.60). Clim ate Change im pact on hydro-energy Njock Clim ate Change im pact on hydro-energy Mouila GWh/yr 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Min 851 824 787 742 697 641 1,449 1,421 1,377 1,310 1,226 1,142 Max 999 999 994 982 956 922 1,660 1,660 1,660 1,653 1,630 1,570 0.10 860 836 816 783 743 698 1,475 1,441 1,394 1,356 1,296 1,221 0.20 895 877 847 812 765 715 1,515 1,493 1,464 1,409 1,345 1,256 0.50 931 913 893 865 831 782 1,579 1,555 1,520 1,485 1,431 1,368 Average 927 911 889 861 826 781 1,568 1,546 1,518 1,478 1,426 1,359 Load factor 89.1% 87.5% 85.4% 82.7% 79.4% 75.0% 90.7% 89.4% 87.8% 85.5% 82.5% 78.6% Guaranteed MW 77 71 64 58 51 45 143 131 119 107 95 83 Elasticities: Average (GWh/yr) 0.22 0.25 0.28 0.31 0.35 0.41 0.16 0.19 0.23 0.26 0.30 0.35 Guarant. MW 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 Clim ate Change im pact on hydro-energy Mem veele GWh/yr 20% 10% 0% -10% -20% -30% Min 889 839 788 738 673 596 Max 1,526 1,488 1,436 1,387 1,340 1,272 0.10 1,110 1,059 994 923 852 781 0.20 1,137 1,090 1,043 991 928 849 0.50 1,255 1,188 1,122 1,052 972 898 Average 1,259 1,211 1,157 1,097 1,028 949 Load factor 71.2% 68.5% 65.4% 62.0% 58.1% 53.7% Guaranteed MW 62 57 52 47 42 36 Elasticities: Average (GWh/yr) 0.44 0.47 0.50 0.52 0.56 0.60 Guarant. MW 1.00 1.00 1.00 1.00 1.00 1.00 Table 5.10: Impact of runoff changes on generated energy for Njock, Mouila and Memve-Ele HP plants Annual generated energy and guaranteed Runoff elasticity annual generated energy power (dashed lines) 2,000 200 Guaranteed power (MW) 0.60 Energy production (GWh) Runoff elasticity 1,500 150 0.40 1,000 100 0.20 500 50 0 0 0.00 20% 10% 0% -10% -20% -30% 20% 10% 0% -10% -20% -30% Runoff change (%) Runoff change (%) Njock Mouila Memve Ele Njock Mouila Memve Ele Fig: 5.13: Impact of runoff changes on generated energy for Njock, Mouila and Memve-Ele HP plants 78 6. Climate change projections and impacts on runoff for the main river basins in Cameroon 6.1. Climate projections for the main river basins of Cameroon This chapter summarizes climate change projections for the five main river basins of Cameroon, i.e. the Coastal, Congo, Sanaga, Niger and Lake Chad Basins as available from the Climate Wizard (CW; CKPCW; Girvetz et al, 2009). Developed through collaboration between The Nature Conservancy, The University of Washington, and The University of Southern Mississippi, the Climate Wizard (Girvetz et al, 2009) enables users – technical and non-technical alike - to easily access leading climate change information and visualize the impacts anywhere on Earth. It allows the user to choose a country and see both the climate change that has occurred to date and the climate change that is predicted to occur. The Climate Wizard can thus be used to assess how climate has changed over time and to project what future changes are likely to occur in a given area. The system uses time series of historical precipitation and temperature data (1901 – 2002) provided by the Climate Research Unit (CRU-TS 2.10; Mitchell et al, 2003 and 2005). Statistical representations of modeled future climate predictions are best achieved by examining a range of time rather that a single year. The Climate Wizard has chosen the time period 2040-2069 and 2070-2099 to provide the user with a range that most accurately describes the predicted conditions for the mid century (2050) and the end of the century (2100) respectively. No projections are provided prior to the 2050 time horizon. The spatial resolution for the Climate Wizard is 50 km. Climate change analysis becomes more complex for the future than for the past because there is not one time-series of climate, but rather many future projections from different GCM runs with a range of CO2 emissions scenarios (Nakicenovic and Swart, 2000). It is thus important not to analyze only one GCM for any given emission scenario, but rather to use ensemble analysis to combine the analyses of multiple GCMs and quantify the range of possibilities for future climates under different emissions scenarios. The Climate Wizard uses results of 15 GCMs, as initially produced for IPCC’s 4th Assessment (Boko et al, 2007; Kundzewicz et al, 2007; Parry et al, 2007), but updated as new results become available for the ongoing 5th Assessment. Figure 6.1 shows for Cameroon and surrounding countries the precipitation projections for 2050 for the 20% driest models (80% of the models agree on greater precipitation increases) and 20% wettest models (80% of the models agree on less precipitation). Figure 6.2 shows the average of all precipitation projections produced by 15 GCMs for the 2050 time horizon (A1B scenario). Clearly, part of the GCMs project a drier future, while others project a wetter future, with on average no significant increase in precipitation by 2050. All models agree instead on a much warmer future (see Annex 7). A screenshot from the World Bank’s Climate Change Knowledge Portal (CCKP; 22 GCM projections) is shown in Figure 6.3. The schematization of Cameroon into 5 river basins, for the purpose of the CW projections, is shown in Figure 6.4. The southern portion of the Lake Chad Basin in Cameroon is included with the Niger Basin. Selected climate change projections for 2050 and 2080 for the entire country are shown in Annex 7. Results of 15 GCM projections from the CW are shown in Table 6.123 and Figure 6.5 for 2050 and in Table 6.2 and Figure 6.6 for 2080. Note that it was beyond the scope of this study to verify the (possibly limited) skills of the 15 GCMs in reproducing the present climate of Cameroon. 23 Note that in this chapter the projected changes in precipitation and runoff are shown in tenths of 1%, which is as such not justified due to the large uncertainty in climate model results, but facilitates intercomparison of results 79 Fig. 6.1: Projections of precipitation changes by 2050 for Central Africa; upper panel: 20% driest models; lower panel: 20% wettest models (source: Climate Wizard; A1B emission scenario) 80 Fig. 6.2: Projections of average precipitation changes by 2050 for Central Africa; (source: Climate Wizard) Fig. 6.3: Screenshot of the Climate Change Knowledge Portal (World Bank) 81 Lake Chad – North North Niger Sanaga Coastal Basins Congo Fig. 6.4: Schematization of river basins for the purpose of Climate Wizard climate change projections Results from the World Bank’s Climate Change Knowledge Portal (CCKP; 22 GCM projections) for 2050 are shown in Tables 6.3 and 6.4 and in Figures 6.7 and 6.8. The spatial resolution for the CCKP portals is 0.50. The results from both portals agree well and provide a good indication of the climate changes which can be expected, and particularly of the wide range of precipitation projections. Differences in projections for the various basins are not significant, particularly when the large spread in projections from different GCMs is considered. Overall we can expect by 2050 across Cameroon a temperature increase of 20C, with a standard deviation of 0.40C; on average an increase of 30C is projected by 2080 with a standard deviation of 0.60C. Most models predict by 2050 a temperature increase in the range of 1.3 to 2.70C (2 to 40C by 2080). Projected increases in temperature are distributed rather homogeneously across the country, with only a slightly larger variation in projected increases across the Niger and Lake Chad basins compared to the central and southern basins. It is noted that this study uses mainly annual projections of climate change, but seasonal projections have also been considered as discussed in Section 6.2, while monthly flow data were used for the system vulnerability analysis discussed in Chapter 5. 82 Coastal Basin Congo Basin Sanaga Basin Niger Basin Lake Chad Basin Cameroon dP dT dP dT dP dT dP dT dP dT dP dT GCMs 0 0 0 0 0 0 (%) ( C) (%) ( C) (%) ( C) (%) ( C) (%) ( C) (%) ( C) bccr_bcm2_0.1 -1.4 1.5 -2.2 1.7 1.7 1.6 4.6 1.7 -4.3 2.0 0.3 1.6 cccma_cgcm3_1.1 -1.4 2.1 1.5 2.1 -0.5 2.2 0.5 2.3 11.7 2.3 1.3 2.2 cnrm_cm3.1 5.5 2.2 4.5 2.2 2.0 2.4 1.8 2.6 9.3 2.6 4.0 2.4 csiro_mk3_0.1 -3.4 1.3 -3.3 1.4 -4.5 1.4 -2.6 1.5 -1.8 1.5 -3.3 1.4 gfdl_cm2_0.1 1.3 2.0 2.9 2.1 -2.8 2.1 -9.2 2.3 -11.7 2.6 -3.1 2.2 gfdl_cm2_1.1 4.9 2.1 1.0 2.2 0.3 2.3 -9.6 2.7 -21.1 3.2 -3.2 2.4 giss_model_e_r.1 14.0 1.7 16.0 1.9 18.5 1.8 10.3 2.1 -1.5 2.4 12.6 2.0 inmcm3_0.1 -10.0 2.0 -10.4 2.1 -7.7 2.0 0.5 2.0 7.3 2.2 -5.5 2.0 ipsl_cm4.1 -0.6 2.1 -0.3 2.2 -2.3 2.2 -6.2 2.3 -3.4 2.4 -2.6 2.2 miroc3_2_medres.1 -7.2 1.7 -4.7 1.7 -3.9 1.7 0.9 1.6 15.1 1.6 -1.8 1.7 miub_echo_g.1 6.0 1.7 5.8 1.8 7.2 1.9 5.9 1.9 9.7 2.0 6.5 1.9 mpi_echam5.1 6.4 1.9 8.4 2.1 9.4 2.1 8.3 2.2 6.3 2.4 8.0 2.1 ncar_ccsm3_0.1 5.5 1.8 13.1 1.7 4.5 1.8 1.2 2.1 19.8 2.3 7.6 1.9 ncar_pcm1.1 6.2 1.2 2.6 1.2 2.2 1.2 2.9 1.3 1.6 1.3 3.2 1.2 ukmo_hadcm3.1 -8.7 1.8 -18.4 2.0 -11.1 2.0 2.2 2.1 8.6 2.1 -7.6 2.0 average 1.1 1.8 1.1 1.9 0.9 1.9 0.8 2.0 3.0 2.2 1.1 1.9 standard deviation 6.7 0.3 8.6 0.3 7.2 0.3 5.8 0.4 10.6 0.5 5.7 0.3 Minimum -10.0 1.2 -18.4 1.2 -11.1 1.2 -9.6 1.3 -21.1 1.3 -7.6 1.2 Maximum 14.0 2.2 16.0 2.2 18.5 2.4 10.3 2.7 19.8 3.2 12.6 2.4 5% -9.1 1.3 -12.8 1.3 -8.8 1.3 -9.4 1.4 -14.5 1.5 -6.1 1.3 20% -4.2 1.7 -3.6 1.7 -4.0 1.7 -3.4 1.6 -3.6 1.9 -3.2 1.7 Table 6.1: Climate wizard projections of relative climate changes for Cameroon by 2050 (A1B scenario) Normal pdf precipitation changes Normal pdf temperature changes 20 3.0 y = 7.3659x + 0.9613 y = 0.3454x + 1.9108 R² = 0.9294 R² = 0.8661 10 2.5 Change (%) Change (0C) 0 2.0 -10 1.5 -20 1.0 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate Coast Congo Sanaga Niger Lake Chad Coast Congo Sanaga Niger Lake Chad Figure 6.5: Normal distribution of projected precipitation and temperature changes for Cameroon river basins (2050, A1B); Source: Climate Wizard (Results for Lake Chad basin excluded from average regression) 83 Coastal Basin Congo Basin Sanaga Basin Niger Basin Lake Chad Basin Cameroon dP dT dP dT dP dT dP dT dP dT dP dT GCMs 0 0 0 0 0 0 (%) ( C) (%) ( C) (%) ( C) (%) ( C) (%) ( C) (%) ( C) bccr_bcm2_0.1 3.2 2.4 2.4 2.6 3.3 2.6 3.4 2.7 -4.7 3.0 2.0 2.6 cccma_cgcm3_1.1 -1.9 3.0 2.0 3.0 -1.9 3.1 -1.1 3.3 4.8 3.4 0.0 3.1 cnrm_cm3.1 13.3 3.1 10.5 3.2 9.6 3.4 6.1 3.7 7.2 3.8 9.6 3.4 csiro_mk3_0.1 -6.0 2.1 -6.7 2.2 -6.6 2.2 -2.7 2.4 6.1 2.5 -4.3 2.3 gfdl_cm2_0.1 1.1 3.0 -0.5 3.1 -7.0 3.2 -16.1 3.5 -22.1 4.1 -7.6 3.3 gfdl_cm2_1.1 5.1 2.9 1.1 3.1 0.6 3.2 -9.5 3.6 -23.6 4.2 -3.3 3.4 giss_model_e_r.1 18.4 2.6 20.4 2.8 22.2 2.7 15.3 3.0 2.0 3.2 17.2 2.8 inmcm3_0.1 -9.9 2.7 -10.4 2.8 -7.0 2.6 5.1 2.7 20.6 2.9 -3.0 2.7 ipsl_cm4.1 -0.3 3.5 4.2 3.6 -4.1 3.6 -10.6 3.8 -7.2 4.0 -3.6 3.7 miroc3_2_medres.1 -9.9 2.7 -4.9 2.7 -4.5 2.6 3.0 2.5 28.0 2.3 -1.1 2.6 miub_echo_g.1 13.4 2.8 17.9 3.0 17.7 3.0 15.0 2.9 16.9 2.9 16.6 3.0 mpi_echam5.1 18.1 3.4 21.2 3.7 24.0 3.7 21.4 3.9 11.3 4.3 20.4 3.8 ncar_ccsm3_0.1 13.2 2.4 18.0 2.4 8.7 2.5 5.0 2.9 32.1 3.0 13.6 2.6 ncar_pcm1.1 4.8 1.6 2.2 1.7 4.3 1.7 2.4 1.9 0.1 2.0 3.3 1.8 ukmo_hadcm3.1 -0.5 2.9 -12.0 3.3 -3.7 3.2 10.2 3.3 15.8 3.3 -0.4 3.2 average 4.1 2.7 4.4 2.9 3.7 2.9 3.1 3.1 5.8 3.3 3.9 3.0 standard deviation 9.4 0.5 11.0 0.5 10.6 0.5 10.2 0.6 16.1 0.7 9.1 0.5 Minimum -9.9 1.6 -12.0 1.7 -7.0 1.7 -16.1 1.9 -23.6 2.0 -7.6 1.8 Maximum 18.4 3.5 21.2 3.7 24.0 3.7 21.4 3.9 32.1 4.3 20.4 3.8 5% -9.9 2.0 -10.9 2.1 -7.0 2.1 -12.2 2.3 -22.5 2.2 -5.3 2.1 20% -2.8 2.4 -5.2 2.6 -4.9 2.6 -4.0 2.6 -5.2 2.8 -3.4 2.6 Table 6.2: Climate wizard projections of relative climate changes for Cameroon by 2080 (A1B scenario) Normal pdf precipitation changes Normal pdf temperature changes 40 4.5 30 y = 10.677x + 3.8348 4.0 y = 0.5526x + 2.8967 20 R² = 0.9354 3.5 R² = 0.9035 Change (%) Change (0C) 10 3.0 0 2.5 -10 2.0 -20 1.5 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate Coast Congo Sanaga Niger Lake Chad Coast Congo Sanaga Niger Lake Chad Figure 6.6: Normal distribution of projected precipitation and temperature changes for Cameroon river basins (2080, A1B); Source: Climate Wizard (Results for Lake Chad basin excluded from average regression) 84 Congo Basin (catchment # 5598) Sanaga basin (catchment # 6086) Mean T Runoff Mean T Runoff Change Change Change Change GCM Change change Change change 0 PCP (%) PET (%) 0 PCP (%) PET (%) ( C) (%) ( C) (%) bccr_bcm2_0 1.7 -7.8 -14.4 5.3 1.7 -5.3 -1.4 5.0 cccma_cgcm3_1 2.2 0.4 7.7 5.1 2.3 3.9 18.4 5.0 cccma_cgcm3_1_t63 2.4 -4.5 -12.6 6.5 2.3 -6.7 -15.0 6.8 cnrm_cm3 2.3 12.9 33.0 3.5 2.3 10.8 28.2 3.7 csiro_mk3_0 1.5 -2.5 -9.1 4.1 1.5 -5.8 -15.8 4.5 csiro_mk3_5 2.4 -5.1 -16.8 6.7 2.3 -1.3 -12.7 5.7 gfdl_cm2_0 2.2 3.9 8.8 4.6 2.2 4.0 10.1 4.6 gfdl_cm2_1 2.2 2.7 -0.6 4.8 2.2 7.7 13.3 3.9 giss_aom 1.5 2.2 2.0 3.4 1.6 1.4 2.9 3.5 giss_model_e_h 1.9 7.3 2.4 3.5 1.7 19.7 57.6 0.8 giss_model_e_r 2.1 10.3 27.8 3.4 1.9 20.4 71.6 1.2 iap_fgoals1_0_g 1.4 -1.7 -6.4 3.8 1.5 1.0 -2.9 3.5 inmcm3_0 2.2 -8.2 -16.3 6.6 2.1 -8.1 -20.9 6.6 ipsl_cm4 2.3 0.8 -2.5 5.6 2.4 -0.9 -7.5 5.9 miroc3_2_hires 2.9 -10.6 -21.0 8.7 2.8 -0.6 8.6 6.6 miroc3_2_medres 1.9 -6.4 -22.3 5.7 1.9 -6.2 -25.5 5.7 mpi_echam5 2.0 7.2 13.2 3.7 1.9 11.9 31.3 2.7 mri_cgcm2_3_2a 1.6 9.8 34.2 2.3 1.6 17.1 64.1 0.6 ncar_ccsm3_0 1.8 15.0 19.0 1.9 1.7 6.8 -2.5 2.8 ncar_pcm1 1.4 -0.7 -5.7 3.5 1.3 10.5 19.9 1.5 ukmo_hadcm3 1.8 -7.0 -21.1 5.5 1.8 -7.7 -25.6 5.9 ukmo_hadgem1 1.5 -1.8 -3.9 3.7 1.4 3.5 6.7 2.6 Average 2.0 0.7 -0.2 4.6 1.9 3.4 9.2 4.0 St.dev. 0.4 7.2 17.2 1.6 0.4 8.8 27.6 1.9 Min 1.4 -10.6 -22.3 1.9 1.3 -8.1 -25.6 0.6 Max 2.9 15.0 34.2 8.7 2.8 20.4 71.6 6.8 5% 1.4 -8.2 -21.1 2.4 1.4 -7.7 -25.3 0.8 20% 1.5 -6.2 -15.9 3.5 1.6 -5.7 -14.6 2.6 Table 6.3: CCK Portal projections of relative climate changes for Congo and Sanaga basins by 2050 (A1B) Normal pdf precipitation changes Normal pdf temperature changes 30 3.5 y = 8.6902x + 1.6575 y = 0.4342x + 2.0716 20 R² = 0.8932 R² = 0.851 3.0 10 2.5 Change (%) Change (0C) 0 2.0 -10 1.5 -20 -30 1.0 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate Congo Sanaga Niger Lake Chad Congo Sanaga Niger Lake Chad Figure 6.7: Normal distribution of projected precipitation and temperature for the Congo, Sanaga, Niger and Lake Chad basins (2050, A1B); Source: CCK Portal. 85 Niger Basin (catchment # 6135) Lake Chad basin (catchment # 5610) Mean T Runoff Runoff Change Change Change Change Change GCM Change change change 0 PCP (%) PET (%) Mean T (0C) PCP (%) PET (%) ( C) (%) (%) bccr_bcm2_0 1.8 3.1 8.2 3.6 2.2 -13.2 -19.8 5.4 cccma_cgcm3_1 2.4 0.9 -2.3 5.4 2.3 5.2 2.2 4.9 cccma_cgcm3_1_t63 2.5 -0.8 -5.3 5.6 2.7 -3.6 -8.3 6.2 cnrm_cm3 2.8 0.2 -5.5 6.6 2.7 12.7 8.6 5.6 csiro_mk3_0 1.7 0.0 -7.6 3.7 1.8 -6.3 -9.8 4.3 csiro_mk3_5 2.4 5.4 -18.1 5.0 2.6 -5.1 -16.0 6.0 gfdl_cm2_0 2.4 -13.9 -40.8 6.5 2.5 -3.9 -13.4 5.7 gfdl_cm2_1 2.7 -18.4 -37.0 7.5 3.1 -29.8 -36.7 8.1 giss_aom 1.7 1.9 -2.3 3.6 1.7 0.8 -3.5 3.8 giss_model_e_h 2.4 3.8 5.1 5.2 2.4 6.1 4.1 5.3 giss_model_e_r 2.5 -0.2 -10.5 5.7 2.6 -1.8 -8.6 5.9 iap_fgoals1_0_g 1.7 0.9 -1.5 3.8 1.7 2.8 0.6 3.7 inmcm3_0 2.1 -0.8 -3.4 5.0 2.4 1.5 -2.1 5.3 ipsl_cm4 2.4 -4.3 -20.0 5.7 2.8 -2.8 -9.0 6.2 miroc3_2_hires 2.7 7.8 3.4 5.2 2.9 4.7 -3.9 6.2 miroc3_2_medres 1.8 2.1 -0.6 3.9 2.0 10.8 13.5 4.1 mpi_echam5 2.2 14.0 30.6 3.8 2.4 6.2 0.9 5.2 mri_cgcm2_3_2a 1.7 -2.4 -7.5 4.1 2.0 -4.5 -9.8 4.6 ncar_ccsm3_0 2.3 0.2 -12.6 5.2 2.5 12.6 12.4 5.1 ncar_pcm1 1.6 0.1 -5.8 3.7 1.7 -0.2 -6.4 4.0 ukmo_hadcm3 2.0 13.7 42.4 3.4 2.1 16.3 20.3 4.1 ukmo_hadgem1 1.2 10.8 44.8 1.4 1.2 21.3 31.3 1.8 Average 2.1 1.1 -2.1 4.7 2.3 1.4 -2.4 5.1 Stdev 0.4 7.4 20.7 1.4 0.5 10.8 14.4 1.3 Min 1.2 -18.4 -40.8 1.4 1.2 -29.8 -36.7 1.8 Max 2.8 14.0 44.8 7.5 3.1 21.3 31.3 8.1 5% 1.6 -13.4 -36.2 3.4 1.7 -12.9 -19.6 3.8 20% 1.7 -0.8 -12.2 3.7 1.9 -4.4 -9.8 4.1 Table 6.4: CCKP projections of relative climate changes for Niger and Lake Chad basins by 2050 (A1B) Normal pdf PET changes Normal pdf runoff changes 10 75 y = 1.5793x + 4.6107 y = 20.193x + 1.1203 8 R² = 0.8723 50 R² = 0.823 6 25 Change (%) Change (%) 4 0 2 -25 0 -50 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate Congo Sanaga Niger Lake Chad Congo Sanaga Niger Lake Chad Figure 6.8: Normal distribution of projected PET and runoff changes for the Congo, Sanaga, Niger and Lake Chad basins (2050, A1B); Source: CCK Portal 86 The rainfall distribution patterns shown in Figure 3 of Annex 7 are consistent with the average 20th century rainfall pattern determined based on the CRU data sets (Figure 3.16). The probability distributions of projected precipitation changes are similar for the various basins; on average no significant changes are projected for 2050, but the standard deviation of the projected long-term average precipitation changes is about 8%. This is slightly less than the present variability of basin average precipitation with a coefficient of variation between 9% and 10% across Cameroon. For 2080 a small increase in precipitation (4%) is projected, with a standard deviation of 12%. Overall, one could thus adopt for impact analysis a worst case scenario of 15% reduction in precipitation by 2050 and 20% by 2080 (projected mean minus two standard deviations). Projected changes in potential evapotranspiration (PET) for 2050 vary mostly between 2% and 7%, with an average of 4.6%. These changes are commensurate with the sensitivity of PET for changes in temperature derived in Annex 6 and Chapter 4 according to the Hargreaves method (1982), i.e. 1/(T+17.8)% per 0C (T = 24 0C), or about 2.4% per 0C. 6.2 Seasonal climate projections for Cameroon Seasonal climate change projections with the CW for the Coastal, Congo, Sanaga and Niger Basins are tabulated in Table 6.5 for 2050. Annual projections are included for comparison. Seasonal projections could not be obtained for the Lake Chad Basin due to system errors. Temperature changes are seen to be uniformly distributed over the year for each GCM. Instead, seasonal precipitation changes can vary significantly from the average annual changes, particularly during the winter season December to February (DJF). However, rainfall is insignificant during this period and runoff during these months consists of recession flow correlated to rainfall during the preceding period September – November (SON). Therefore, we can ignore the projections for the winter period. Runoff during the period March to May (MAM) is also rather small in the Sanaga and Benue basins; hence, climate variations during the periods JJA (June to August) and SON are the most important projections for the purpose of our study. For the Sanaga basin, the largest weight is assigned to the SON precipitation, since runoff during the period September – February is generally more than twice the runoff during JJA. For the Sanaga Basin we weighted the seasonal precipitation projections for each GCM according to the seasonal runoff at Edea, whereby the runoff during DJF was added to the weight for the SON precipitation changes. The weighted average annual change is also included in Table 6.5b for the Sanaga Basin, showing a good agreement between the projected annual changes and weighted seasonal changes. In view of the relatively large storage capacities of the reservoirs in the Sanaga and Benue basins, our CRA study for these basins is primarily concerned with regulated annual flow volumes and total rainy season precipitation and runoff. Therefore, we may conclude that the use of annual projections is adequate for these Basins. As shown in Figures 3.21 and 3.22, the Southern Coastal basins (Nyong and Ntem Rivers) and the Congo basin in the South of Cameroon tend to exhibit two rainy seasons (MAM and SON), while there are as yet no large reservoirs in these basins. Thus, seasonal climate change impacts on precipitation and temperature may need to be taken into account in these basins. This will be addressed in Section 7.3. 87 Precipitation Coastal Basins Congo Basin GCM DJF MAM JJA SON Year DJF MAM JJA SON Year bccr_bcm2_0.1 -6.0 -5.5 -2.3 2.1 -1.4 6.9 -4.9 -6.9 0.8 -2.2 cccma_cgcm3_1.1 -27.6 -10.2 8.9 -0.3 -1.4 -8.9 -11.0 18.8 3.2 1.5 cnrm_cm3.1 8.2 3.4 6.7 6.2 5.5 7.4 3.4 3.2 5.7 4.5 csiro_mk3_0.1 -13.1 -8.5 0.8 -2.4 -3.4 -13.6 -3.3 0.0 -2.9 -3.3 gfdl_cm2_0.1 -11.8 0.0 5.4 1.7 1.3 -3.1 0.7 10.5 1.7 2.9 gfdl_cm2_1.1 14.8 -3.6 5.5 10.1 4.9 17.9 -0.9 -0.9 -0.3 1.0 giss_model_e_r.1 3.3 9.8 13.1 19.5 14.0 22.7 8.1 15.4 20.9 16.0 inmcm3_0.1 -15.0 -7.4 -19.3 -3.1 -10.0 -16.5 -7.2 -22.0 -5.2 -10.4 ipsl_cm4.1 -18.2 -5.7 -0.7 5.0 -0.6 -11.8 -1.3 -2.0 3.4 -0.3 miroc3_2_medres.1 -3.8 -2.2 -11.8 -7.5 -7.2 16.5 0.8 -13.4 -7.8 -4.7 miub_echo_g.1 -17.3 8.5 5.2 7.3 6.0 -10.8 6.3 7.7 7.1 5.8 mpi_echam5.1 -12.1 0.8 8.2 10.5 6.4 -7.9 1.8 6.2 18.3 8.4 ncar_ccsm3_0.1 76.6 4.5 -6.2 8.0 5.5 85.4 9.3 2.5 10.6 13.1 ncar_pcm1.1 5.6 6.2 10.8 2.3 6.2 -6.6 2.6 6.1 2.8 2.6 ukmo_hadcm3.1 -20.7 -1.1 -13.9 -10.4 -8.7 -22.3 -4.0 -37.6 -17.8 -18.4 Average -2.5 -0.7 0.7 3.3 1.1 3.7 0.0 -0.8 2.7 1.1 Standard deviation 24.9 6.2 9.7 7.6 6.7 26.3 5.6 14.6 9.6 8.6 Min -27.6 -10.2 -19.3 -10.4 -10.0 -22.3 -11.0 -37.6 -17.8 -18.4 Max 76.6 9.8 13.1 19.5 14.0 85.4 9.3 18.8 20.9 16.0 10% -19.7 -8.0 -13.0 -5.8 -8.1 -15.3 -6.3 -18.6 -6.8 -8.1 20% -17.5 -6.0 -7.3 -2.6 -4.2 -12.2 -4.2 -8.2 -3.3 -3.6 50% -11.8 -1.1 5.2 2.3 1.3 -6.6 0.7 2.5 2.8 1.5 Temperature Coastal Basins Congo Basin GCM DJF MAM JJA SON Year DJF MAM JJA SON Year bccr_bcm2_0.1 1.5 1.6 1.4 1.4 1.5 1.8 1.8 1.5 1.5 1.7 cccma_cgcm3_1.1 2.4 2.2 1.8 2.0 2.1 2.3 2.3 1.9 2.1 2.1 cnrm_cm3.1 2.3 2.1 2.0 2.1 2.2 2.5 2.1 2.1 2.1 2.2 csiro_mk3_0.1 1.4 1.3 1.2 1.2 1.3 1.5 1.4 1.3 1.3 1.4 gfdl_cm2_0.1 2.1 2.1 1.9 1.8 2.0 2.1 2.2 2.1 1.9 2.1 gfdl_cm2_1.1 2.2 2.0 2.2 2.0 2.1 2.2 2.1 2.5 2.1 2.2 giss_model_e_r.1 1.6 2.0 1.7 1.7 1.7 1.6 2.0 2.0 2.0 1.9 inmcm3_0.1 2.1 2.0 2.0 1.9 2.0 2.1 2.0 2.1 1.9 2.1 ipsl_cm4.1 2.3 2.1 2.0 2.1 2.1 2.3 2.2 2.1 2.1 2.2 miroc3_2_medres.1 1.6 1.9 1.8 1.8 1.7 1.6 1.8 1.8 1.8 1.7 miub_echo_g.1 1.9 1.7 1.6 1.6 1.7 2.1 1.8 1.7 1.7 1.8 mpi_echam5.1 2.0 2.1 1.9 1.7 1.9 2.1 2.3 2.1 1.8 2.1 ncar_ccsm3_0.1 1.5 1.6 2.1 1.8 1.8 1.4 1.5 2.1 1.9 1.7 ncar_pcm1.1 1.2 1.1 1.3 1.1 1.2 1.2 1.1 1.2 1.2 1.2 ukmo_hadcm3.1 1.9 1.9 1.8 1.6 1.8 2.1 2.1 2.1 1.8 2.0 Average 1.9 1.8 1.8 1.7 1.8 1.9 1.9 1.9 1.8 1.9 Standard deviation 0.4 0.3 0.3 0.3 0.3 0.4 0.3 0.4 0.3 0.3 Min 1.2 1.1 1.2 1.1 1.2 1.2 1.1 1.2 1.2 1.2 Max 2.4 2.2 2.2 2.1 2.2 2.5 2.3 2.5 2.1 2.2 10% 1.4 1.4 1.3 1.3 1.4 1.4 1.4 1.4 1.3 1.5 20% 1.5 1.6 1.6 1.6 1.7 1.6 1.7 1.7 1.7 1.7 50% 1.9 2.0 1.8 1.8 1.8 2.1 2.0 2.1 1.9 2.0 Table 6.5a: Seasonal climate projections for the Coastal and Congo Basins (A1B, 2050); Source: CW 88 Precipitation Sanaga Basin Weighted Niger Basin Year GCM DJF MAM JJA SON Year Sanaga DJF MAM JJA SON Year bccr_bcm2_0.1 18.3 -0.8 1.0 3.4 1.7 2.5 23.0 10.0 1.4 1.4 4.6 cccma_cgcm3_1.1 -18.1 -10.3 8.2 -0.9 -0.5 1.0 43.2 -12.1 3.5 0.9 0.5 cnrm_cm3.1 8.7 -1.0 0.9 5.5 2.0 3.9 29.4 -0.7 0.7 4.3 1.8 csiro_mk3_0.1 -16.6 -5.0 -2.7 -4.9 -4.5 -4.3 9.1 -4.4 -2.1 -3.7 -2.6 gfdl_cm2_0.1 -17.1 2.6 -1.4 -6.6 -2.8 -4.6 -39.0 3.1 -9.8 -16.8 -9.2 gfdl_cm2_1.1 19.2 -5.5 0.8 2.7 0.3 1.6 41.7 -21.2 -7.3 -5.4 -9.6 giss_model_e_r.1 17.7 16.2 20.9 18.2 18.5 18.8 27.0 9.5 11.4 11.1 10.3 inmcm3_0.1 -18.3 -7.9 -14.4 -1.4 -7.7 -5.3 8.6 -10.4 0.9 5.6 0.5 ipsl_cm4.1 -21.7 -7.4 -0.8 1.1 -2.3 0.1 -12.7 -10.9 -5.9 -2.9 -6.2 miroc3_2_medres.1 25.1 -2.2 -8.0 -2.2 -3.9 -3.8 42.7 -5.1 -1.2 5.2 0.9 miub_echo_g.1 -31.2 6.6 9.4 7.4 7.2 7.9 -34.0 1.6 8.0 5.4 5.9 mpi_echam5.1 -9.3 0.3 8.9 18.3 9.4 14.7 15.2 -1.6 6.3 20.6 8.3 ncar_ccsm3_0.1 141.2 5.1 -5.4 6.7 4.5 3.3 117.3 4.7 -2.8 4.8 1.2 ncar_pcm1.1 -1.5 -0.4 2.4 3.3 2.2 2.9 56.7 -2.3 2.4 7.7 2.9 ukmo_hadcm3.1 -34.8 -3.7 -18.0 -8.6 -11.1 -10.8 -35.1 2.5 -0.9 3.2 2.2 Average 4.1 -0.9 0.1 2.8 0.9 1.8 19.5 -2.5 0.3 2.8 0.8 Standard deviation 42.6 6.7 9.7 7.8 7.2 7.7 40.8 8.5 5.7 8.3 5.8 Min -34.8 -10.3 -18.0 -8.6 -11.1 -10.8 -39.0 -21.2 -9.8 -16.8 -9.6 Max 141.2 16.2 20.9 18.3 18.5 18.8 117.3 10.0 11.4 20.6 10.3 10% -27.4 -7.7 -11.8 -5.9 -6.4 -5.0 -34.7 -11.6 -6.8 -4.7 -8.0 20% -19.0 -5.9 -6.0 -2.7 -4.0 -4.4 -17.0 -10.5 -3.4 -3.0 -3.4 50% -9.3 -1.0 0.8 2.7 0.3 1.6 23.0 -1.6 0.7 4.3 1.2 Temperature Sanaga Basin Niger Basin GCM DJF MAM JJA SON Year DJF MAM JJA SON Year bccr_bcm2_0.1 1.7 1.7 1.5 1.6 1.6 1.7 1.5 1.6 1.8 1.7 cccma_cgcm3_1.1 2.5 2.3 1.8 2.2 2.2 2.6 2.6 1.9 2.4 2.3 cnrm_cm3.1 2.7 2.3 2.1 2.1 2.4 3.2 2.7 2.2 2.2 2.6 csiro_mk3_0.1 1.6 1.5 1.3 1.3 1.4 1.6 1.7 1.4 1.4 1.5 gfdl_cm2_0.1 2.2 2.2 2.1 1.9 2.1 2.3 2.2 2.3 2.3 2.3 gfdl_cm2_1.1 2.4 2.2 2.5 2.1 2.3 2.8 2.7 2.9 2.2 2.7 giss_model_e_r.1 1.5 2.1 1.8 1.8 1.8 1.7 2.5 2.1 2.2 2.1 inmcm3_0.1 2.1 2.0 1.9 1.9 2.0 2.2 2.0 1.8 2.0 2.0 ipsl_cm4.1 2.2 2.2 2.1 2.2 2.2 2.1 2.3 2.4 2.4 2.3 miroc3_2_medres.1 1.4 1.8 1.8 1.8 1.7 1.2 1.8 1.7 1.6 1.6 miub_echo_g.1 2.1 1.9 1.7 1.7 1.9 2.1 2.2 1.7 1.8 1.9 mpi_echam5.1 2.1 2.3 2.0 1.8 2.1 2.2 2.5 2.1 2.0 2.2 ncar_ccsm3_0.1 1.7 1.5 2.1 1.8 1.8 2.6 1.9 2.1 1.8 2.1 ncar_pcm1.1 1.2 1.2 1.3 1.1 1.2 1.2 1.3 1.3 1.2 1.3 ukmo_hadcm3.1 2.1 2.1 1.9 1.8 2.0 2.0 2.4 2.0 2.1 2.1 Average 2.0 2.0 1.9 1.8 1.9 2.1 2.1 2.0 2.0 2.0 Standard deviation 0.4 0.4 0.3 0.3 0.3 0.6 0.4 0.4 0.4 0.4 Min 1.2 1.2 1.3 1.1 1.2 1.2 1.3 1.3 1.2 1.3 Max 2.7 2.3 2.5 2.2 2.4 3.2 2.7 2.9 2.4 2.7 10% 1.4 1.5 1.4 1.4 1.5 1.3 1.6 1.5 1.5 1.5 20% 1.6 1.6 1.6 1.7 1.7 1.7 1.7 1.7 1.8 1.6 50% 2.1 2.1 1.9 1.8 2.0 2.1 2.2 2.0 2.0 2.1 Table 6.5b (cont.): Seasonal climate projections for the Sanaga and Niger Basins (A1B, 2050); Source: CW 89 6.3 Runoff projections from the Climate Portal The CCK Portal provides estimates of future changes in long-term average basin runoff. Overall, the average projected changes are insignificant, but with a significant standard deviation in the order of 20% and a worst case scenario of an average decrease in annual runoff of about 35% (Figure 6.8, right panel). The relationship between changes in precipitation (P) and runoff (Q) as projected by the CCK Portal is shown in Figure 6.9. The graph suggests that the precipitation elasticity will be in the order of 2.2 for the river basins of Cameroon, which compares well to the values derived in Chapter 4 on the basis of historical annual precipitation and runoff data (Table 4.5). Based on a precipitation elasticity of runoff of 2.2 and a temperature sensitivity of runoff of -3% per 0C, one can estimate the average projected runoff decrease by 2050 due to a temperature increase of Runoff vs precipitation changes 75 2.0 0C at 6% and the runoff increase due to an y = 2.1743x average 1.7% increase in average rainfall at 4%, in R² = 0.8343 50 total a decrease of 2%, with a standard deviation of 19% (2.2 times standard deviation of 8.7% of the Runoff change (%) 25 projected precipitation changes; see Figure 6.7). This 0 agrees well with the fitted normal distribution of projected changes in annual runoff (right panel of -25 Figure 6.8). -50 -40.0 -20.0 0.0 20.0 40.0 Precipitation change (%) Fig. 6.9: Regression of projected changes in runoff Congo Sanaga Niger Lake Chad and precipitation; Source: CCK Portal, 2050 6.4 Runoff projections based on climate projections from the Climate Wizard The GCM projections of temperature and precipitation changes for 2050 and 2080 provided by the Climate Wizard, shown in Tables 6.1 and 6.2, are used to project runoff changes for these time horizons. Runoff changes are estimated for each basin based on the precipitation elasticity and temperature sensitivity of runoff derived in Chapter 4 (Table 4.5), from the equation: dQ/Q = εP dP/P + ST dT Results are shown in Table 6.6 and Figure 6.10. Differences between the projections for various basins are minor compared to the large spread in projections for each basin. As expected, projections are normally distributed. On average the adopted 15 GCMs do not project a significant change in runoff for 2050 and 2080, but the spread between individual projections is significant, with a standard deviation of 17% by 2050 and 24% by 2080, reflecting also that the uncertainty in climate projections increases with time. Barring a few outliers for the Lake Chad basin, projections of average runoff vary between -35% and +30% by 2050 and between -45% and +45% by 2080. 90 Coastal Basins Congo Basin Sanaga Basin Niger Basin Lake Chad Basin GCMs 2050 2080 2050 2080 2050 2080 2050 2080 2050 2080 bccr_bcm2_0.1 -6.9 0.4 -9.7 -2.3 -1.2 -0.5 6.2 -0.6 -19.7 -25.0 cccma_cgcm3_1.1 -8.5 -11.8 -2.9 -4.4 -7.7 -13.5 -6.7 -14.2 23.5 0.3 cnrm_cm3.1 5.9 19.9 3.3 13.7 -2.6 11.1 -4.4 2.9 15.9 5.2 csiro_mk3_0.1 -10.6 -18.1 -11.2 -21.3 -14.0 -21.2 -12.1 -15.4 -11.0 7.3 gfdl_cm2_0.1 -2.4 -5.5 0.3 -10.1 -12.4 -24.9 -32.0 -54.2 -42.8 -77.7 gfdl_cm2_1.1 4.9 3.1 -4.3 -6.6 -6.2 -8.3 -34.4 -37.4 -71.5 -82.4 giss_model_e_r.1 24.9 32.0 29.7 36.8 35.3 40.8 19.3 29.3 -13.5 -6.8 inmcm3_0.1 -26.1 -27.8 -28.9 -30.9 -23.0 -23.4 -5.7 3.8 11.6 46.4 ipsl_cm4.1 -6.7 -9.5 -7.0 -1.0 -11.7 -19.9 -24.1 -40.9 -19.0 -35.9 miroc3_2_medres.1 -19.7 -28.0 -15.3 -18.7 -13.6 -17.8 -3.1 -0.9 36.0 69.3 miub_echo_g.1 8.2 20.8 7.4 30.5 10.2 29.9 8.6 28.6 19.3 36.0 mpi_echam5.1 8.3 29.1 12.5 36.0 14.5 41.8 13.8 41.9 8.4 14.8 ncar_ccsm3_0.1 6.9 21.6 23.8 32.6 4.5 11.8 -4.3 2.8 46.7 77.9 ncar_pcm1.1 9.9 5.9 2.3 0.1 1.1 4.3 3.2 -0.4 -0.6 -7.6 ukmo_hadcm3.1 -22.9 -8.7 -46.4 -35.9 -30.5 -17.8 -1.5 15.1 15.6 31.1 average -2.3 1.6 -3.1 1.2 -3.8 -0.5 -5.1 -2.6 -0.1 3.5 standard deviation 14.0 19.6 19.0 23.9 16.0 23.1 15.5 26.8 30.6 46.5 Minimum -26.1 -28.0 -46.4 -35.9 -30.5 -24.9 -34.4 -54.2 -71.5 -82.4 Maximum 24.9 32.0 29.7 36.8 35.3 41.8 19.3 41.9 46.7 77.9 5% -23.9 -27.9 -34.2 -32.4 -25.2 -23.9 -32.7 -44.9 -51.4 -79.1 20% -12.4 -13.1 -12.0 -19.2 -13.7 -20.1 -14.5 -19.8 -19.1 -27.2 Table 6.6: Projections of relative runoff changes (in %) based on Climate Wizard projections of precipitation and temperature changes (A1B scenario) Normal pdf runoff changes 2050 Normal pdf runoff changes 2080 60 60 40 40 y = 16.805x - 3.5959 y = 24.195x - 0.0822 20 20 R² = 0.9334 R² = 0.9221 Change (0C) Change (%) 0 0 -20 -20 -40 -40 -60 -60 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate Coast Congo Sanaga Niger Lake Chad Coast Congo Sanaga Niger Lake Chad Fig. 6.10: Normal distribution of projected runoff changes for 2050 and 2080 (Results for Lake Chad basin excluded from average regression) 91 6.5 Runoff scenarios for the economic analysis of water and hydropower projects in Cameroon Climate change confronts decision makers with deep uncertainties, requiring robust decision analysis to inform good decisions by identifying system vulnerabilities and assessing alternatives for ameliorating those vulnerabilities. Hence, the economic performance of water and energy infrastructure projects must also be tested for worst case scenarios. Based on the results of this study as reflected in Figure 6.10 (left panel), the following climate change induced runoff scenarios may be used for further analysis of the robustness of such projects in Cameroon, for the 2050 investment horizon: ï‚· Average future (2050) scenario under climate change: no significant change in annual runoff ï‚· Medium 2050 dry scenario of climate change: decrease of annual runoff with 15% (25% probability) ï‚· Worst case/Dry scenario by 2050: decrease of annual runoff with 35% (3% probability) ï‚· Medium wet scenario of climate change: increase of annual runoff with 10% (20% exceedance probability) ï‚· Extreme wet scenario: increase of annual runoff with 30% (2% probability). The results of all available climate projections shown in this Chapter demonstrate the importance of taking all GCM projections into consideration, rather than building the analysis on the projections of only a few GCMs. Models which performed well for the present climate conditions may not necessarily give the same performance for future situations. Cook and Vizy (2006), for example, selected 3 GCMs out of 18 for West Africa climate change studies, based on how well they reproduced the features of the West African monsoon system. The three models, however, displayed significantly different projections for the 21th century, meaning at least one and possibly two of the three “bestâ€? models are wrong. Similarly, UCI (2011) selected three GCMs which – after downscaling through a Regional Climate Model (RCM) – best reproduced the historic climate over the Atlas Mountains in the Oum-Er-Rbia river basin in Morocco. However, one of these 3 models projected an extremely dry future, while another projected only a small decrease in precipitation by 2050 and beyond. In such circumstances, multi-model ensembling is found to be the appropriate approach for assessing the impacts of climate change on the water resources in Cameroon. This helps reducing the effects of model errors in one particular model and the natural variability in any particular run. The huge range in climate change and runoff projections, in other words the large uncertainty of these projections, also supports the simplified approach whereby the input of water resources system models is varied parametrically, and the runoff – precipitation – temperature relationships are linearized. 92 6.6 CMIP-5 climate projections Whereas IPCC’s 4th Assessment Report used emissions and greenhouse concentrations developed by the Special Report on Emission Scenarios (SRES) as “plausible descriptions of a possible future state of the worldâ€?, the 5th Assessment Report (expected to be issued in 2014) uses Representative Concentration Pathways (RCP). These RCPs are projections of consistent sets of “radiative forcingâ€? or the changes in the earth’s energy balance (incoming versus outgoing energy) based on selected greenhouse gas concentrations. RCPs do not calculate greenhouse concentrations using socioeconomic drivers as in AR4. Four greenhouse gas concentration (not emissions) trajectories are adopted by the IPCC as the RCP’s shown in Figure 6.11. The pathways are used for climate modeling and research. They describe four possible climate futures, all of which are considered possible depending on how much greenhouse gases are emitted in the years to come. The four RCPs, RCP2.6, RCP4.5, RCP6, and RCP8.5, are named after a possible range of radiative forcing values in the year 2100 relative to pre-industrial values (+2.6, +4.5, +6.0, and +8.5 Watt/m2, respectively. Fig. 6.11: Representative Concentration Pathways adopted by ICPP for AR5 Climate models have also evolved since AR4 (2007), so a larger and up-to-date suit of model projections (referred to as CMIP5) is now available with climate research institutions. However, those new results are not yet readily accessible for the public at large. Nonetheless, we obtained access to a beta-version of the new Climate Wizard tool, which will likely be released in 2014, and we were able to download results for 26 of the latest GCM climate projections for the Sanaga Basin, for 2050 (2036-2065) and 2080 (2066-2095), and for the RCP4.5 and RCP8.5. Results are shown in Figure 6.12 in terms of probability distributions of projected changes in precipitation and temperature for 2050 and 2080, and for both RCPs. For comparison we have included the results shown in Figures 6.5 and 6.6 for the Sanaga Basin, which were based on climate projections from 15 GCMs for the A1B emission scenario (AR4). For all practical intents and purposes, it is seen that probability distributions of the precipitation projections for the RCP4.5 and RCP8.5 pathways are nearly identical and similar to the results for the A1B emission scenarios, both for 2050 and 2080. The temperature projections for the RCP4.5 and 93 RCP8.5 pathways show larger differences, particular for 2080, but the previous results for the A1B scenario are suitably located within the range of results for both pathways. Keeping in mind that precipitation changes dominate changes in runoff - and therefore changes in generated hydro-energy – much more than changes in temperature, we may conclude that despite the huge recent investments in climate research, results for Cameroon have not changed significantly to the extent it concerns projected annual changes in precipitation and – by proxy – changes in runoff. It is particularly significant that the present suit of models has not yet narrowed down the wide spread between individual model projections, and therefore, does not appear to have reduced the GCM model uncertainties significantly. Therefore, the results presented in this report would appear to remain fully valid for Cameroon under the newest available climate change projections. Normal pdf precipitation changes Normal pdf temperature changes Sanaga Basin 2050 Sanaga Basin 2050 20 3.0 10 y = 0.4495x + 2.3279 Change (0C) 2.5 Change (%) y = 7.1095x + 3.7342 0 2.0 y = 6.872x + 2.9718 y = 0.3439x + 1.9032 -10 1.5 y = 7.5269x + 0.8578 y = 0.3753x + 1.7972 -20 1.0 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate RCP 4.5 RCP 8.5 AR4-A1B RCP 4.5 RCP 8.5 AR4-A1B Normal pdf precipitation changes Normal pdf temperature changes Sanaga Basin 2080 Sanaga Basin 2080 20 5.0 y = 10.171x + 7.8837 y = 0.7682x + 3.8974 10 4.0 y = 0.5486x + 2.8885 Change (0C) Change (%) 0 3.0 y = 8.0383x + 3.4899 -10 2.0 y = 10.696x + 3.7091 y = 0.5512x + 2.285 -20 1.0 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate RCP 4.5 RCP 8.5 AR4-A1B RCP 4.5 RCP 8.5 AR4-A1B Fig. 6.12: Comparison of climate change projections for RCP4.5, RCP8.5 and the A1B (SRES) scenario 94 7 Climate risks for hydro-energy generation in Cameroon 7.1 Hydro-energy Sanaga basin and EIRR Lom Pangar project The runoff changes projected in Table 6.6 for the Sanaga basin have been translated into changes in hydro-energy generation and changes in the Economic Internal Rate of Return (EIRR) of the Lom Pangar and Nachtigal projects, based on the runoff elasticities shown in Tables 5.5 and 5.7. Results are shown in Table 7.1 and Figure 7.1 for two performance indicators, i.e. for the change in total annual hydro-energy (GWh/yr) generated by Edea, Song Loulou, Lom Pangar and Nachtigal HP stations (denominated E-4HP) and for the EIRR of the combined Lom Pangar and Nachtigal projects (denominated EIRR-4HP). In the latter case the total cost and benefits of the Nachtigal HP project are included in the EIRR, as well as the benefits of the incremental hydro-energy generated at Edea, Song Loulou and Lom Pangar and the cost of the Lom Pangar project. Changes in the selected performance indicators are estimated by multiplying the projected changes in runoff with the runoff elasticities of the indicators, as approximated by: for E-4HP: ε = 0.34 – 0.7 * dQ/Q (Table 5.5) for EIRR-4HP: if dQ/Q>0: ε = -0.3 * dQ/Q; else: ε = - dQ/Q (Table 5.7) Changes in Runoff Changes (%) in Changes (%) in GCM (%) hydro-energy 4 HP EIRR LP + Nachtigal 2050 2080 2050 2080 2050 2080 bccr_bcm2_0.1 -1.2 -0.5 -0.42 -0.17 -0.01 0.00 cccma_cgcm3_1.1 -7.7 -13.5 -3.05 -5.89 -0.60 -1.83 cnrm_cm3.1 -2.6 11.1 -0.93 2.91 -0.07 -0.37 csiro_mk3_0.1 -14.0 -21.2 -6.12 -10.33 -1.96 -4.48 gfdl_cm2_0.1 -12.4 -24.9 -5.29 -12.84 -1.53 -6.22 gfdl_cm2_1.1 -6.2 -8.3 -2.38 -3.30 -0.38 -0.69 giss_model_e_r.1 35.3 40.8 3.28 2.21 -3.74 -5.00 inmcm3_0.1 -23.0 -23.4 -11.52 -11.79 -5.29 -5.48 ipsl_cm4.1 -11.7 -19.9 -4.93 -9.52 -1.37 -3.95 miroc3_2_medres.1 -13.6 -17.8 -5.93 -8.30 -1.85 -3.19 miub_echo_g.1 10.2 29.9 2.74 3.91 -0.31 -2.68 mpi_echam5.1 14.5 41.8 3.45 1.97 -0.63 -5.25 ncar_ccsm3_0.1 4.5 11.8 1.39 3.03 -0.06 -0.42 ncar_pcm1.1 1.1 4.3 0.37 1.33 0.00 -0.06 ukmo_hadcm3.1 -30.5 -17.8 -16.86 -8.24 -9.29 -3.15 average -3.8 -0.5 -3.1 -3.7 -1.8 -2.9 st.dev. 16.0 23.1 5.7 6.1 2.6 2.2 Min -30.5 -24.9 -16.9 -12.8 -9.3 -6.2 Max 35.3 41.8 3.5 3.9 0.0 0.0 5% -25.2 -23.9 -13.1 -12.1 -6.5 -5.7 20% -13.7 -20.1 -6.0 -9.7 -2.3 -5.0 50% -6.2 -8.3 -2.4 -3.3 -0.6 -3.2 Table 7.1: Projected changes in hydro-energy generation and EIRR of Lom Pangar and Nachtigal projects 95 Normal pdf hydro-energy changes (E-4HP) Normal pdf EIRR changes (EIRR-4HP) 10 4 2050 2080 5 2 2050 2080 0 0 Change (%) Change (%) -5 y = 5.7681x - 3.0801 -2 -10 y = 2.3036x - 1.8062 y = 6.2416x - 3.6665 -4 -15 y = 2.2584x - 2.8501 -20 -6 -25 -8 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate Fig. 7.1: Projected changes in hydro-energy generation and EIRR of Lom Pangar and Nachtigal projects Figure 7.1 (left panel) shows that by 2050 the total long-term average hydro-energy generation by the Edea, Song Loulou, Lom Pangar and Nachtigal power plants could vary between -15% and +5% of the base case value (present hydrology). Results for 2080 are similar. Given the present climate projections, it is highly unlikely that hydro-energy generation would decrease more than 20% due to climate change. Changes in the EIRR for Lom Pangar and Nachtigal due to climate change are also limited. By 2050 it is not likely that the EIRR would change significantly (decrease more than 5%), while in the worst case the EIRR would be reduced with less than 10% of its base case value, e.g. from about 14.5% in the base case to 13% under severe climate change impacts. Hence, the Lom Pangar and Nachtigal projects are economically robust and resilient to long-term climate changes (note that short-term climate variability can severely impact reservoir operations and hydro-energy generation). As seen in Chapter 5, the EIRR of Lom Pangar and Nachtigal projects are at their maximum under the current hydrology. Runoff decreases due to climate change reduce the effectiveness of the chosen storage volume in Lom Pangar reservoir in terms of incremental hydro-energy generation at Edea and Song Loulou (a smaller reservoir would be more economic at lower inflows), while runoff increases render the reservoir less necessary due to increased natural flow conditions and limited maximum turbine capacities at the considered hydropower stations. Hence, the EIRR decreases both under long- term increasing and decreasing runoff conditions; as a result the normal distribution does not adequately fit the range of estimated EIRR values. 7.2 Lagdo dam in Niger Basin The runoff changes projected in Table 6.6 for the Niger/Benue basin in Cameroon have been translated into changes in hydro-energy generated at Lagdo dam, based on the runoff elasticities shown in Table 5.8. Results are shown in Figure 7.2 and Table 7.2 for the change in total annual hydro-energy (GWh/yr) generated by Lagdo dam. Changes in this performance indicator are estimated by multiplying the projected changes in runoff with the runoff elasticity of the indicator, approximated as per Table 5.9 by: ε = 0.82 – 0.76 * dQ/Q 96 Figure 7.2 shows that by 2050 total long-term average hydro-energy generation at Lagdo dam could vary between -35% and +10% of the base case value (present hydrology). Results for 2080 show even a larger variation. Under the 2050 climate conditions there is nearly 20% probability that the annual hydro-energy generation would reduce with 20% or more, i.e. one out of 5 models project such reductions. Hence, Lagdo dam hydro-energy generation may suffer a significant decrease due to climate change, and is less climate resilient than Lom Pangar and Nachtigal. Normal pdf hydro-energy changes at Change Runoff Changes (%) in GCMs Lagdo dam power plant (%) hydro-energy 4 HP 2050 2080 2050 2080 20 bccr_bcm2_0.1 6.2 -0.6 4.8 -0.5 2050 2080 cccma_cgcm3_1.1 -6.7 -14.2 -5.9 -13.2 0 cnrm_cm3.1 -4.4 2.9 -3.7 2.3 Change (%) y = 15.027x - 6.1116 csiro_mk3_0.1 -12.1 -15.4 -11.0 -14.5 -20 gfdl_cm2_0.1 -32.0 -54.2 -34.0 -66.7 gfdl_cm2_1.1 -34.4 -37.4 -37.2 -41.3 y = 25.181x - 7.2849 giss_model_e_r.1 19.3 29.3 13.0 17.5 -40 inmcm3_0.1 -5.7 3.8 -4.9 3.0 ipsl_cm4.1 -24.1 -40.9 -24.2 -46.2 -60 miroc3_2_medres.1 -3.1 -0.9 -2.6 -0.7 -2.0 -1.0 0.0 1.0 2.0 miub_echo_g.1 8.6 28.6 6.5 17.2 Normal reduced variate mpi_echam5.1 13.8 41.9 9.9 21.0 ncar_ccsm3_0.1 -4.3 2.8 -3.7 2.3 ncar_pcm1.1 3.2 -0.4 2.6 -0.3 ukmo_hadcm3.1 -1.5 15.1 -1.3 10.6 Figure 7.2: Projected changes in hydro- energy generation at Lagdo dam average -5.1 -2.6 -6.1 -7.3 st.dev. 15.5 26.8 14.9 25.4 Min -34.4 -54.2 -37.2 -66.7 Max 19.3 41.9 13.0 21.0 Table 7.2: Projected changes in hydro- 5% -32.7 -44.9 -35.0 -52.3 energy generation at Lagdo dam 20% -14.5 -19.8 -13.6 -19.8 50% -4.3 -0.4 -3.7 -0.3 7.3 Nyong and Ntem River Basins The runoff changes projected for the Coastal Basins (Table 6.6) have been translated into changes in annual hydro-energy generated at Njock and Mouila HP stations on Nyong River and Memve-Ele on Ntem River (see Figure 3.1 for locations), based on the runoff elasticities shown in Table 5.10. Results are shown in Table 7.3 and Figure 7.3 for the estimated future changes in total annual generated hydro- energy. Changes in this performance indicator are estimated by multiplying the projected changes in runoff with the runoff elasticities of the indicator for each power plant, approximated by (Table 5.10): for Njock: ε = 0.30 – 0.38 * dQ/Q for Mouila: ε = 0.24 – 0.37 * dQ/Q for Memve-Ele: ε = 0.50 – 0.31 * dQ/Q 97 Changes in runoff Changes in hydro-energy by Changes in hydro-energy by GCMs (%) 2050 (%) 2080 (%) 2050 2080 Njock Mouila Memve-Ele Njock Mouila Memve-Ele bccr_bcm2_0.1 -6.9 0.4 -2.2 -1.8 -3.6 0.1 0.1 0.2 cccma_cgcm3_1.1 -8.5 -11.8 -2.8 -2.3 -4.5 -4.1 -3.3 -6.3 cnrm_cm3.1 5.9 19.9 1.6 1.3 2.8 4.5 3.3 8.7 csiro_mk3_0.1 -10.6 -18.1 -3.6 -3.0 -5.7 -6.7 -5.6 -10.1 gfdl_cm2_0.1 -2.4 -5.5 -0.7 -0.6 -1.2 -1.7 -1.4 -2.8 gfdl_cm2_1.1 4.9 3.1 1.4 1.1 2.4 0.9 0.7 1.5 giss_model_e_r.1 24.9 32.0 5.1 3.7 10.5 5.7 3.9 12.8 inmcm3_0.1 -26.1 -27.8 -10.4 -8.8 -15.2 -11.3 -9.5 -16.3 ipsl_cm4.1 -6.7 -9.5 -2.2 -1.8 -3.5 -3.2 -2.6 -5.1 miroc3_2_medres.1 -19.7 -28.0 -7.4 -6.2 -11.1 -11.4 -9.6 -16.4 miub_echo_g.1 8.2 20.8 2.2 1.7 3.9 4.6 3.4 9.1 mpi_echam5.1 8.3 29.1 2.2 1.7 3.9 5.5 3.8 11.9 ncar_ccsm3_0.1 6.9 21.6 1.9 1.5 3.3 4.7 3.5 9.4 ncar_pcm1.1 9.9 5.9 2.6 2.0 4.6 1.6 1.3 2.9 ukmo_hadcm3.1 -22.9 -8.7 -8.9 -7.4 -13.1 -2.9 -2.4 -4.6 average -2.3 1.6 -1.4 -1.3 -1.7 -0.9 -1.0 -0.3 st.dev. 14.0 19.6 4.6 3.7 7.3 5.7 4.6 9.6 Min -26.1 -28.0 -10.4 -8.8 -15.2 -11.4 -9.6 -16.4 Max 24.9 32.0 5.1 3.7 10.5 5.7 3.9 12.8 5% -23.9 -27.9 -9.3 -7.8 -13.7 -11.3 -9.6 -16.3 20% -12.4 -13.1 -4.4 -3.6 -6.7 -4.6 -3.8 -7.1 50% -2.4 0.4 -0.7 -0.6 -1.2 0.1 0.1 0.2 Table 7.3: Projected changes in hydro-energy generation at Njock, Mouila and Memve-Ele HP stations Normal pdf hydro-energy changes 2050 Normal pdf hydro-energy changes 2080 20 20 Njock Njock Mouila Mouila 10 10 Memve_Ele Memve_Ele Change (%) Change (%) 0 0 y = 7.591x - 1.7478 y = 10.058x - 0.3419 -10 -10 y = 4.7092x - 1.4161 y = 5.8818x - 0.907 y = 3.82x - 1.2578 y = 4.6533x - 0.9644 -20 -20 -30 -30 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate Fig. 7.3: Projected changes in hydro-energy generation at Njock, Mouila and Memve-Ele HP stations 98 Results are similar for Njock and Mouila HP stations, which are both located close to Eseka on Nyong River. Figure 7.3 shows that by 2050 and 2080 total long-term average hydro-energy generation at both stations could vary between -10% and +5% of the base case value (present hydrology).The standard deviation of future changes projected for Memve-Ele station is slightly larger, and annual hydro-energy could vary for this plant between -15% and +10%, due to its higher ratio between maximum turbine capacity and average flow. The probability that the annual hydro-energy generation at anyone of these power stations would reduce with 20% or more is negligible. Hence, climate change impacts on hydro- energy generation at these stations will only be minor to moderate. As shown in Figures 3.21 and 3.22 and discussed in Section 6.2, the Southern Coastal basins (Nyong and Ntem Rivers) and the Congo basin in the South of Cameroon tend to exhibit two rainy seasons (MAM and SON), with lower rainfall and runoff during the intermediate summer period JJA. As yet there are no large reservoirs in these basins for runoff regulation and it may thus be of some importance to consider for these basins seasonal variations in climate change impacts on (particularly) precipitation. The seasonal precipitation and temperature projections for the coastal basins (Table 6.5a) have accordingly been translated into seasonal runoff changes and subsequently into changes in seasonal hydro-energy generated at Njock, Mouila and Memve-Ele power stations. Results are shown in Figure 7.4 and Table 7.4 for the estimated changes (in %) by 2050 in generated hydro-energy during the JJA and SON seasons. For the SON season (Fig. 7.4, right panel) results are similar to the annual results shown in Figure 7.3; results for the MAM season (not shown here) are equally similar to the annual results. The larger standard deviation in projected precipitation changes for the JJA quarter compared to the standard deviation in annual changes translates into a potentially larger variation in future energy output for this season. However, runoff during the JJA season is only about 50% of the runoff during the SON season and 20% of the annual runoff. Thus, these potentially larger variations in climate and runoff changes have only a minor impact on annual generated hydro-energy in these basins, also keeping in view the large uncertainties in the presently available climate change projections. Therefore, the analysis based on projected annual changes in climate parameters is considered to adequate and representative for the southern basins in Cameroon. 99 Normal pdf hydro-energy changes JJA-2050 Normal pdf hydro-energy changes SON-2050 20 20 Njock Njock Mouila Mouila 10 10 Memve_Ele Memve_Ele Change (%) Change (%) 0 0 y = 11.639x - 2.8385 y = 7.6698x - 2.4808 y = 8.0415x + 0.4432 -10 -10 y = 6.3693x - 2.2492 y = 4.6321x - 0.1974 y = 3.6401x - 0.3156 -20 -20 -30 -30 -2.0 -1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0 Normal reduced variate Normal reduced variate Figure 7.4: Projected seasonal changes in hydro-energy generation at Njock, Mouila and Memve-Ele HP stations for 2050 Changes in runoff Changes in hydro-energy by Changes in hydro-energy by GCMs (%) by 2050 JJA-2050 (%) SON-2050 (%) JJA SON Njock Mouila Memve-Ele Njock Mouila Memve-Ele bccr_bcm2_0.1 -8.5 0.7 -2.8 -2.3 -4.5 0.2 0.2 0.3 cccma_cgcm3_1.1 13.9 -6.0 3.4 2.6 6.3 -1.9 -1.6 -3.1 cnrm_cm3.1 8.7 7.5 2.3 1.8 4.1 2.0 1.6 3.6 csiro_mk3_0.1 -1.5 -8.3 -0.4 -0.4 -0.7 -2.8 -2.3 -4.4 gfdl_cm2_0.1 6.4 -1.0 1.8 1.4 3.1 -0.3 -0.2 -0.5 gfdl_cm2_1.1 5.8 16.0 1.6 1.3 2.8 3.8 2.9 7.2 giss_model_e_r.1 23.1 36.4 4.9 3.6 9.9 5.9 3.8 14.1 inmcm3_0.1 -45.8 -11.4 -21.7 -18.7 -29.4 -3.9 -3.2 -6.1 ipsl_cm4.1 -6.6 5.2 -2.1 -1.7 -3.4 1.5 1.2 2.5 miroc3_2_medres.1 -29.4 -20.5 -12.1 -10.3 -17.4 -7.7 -6.5 -11.5 miub_echo_g.1 6.7 11.1 1.8 1.4 3.2 2.9 2.2 5.2 mpi_echam5.1 12.2 17.8 3.1 2.4 5.6 4.1 3.1 7.9 ncar_ccsm3_0.1 -18.5 12.0 -6.9 -5.7 -10.3 3.1 2.4 5.6 ncar_pcm1.1 19.5 1.8 4.4 3.3 8.6 0.5 0.4 0.9 ukmo_hadcm3.1 -33.8 -26.0 -14.5 -12.4 -20.5 -10.3 -8.7 -15.1 average -3.2 2.4 -2.5 -2.2 -2.8 -0.2 -0.3 0.4 st.dev. 20.5 15.8 7.9 6.6 11.6 4.5 3.6 7.6 Min -45.8 -26.0 -21.7 -18.7 -29.4 -10.3 -8.7 -15.1 Max 23.1 36.4 4.9 3.6 9.9 5.9 3.8 14.1 5% -37.4 -22.1 -16.7 -14.3 -23.1 -8.5 -7.1 -12.6 20% -20.7 -8.9 -7.9 -6.6 -11.7 -3.0 -2.4 -4.7 50% 5.8 1.8 1.6 1.3 2.8 0.5 0.4 0.9 Table 7.4: Projected seasonal changes in hydro-energy generation at Njock, Mouila and Memve-Ele HP stations for 2050 100 8. Conclusions and Recommendations Central and West Africa have experienced a significant climate variability in the 20th century and more distant past. Most recently, around 1970 an abrupt downward shift in precipitation is believed to have occurred, which reduced for example runoff from the Sanaga basin with about 16%. In this context, this report presents a robust bottom-up, risk-based Climate Risk Assessment (CRA) for the five main river basins of Cameroon, which focuses on potential climate change impacts on water resources availability for hydro-energy generation, particularly in the Sanaga, Benue, Nyong and Ntem River Basins. Whereas multiple Global Circulation Models (GCM) project on average no significant changes in annual precipitation across Cameroon by 2050 for the A1B emission scenario, it is more important to note that individual model projections vary between -10% and +15%, with a standard deviation of 7%. Equally, for 2080 only a small increase in precipitation (4%) is projected, but the standard deviation of individual model projections is a significant 11%. Since extreme model projections may become reality in the future, it is recommended to adopt for impact analysis a worst case scenario of 15% reduction in precipitation by 2050 and 20% reduction by 2080. All GCMs project significant increases in temperature, on average 2.00C by 2050 and 3.00C by 2080. The temperature sensitivity of the potential evapotranspiration varies between about 2% per 0C in the North of Cameroon and 2.4% per 0C in Central and South Cameroon. Thus, a 20C increase in temperature by 2050 will increase potential evapotranspiration, and thus crop water requirements, by 4 to 5%. The assessed precipitation elasticities of runoff vary between 1.9 for the Coastal Basins, 2.2 for the Sanaga and Congo Basins, 2.6 for the Benue basin upstream of Garoua and nearly 3 for the extreme North of Cameroon; similarly, the temperature sensitivity of runoff varies between -2.5% per 0C for the Coastal basins, -3% per 0C for the Sanaga and Congo Basins, -3.5% per 0C for the Benue basin upstream of Garoua, and -4% per 0C for the extreme North. Thus, runoff is the least sensitive to climate changes in the high rainfall regions and most vulnerable in the arid extreme North. This study does not project significant changes in average annual runoff by 2050 (-4%), or by 2080. However, the spread between individual projections is significant, with a standard deviation of 17% by 2050 and 24% by 2080. Projections of average runoff vary mostly between -35% and +30% by 2050 and between -40% and +40% by 2080. The conclusions of this study are primarily based on projected changes in annual precipitation, temperature, potential evapotranspiration and runoff. However, projected seasonal changes in climate parameters may differ significantly from projected annual changes, particularly for the dry season. Nonetheless, it has been demonstrated that projected annual climate changes, used in conjunction with monthly flow data, are adequate for the purpose of this Climate Risk Assessment. It is found that the new Lom Pangar reservoir (under construction) in the Sanaga Basin significantly improves the guaranteed capacity and dry season hydro-energy generation at the Nachtigal, Edea and Song Loulou hydropower stations, as well as at other future planned Run-of-River hydropower stations 101 on the Sanaga River. For the Nachtigal, Edea and Song Loulou plants and including the energy generated by Lom Pangar, the new reservoir enables a 46% increase of the dry season guaranteed capacity, and an 18% increase of the overall annual hydro-energy generation. Its significant contribution to hydro-energy generation in the Sanaga Basin is fairly constant for runoff variations in the range of -20% to +20%. Without climate change induced runoff changes, the EIRR varies between 14 and 15.5%, depending on the overall hydropower development scenario. The EIRR of the Lom Pangar project is not very sensitive to runoff changes, unless decreases in runoff exceed 20%. Under the 2050 climate conditions it is highly unlikely that the EIRR would change significantly (more than 5% change), while in the worst case the EIRR would be reduced with less than 10% of the base case value, i.e. from about 14.5% in the base case (including LP and Nachtigal) to 13% under severe climate change impacts. Overall, the Lom Pangar and Nachtigal projects are economically robust and climate resilient projects. On the contrary, by 2050 total long-term average hydro-energy generation at Lagdo dam in the Niger Basin could vary between -35% and +15% of the base case value (present hydrology). There is nearly 20% probability that by 2050 the annually generated hydro-energy would reduce with 20% or more, i.e. one out of 5 GCMs project such reductions. Thus, Lagdo dam hydro-energy generation may suffer a significant decrease due to climate change, and is less climate resilient than Lom Pangar and Nachtigal power stations. By 2050 and 2080 total long-term average hydro-energy generation at Njock and Mouila stations on Nyong River could vary between -10% and +5% of the base case value (present hydrology), and between -15% and +10% for Memve-Ele station on Ntem River. The probability that the annual hydro-energy generation at anyone of these three plants would reduce with 20% or more is negligible. Hence, climate change impacts on hydro-energy generation at these stations will be minor to moderate. From a hydrological perspective it may be worthwhile to re-evaluate the proposed installed capacity for the future Nachtigal power station, presently designed at 360 MW for a head of 50 m. Hydro-energy generation during the rainy season as well as daily flow modulation in conjunction with a small regulating reservoir upstream of Nachtigal could possibly benefit from an increase in installed capacity. It would also enable Cameroon to benefit from increased hydro-energy generation under conditions of increased flow due to future climate changes, which according to our analysis has a nearly 50% probability. For the present hydrology, Nachtigal’s optimal target turbine discharge has been assessed at 700 m3/s, enabling a guaranteed flow of 600 m3/s during the dry season. Climate change confronts decision makers with deep uncertainties, requiring Robust Decision Analysis to inform good decisions by identifying system vulnerabilities and assessing alternatives for ameliorating those vulnerabilities. These Decision Analysis Techniques and various approaches for economic project analysis under conditions of medium- to long-term climate change are outlined in World Bank (2010), and can be useful for a more in-depth analysis of e.g. the Nachtigal hydropower project, finalization of its installed capacity, and optimization of the target discharge for its turbine flow. 102 The economic performance of water and energy infrastructure projects must also be tested for worst case scenarios. It is recommended to use the following climate change induced runoff scenarios for analysis of the robustness of water and energy infrastructure projects in Cameroon, for the 2050 investment horizon: ï‚· Average future scenario under climate change: no significant changes in annual runoff ï‚· Medium dry scenario of climate change: decrease of annual runoff by 15% (25% probability) ï‚· Worst case/Dry scenario: decrease of annual runoff by 35% (3% probability) ï‚· Medium wet scenario of climate change: increase of annual runoff with 10% (20% exceedance probability) ï‚· Extreme wet scenario: increase of annual runoff with 30% (2% probability). The results of all available climate projections shown in this report demonstrate the large uncertainty in individual climate projections and the importance of taking all GCM projections into consideration, rather than building a Climate Risk Analysis on the projections of only a few selected GCMs. Multi-model ensembling is found to be the appropriate approach for climate change impact assessment. Anticipatory adaptation is most important for investments or decisions that are inflexible or irreversible, and have long lifetimes or lead times. Lom Pangar dam with multiple future downstream hydropower developments is an example of long-lived, climate-insensitive infrastructure investments. The methodology applied under this CRA study provides a powerful tool to identify system vulnerabilities and future climate change scenarios where the proposed developments might fail to meet their goals. These scenarios can then be used to address potential actions to address such climate vulnerabilities. This study has shown significant gaps in hydrometeorological data available for Cameroon, which particularly hampered the Climate Risk Assessments for basins other than the Sanaga Basin. No runoff data were available beyond 1980 for these other river basins. High priority should be given to the monitoring of the present status of the country’s water resources, current runoff trends, minimum flows and similar performance metrics. Moreover, consistent records of long-term recorded rainfall could not be located for this study, other than for a few stations in the Sanaga basin. Whereas the CRU TS 3.10 gridded precipitation, temperature and potential evapotranspiration data sets for the period 1901 – 2009 provided a useful alternative, actual precipitation (and runoff) data are required for more enhanced and more accurate future analyses of rainfall – runoff relationships and hydrological modeling. Therefore, it is recommended to revive, upgrade and possibly expand the previously existing hydrometeorological networks in Cameroon (see also MINEE and GWP, 2009); for Climate Risk Assessments the priority would be for the collection of runoff data. High importance should also be given to the preparation of a comprehensive and nation-wide data base of already available (historical) hydrometeorological data. It is imperative to develop and operationalize a comprehensive Hydrological or Water Information System (HIS/WIS) for the country. 103 It is important to emphasize that in this report climate risks are calculated based on the long term (i.e. 30 years) shift in mean precipitation and temperature values; they do not account for inter-annual or decadal variability. We do not consider this a serious limitation since the hydrological baseline period (1972 - 2003) used for these studies comprises arguably a period with relatively low flows, after an abrupt downward shift in runoff around 1970. It is noteworthy that historically the region may have experienced runoff changes greater than the projected levels of change for the 21st century. Thus, the historical experience provides an analogue for dealing with future climate; for water managers and farmers who do not know what to expect in the upcoming rainy season, managing the impacts of the present intra-seasonal and inter-annual climate variability (and particular its potential impacts on rainfed agriculture) would be the priority to start with. Managing the near-term climate variability also has the potential to better prepare the country for dealing with long term-climate change impacts. Therefore, the use of seasonal hydrologic forecasts for reservoir operations can be an important adaptation opportunity. It may be noted that many key factors relevant to water resources management remain as yet unexplored, e.g. in the case of irrigated agriculture factors related to how climate change impacts crop evapotranspiration, plant development and yield production, and factors related to rainfed agriculture, such as the onset of the rainy season, and length and frequency of dry spells. GCM projections offer little guidance for the assessment of such impacts, which for the time being may present the threat of greatest concern to basin development objectives. Finally, the Rapid Assessment Method for CRA adopted under this study as a reliable tool under conditions of limited technical capacities and data deficiencies, is summarized as follows: ï‚· Derive P and T projections from the Climate Wizard and/or Climate Change Knowledge Portal (WB); determine the (normal) probability distributions of the relative (in %) long-term changes in annual, seasonal and/or monthly P and T and estimate the means and Coefficients of Variation of projected changes ï‚· Estimate the climate elasticity and sensitivity of runoff (εP, ST) based on historical runoff (Q) data, and hydrometeorological or gridded data sets of Precipitation (P) and Temperature (T), and runoff coefficients (Q/P) ï‚· Derive the probability distribution function of future runoff changes from the projected Precipitation and Temperature changes (dP/P and dT), as follows: o E{dQ/Q0} = εP E{dP/P0} + ST E{dT}; shift in mean value (%) o Cv(dQ/Q0) = εP Cv(dP/P0); coefficient of variation of shift (the variability of T projections can be ignored compared to the variability P projections ï‚· Estimate the system response to changes in runoff, i.e. the runoff elasticity (εQ) of key Performance Indicators (PI) for (sub-)basins such as hydro-energy generation, flooding and irrigation potential; climate sensitivity analysis can be performed through water resources system modeling (e.g. WEAP model), regression analysis, energy simulation and reservoir simulation models (e.g. HEC-RAS). 104 ï‚· Estimate the mean and standard deviation of the projected future changes in Performance Indicators (PI) as follows: o E{dPI/PI0} = εQ E{dQ/Q0}; εQ may be a non-linear function of dQ/Q0 o Cv(dPI/PI0) = εQ Cv(dQ/Q0) ï‚· Calculate probabilities of non-exceedance for specified risk levels The system for transforming ΔP&ΔT ΔQ ΔPI is treated as a quasi-non-linear system, which is justified in view of the huge variability in precipitation projections. Since projections generally concern relative changes (in %) in long-term average climate conditions (typically averaged over 30 years), the adoption of the normal distribution to describe probabilities of future changes in climate and runoff and in basin performance indicators is justified. 105 References Adam, J. C. and D.P Lettenmaier, 2003: Adjustment of global gridded precipitation for systematic bias, Journal of Geophysics Research, 108, 1–14. Alavian, V., H. Qaddumi, E. Dickson, S. Diez, A. Danilenko, R. Hirji, G. Puz, C. Pizarro, M. Jacobsen and B. Blankespoor, 2009: Water and Climate Change: Understanding the Risks and Making Climate- Smart Investment Decisions, World Bank. Allen, R.G., M. Smith, L.S. Pereira and A. Perrier, 1994: An update for the calculation of reference evapotranspiration, ICID Bulletin 43, 35-92. Allen, R.G., L. S. Pereira, D. Raes and M. Smith, 1998: Crop evapotranspiration - Guidelines for computing crop water requirements, FAO Irrigation and Drainage, Paper 56, Rome (FAO). Arora, V.K., 2002: The use of the aridity index to assess climate change effect on annual runoff, Journal of Hydrology, 265, p. 164–177. Beaulieu, C., J. Chen, and J.L. Sarmiento, 2012: Change-point analysis as a tool to detect abrupt climate variations, Phil. Trans. R. Soc. A 2012 370, 1228-1249. doi: 10.1098/rsta.2011.0383. Blaney, H.F. and W.D. Criddle, 1950: Determining Water Requirements in Irrigated Areas from Climatological Irrigation Data; Technical Paper No. 96, US Department of Agriculture, Soil Conservation Service, Washington, D.C. Boko, M., I. Niang, A. Nyong, C. Vogel, A. Githeko, M. Medany, B. Osman-Elasha, R. Tabo and P. Yanda, 2007: Africa. Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F. Canziani, J.P. Palutikof, P.J. van der Linden and C.E. Hanson, Eds., Cambridge University Press, Cambridge UK, 433-467. BRLi and DHI, 2007: Establishment of a Water Management Model for the Niger River Basin, Final Report. Brown, C., Y. Ghile, M. Laverty and K. Li, 2012: Decision scaling: linking bottom-up vulnerability analysis with climate projections in the water sector, Water Resources Research, 48(9), 1-12. Budyko, M.I., 1948: Evaporation under natural conditions, Gidrometeorizdat, Leningrad (English translation by IPST, Jerusalem). Chiew, F.H.S, 2006: Estimation of rainfall elasticity of streamflow in Australia, Hydrological Sciences Journal 51:4, 613-625 (Chiew-2006). Chiew, F.H.S., M.C. Peel, T.A. McMahon and L.W. Siriwardena, 2006: Precipitation elasticity of streamflow in catchments across the world, Climate Variability and Change, Proceedings of the 5th FRIEND World Conference, Havana, Cuba, 26 Nov - 1 Dec 2006; IAHS Publication 308; CSIRO and University of Melbourne, Australia. CLIMWAT: CLIMWAT is a climatic database to be used in combination with CROPWAT for the calculation of crop water requirements, irrigation supply and irrigation scheduling for various crops for a 106 range of climatological stations worldwide; Water Development and Management Unit and the Climate Change and Bio-energy Unit of FAO. Cook, K and E. Vizy, 2006: Coupled Model Simulations of the West African Monsoon System; Twentieth- and Twenty-First-Century Simulations, J. Clim 19, 3681-3703. Deksyos T. and T. Abebe, July 2006: Assessing the impact of climate change on the water resources of the Lake Tana sub-basin using the Watbal model, CEEPA Discussion Paper No. 30, Special Series on Climate Change and Agriculture in Africa; Centre for Environmental Economics and Policy in Africa, University of Pretoria, Pretoria, South Africa. Dzana, J.G., J.R. N. Ngoupayou and P. Tchawa, 2011: The Sanaga discharge at the Edea catchment outlet (Cameroon): An example of hydrologic responses of a tropical rain-fed river system to changes in precipitation and groundwater inputs and to flow regulation, River Research and Applications 27: 754 – 771. Droogers, P. and R. Allen, 2002: Estimating reference evapotranspiration under inaccurate data conditions, Irrigation and Drainage Systems 16 (1): 33-45. Ekström M., P.D. Jones, H.J. Fowler, G. Lenderink, T.A. Buishand and D. Conway, 2007: Regional climate model data used within the SWURVE project 1: Projected changes in seasonal patterns and estimation of PET, Hydrol. Earth Syst. Sci., 11(3), 1069-1083. Gedney, N., P.M. Cox, R.A. Betts, O. Boucher, C. Huntingford, and P.A. Stott, P. A., 2006: Detection of a direct carbon dioxide effect in continental river runoff records, Nature, 439, 835–838. Ghile Y.B., M.Ãœ. Taner, C. Brown, Johan Grijsen, 2013: Bottom-up Climate Risk Assessment of Infrastructure Investment in the Niger River Basin, Climate Change, DOI 10.1007/s10584-013- 1008-9. Giannini, A., M. Biasutti, I. Held and A.H. Sobel, 2008: A global perspective on African Climate, Climate Change, 90(4), 359-383. Girvetz, E.H, C. Zganjar, G. T. Raber, E. P. Maurer, P. Kareiva and J.J. Lawler, 2009: Applied Climate- Change Analysis: The Climate Wizard Tool, PLoS ONE 4(12): e8320; doi: 10.1371/journal.pone. 0008320 (CW publication) Gleick, E. H., 1986: Methods for Evaluating the Regional Hydrological Impact of Global Climatic Changes, J. of Hydrology 88, 97-116. Gleick, E. H., 1987: Regional Hydrologic Consequences of Increases of Atmospheric CO2 and Other Trace Gases, Climatic Change 110, 137-161. Grijsen J.G., C. Brown, A. Tarhule, Y.B. Ghile and Ãœ. Taner, 2013: Climate Risk Assessment for Water Resources Development in the Niger River Basin, HydroPredict’12, Special Session 3: Choosing Models for Resilient Water Resources Management, Water Partnership Program (WPP)/TWIWA, World Bank, Vienna September 2012 (HydroPredict Vienna); also published in: Climate Risk Assessment for Water Resources Development in the Niger River Basin: Part I (Chapter 2): Context and Climate Projections; Part II (Chapter 3): Runoff elasticity and 107 probabilistic analysis; in: Climate Variability - Regional and Thematic Patterns, Published by InTech, Croatia (intechopen.com/books/). Grijsen, J.G. and C. Brown, 2013: Climate elasticity of runoff and climate change impacts in the Niger River Basin; Niger River Basin Climate Risk Assessment, a joint initiative of the Niger Basin Authority (NBA) and the World Bank (not yet disclosed). Hamon, W.R., 1961: Estimating potential evapotranspiration, Journal of Hydraulics Division, Proceedings of the American Society of Civil Engineers 871: 107–120. Hargreaves, G.H. and Z.A. Samni, 1982: Estimation of potential evapotranspiration: Journal of Irrigation and Drainage Division, Proceedings of the American Society of Civil Engineers 108: 223–230. Hargreaves, G.H. and Z.A. Samni, 1985: Reference crop evapotranspiration from temperature; Transactions of the American Society of Agricultural Engineers. Hargreaves, G. H., 1994: Defining and using reference evapotranspiration, Journal of Irrigation and Drainage, Engineering, ASCE 120(6): 1132-1139. Hargreaves, G. H. and R.G. Allen, 2003: History and evaluation of Hargreaves evapotranspiration equation, Journal of Irrigation and Drainage Engineering, ASCE 129(1): 53-63. Harris I., P.D. Jones, T.J. Osborn and D.H. Lister, 2013: Updated high-resolution grids of monthly climatic observations – the CRU TS3.10 Dataset; International Journal of Climatology, Doi: 10.1002/joc.3711, Harris et al, 2013 (in press). ISL- Oreade-Breche and Sogreah, 2005a : ETUDE ENVIRONNEMENTALE DU BARRAGE DE LOM PANGAR, THEME 1 : ETUDE DES ALTERNATIVES, Rapport Final, Révision et Actualisation. ISL- Oreade-Breche and Sogreah, 2005b : ETUDE ENVIRONNEMENTALE DU BARRAGE DE LOM PANGAR, THEME 14 : Impact hydraulique a l’aval, Rapport Final. ISL- Oreade-Breche and Sogreah, June 2007 : ETUDE D’IMPACT SUR L’ENVIRONNEMENT DU BARRAGE DE LOM PANGAR, Optimisation de la capacité de la retenue, Rapport Final. Karl, T.R. and W.E. Riebsame, 1989: The impact of decadal fluctuations in mean precipitation and temperature on runoff: A sensitivity study over the United States, Climatic Change 15: 423-447. Kpoumie A., 2010: Impact des changements climatiques et des activités anthropiques sur les ressources en eau, les activités agropastorales et la production hydroélectrique du bassin versant de la Sanaga- Cameroun, Rapport Final ACCFP 2009/2010. Krakauer N.Y. and I. Fung, 2008: Mapping and attribution of change in streamflow in the coterminous United States, Hydrology and Earth System Sciences, 12, 1111–1120. Kuhnel, V., J.C.I. Dooge, P.J. O’Kane, and R.J. Romanowicz, 1991: Partial analysis applied to scale problems in surface moisture fluxes, Surveys in Geophysics, Vol. 12, 221–247. Kundzewicz, Z.W., L.J. Mata, N.W. Arnell, P. Döll, P. Kabat, B. Jiménez, K.A. Miller, T. Oki, Z. Sen and I.A. Shiklomanov, 2007: Freshwater resources and their management. Climate Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group II (Chapter 3) to the 108 Fourth Assessment Report of the Intergovernmental Panel on Climate Change, M.L. Parry, O.F. Canziani, J.P. Palutikof,P.J. van der Linden and C.E. Hanson, Eds., Cambridge University Press, Cambridge, UK, 173-210. Labat, D., Y. Godderis, J.L. Probst, and J.L. Guyot, 2004: Evidence for global runoff increase related to climate warming, Advances in Water Resources, 27, 631–642, 2004. Langbein, W. B., 1949: Annual Runoff in the United States, U.S. Geological Survey Circular 5. U.S. Dept. of the Interior, Washington, DC (reprinted, 1959). 1989: Last 80 years in relation to world temperature change, American Journal of Science, 289, 267–285 Legates, D., H. Lins, H., and G. McCabe, 2005: Comments on “Evidence for global runoff increase related to climate warmingâ€? by Labat et al., Advances in Water Resources, 28, 1310–1315, 2005. Lempert, R. and N. Kalra, 2010-2011: “Managing Climate Risks in Developing Countries with Robust Decision Makingâ€?, World Resources Report, Washington DC. Liénou G., G. Mahé, J. E. Paturel, E. Servat, D. Sighomnou, L. Sigha-Nkamdjou, 2008 : Impact de la variabilité climatique en zone équatoriale : exemple de modification de cycle hydrologique des rivières du sud-Cameroun, 13th IWRA World Water Congress, Montpellier, France (Lienou 2008). Maurer, E.P., J.C. Adam, and A.W. Wood, 2009: Climate Model based consensus on the hydrologic impacts of climate change to the Rio Lempa basin of Central America, Hydrology and Earth System Sciences13, 183-194. Meehl, G. A., C. Covey, T. Delworth, M. Latif, B. McAvaney, J. F. B. Mitchell, R. J. Stouffer, and K. E. Taylor, 2007: The WCRP CMIP3 multi-model dataset: A new era in climate change research, Bulletin of the American Meteorological Society, 88, 1383-1394. MINEE and GWP, 2009: Plan d’Action Nationale de Gestion Intégrée des Ressources en Eau (PANGIRE): Volume 1: État des lieux du secteur - Connaissance et usages des ressources en eau Volume 2: Eau et environnement Volume 3: Cadre financier, économique et social Volume 4 : Cadre législatif, réglementaire, institutionnel et ressources humaines MINEP and UNDP, 2012: Évaluation des risques, de la vulnérabilité et adaptations aux changements climatiques Cameroun, Final report. Mitchell, T.D., T.R. Carter, P.D. Jones, M. Hulme and M. New, 2003: A comprehensive set of high- resolution grids of monthly climate for Europe and the globe: the observed record (1901-2000) and 16 scenarios (2001-2100); Journal of Climate. Mitchell T. D. and P. D. Jones, 2005: An improved method of constructing a database of monthly climate observations and associated high-resolution grids; International Journal of Climatology 25,693- 109 712. http://cru.csi.cgiar.org/PDF/mitchelljones.pdfMonteith J.L., 1965: Radiation and crops. Experimental Agricultural Review 1: 241–251. Monteith J.L., 1965: Radiation and crops. Experimental Agricultural Review 1: 241–251. Munang M. and S.N. Ayongue, 2010: Climate change impacts on the hydroelectricity potentials of the Sanaga basin: past and future trends, M Sc thesis, University of Buea. Nakicenovic and Swart, 2000: Special Report on Emissions Scenarios, Cambridge University Press, Cambridge, UK; available at http://www.grida.no/climate/ipcc/emission/023.htm. Parry, M.L., O.F. Canziani, J.P. Palutikof, P.J. van der Linden and C.E. Hanson (eds), 2007: Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change; Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Olivry J.C., 1986 : Fleuves et rivières du Cameroun, Monographies hydrologiques, MESRES/ORSTOM, n° 9, 733 p. Pike, J. G., 1964: The estimation of annual runoff from meteorological data in a tropical climate. Journal of Hydrology, Amsterdam, 2, 116–123. Probst, J. L. and Y. Tardy, 1987: Long range streamflow and world continental runoff fluctuations since the beginning of this century. Journal of Hydrology., 94, p. 289-311. Probst, J.L. and Y. Tardy, 1989: Global runoff fluctuations during the last 80 years in relation to world temperature change. American Journal of Science, 289, p. 267-285. Prudhomme C., Wilby R.L., Crooks S., Kay L, Reynard N.S., 2010: Scenario-neutral approach to climate change impact studies: Application to flood risk. Journal of Hydrology 390:198–209. doi:10.1016/j.jhydrol.2010.06.043. Revelle, R. R. and R.E. Waggoner, 1983: Effects of a Carbon Dioxide-Induced Climatic Change on Water Supplies in the Western United States', in Changing Climate, Report of the Carbon Dioxide Assessment Committee, National Academy Press, Washington, DC, pp. 419-431. Runoff fluctuations since the beginning of this century, J. Hydrol., 94, 289–311, doi: 10.1016/0022-1694(87)90057-6. Risbey, J. S., and Entekhabi, D, 1996: Observed Sacramento basin streamflow response to precipitation and temperature changes and its relevance to climate impact studies.’’ Journal of Hydrology, p. 184, 209–223.Schaake, J. C., 1990: From climate to flow, in Climate Change and U.S. Water Resources, edited by P. E. Waggoner, chap. 8, pp. 177–206, John Wiley & Sons Inc., New York, 1990. Rodionov, S. N., 2004: A sequential algorithm for testing climate regime shifts, Geophys. Research. Sankarasubramanian, A., R.M. Vogel and J.F. Limbrunner, 2001: Climate elasticity of streamflow in the United States, Water Resources Research, Vol. 37, No. 6, pp 1771-1781. Schaake, J. C., 1990: From climate to flow, in Climate Change and U.S. Water Resources, edited by P. E. Waggoner, chap. 8, pp. 177–206, John Wiley & Sons Inc., New York, 1990. 110 Segui, P.Q., A. Ribes, E. Martin, F. Habtes, and J. Boe, 2010: Comparison of three downscaling methods in simulating the impact of climate change on the hydrology of Mediterranean basins, Journal of Hydrology 383, 111-124. Shuttleworth, W.J., 1993: Chapter 4 “Evaporationâ€? of the Handbook of Hydrology by D.R. Maidment. Sighomnou, D. 2004 : Analyse et redéfinition des régimes climatiques et hydrologiques du Cameroun : perspectives d’évolution des ressources en eau. Thèse d’Etat en Sciences de l’Eau. Sighomnou, D., L. Sigha Nkamdjou, G. Lienou, A. Dezetter, G. Mahe, E. Servat, J.E. Paturel, J.C. Olivry, F. Tchoua, G.E. Ekodeck, 2007: Impact des fluctuations climatiques sur le régime des écoulements du fleuve Sanaga au Cameroun, prospectives pour le XXIème siècle. Document technique en hydrologie n° 80; UNESCO, Paris/UMR 5569, Hydro Sciences Montpellier, pp 173 – 181. SNC Lavalin International, February 2007: Strategic/Sectoral, Social and Environmental Assessment of Power Development Options in the Nile Equatorial Lakes Region, Final Report, Chapter 12 and Appendix K. Strzepek K. M. and A. McCluskey, July 2006: District level hydroclimatic time series and scenario analyses to assess the impacts of climate change on regional water resources and agriculture in Africa; CEEPA Discussion Paper No. 13, Special Series on Climate Change and Agriculture in Africa. Strzepek, K. M., and C. W. Fant IV, 2010: Water and Climate Change: Modeling the Impact of Climate Change on Hydrology and Water Availability, University of Colorado and Massachusetts Institute of Technology. Strzepek, K.M., A. McCluskey, B. Boehlert, M. Jacobsen, and C.W. Fant IV, 2011: Climate Variability and Change: A Basin Scale Indicator Approach to Understanding the Risk to Water Resources Development and Management, Water Anchor of the World Bank Group, series Water Papers: WB Water Paper CCK Portal. Suchel, J.B., 1987: Les climats du Cameroun, Thèse Doctoral d’État, Université Bordeaux III, France. Tarhule, A. and J.G. Grijsen, 2013: Water Resources Profile and Historical Climate Variability; the NRB Climate Risk Assessment, a joint initiative of the Niger Basin Authority (NBA) and the World Bank (not yet disclosed). Tarhule, A., J. T. Zume and J.G. Grijsen, 2013: Exploring temporal hydroclimatic variability in the Niger Basin (1901-2006) using observed and reanalysis data, International Journal of Climatology (in revision) TECSULT and SOGREAH, April 2006 : Aménagement hydroélectrique de Nachtigal - Productible de l’Aménagement, Final Report. Teng, J., F. H. S. Chiew, J. Vaze, S. Marvanek and D.G.C. Kirono, 2012: Estimation of climate change impact on mean annual runoff across continental Australia using Budyko and Fu equations and hydrological models, American Meteorological Society, AMS Journals online, Volume 13, Issue 3. Turc, L., 1954 : Le Bilan d’eau des sols. Relation entre les précipitation, l’évaporation et l’écoulement. Annales Agronomique, 5, 491–595. 111 UCI, 2011: Impacts du changement climatique global sur les ressources en eau du Maroc ; Phase I: Descente d’échelle des résultats des Modèles Climatiques Globaux au Maroc, Rapport Final, Centre d’Hydrométéorologie et de Télédétection, Université de Californie d’Irvine, CA, USA. Vano J., T. Das and D. Lettenmaier, 2012: Hydrologic sensitivities of Colorado River runoff to changes in precipitation and temperature, Journal of Hydrometeorology. Vogel, R.M., I. Wilson and C. Daly, May/June 1999: Regional regression models of annual stream flow for the United States, Journal of Irrigation and Drainage Engineering. Wigley, T. M. L. and E.D. Jones, 1985: Influences of Precipitation Changes and Direct CO 2 Effects on Streamflow, Nature 314, 140-152. Willmott, Cort J., and Matsuura Kenji, 2001: Terrestrial Water Budget Data Archive: Monthly Time Series (1950-1999); http://climate.geog.udel.edu/~climate/html_pages/README.wb_ts2.html. Wood, A. W., E. P. Maurer, A. Kumar and D.P. Lettenmaier, 2002: Long-range experimental hydrologic forecasting for the eastern United States, Journal of Geophysics Research 107(D20), 4429. Wood, A.W., L.R. Leung, V. Sridhar and D.P. Lettenmaier, 2004: Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs, Climate Change 62, 189-216. World Bank, 2010: Economic Evaluation of Climate Change Adaptation Projects; approaches for the Agricultural Sector and beyond; published in Development and Climate Change Series World Bank, 2012a: Project Appraisal Document on the Lom Pangar Hydropower Project. World Bank, 2012b: Project Appraisal Document on the 1st part of the 2nd phase of the Niger Basin Water Resources Development and Sustainable Ecosystems Management Project (WRD-SEM APL 2A). Xu C.Y. and V.P. Singh, 2001: Evaluation and generalization of temperature-based methods for calculating evaporation, Hydrological Processes 15, 305–319. Yates D., 1996: WatBal - An integrated water balance model for climate impact assessment of river basin runoff, International Journal of Water Resources Development 12(2): 121–139. Zomer R.J., A. Trabucco, O. van Straaten and D.A. Bossio, 2006: Carbon, Land and Water: A Global Analysis of the Hydrologic Dimensions of Climate Change Mitigation through Afforestation/Reforestation, Colombo, Sri Lanka: International Water Management Institute Research Paper 101 (Zomer-2006). Zomer R.J., A. Trabucco, D.A. Bossio, O. van Straaten and L.V. Verschot, 2008: Climate Change Mitigation: A spatial analysis of global land suitability for clean development mechanism afforestation and reforestation; Agriculture, Ecosystems and Environment, 126: 67-80 (Zomer-2008). 112 Annex 1: Terms of Reference Hydrologist/water resources management specialist for TFESSD financed activity “Understanding the Impact of Climate Change on Hydropower: the case of Cameroonâ€? 1. Background AFTEG has obtained financing from the Trust Fund for Environmentally and Socially Sustainable Development (TFESSD) for an activity called “Understanding the Impact of Climate Change on Hydropower: the Case of Cameroonâ€?. The development objective of the activity is: (i) to develop tools for the assessment of climate change impacts on the operation of hydraulic infrastructure such as regulating dams and hydropower plants in the Sanaga river basin, and (ii) to take steps towards a climate resilient institutional framework for water resources management in Cameroon. The activity includes the following components: i. Develop suitable climate change scenarios for the Sanaga basin and support the Electricity Development Corporation (EDC) to develop a reliable hydrological model for the Sanaga River basin in Cameroon. Derive impacts on potential generation capacity in the context of changing hydrology caused by climate change; ii. Assess the impact of climate change on the future operation of the Lom Pangar dam and the other three regulating dam in the Sanaga basin and support the establishment of an operational regime of hydrological infrastructures in the Sanaga River basin in a consultative manner with water users in the basin and taking into account equitable sharing of resources between users and environmental flows. iii. Assess future impacts of climate change on water resources management in Cameroon. This terms of reference is related to activity iii) above. The objective of this activity is to carry out a preliminary Climate Risk Assessment to better understand the dynamics of future climate in the five river basins in Cameroon and assess the potential impacts on water resources, hydro-energy generation, navigation, agriculture and the environment. This assessment will provide an analytical base for increased dialogue on climate variability and change and on integrated management of water resources in Cameroon. The purpose of this assessment is to add value to the work that is currently being undertaken in Cameroon on the topic of water and climate change including the work being undertaken under the Water, Climate and Development Program for Africa. The assessment will also identify information and knowledge gaps and priorities for future studies/activities. The final output will be a scoping report that identifies the potential impacts of future climate on water resources, energy, navigation, agriculture and the environment in the five river basins in Cameroon. 113 2. Objective The overall objective of this consultancy is to provide hydrology/water resources management expertise and advice with regards to a preliminary Climate Risk Assessment for the five river basins in Cameroon. 3. Description of the specific tasks of the Consultant The specific activities of the hydrologist/water resources management specialist will include: 1. Review existing reports and other materials, as requested by the task team leader; 2. Conduct a literature/data review to map the past and current analytical work and the primary data on water resources availability, present water uses and historic climate variability in each of the five basins in Cameroon. This sub-component will also analyze projections of future water uses in the basins, based on sectoral plans and existing infrastructure plans; 3. Carry out a three-step methodology for the climate risk assessment of the five river basins in Cameroon as described in the concept note “Assess future impacts of climate change on water resources management in Camerounâ€?. The three step methodology includes: step 1: Problem Definition: Identification and thresholds and climate hazards, step 2: Hazard Discovery: Assessing the system response to changes in climate and step 3: Climate Informed Risks: Estimating likelihood of climate conditions and hazards; 4. Draft a climate risk assessment scoping report, including inputs from other team members if available; 5. Prepare presentations for discussions and stakeholder workshops, as necessary; 6. Other related tasks, as requested by the task team. 114 Annex 2: Principles of climate change projections General Circulation Models (GCMs) are the tools often used for future climate studies and they provide substantial information about the possible climate changes in the years ahead. Many GCMs are available research institutions around the world and produce different projections about what may occur in the future. For West and Central Africa, the ensemble of all available climate change projections yields strong agreement on the direction of change in temperature but is inconclusive with respect to precipitation. Figure 1 shows ensembles of rainfall projections for West and East; for West Africa nearly as many GCMs project increases in precipitation as decreases, while for East Africa on average a slight increase in precipitation is projected. Given the large differences in projections by individual GCMs, it is important to look at the range of projections from multi-model simulations, rather than just relying on a single run chosen from many options. Their mutual independence provides credibility to results upon which the models agree. That is, if independently developed models agree on a particular climate change, the “consensusâ€? is more likely due to a true climate signal than due to coincidental model errors. Fig. 1: Range of projected relative changes in precipitation from the baseline for an ensemble of GCMs. st Note: Regional precipitation averages as simulated by multiple GCMs for the A1B emission scenario for the 21 century. Each grey line represents one model simulation and the bold black line represents the multi-model mean. (Source: Giannini et al., 2008). 115 The IPCC reports that no single model can be regarded as “bestâ€? and this means that multi-model simulations with different scenarios must be used to capture the full envelope of future climate uncertainty. Screening of a GCM that does not demonstrate any skill for a particular region might also be recommended to reduce the biases in the multi ensembles approach. However, how a selection can be made is a challenge by itself. More importantly, models which performed well for the present climate conditions may not necessarily give the same performance for future situations. Cook and Vizy (2006), for example, selected 3 GCMs out of 18 for West Africa climate change studies, based on how well they reproduced the features of the West African monsoon system. The three models, however, displayed significantly different projections for the 21th century, meaning at least one and possibly two of the three “bestâ€? models are wrong. Similarly, UCI (2011) selected three GCMs which – after downscaling through a Regional Climate Model (RCM) – best reproduced the historic climate over the Atlas Mountains in the Oum-Er-Rbia river basin in Morocco. However, one of these 3 models projected an extremely dry future, while another projected only a small decrease in precipitation by 2050 and beyond. In such circumstances, multi-model ensembling is found to be the appropriate approach for assessing the impacts of climate change on the water resources in Cameroon. This helps reduce the effects of model errors in one particular model and the natural variability in any particular run. Fig. 2: Greenhouse gas (GHG) emissions and estimated global surface warming for SRES scenarios th (Source: IPCC 4 Assessment report AR4, 2007; the grey bars at right indicate the best estimate and the likely range assessed for the six emissions marker scenarios) Figure 2 above shows Greenhouse gas (GHG) emissions and estimated global surface warming during the 21st century for a range of SRES emission scenarios (Nakicenovic and Swart, 2000). Overall Global Circulation Models (GCM) project for most emission scenarios similar temperature increases for 2050, while temperature projections divert towards 2100. Therefore, we use in this report mostly GCM results for the medium CO2 A1B emission scenario, which represents a relatively high economic growth worldwide and a relatively low growth in population. In view of the economic life time of currently planned HP interventions and the huge uncertainties embedded in available long-term climate projections, this report focuses mainly on potential climate and runoff changes by 2050. 116 Because of their enormous mathematical complexity, GCMs generally operate at relatively coarse spatial scales, and their skill is limited to larger spatial areas (e.g. sub-continental scale). Downscaling GCM outputs to a finer scale is a common practice for local impact studies, achieved either by dynamical downscaling through applying a Regional Climate Model (RCM) which is calibrated for historical climate conditions, or by statistical downscaling through commonly used methods such as the delta method and quantile mapping. However, the uncertainty in downscaling techniques often reduces the skill observed over larger areas. GCMs are also often subjected to large systematic errors that can seriously distort the quality of climate projections. Biases need to be removed before the projections are used as input, and the basis for bias correction is to shift GCM output to a reasonable range, based on observed conditions. Typically statistical downscaling involves matching monthly or seasonal average GCM output with observed averages. The climate projections provided by the Climate Wizard (climatewizard) and Climate Change Knowledge Portal (CCKP; Climate Portal; CKP-CW ) are bias corrected through quantile mapping. The quantile mapping approach, as described in Wood et al. (2002) and Wood et al. (2004), utilizes a mapping between the observed cumulative density function (CDF) of precipitation and temperature and that produced by GCMs. It accomplishes downscaling and bias correction in the same step. Transfer functions are calculated based on the assumption that the CDF of GCM output is the same as that of observed data. CDFs are constructed for each month of the year for all available individual GCM projections as well as for the available precipitation and temperature observations. The quantile mapping between GCM and observed values is generally trained for the period from 1901-1999 and the transfer functions are then used during the future projections for the entire GCM monthly time-series (2001-2099). Quantile mapping has been widely used (e.g. Segui et al., 2010) and it is conceptually a sound method as it incorporates the large-scale spatial and temporal variability information from the GCM simulations. Figure 3 demonstrates the bias correction using quantile mapping. Generally GCMs do a better job in simulating temperature than precipitation. The skill of GCM projections of temperature and precipitation generally decreases along with the spatial and temporal scales under consideration; the models have relatively more skill over larger spatial areas. Given the scale – skill relationship and basin homogeneity (Chapter 3), we assessed the climate risks over each of the river basins in Cameroon as the most skillful and useful scale of GCM projections. 150 Figure 3: Bias correction using quantile mapping Precipitation (mm) 100 Climate change may also alter the frequency and timing 50 of extreme events such as floods and droughts, but the 0 present GCMs are less successful at predicting such 0 0.2 0.4 0.6 0.8 1 features as their skill is generally limited to predicting Probability of non-exceedence (%) shifts in mean values of precipitation and temperature. Uncorrected Corrected 117 Annex 3: Climate Risk Assessment for water resources development in the Niger Basin24 1. The Niger Basin The Niger River Basin (NRB) is shared among nine riparian countries, Guinea, Mali, Cote d’Ivoire, Burkina Faso, Niger, Benin, Nigeria, Cameroon and Chad (Figure 1). The Niger River, with a total length of 4,100 km, is the third longest river in Africa. The Basin extends 3,000 km from east to west and 2,000 km from south to north; the total basin area is 2,200,000 km2, of which 1,400,000 km² is hydrologically active. In 2005 an estimated 92 million people lived in the basin, projected to rise to 155 million by 2025 and to 250 million by 2050. Forty-five percent of the people in the basin depend on surface water for drinking and other needs and 65% of the people depend on rainfed agriculture, which is the major source of income in the NRB; crop production comprises 25 - 35% of GDP. The hydro-energy potential in the NRB is estimated at 30,000 GWh/yr, of which only 6,000 GWh/yr has been developed. The Inner Delta of the Niger is a prominent wetland in Africa. The key development challenges in the NRB are rapid deterioration of land and water resources, poor performance and inadequate investments in water infrastructure, food insecurity, poverty and high rates of population growth and urbanization. The region has a history of marked climate variability with significant societal and environmental impacts. There is concern that in the 21st century climate change may exacerbate and accelerate the existing trends in poverty, underdevelopment and environmental degradation. Figure 1: The Niger Basin – Distribution of precipitation (1948 – 2002) The Niger Basin Authority (NBA) and the World Bank have undertaken a joint initiative to assess the climate risks associated with implementation of the investment components of the Sustainable Development Action Plan (SDAP) and the Investment Program (IP), respectively adopted in July 2007 by the Council of Ministers and in April 2008 by the Heads of States of the nine member countries of the NBA. The SDAP and IP involve the investment of about US$8 billion over at least 20 years, and aim to (i) provide new social and economic opportunities for the people living in the Niger Basin, (ii) contribute to capacity building in integrated water resources management, and (iii) protect natural resources and ecosystems. The SDAP features a comprehensive infrastructure and investment plan, including (i) the construction of three large new dams with hydro-energy generation; (ii) the construction of new hydro-agriculture infrastructure; (iii) support to economic development other than through large-scale infrastructure, including agriculture, 24 Source: Brown et al. 2012; Ghile et al., 2013; Grijsen et al., 2013. 118 fisheries, livestock, and tourism; and (iv) ecosystem conservation including the protection of biodiversity, erosion control, sand/silt control, and the prevention of pollution. The Niger Basin is prone to a large climatic variability at different time scales i.e. at inter-annual, multi- years and decadal scales, as well as spatial variability from North to South. Precipitation projections for West Africa vary widely, such that GCMs even lack agreement on the direction of future changes in precipitation (Figure 2). By the 2050 investment horizon most GCMs project changes in annual average precipitation between -10% and +10%, on average only an insignificant increase. All GCMs project significant increases in temperature, about 2 0C by 2050 (Figure 3). What seems clear is that it will be hotter and potential evapotranspiration and crop water demands will increase with about 5% by 2050. The projected trends in precipitation are small compared to the historical inter-annual rainfall variability. 30% 20% 10% 0% 1900 1925 1950 1975 2000 -10% -20% -30% Figure 2: Ensemble of regional climate projections for West Africa. Note: Regional precipitation averages as st simulated by multiple GCMs for the A1B emission scenario for the 21 century. Each grey line represents one model th simulation and the bold black line represents the multi-model mean. The bold blue line reflects the historical 20 century rainfall variability. Source: Giannini et al., 2008. This lack of reliability of climate change models for West Africa, particularly the unreliable precipitation projections for the 2050 SDAP investment horizon, favored the adoption of the bottom-up decision scaling approach. Key in this Climate Risk Assessment (CRA) was that attention focused first on the Niger Basin’s water resources system through extensive water resources system modeling, to understand how the system would respond to future climate changes, followed by an assessment of the response of runoff to climate change and a risk assessment on the basis of a large number of GCM projections. 119 Projected rainfall changes in NRB Average temperature projections in NRB 20 31 Median 30 Median 10 Temperature (°C) Change (%) 29 0 28 -10 27 -20 26 20th C 2030 2050 2070 20th C 2030 2050 2070 Normal pdf future mean rainfall in NRB Normal pdf future mean temperature in NRB 20 16 Temperature changes (%) 15 14 Rainfall changes (%) 12 10 10 5 8 0 6 4 -5 2 -10 0 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 Cumulative probability Cumulative probability 2030 2070 2070 - npd 2030 2070 2070 - npd 2030 - npd 2050 - npd 2050 2030 - npd 2050 - npd 2050 Figure 3: Rainfall and temperature quantiles and normal pdf for 20th century and 38 GCM projections for 2030, 2050 and 2070 for the Niger Basin. Note: Figure 3 is based on 38 model runs with 15 GCMs; results may therefore slightly vary from Figure 2. 2. Performance Matrix The development of the climate risk assessment framework for the Niger Basin focused on the climate sensitivity of the basin’s performance metrics and thresholds of significant climate impacts, for irrigated agriculture, hydro-energy generation, navigation, flooding of the Inner Delta, and the sustenance of minimum environmental flows throughout the Niger River system. Vulnerabilities were identified by simulating the basin’s water resources system for +10% increase and 10, 20 and 30% decrease in the 1966 to 1989 historical stream flow records, for all sub-catchments and for a 5% increase in crop water demands by 2050, under two infrastructure scenarios: (i) existing infrastructure and development, and (ii) with the three new large dams (Fomi, Taoussa and Kandadji dams) in place and the associated irrigation infrastructure developed. A sample of results is shown in Table 1 in terms of a Performance Matrix, which shows in the most right column the approximate runoff elasticity (εQ) of the performance indicators. Results were assessed for 1 out of 2 (median – 50%) and 1 out of 5 years (20% dry) performance. 120 Performance Probability Reference Impacts 20thC + 5% water demands; FO-TA-KD Average of non- runoff Metrics 2005 SDAP 20thC-R +10% R 0% R -10% R -20% R -30% R exceedance elasticity Value % change Value Percentage changes (%) Hydro-energy Basin energy 1/2(50%) 7,376 12.6 8,303 5.2 -3.0 -13.9 -24.3 -33.8 1.0 (GWh/yr) 1/5(20%) 5,771 3.4 5,969 8.9 -3.5 -16.0 -28.5 -38.7 1.3 Irrigated Agriculture Total irrigation mean 228,138 435 1,220,591 0.1 -0.3 -0.9 -1.8 -3.6 0.1 RS (ha) 1/5(20%) 228,138 424 1,194,537 -0.1 -0.5 -2.1 -4.2 -8.1 0.2 Total irrigation mean 111,744 471 637,537 -0.4 -0.8 -1.2 -1.5 -2.2 0.1 DS (ha) 1/5(20%) 105,130 500 630,890 -0.7 -0.9 -1.4 -5.5 -15.7 0.3 Navigation for various reaches (average number of days) Average Large boats 171 -20.9 135 7.9 -1.4 -10.9 -19.7 -30.2 1.0 Flooding (km2) mean 12,117 -9.7 10,940 5.4 -1.5 -10.9 -18.7 -28.6 0.9 Inland Delta 1/5(20%) 10,342 -14.1 8,887 7.1 -1.6 -13.9 -24.8 -37.3 1.2 Sustenance of 10-day average minimum flows (m 3/s) Markala 1/2(50%) 70 -13 61 -25.0 -38.6 -54.7 -83.5 -100.0 4 1/5(20%) 51 -2 50 -31.1 -40.6 -69.8 -99.6 -100.0 5 Niamey 1/2(50%) 55 78 99 2.0 -9.2 -16.9 -37.5 -80.9 2 1/5(20%) 9 818 85 -0.9 -11.3 -33.6 -73.7 -97.0 3.5 Table 1: Performance matrix for basin-wide Performance Indicators for the Niger Basin (limited sample of results for the purpose of illustrations only) 3. Runoff response to projected climate changes Analytic hydrological techniques and results from literature for comparable basins in Africa (as also discussed in Chapter 4 of this report) were subsequently used to estimate the response of runoff to changes in precipitation and temperature across the Basin, in terms of the climate elasticity of runoff. The precipitation elasticity was assessed at 2.5 and the temperature elasticity at -0.75; see Figure 4 for an example of the precipitation elasticity of runoff from the Upper Niger Basin. Figure 4: Relative changes in runoff related to relative Koulikoro changes in rainfall at Koulikoro station, representing the 0.75 runoff of the Upper Niger Basin 0.50 0.25 dQ/Q 0.00 -0.25 y = 2.4324x R² = 0.8391 -0.50 -0.75 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 dP/P 121 Climate projections were then used to assess the plausibility of the identified vulnerabilities to climate change. An ensemble of 38 climate projections for the 21st century, generated by 15 Global Circulation Models (GCM) for the A1B emission scenario25, was initially used to capture the possible future annual average climate across the Basin in terms of probability distributions of precipitation and temperature changes for 2030, 2050 and 2070 (see Figure 3). Table 2 compares the results for the 38 downscaled GCM projections for the Niger Basin with the results of climate projections for 2050 for the Upper Niger Basin in Guinea and the Lower Niger and Benue Basins in Nigeria (the “water towersâ€? of the Niger Basin) obtained from the Climate Wizard (climatewizard) and the WBG’s Climate Change Knowledge Portal (Climate Portal; Strzepek et al., 2011). Overall, results from the various sources of climate projections agree very well. Variable Min 20% Mean 80% Max St. dev. Table 2: Summary of projected climate Climatewizard.org Guinea changes from various sources for the Niger Temperature (0 C) 1.8 2.0 2.3 2.8 3.0 Basin (2050; A1B scenario) Precipitation (%) -20.0 -6.0 0.0 6.0 10.0 Nigeria Temperature (0 C) 1.5 1.8 2.1 2.5 2.8 Precipitation (%) -15.0 -4.0 2.0 10.0 15.0 WB Climate Change Knowledge Portal The climate elasticities were used to Guinea Temperature ( C) 0 1.2 1.8 2.1 2.6 3.0 0.5 translate projected relative changes in Precipitation (%) -12.2 -5.2 0.5 5.6 12.9 6.8 annual temperature and precipitation into Annual runoff (%) -23.8 -13.5 -0.3 12.0 38.7 16.5 Annual PET (%) 0.7 3.9 5.0 6.7 8.1 1.7 projected relative changes in annual runoff, Nigeria and derive a probability distribution of Temperature (0 C) 1.2 1.6 2.0 2.4 2.7 0.4 future runoff changes from the probability Precipitation (%) -13.4 -4.4 1.2 7.0 10.9 6.4 Annual runoff (%) -31.0 -11.3 -0.2 15.1 29.9 17.0 distributions of projected precipitation and Annual PET (%) 1.5 3.7 4.6 6.1 7.4 1.4 temperature changes (Figure 5). The Projections 38 GCM model runs for Niger River Basin projected mean flows are essentially Temperature (0 C) 1.2 1.6 2.1 2.6 2.9 0.5 Precipitation (%) -5.8 -3.5 1.4 4.5 13.7 4.5 constant over the 21st century at an Annual runoff (%) -19.5 -13.2 -1.9 4.7 32.3 10.9 average of only 2% below the 20th century Annual PET (%) 2.6 3.6 4.7 5.8 6.7 1.1 mean till 2050, with an insignificant increase by 2070. Projected climate change impacts on runoff from the Niger Basin are thus moderate. The standard deviation of projected mean annual flows varies between 7% by 2030 and 13% by 2070, reflecting the increasing uncertainty in climate projections for the distant future. The probability of a decrease in average annual runoff by 2050 is just over 50%. The probability of a 20% decrease in average runoff is only 5%, which has subsequently been adopted as the worst case scenario for standard project economic analyses for the SDAP. 25 It is general practice to use the median A1B economic development and emission scenario – proposed by the IPCC - as the basis for projecting future climate changes. The A1B scenario represents a relatively high economic growth worldwide and a relatively low growth in population. It has been shown that the differences between various emissions scenarios have little impact on projected climate changes until 2050 to 2070. Hence, there was no compelling need to test the vulnerability of the SDAP and IP for a variety of emission scenarios. 122 45 30 20 Flow changes (%) 30 Median 10 15 Runoff change (%) 0 0 -10 -20 -15 -30 -30 0.0 0.2 0.4 0.6 0.8 1.0 Cumulative probability (normal pdf) -45 2030 2070 2070 - npd 20th C 2030 2050 2070 2030 - npd 2050 - npd 2050 Figure 5: Quantiles of probability distributions of projected runoff changes for the Niger Basin (A1B) 4 Risk assessment for selected sectors Finally, based on the derived probability distributions of projected runoff changes for 2030, 2050 and 2070 and a more detailed version of the Performance Matrix shown in Table 1, the probabilities of exceedance of identified risk levels (climate hazards) were estimated for key performance metrics, such as hydro-energy generation, irrigated agriculture, navigation, flooding of the Inner Delta and the sustenance of environmental flows (see Figure 6 for examples). Irrigated agriculture was assessed to be insensitive to projected climate changes. Current water allocation rules in the NRB prioritize irrigated agriculture to secure food production and alleviate poverty, making it insensitive to decreased runoff due to the projected climate changes. Only mild agricultural production decreases are likely to occur, but would generally be less than a few percents of the output projected for SDAP. By regulating the variability of the Upper Niger flow, SDAP proves to be good insurance for the protection of irrigated agriculture in the NRB against the potential negative impacts of climate change. Climate change impacts on hydro-energy, navigation and flooding of the Inner Delta are projected to be mild (< 10% decrease) to moderate (< 20% decrease). Percentage wise, these performance indicators vary similarly to runoff variations (runoff elasticities are +1.0 to +1.2). There is 25 to 35% probability that by 2050 these sectors could suffer performance decreases more than 10%, and the probability of performance decreases exceeding 20% is only between 5 and 15%. Climate change impacts on minimum flows during the dry season can be severe and these indicators are the most sensitive to climate change and variability under the present water allocation rules. Under SDAP - and even more under the present development conditions – it may not be possible to maintain by 2050 without adaptation measures the agreed (transboundary) minimum flows at the entry of the Niger’s Inner Delta and throughout the Middle Niger under the projected climate change conditions. 123 30 4 Hydro-energy NRB Dry season irrig. agriculture NRB 15 2 Median 0 Change (%) Change (%) 0 -15 -2 -30 Median -45 -4 20th C 2030 2050 2070 20th C 2030 2050 2070 100 100 Dry Season Irrigated Agriculture Hydro-Electricity 2030-A 80 80 2050-A 2030-A 2070-A Probability (%) Probability (%) 60 2050-A 60 2070-A 2030-1/5 40 40 2050-1/5 2030-1/5 2070-1/5 2050-1/5 20 2070-1/5 20 0 0 Mild Moderate Mild Moderate Significant Severe Figure 6: Percentiles (5, 25, 50, 75 and 95%) for an average year (upper panels) and probabilities of risks for average (A) and 1/5 years (20% dry) performance of selected indicators Should higher priority be accorded to the sustenance of minimum flows, dry season irrigated agriculture would become more sensitive to climate change impacts and the planned abstractions for irrigation would need to be slightly reduced. This potential problem can be adequately addressed by measures to increase the present low dry season irrigation efficiencies in the region and/or by slightly reducing the future (and not yet developed) dry season irrigated areas. The vulnerabilities to climate change of rainfed agriculture are significant. Food security and climate risk assessments for the Niger Basin must also take into account potential climate change impacts on rainfed agriculture. Rainfed agriculture is presently the dominant agricultural system in the Niger basin and will remain so in the foreseeable future. Decreasing precipitation and increasing temperature present a worst-case scenario for crop yields. 124 Potential climate change impacts on the economic performance of SDAP were assessed to be modest. Under the proposed ‘worst case’ scenario of 20% reduction in the long-term average basin runoff, the projected climate change impacts on irrigated agriculture, navigation, minimum flows and flooding of the Inner Delta have only a negligible impact on the overall economic performance of the SDAP. In economic terms the only significant impact of reduced runoff due to climate changes would stem from its impact on hydro-energy generation. In the ‘worst case scenario’, one can expect a reduction of 1.6% in the EIRR; the runoff elasticity of the EIRR of the SDAP investment plan is about 0.6. Overall, the interventions planned under SDAP and IP for the Niger Basin constitute good adaptations to future climate risks, due to the abundance of water in the rainy season and the creation of large water storages for dry season water supply. With (i) the implementation of 3 dams on the Upper and Middle Niger (Fomi, Taoussa and Kandadji dams), along with (ii) optimal basin-wide reservoir management, (iii) increased irrigation efficiencies, (iv) changes to less water demanding non-rice dry season crops, and (v) other similar adaptation measures to improve water use efficiencies in (particularly dry season) irrigated agriculture, the ‘irrigation sector’ can be well protected from the impacts of severe climate change. Severe reductions in runoff would cause equally severe reductions in generated hydro-energy, navigation and flooding of the Inner Delta. Such severe future impacts can only be minimized by reducing rainy season irrigated agriculture in the Basin (not realistic, nor practical) or by the construction of additional storage reservoirs along with hydro-energy generation facilities in the Upper Niger Basin and in Nigeria (particularly in the Benue basin), and by the construction of economically beneficial Run-of-River hydropower plants, which do not cause additional water losses and negative impacts on downstream hydropower generation, while the negative social and environmental impacts are generally much less than those caused by the construction of large reservoirs. Significantly, the historical variability of Niger Basin precipitation and runoff is greater in magnitude than any climate trend projected for the 21st century. Therefore, learning lessons from managing the present impacts of intra-seasonal and inter-annual variability of the NRB climate has the potential to better prepare water managers for dealing with long-term climate change impacts. On the short-term successfully managing the historical climate variability and droughts experienced in the Basin is the best adaptation strategy that can be recommended. 125 Annex 4: Monthly runoff data series for key hydrometric stations Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr 1971 734 409 404 553 542 768 1,738 2,259 4,277 4,407 1,605 965 1,562 375 1972 481 271 301 417 639 1,255 1,173 1,587 2,934 4,564 2,124 741 1,380 331 1973 526 265 181 348 917 1,293 1,387 2,423 3,475 3,819 2,242 966 1,493 358 1974 500 278 206 455 763 1,298 1,446 2,708 4,646 5,818 3,285 1,167 1,888 453 1975 662 418 310 405 579 994 1,695 2,192 3,461 5,441 3,109 1,493 1,738 417 1976 724 555 460 582 878 1,298 1,760 3,060 4,393 5,656 4,210 1,482 2,095 502 1977 804 417 203 232 498 875 1,787 2,511 4,587 4,501 1,552 700 1,562 375 1978 356 199 176 494 1,105 1,477 2,316 2,720 5,416 5,617 3,829 1,250 2,087 501 1979 644 352 301 378 1,127 1,447 2,223 2,980 3,884 4,116 3,115 1,214 1,823 437 1980 625 318 229 291 729 1,039 1,400 2,214 4,298 6,853 3,620 1,277 1,916 460 1981 692 358 218 295 816 1,061 1,924 3,067 4,781 4,231 3,020 983 1,793 430 1982 589 312 328 433 1,027 1,078 1,633 2,897 4,660 6,216 3,229 1,165 1,973 473 1983 584 282 127 148 388 744 1,096 1,903 3,214 3,269 1,314 576 1,142 274 1984 259 121 149 365 507 914 1,988 3,166 3,828 4,679 2,527 914 1,627 390 1985 532 215 224 825 637 1,191 1,958 3,090 4,354 5,124 2,963 1,245 1,872 449 1986 621 357 352 451 573 856 1,114 1,833 2,966 3,899 2,337 834 1,354 325 1987 414 192 111 280 279 759 1,013 1,639 3,858 4,380 1,780 722 1,290 309 1988 360 179 122 305 812 1,143 1,321 2,272 3,789 5,332 2,510 1,107 1,612 387 1989 458 176 102 241 794 1,143 1,602 3,247 4,638 4,876 2,395 839 1,718 412 1990 431 218 59 136 580 1,022 1,590 3,351 4,439 4,979 3,582 1,506 1,832 439 1991 675 290 249 541 1,068 1,723 2,188 3,416 4,131 4,347 3,022 1,074 1,902 456 1992 451 202 136 340 534 1,054 1,645 3,243 4,987 5,117 3,819 1,178 1,899 455 1993 565 214 236 233 562 1,265 2,194 3,647 4,735 4,928 2,821 1,275 1,899 455 1994 516 242 137 250 749 1,110 1,974 2,845 5,452 5,687 2,986 936 1,915 459 1995 423 180 176 271 588 944 1,308 2,381 3,139 3,753 3,068 861 1,430 343 1996 360 148 214 401 612 1,407 2,223 2,906 3,812 6,887 2,389 901 1,867 448 1997 446 176 87 420 743 883 1,600 2,702 3,936 3,296 2,698 922 1,498 359 1998 470 199 73 136 492 807 1,494 3,351 4,207 5,823 2,379 960 1,710 410 1999 477 349 314 371 1,061 1,285 1,740 2,644 4,223 5,879 4,310 1,448 2,016 483 2000 658 331 182 444 819 1,282 1,777 3,199 3,715 4,790 2,047 764 1,676 402 2001 365 175 189 269 602 901 1,697 2,610 3,495 4,636 1,492 574 1,426 342 2002 305 138 156 363 588 868 1,792 2,667 5,315 4,990 2,834 879 1,748 419 2003 443 249 202 294 391 609 1,619 3,094 4,709 5,953 3,447 961 1,839 441 Average 513 262 203 357 702 1,094 1,677 2,736 4,171 4,983 2,814 1,029 1,719 412 StDev 131 96 90 139 219 247 350 529 682 917 755 260 241 58 CV 0.26 0.37 0.44 0.39 0.31 0.23 0.21 0.19 0.16 0.18 0.27 0.25 0.14 0.14 Min 259 121 59 136 279 609 1,013 1,587 2,934 3,269 1,314 574 1,142 274 Max 804 555 460 825 1,127 1,723 2,316 3,647 5,452 6,887 4,310 1,506 2,095 502 5% 333 144 81 143 445 764 1,106 1,746 3,061 3,548 1,525 644 1,326 318 10% 360 176 103 232 492 812 1,186 1,931 3,239 3,827 1,806 724 1,384 332 20% 416 179 129 254 540 876 1,390 2,294 3,539 4,254 2,261 835 1,494 358 50% 491 245 195 355 638 1,070 1,670 2,782 4,215 4,953 2,898 964 1,771 425 Table 1: Naturalized monthly flow data (m3/s) for Edea in the Sanaga Basin 126 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr 1971 543 1,013 1,145 2,496 2,260 819 448 1972 244 102 164 254 346 757 644 1,005 1,618 2,111 1,109 414 733 304 1973 277 113 82 190 507 799 997 1,506 2,011 2,166 1,205 553 871 362 1974 266 142 98 242 391 788 873 1,721 2,517 2,801 1,671 592 1,012 420 1975 337 209 171 213 325 607 1,093 1,307 1,994 2,628 1,661 776 947 393 1976 404 294 207 307 441 682 893 1,597 2,258 3,013 2,217 806 1,097 455 1977 448 207 75 133 286 496 962 1,332 2,426 2,251 825 412 825 342 1978 198 63 59 342 543 963 1,239 1,603 2,653 2,886 1,726 642 1,081 449 1979 295 133 108 191 555 758 1,208 1,738 2,233 2,452 1,692 620 1,003 416 1980 284 167 155 216 458 672 867 1,138 1,514 1,855 2,131 664 846 351 1981 362 221 202 188 351 910 1,520 1,858 2,355 2,324 1,310 546 1,017 422 1982 370 193 200 273 505 466 433 749 2,557 1,684 1,025 540 750 311 1983 317 164 101 88 221 442 731 1,025 1,210 1,564 722 288 576 239 1984 214 136 20 128 327 635 1,241 1,860 1,805 2,413 1,223 415 873 362 1985 264 103 110 294 429 638 961 1,530 1,747 2,015 1,425 581 846 351 1986 341 133 137 290 423 445 693 928 1,446 1,889 1,126 397 690 286 1987 282 154 99 164 226 445 658 715 2,026 2,357 1,089 449 724 301 1988 162 20 20 100 471 742 927 1,333 1,976 2,677 1,433 630 879 365 1989 208 20 20 70 462 742 1,060 1,791 2,374 2,454 1,379 504 929 385 1990 195 20 20 20 361 685 1,180 1,718 2,330 2,476 1,973 848 991 411 1991 347 35 20 132 627 1,144 1,359 1,710 2,131 2,320 1,684 613 1,015 421 1992 152 24 20 81 336 698 1,211 1,716 2,538 2,631 2,040 706 1,017 422 1993 295 34 20 29 299 856 1,377 1,886 2,506 2,527 1,564 736 1,016 422 1994 212 31 20 49 409 657 1,088 1,800 2,842 2,889 1,666 542 1,022 424 1995 174 20 20 71 322 610 951 1,387 1,857 1,914 1,653 515 795 330 1996 132 20 20 166 344 990 1,347 1,599 1,987 3,415 1,400 509 1,001 415 1997 127 20 20 283 410 598 1,203 1,556 2,050 1,687 1,392 471 822 341 1998 226 98 94 20 303 488 1,132 2,029 2,141 3,062 1,297 521 958 397 1999 227 60 20 119 635 700 1,051 1,693 2,201 3,051 2,259 788 1,073 445 2000 346 84 20 98 476 747 1,019 1,899 2,047 2,473 1,190 368 902 374 2001 130 20 20 53 371 583 1,120 1,516 2,042 2,426 911 284 795 330 2002 49 20 20 204 383 576 1,165 1,502 2,797 2,669 1,521 479 953 395 2003 198 20 20 41 274 392 1,067 1,719 2,448 2,987 1,873 561 972 403 Average 253 96 74 158 401 679 1,040 1,514 2,145 2,440 1,481 555 907 376 StDev 91 76 65 94 104 174 238 340 386 452 395 144 127 53 CV 0.36 0.79 0.87 0.60 0.26 0.26 0.23 0.22 0.18 0.19 0.27 0.26 0.14 0.14 Min 49 20 20 20 221 392 433 715 1,210 1,564 722 284 576 239 Max 448 294 207 342 635 1,144 1,520 2,029 2,842 3,415 2,259 848 1,097 455 5% 129 20 20 25 253 443 652 848 1,483 1,686 872 332 709 294 10% 134 20 20 42 287 447 697 1,007 1,631 1,858 1,032 399 735 305 20% 179 20 20 70 323 512 877 1,312 1,881 2,034 1,139 422 800 332 50% 254 91 40 148 387 677 1,064 1,598 2,136 2,453 1,429 544 938 389 Table 2: Naturalized monthly flow data (m3/s) for Nachtigal in the Sanaga Basin (Notes: discharges corrected for flow regulation by Mbakaou reservoir; data from September 1987 till 2003 determined through regression with flow data for Edea) 127 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr 1971 102 66 89 108 79 113 382 443 646 620 257 113 253 404 1972 49 21 7 34 63 205 248 405 506 533 174 78 194 311 1973 58 32 9 15 99 222 376 492 565 443 143 64 211 338 1974 38 18 3 23 92 186 362 559 738 713 269 100 260 416 1975 71 35 2 21 90 155 392 497 757 690 282 145 263 421 1976 58 37 13 9 62 171 314 522 640 802 441 162 271 433 1977 103 49 10 5 50 156 416 586 718 583 213 93 250 400 1978 45 15 2 69 119 334 379 566 819 550 253 113 273 437 1979 73 35 14 28 200 249 525 663 755 629 395 156 312 499 1980 100 53 27 11 118 160 436 593 972 929 386 152 330 528 1981 90 47 7 17 128 224 565 807 1,080 837 368 150 362 579 1982 56 26 27 19 108 149 353 652 819 810 281 120 287 459 1983 64 35 8 2 31 175 303 451 731 391 120 55 198 317 1984 19 9 5 25 58 82 238 404 544 494 193 71 179 287 1985 50 15 6 37 44 139 401 625 802 478 256 108 248 397 1986 65 32 39 34 43 108 237 467 588 608 281 99 218 349 1987 46 24 4 13 15 146 206 442 630 626 175 73 201 322 1988 57 19 4 23 115 161 317 716 725 724 215 92 266 425 1989 36 16 1 8 105 231 312 745 966 778 228 96 295 472 1990 45 20 6 6 140 220 430 675 726 544 353 139 277 443 1991 54 26 18 33 155 310 468 620 682 616 285 112 283 453 1992 55 28 17 16 80 162 392 617 768 600 339 121 268 428 1993 60 32 23 21 82 237 421 671 708 524 231 102 261 418 1994 48 24 14 16 66 184 363 581 896 663 259 101 269 431 1995 56 29 10 13 56 136 298 393 494 434 250 88 189 303 1996 36 16 11 18 60 215 328 441 483 676 208 88 217 347 1997 49 21 4 37 101 141 313 416 515 355 265 88 193 309 1998 48 20 3 4 32 89 226 636 733 707 212 91 235 376 1999 44 23 11 18 113 133 275 504 584 609 297 109 228 365 2000 54 29 13 27 91 147 253 516 563 463 140 60 197 316 2001 28 15 2 4 37 101 277 466 462 456 124 53 170 272 2002 31 12 2 18 67 162 339 513 714 557 250 93 231 370 2003 45 26 8 9 26 91 349 700 759 675 294 106 259 414 Average 54 26 11 19 83 173 351 566 708 612 258 103 248 398 StDev 19 11 9 13 42 60 85 108 148 139 79 30 45 72 CV 0.35 0.41 0.85 0.69 0.50 0.34 0.24 0.19 0.21 0.23 0.31 0.29 0.18 0.18 Min 19 9 1 2 15 82 206 393 462 355 120 53 170 272 Max 103 53 39 69 200 334 565 807 1,080 929 441 162 362 579 10% 36 15 2 5 32 101 238 441 515 443 143 64 193 309 20% 44 16 3 9 44 136 277 466 565 478 208 88 201 322 50% 54 26 8 18 82 161 349 566 725 609 256 100 259 414 Table 3: Monthly flow data (m3/s) for Lom Pangar in the Sanaga Basin 128 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr MCM/yr 1971 105 61 39 67 55 176 644 752 1028 601 201 90 320 499 10,086 1972 65 27 9 45 83 270 327 534 667 726 229 103 258 403 8,148 1973 72 40 11 18 123 276 467 611 702 550 178 80 262 409 8,267 1974 59 28 5 35 143 291 565 871 1,151 1,112 419 156 405 633 12,777 1975 92 46 3 27 117 201 507 642 978 891 364 188 340 530 10,712 1976 95 61 21 14 101 280 514 854 1,047 1,312 721 264 443 691 13,960 1977 142 67 14 7 69 215 574 809 990 805 294 128 345 538 10,873 1978 70 24 3 108 187 524 595 889 1,285 863 397 177 428 669 13,512 1979 94 45 19 36 258 321 678 856 975 812 510 202 403 629 12,701 1980 106 56 29 11 124 169 460 626 1,026 981 408 161 348 544 10,981 1981 86 45 7 16 122 214 539 769 1,029 798 351 143 345 539 10,878 1982 79 37 38 27 152 210 497 918 1,152 1,139 395 169 404 630 12,729 1983 85 46 11 3 41 231 402 597 967 518 159 72 262 409 8,263 1984 31 14 9 41 94 133 387 657 884 804 313 115 292 456 9,205 1985 72 22 9 53 63 200 576 898 1,152 686 367 155 356 556 11,236 1986 77 38 47 41 52 130 285 560 706 730 338 119 262 408 8,248 1987 62 32 5 17 21 197 277 595 849 843 235 99 271 422 8,534 1988 51 17 3 21 103 145 285 644 652 651 194 82 239 373 7,538 1989 41 18 1 10 121 267 360 861 1,116 899 263 111 341 532 10,746 1990 57 25 7 7 175 274 536 842 905 679 440 173 345 539 10,891 1991 78 37 26 49 225 451 681 903 992 896 415 163 412 643 12,991 1992 82 41 26 24 120 242 586 921 1,146 896 506 180 399 624 12,598 1993 92 49 36 33 125 361 641 1,021 1,077 798 352 156 397 620 12,521 1994 75 37 21 24 102 284 561 899 1,385 1,025 400 157 416 650 13,124 1995 89 45 16 20 88 214 468 617 776 682 392 139 297 464 9,364 1996 73 33 23 37 122 436 667 896 981 1,374 422 179 440 687 13,874 1997 91 39 8 68 187 260 577 768 950 655 489 163 356 556 11,237 1998 82 34 5 7 54 152 385 1,084 1,251 1,206 361 155 401 626 12,635 1999 75 39 19 31 195 229 473 864 1,002 1,045 510 186 391 611 12,343 2000 94 51 22 46 157 253 437 891 972 800 242 104 341 532 10,751 2001 50 27 3 7 67 181 494 834 826 815 221 94 304 474 9,581 2002 43 17 3 25 94 229 473 724 1,009 786 354 130 326 508 10,266 2003 64 37 12 13 36 128 491 983 1,067 947 412 148 364 568 11,467 Average 77 37 15 29 116 249 493 795 990 866 364 145 350 546 11,030 StDev 22 13 12 21 56 92 112 146 175 203 117 41 59 92 1,866 CV 0.28 0.35 0.79 0.74 0.48 0.37 0.23 0.18 0.18 0.23 0.32 0.28 0.17 0.17 0.17 Min 31 14 1 3 21 128 277 534 652 518 159 72 239 373 7,538 Max 142 67 47 108 258 524 681 1,084 1,385 1,374 721 264 443 691 13,960 10% 50 19 3 7 52 146 330 598 713 657 222 94 262 409 8,264 20% 60 26 5 12 67 184 390 629 856 694 246 105 293 457 13,110 50% 77 37 11 25 117 229 497 842 992 812 364 155 345 539 10,891 Table 4: Monthly inflow data (m3/s) for Mbakaou reservoir in the Sanaga Basin 129 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr MCM/yr 1971 11 4 4 8 9 14 64 105 108 124 62 20 45 644 1,410 1972 7 3 2 3 6 18 43 70 92 99 56 15 35 500 1,094 1973 6 2 1 2 10 23 29 47 81 81 54 15 29 423 925 1974 5 2 1 7 11 68 107 103 169 138 65 10 57 826 1,809 1975 15 12 10 10 55 39 81 77 140 173 57 24 58 836 1,832 1976 10 17 18 24 31 43 111 153 152 172 90 24 71 1020 2,234 1977 15 8 6 4 11 61 119 118 170 112 27 12 56 800 1,752 1978 3 1 7 24 31 65 89 95 186 123 56 19 58 841 1,843 1979 11 6 8 9 42 52 98 139 139 113 67 22 59 850 1,862 1980 13 3 9 6 21 42 65 105 181 186 62 24 60 863 1,889 1981 10 5 12 5 37 42 97 141 199 88 40 16 58 835 1,828 1982 19 8 19 15 31 52 78 119 136 143 48 21 58 833 1,825 1983 9 10 6 8 16 31 58 108 128 63 17 12 39 562 1,231 1984 3 6 15 11 9 23 84 126 152 130 52 20 53 761 1,666 1985 13 3 24 33 26 46 90 114 157 111 45 20 57 823 1,802 1986 9 5 17 12 14 23 61 83 126 126 58 18 46 666 1,459 1987 7 5 4 5 0 49 67 86 121 108 22 7 40 578 1,266 1988 4 0 5 7 20 24 50 60 108 109 33 13 36 523 1,146 1989 3 1 7 9 31 57 81 104 183 123 37 13 54 784 1,716 1990 7 5 10 11 28 30 80 136 138 124 52 26 54 781 1,710 1991 14 8 8 20 48 60 86 86 105 91 34 12 48 691 1,513 1992 8 3 16 12 23 44 92 95 119 105 46 14 48 695 1,522 1993 8 2 10 16 21 41 79 145 148 140 57 18 57 827 1,810 1994 10 1 7 18 27 51 93 110 168 154 50 15 59 850 1,861 1995 11 5 7 13 27 36 75 121 141 141 47 13 53 770 1,686 1996 9 3 13 14 21 69 125 116 111 134 28 12 55 791 1,731 1997 7 3 15 30 24 38 73 135 125 74 68 14 51 732 1,603 1998 6 3 8 4 21 39 73 98 122 140 35 12 47 679 1,487 1999 9 5 7 14 35 42 66 68 143 177 88 16 56 807 1,767 2000 13 6 8 15 16 50 83 142 165 161 39 14 60 862 1,889 2001 4 4 8 7 13 40 69 94 104 77 19 12 38 546 1,195 2002 4 6 4 3 10 41 122 141 152 131 49 14 57 817 1,790 2003 8 4 9 4 6 51 129 142 169 142 55 13 61 883 1,935 Average 9 5 9 12 23 43 83 109 142 125 49 16 52 752 1,646 StDev 4 3 5 8 13 13 23 28 29 31 17 5 9 132 290 CV 0.45 0.72 0.58 0.67 0.56 0.31 0.28 0.25 0.20 0.25 0.36 0.29 0.18 0.18 0.18 Min 3 0 1 2 0 18 29 47 81 63 17 7 29 423 925 Max 19 17 24 33 55 69 129 153 199 186 90 26 71 1,020 2,234 10% 4 1 4 4 9 23 58 71 106 82 27 12 38 547 1,199 20% 5 2 6 5 11 32 66 86 120 100 35 12 46 669 1,465 50% 9 5 8 10 21 42 81 109 141 125 49 15 55 795 1,742 Table 5: Monthly inflow data (m3/s) for Bamendjing reservoir in the Sanaga Basin 130 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr MCM/yr 1971 11 4 5 11 8 49 133 128 190 137 47 26 63 492 1,979 1972 8 3 6 23 28 81 96 139 211 255 74 29 80 626 2,517 1973 10 3 2 8 29 47 94 149 179 168 74 31 67 523 2,101 1974 10 3 2 22 34 79 152 175 297 296 150 48 106 833 3,350 1975 18 6 5 11 26 58 129 107 175 230 105 48 77 604 2,428 1976 18 9 6 18 47 86 166 228 234 271 166 56 109 856 3,443 1977 26 8 2 5 22 50 150 170 268 249 72 28 88 690 2,773 1978 8 3 2 20 35 78 110 146 310 318 178 48 105 825 3,315 1979 17 5 3 9 54 63 153 192 229 195 138 51 93 729 2,929 1980 21 5 3 4 30 63 91 138 315 365 158 49 104 816 3,279 1981 21 5 3 11 45 58 145 179 252 182 85 30 85 668 2,686 1982 12 5 7 7 24 68 117 155 228 275 114 33 87 686 2,759 1983 11 4 1 3 3 37 93 175 206 101 28 19 57 448 1,801 1984 4 10 25 17 13 36 135 204 246 210 83 31 85 667 2,681 1985 21 4 38 53 42 75 145 183 253 179 72 32 92 721 2,900 1986 14 8 27 19 23 37 99 133 204 203 93 28 74 584 2,346 1987 10 7 5 8 0 78 65 88 147 260 61 26 63 497 1,999 1988 14 10 38 50 24 35 51 130 169 230 88 33 73 573 2,304 1989 15 12 23 45 58 74 103 150 284 240 84 28 93 732 2,943 1990 18 4 4 43 60 55 116 212 225 321 184 96 112 880 3,537 1991 23 3 4 35 71 107 194 228 268 256 125 45 114 894 3,594 1992 9 7 13 35 49 92 110 228 311 337 201 60 122 953 3,832 1993 13 3 17 34 50 104 208 291 238 315 160 60 125 983 3,952 1994 15 11 6 56 94 123 175 190 325 349 127 46 127 997 4,007 1995 3 5 12 35 76 77 127 127 167 301 166 38 95 746 2,999 1996 1 8 25 20 49 110 182 267 224 384 119 35 119 937 3,768 1997 5 1 9 82 39 58 85 129 219 231 191 38 91 713 2,867 1998 7 2 35 85 67 45 89 193 209 298 93 29 97 759 3,050 1999 6 10 24 58 60 76 119 195 216 384 257 61 123 962 3,869 2000 1 4 13 43 51 66 125 196 304 384 123 30 112 882 3,545 2001 3 15 73 93 106 28 45 80 199 189 66 24 77 604 2,428 2002 45 13 6 66 54 23 75 130 176 213 113 44 80 629 2,529 2003 9 13 67 41 111 40 93 170 204 257 154 45 101 793 3,187 Average 13 7 16 33 46 66 120 171 234 264 122 41 95 744 2,991 StDev 9 4 18 25 26 25 40 48 49 70 50 15 19 148 597 CV 0.68 0.55 1.15 0.76 0.57 0.38 0.33 0.28 0.21 0.27 0.41 0.37 0.20 0.20 0.20 Min 1 1 1 3 0 23 45 80 147 101 28 19 57 448 1,801 Max 45 15 73 93 111 123 208 291 325 384 257 96 127 997 4,007 10% 4 3 2 7 22 36 76 127 175 183 72 28 73 574 2,308 20% 6 3 3 10 24 41 91 131 200 205 76 29 78 608 2,446 50% 11 5 7 28 46 65 116 172 226 257 116 36 93 730 2,936 Table 6: Monthly inflow data (m3/s) for Mapé reservoir at Magba in the Sanaga Basin 131 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr MCM/yr 1945 23 10 4 2 12 70 190 586 1,820 701 94 24 295 145 9,297 1946 7 2 1 2 12 79 312 666 2,280 2,340 242 89 505 249 15,931 1947 23 10 4 2 12 50 234 1,340 1,940 533 79 6 354 174 11,156 1948 23 10 4 2 1 142 364 2,460 2,580 911 130 29 557 275 17,578 1949 6 1 0 0 6 30 240 1,075 1,420 460 122 41 285 140 8,973 1950 11 3 0 1 26 50 202 1,045 1,434 671 188 74 310 153 9,778 1951 32 15 7 4 34 51 205 968 1,351 843 209 36 314 155 9,914 1952 12 7 1 0 10 60 213 731 1,208 916 183 78 286 141 9,030 1953 31 14 6 2 39 87 422 818 1,160 557 97 32 273 135 8,625 1954 14 5 1 1 8 108 449 584 1,980 1,048 218 98 377 186 11,890 1955 41 18 6 4 9 95 347 1,357 2,345 1,561 342 145 525 259 16,548 1956 40 16 8 5 4 33 238 958 1,874 1,161 157 74 382 188 12,052 1957 40 20 10 3 15 161 393 1,037 1,684 1,082 189 73 394 194 12,426 1958 35 19 7 1 15 82 294 689 1,392 670 108 49 281 138 8,860 1959 30 17 8 5 26 121 223 452 2,157 617 161 41 321 158 10,118 1960 26 14 7 5 12 90 579 1,864 2,803 1,204 278 121 586 289 18,479 1961 44 15 6 2 2 58 739 992 2,523 572 158 48 431 212 13,577 1962 30 15 8 0 3 104 203 1,126 2,373 1,091 162 79 434 214 13,680 1963 35 16 6 3 15 39 357 1,724 1,833 958 225 63 442 218 13,940 1964 31 16 8 10 13 49 283 681 1,720 851 188 71 327 161 10,328 1965 33 17 6 1 2 98 307 1,770 1,600 432 82 30 367 181 11,568 1966 15 6 2 1 35 158 234 913 2,380 555 186 57 378 186 11,928 1967 29 15 4 2 4 39 356 735 1,440 598 88 50 281 138 8,860 1968 24 11 3 1 10 119 438 1,080 1,810 515 90 37 346 170 10,906 1969 16 8 3 4 16 95 414 1,910 2,270 1,220 290 82 530 261 16,717 1970 31 13 3 2 1 23 202 1,550 2,330 896 222 51 445 219 14,034 1971 26 11 3 1 0 15 276 1,080 1,670 300 65 27 290 143 9,151 1972 15 7 3 1 9 120 256 584 516 319 69 26 162 80 5,093 1973 4 1 0 0 1 32 246 1,250 1,340 346 58 14 276 136 8,693 1974 6 2 0 0 10 6 224 935 1,045 792 111 32 265 131 8,371 1975 17 9 4 1 3 19 276 1,637 2,382 907 109 40 452 223 14,255 1976 36 16 6 2 9 40 295 895 713 673 235 73 251 124 7,927 1977 24 9 3 0 10 22 274 984 1,520 276 28 9 264 130 8,323 1978 3 1 1 0 22 56 366 1,042 2,043 609 160 57 364 179 11,482 1979 12 3 1 6 11 48 242 706 692 226 40 13 168 83 5,288 1980 9 3 1 2 12 70 410 1,575 1,374 443 103 20 337 166 10,638 Average 23 10 4 2 12 70 314 1,106 1,750 774 152 52 357 176 11,262 StDev 12 6 3 2 10 41 115 454 562 409 75 31 102 50 3,215 CV 0.51 0.56 0.70 0.98 0.82 0.58 0.37 0.41 0.32 0.53 0.49 0.60 0.29 0.29 0.29 Min 3 1 0 0 0 6 190 452 516 226 28 6 162 80 5,093 Max 44 20 10 10 39 161 739 2,460 2,803 2,340 342 145 586 289 18,479 10% 6.5 3 1 1 3 33 224 706 1,351 460 88 27 281 138 8,860 20% 12 3 1 1 3 33 224 706 1,351 460 88 27 281 138 8,860 50% 24 11 4 2 10 59 280 1,015 1,765 672 158 49 342 168 10,772 Table 7: Monthly flow data (m3/s) for Garoua in the Niger-Benue Basin 132 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr MCM 1950 8 4 2 1 5 33 130 878 1213 631 74 16 251 258 7,904 1951 8 3 0 0 2 30 151 697 992 619 157 39 226 232 7,124 1952 9 5 1 0 7 43 152 522 862 654 131 56 204 210 6,444 1953 9 3 1 0 12 47 285 682 869 430 46 11 201 207 6,331 1954 6 2 1 0 2 31 305 540 1533 785 98 23 278 286 8,764 1955 8 2 0 0 3 40 259 1034 1826 1148 156 49 379 390 11,945 1956 25 12 6 0 5 50 223 670 1493 928 101 37 297 305 9,363 1957 17 7 1 0 7 113 264 768 1243 769 91 16 276 284 8,701 1958 1 0 0 0 0 15 194 484 937 395 45 7 174 179 5,480 1959 1 0 0 0 1 20 84 327 1735 262 27 6 204 210 6,443 1960 8 4 2 0 5 33 318 1282 2162 875 121 24 404 416 12,745 1961 8 4 2 0 5 33 477 808 1575 278 90 24 276 284 8,700 1962 8 4 2 1 0 38 82 966 1913 680 70 10 315 324 9,929 1963 2 0 0 0 1 7 268 1238 1116 558 88 18 277 285 8,723 1964 5 2 0 2 6 23 206 539 1520 547 109 28 249 256 7,857 1965 7 3 1 0 1 35 143 985 1130 250 39 8 218 224 6,864 1966 2 0 0 0 14 86 134 806 1710 367 72 15 267 275 8,426 1967 7 3 1 1 3 19 188 476 947 413 40 11 176 181 5,561 1968 5 2 1 0 3 12 319 907 1135 428 93 45 247 254 7,794 1969 8 4 2 4 11 56 346 1380 1562 717 145 49 359 369 11,321 1970 7 3 1 0 0 10 159 1370 1660 361 77 63 310 319 9,788 1971 9 5 2 0 0 13 199 844 1190 168 28 9 206 212 6,500 1972 5 3 1 1 5 29 133 386 328 239 28 6 98 101 3,084 1973 17 11 6 0 9 36 145 704 925 210 34 10 176 181 5,555 1974 3 0 0 0 5 5 233 748 805 691 122 48 223 230 7,044 1975 5 4 2 1 2 5 290 1283 1678 643 142 63 345 355 10,871 1976 8 5 4 3 4 43 268 821 609 583 161 17 212 218 6,694 1977 6 4 3 1 1 15 231 852 1154 249 30 10 214 220 6,741 1978 6 4 3 4 16 42 217 1002 1762 487 72 30 304 313 9,599 1979 13 8 6 4 16 49 143 515 459 147 26 8 117 120 3,685 1980 8 4 2 1 5 33 275 1104 844 306 64 24 224 231 7,067 Average 8 4 2 1 5 34 220 826 1,254 510 83 25 249 256 7,840 StDev 5 3 2 1 5 23 86 290 449 246 43 18 71 73 2,223 CV 0.64 0.78 1.06 1.70 0.92 0.68 0.39 0.35 0.36 0.48 0.52 0.71 0.28 0.28 0.28 Min 1 0 0 0 0 5 82 327 328 147 26 6 98 101 3,084 Max 25 12 6 4 16 113 477 1,380 2,162 1,148 161 63 404 416 12,745 10% 2 0 0 0 0 10 133 484 805 239 28 8 176 181 5,555 20% 5 2 0 0 1 15 143 539 869 262 39 10 204 210 6,443 50% 8 4 1 0 5 33 217 808 1,190 487 77 18 247 254 7,794 Table 8: Monthly flow data (m3/s) for Riao (Lagdo reservoir) in the Niger-Benue Basin 133 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr MCM/yr 1951 171 98 115 152 170 232 232 136 302 621 631 335 267 390 8,420 1952 145 87 110 157 363 450 320 249 261 473 577 336 295 431 9,302 1953 169 92 180 176 200 205 176 115 220 340 548 273 225 328 7,092 1954 132 188 152 240 284 376 241 131 204 433 459 338 265 387 8,353 1955 161 86 129 196 216 295 235 178 228 376 524 294 244 356 7,686 1956 144 85 242 406 446 379 273 159 224 531 613 431 329 480 10,372 1957 198 90 84 151 174 270 229 197 248 521 651 452 273 399 8,608 1958 197 97 69 189 292 292 178 81 94 338 449 299 215 314 6,783 1959 142 54 41 122 255 263 216 147 259 552 664 418 262 383 8,264 1960 202 95 82 137 263 295 193 212 308 531 673 381 282 411 8,885 1961 260 128 68 122 156 171 117 67 181 456 527 236 208 303 6,547 1962 100 47 117 326 455 459 310 198 400 586 849 488 362 529 11,423 1963 223 191 154 197 251 293 296 223 391 562 596 372 313 457 9,864 1964 168 102 105 257 260 349 242 139 309 553 826 429 312 455 9,838 1965 225 174 194 223 297 346 232 218 313 511 478 288 292 426 9,209 1966 167 103 58 178 460 458 448 287 312 618 835 486 369 538 11,631 1967 229 112 73 110 165 265 177 133 254 520 596 272 243 354 7,649 1968 146 70 117 165 306 276 202 151 292 414 594 354 258 377 8,135 1969 173 119 248 265 301 289 274 300 508 601 625 341 338 493 10,655 1970 174 94 174 242 302 302 213 251 427 680 793 357 335 489 10,561 1971 183 71 134 154 227 246 201 189 308 704 543 335 276 403 8,699 1972 159 79 83 161 172 244 124 123 286 667 544 266 243 355 7,662 1973 161 55 68 159 316 397 292 228 247 474 507 299 268 391 8,451 1974 130 58 60 145 212 242 124 150 315 580 587 322 244 357 7,708 1975 156 100 90 161 211 182 211 145 171 437 616 355 237 346 7,470 1976 186 129 130 195 218 250 127 101 163 407 502 240 221 322 6,963 1977 122 47 35 Average 171 98 115 192 268 301 226 173 278 519 608 346 276 403 8,701 StDev 36 38 56 67 87 80 73 60 88 101 112 71 45 65 1,411 CV 0.21 0.39 0.49 0.35 0.32 0.27 0.32 0.35 0.32 0.19 0.18 0.20 0.16 0.16 0.16 Min 100 47 35 110 156 171 117 67 94 338 449 236 208 303 6,547 Max 260 191 248 406 460 459 448 300 508 704 849 488 369 538 11,631 10% 131 55 59 130 171 219 126 108 176 392 490 269 223 325 7,028 20% 144 70 68 151 200 244 177 131 220 433 524 288 243 354 7,649 50% 168 94 110 171 258 291 223 155 274 526 595 337 267 391 8,436 Table 9: Monthly flow data (m3/s) for Eseka in the Nyong Basin 134 Month 1 2 3 4 5 6 7 8 9 10 11 12 Annual Year m 3/s m m /yr MCM/yr 1953 95 40 139 321 523 152 1954 61 122 141 190 294 256 121 31 133 442 505 209 209 364 6,584 1955 77 66 173 359 264 287 131 76 248 342 521 218 230 401 7,258 1956 177 72 250 402 506 481 181 71 178 577 670 424 333 581 10,516 1957 207 97 111 213 315 347 197 77 188 709 584 419 290 505 9,138 1958 134 82 67 129 241 146 31 15 59 409 347 197 155 270 4,894 1959 87 85 56 120 345 209 99 67 294 737 687 471 272 474 8,588 1960 304 213 127 266 311 350 170 138 479 664 826 382 352 614 11,108 1961 207 218 144 255 163 214 100 102 127 430 529 253 228 397 7,190 1962 72 65 196 519 520 384 165 80 168 563 523 418 307 535 9,683 1963 152 137 432 270 418 254 323 169 419 738 503 294 344 599 10,845 1964 159 112 161 415 439 270 143 74 175 594 790 792 345 601 10,874 1965 199 163 279 363 335 340 165 123 281 704 639 302 325 566 10,245 1966 124 116 125 420 624 717 534 247 360 564 674 363 407 708 12,821 1967 137 108 90 122 173 407 186 72 219 784 729 377 284 495 8,962 1968 155 146 186 273 579 419 157 88 254 533 546 640 332 579 10,483 1969 108 180 410 514 336 275 157 121 301 654 581 280 327 569 10,300 1970 106 85 176 195 233 440 199 117 264 659 957 262 308 536 9,709 1971 132 67 105 232 168 131 890 80 225 558 556 242 284 495 8,953 1972 71 71 108 290 230 206 88 76 205 597 639 202 232 404 7,320 1973 143 97 115 204 292 384 148 111 182 411 412 177 223 389 7,040 1974 79 84 114 188 434 303 111 125 228 506 572 268 252 438 7,934 1975 84 115 115 334 259 172 156 54 84 430 646 385 236 412 7,454 1976 120 115 136 260 278 355 173 79 135 583 582 337 263 459 8,300 1977 233 112 84 101 112 172 58 59 330 644 536 225 222 387 7,013 1978 72 43 109 252 544 380 174 61 181 455 448 179 242 422 7,642 1979 93 61 131 221 330 305 168 89 216 458 519 218 235 409 7,399 Average 134 109 159 273 336 316 190 90 225 558 594 322 278 485 8,779 StDev 59 45 92 114 136 125 167 46 97 127 129 146 57 100 1,807 CV 0.44 0.41 0.58 0.42 0.40 0.40 0.88 0.51 0.43 0.23 0.22 0.45 0.21 0.21 0.21 Min 61 43 56 101 112 131 31 15 59 321 347 152 155 270 4,894 Max 304 218 432 519 624 717 890 247 479 784 957 792 407 708 12,821 10% 72 66 87 126 171 172 92 48 131 410 481 190 223 388 7,027 20% 79 71 108 190 233 209 102 62 145 432 519 211 230 401 7,258 50% 128 103 129 258 313 304 157 79 216 564 572 280 278 485 8,771 Table 10: Monthly flow data (m3/s) for Ngoazik in the Ntem Basin 135 Annex 5: Annual and monthly data series of rainfall, temperature (CRU-TS 3.10), PET and runoff Precipitation Betare- Lake Edea Nachtigal Goura Mbakaou Garoua Riao Eseka Ngoazik Ngbala Melong Mundame Yabassi Gouri (mm/year) Oya Chad Nyong Ntem Congo Niger - Basin - Year Sanaga Basin Niger Basin North-Coastal Basins North Basin Basin Basin South 1944 1,831 1,626 2,041 1,562 1,324 1,014 1,130 1,876 1,799 1,634 2,286 2,436 2,292 2,207 567 1945 1,548 1,398 1,633 1,322 1,189 1,100 1,133 1,767 2,020 1,921 1,896 2,259 1,814 1,952 724 1946 1,458 1,396 1,490 1,556 1,199 1,089 1,168 1,520 1,686 1,491 1,955 2,424 1,877 1,901 859 1947 1,635 1,585 1,777 1,548 1,729 1,082 1,198 1,544 1,953 1,633 2,188 2,412 2,086 2,316 653 1948 1,634 1,488 1,849 1,565 1,223 1,221 1,342 1,647 1,819 2,070 2,283 2,433 2,186 2,369 602 1949 1,851 1,669 2,056 1,582 1,619 989 1,094 1,871 1,809 1,857 2,219 2,503 2,203 2,228 527 1950 1,677 1,490 1,798 1,406 1,330 1,153 1,337 1,805 1,817 1,720 2,053 2,617 2,125 1,872 764 1951 1,769 1,582 1,975 1,360 1,633 1,117 1,163 1,773 2,127 1,692 2,320 2,697 2,282 2,328 736 1952 1,787 1,625 1,867 1,452 1,847 984 1,042 1,874 2,320 1,776 2,254 2,636 2,304 2,186 785 1953 1,742 1,619 1,855 1,562 1,474 1,086 1,184 1,685 1,616 1,506 2,148 2,839 2,298 2,007 750 1954 2,053 1,870 2,419 1,969 1,797 1,058 1,152 1,840 1,787 1,651 2,544 3,120 2,558 2,649 786 1955 1,935 1,718 2,135 1,616 1,520 1,163 1,253 1,967 1,588 1,695 2,208 3,150 2,437 2,175 747 1956 1,826 1,692 1,853 1,688 1,483 1,084 1,220 1,858 2,080 1,596 2,309 3,325 2,547 1,929 797 1957 1,901 1,711 2,075 1,620 1,601 1,162 1,260 1,851 1,752 1,978 2,698 3,198 2,684 2,579 854 1958 1,583 1,472 1,726 1,490 1,464 1,100 1,188 1,496 1,467 1,435 1,998 2,636 2,029 2,043 747 1959 1,650 1,460 1,812 1,516 1,595 1,106 1,139 1,664 1,748 1,677 2,266 2,970 2,332 2,072 885 1960 1,850 1,675 1,961 1,696 1,632 1,290 1,392 1,871 1,912 1,711 2,337 2,836 2,332 2,292 802 1961 1,613 1,428 1,896 1,432 1,539 969 1,009 1,463 1,410 1,384 1,789 2,823 2,191 2,081 844 1962 1,896 1,758 1,946 1,716 1,728 1,181 1,298 2,021 1,813 1,934 2,007 3,252 2,388 2,164 694 1963 1,582 1,500 1,648 1,536 1,453 1,201 1,262 1,678 1,812 1,775 1,739 2,575 1,979 2,274 663 1964 1,889 1,717 1,964 1,889 1,557 976 1,072 1,991 1,777 1,637 2,083 2,698 2,303 2,187 582 1965 1,706 1,567 1,799 1,604 1,537 1,053 1,192 1,780 1,755 1,614 2,224 3,125 2,196 2,045 668 1966 1,899 1,707 2,012 1,716 1,738 1,071 1,198 2,033 1,842 1,583 2,402 2,901 2,463 2,236 658 1967 1,713 1,539 1,763 1,519 1,650 1,035 1,112 1,801 1,721 1,521 2,416 2,651 2,367 2,102 751 1968 1,692 1,548 1,822 1,478 1,620 1,068 1,167 1,691 1,843 1,553 2,184 2,956 2,261 2,153 694 1969 1,963 1,775 2,076 1,688 1,836 1,245 1,351 1,987 1,818 1,722 2,698 3,291 2,645 2,362 662 1970 1,762 1,542 1,823 1,609 1,439 1,011 1,103 1,980 1,819 1,655 1,994 2,735 2,254 1,931 710 1971 1,736 1,561 1,792 1,490 1,488 966 1,034 1,836 1,514 1,633 2,490 2,704 2,407 1,699 556 1972 1,719 1,596 1,842 1,498 1,476 1,077 1,142 1,800 1,724 1,439 1,914 2,596 2,057 1,908 636 1973 1,728 1,587 1,741 1,561 1,457 1,042 1,112 1,911 1,645 1,574 1,838 2,125 1,947 1,694 469 1974 1,793 1,627 1,867 1,582 1,542 1,010 1,110 1,812 1,656 1,837 2,343 2,425 2,392 2,033 676 1975 1,692 1,514 1,850 1,520 1,370 1,158 1,250 1,718 1,542 1,573 2,229 2,183 2,244 1,929 756 1976 1,777 1,630 1,959 1,667 1,469 1,092 1,193 1,669 1,739 1,532 2,727 2,428 2,480 2,150 740 1977 1,533 1,393 1,638 1,435 1,398 919 1,031 1,523 1,663 1,314 2,355 2,336 2,127 1,906 587 1978 1,889 1,784 1,993 1,764 1,604 1,256 1,394 1,754 1,654 1,573 2,156 2,807 2,315 1,947 753 1979 1,693 1,518 1,904 1,596 1,572 1,068 1,203 1,641 1,548 1,514 2,198 2,872 2,227 2,252 666 1980 1,743 1,639 1,804 1,630 1,439 1,085 1,179 1,800 1,699 1,635 2,201 3,031 2,207 2,119 619 1981 1,610 1,439 1,704 1,446 1,320 1,031 1,110 1,681 1,626 1,523 2,168 2,587 2,191 2,066 623 1982 1,787 1,620 1,898 1,737 1,636 1,057 1,169 1,755 1,718 1,701 2,548 2,644 2,431 1,997 503 1983 1,371 1,194 1,519 1,119 1,175 794 895 1,401 1,457 1,346 1,903 2,419 1,869 1,779 434 1984 1,748 1,525 1,862 1,437 1,457 788 916 1,988 1,822 1,645 2,008 2,044 2,016 1,965 442 1985 1,810 1,665 1,771 1,591 1,680 929 1,046 2,046 1,893 1,839 2,051 2,430 2,062 1,939 510 1986 1,487 1,306 1,701 1,427 1,242 951 1,022 1,472 1,471 1,427 2,111 2,486 2,077 1,893 527 1987 1,583 1,377 1,723 1,389 1,303 865 969 1,680 1,668 1,433 2,121 2,172 1,991 1,589 480 1988 1,746 1,598 1,786 1,501 1,491 1,120 1,200 1,799 1,877 1,477 2,177 2,487 2,235 1,817 676 1989 1,486 1,378 1,580 1,456 1,340 992 1,093 1,504 1,681 1,534 1,882 2,419 1,894 1,976 709 1990 1,702 1,538 1,864 1,591 1,518 1,051 1,185 1,703 1,844 1,785 2,173 2,600 2,210 2,048 464 1991 1,579 1,496 1,658 1,556 1,485 1,117 1,167 1,508 1,498 1,618 2,129 2,451 2,093 1,830 727 1992 1,678 1,592 1,844 1,684 1,477 1,102 1,223 1,462 1,518 1,493 2,100 2,614 2,186 2,052 726 1993 1,751 1,534 1,955 1,546 1,390 1,050 1,168 1,852 1,563 1,746 2,319 2,689 2,228 2,231 542 1994 1,585 1,467 1,786 1,548 1,331 1,270 1,317 1,436 1,729 1,633 2,071 2,400 2,127 1,927 860 1995 1,483 1,379 1,520 1,394 1,547 1,141 1,227 1,460 1,541 1,577 1,873 2,465 1,932 1,610 680 1996 1,754 1,582 2,039 1,693 1,475 1,104 1,196 1,607 1,632 1,525 2,365 2,490 2,354 2,279 596 1997 1,883 1,750 2,032 1,825 1,663 1,187 1,322 1,772 1,458 1,532 2,444 3,129 2,509 2,025 569 1998 1,680 1,557 1,781 1,521 1,554 1,023 1,115 1,648 1,688 1,487 2,139 2,525 2,212 1,966 747 1999 1,739 1,572 1,883 1,604 1,515 1,045 1,117 1,750 1,713 1,675 2,240 2,682 2,259 2,096 831 2000 1,869 1,713 1,959 1,710 1,703 1,179 1,288 1,996 2,234 1,936 2,270 2,482 2,228 2,141 744 2001 1,683 1,490 1,808 1,449 1,361 1,038 1,102 1,754 1,584 1,524 2,165 2,477 2,205 2,009 727 2002 1,834 1,734 1,892 1,793 1,702 1,188 1,327 1,863 1,812 1,574 2,183 2,478 2,235 2,051 651 2003 1,792 1,611 1,917 1,564 1,464 1,386 1,466 1,838 1,581 1,780 2,329 3,384 2,418 2,313 950 2004 1,596 1,389 1,726 1,375 1,283 1,002 1,082 1,701 1,691 1,432 2,155 2,849 2,222 1,949 643 2005 1,588 1,450 1,655 1,424 1,403 991 1,088 1,673 1,596 1,495 1,954 2,326 1,994 1,787 640 2006 1,720 1,501 1,823 1,461 1,381 1,110 1,184 1,865 1,885 1,584 2,283 2,953 2,368 2,026 790 2007 1,655 1,506 1,741 1,516 1,510 1,158 1,241 1,714 1,888 1,554 2,083 2,727 2,162 1,940 818 2008 1,638 1,457 1,717 1,433 1,395 1,061 1,124 1,793 1,944 1,652 2,094 2,653 2,157 1,945 769 2009 1,631 1,506 1,690 1,491 1,501 1,046 1,132 1,684 1,690 1,582 2,024 2,520 2,053 1,909 694 Surf. Area (km2) 131,500 76,000 42,300 20,200 11,100 64,000 30,650 21,600 18,100 38,600 2,280 2,420 8,026 2,240 27,470 Avg 1944-2009 1,715 1,560 1,838 1,556 1,498 1,077 1,171 1,746 1,736 1,623 2,186 2,660 2,224 2,052 682 St. dev 133 126 158 139 155 107 111 164 181 154 209 304 187 203 116 Coeff of Var. 0.077 0.081 0.086 0.090 0.103 0.099 0.095 0.094 0.104 0.095 0.096 0.114 0.084 0.099 0.170 E0 1,631 1,658 1,609 1,710 1,675 1,934 1,901 1,548 1,572 1,632 1,472 1,499 1,526 1,526 2,061 Aridity index 0.95 1.06 0.88 1.10 1.12 1.80 1.62 0.89 0.91 1.01 0.67 0.56 0.69 0.74 3.02 Avg (Q-data) 1,723 1,575 1,855 1,559 1,524 1,091 1,185 1,810 1,717 1,636 2,237 2,814 2,322 2,123 St. dev (Q-data) 136 130 156 138 143 87 102 149 147 153 271 333 188 221 Coeff of Var. 0.079 0.083 0.084 0.088 0.094 0.080 0.086 0.082 0.086 0.093 0.121 0.118 0.081 0.104 Aridity index 0.95 1.05 0.87 1.10 1.10 1.77 1.60 0.86 0.92 1.00 0.66 0.53 0.66 0.97 1944- 1951- 1951- 1959- 1951- 1945- 1950- 1951- 1954- 1956- 1951- 1953- 1951- 1951- Period Q-data 2003 2003 2003 2008 2003 1980 1980 1976 1979 1976 1976 1976 1976 1976 Table 1: Annual precipitation data for selected sub-catchments for the period 1944-2009 (data represent the annual precipitation for the catchments upstream of the indicated flow gauging stations) 136 Annual Betare- Lake Edea Nachtigal Goura Mbakaou Garoua Riao Eseka Ngoazik Ngbala Melong Mundame Yabassi Gouri temperature (0 C) Oya Chad Basin - Nyong Ntem Congo Niger - Year Sanaga Basin Niger Basin North-Coastal Basins North Basin Basin Basin South 1944 24.1 23.9 23.4 23.5 23.5 27.2 26.8 24.6 24.4 24.4 21.7 25.0 23.9 22.5 27.9 1945 24.0 23.8 23.2 23.4 23.4 27.0 26.6 24.4 24.3 24.3 21.5 24.9 23.8 22.3 27.6 1946 24.0 23.9 23.2 23.6 23.6 27.2 26.8 24.5 24.2 24.3 21.5 24.8 23.8 22.4 27.8 1947 24.4 24.2 23.6 23.8 23.8 27.4 27.0 24.8 24.7 24.6 21.9 25.3 24.2 22.8 28.0 1948 23.7 23.5 22.9 23.1 23.1 26.7 26.3 24.2 24.1 24.0 21.2 24.7 23.5 22.1 27.4 1949 24.1 23.9 23.4 23.5 23.4 27.1 26.7 24.6 24.5 24.3 21.7 25.2 24.0 22.6 27.8 1950 23.7 23.4 22.9 23.1 23.0 26.6 26.2 24.2 24.1 24.0 21.3 24.7 23.6 22.1 27.3 1951 23.8 23.8 22.9 23.4 23.7 27.1 26.7 24.3 24.2 24.3 21.1 24.5 23.4 21.9 27.7 1952 23.8 23.8 22.8 23.4 23.8 27.2 26.9 24.2 23.8 24.1 21.0 24.3 23.3 21.9 27.7 1953 23.6 23.6 22.6 23.2 23.5 27.0 26.6 24.0 23.9 24.2 20.8 24.2 23.1 21.7 27.7 1954 23.6 23.6 22.8 23.3 23.4 27.2 26.8 24.0 24.0 24.2 21.0 24.2 23.2 21.9 27.8 1955 23.4 23.3 22.5 23.0 23.1 26.8 26.4 23.8 23.8 23.8 20.7 24.1 23.0 21.6 27.4 1956 23.6 23.6 22.6 23.2 23.5 27.1 26.7 23.9 23.8 23.9 20.8 24.1 23.1 21.8 27.6 1957 23.8 23.8 22.8 23.4 23.9 27.2 26.8 24.3 24.2 24.3 21.0 24.4 23.3 21.9 27.7 1958 24.0 24.0 23.0 23.7 24.1 27.5 27.1 24.4 24.4 24.5 21.2 24.6 23.5 22.2 27.9 1959 23.9 23.9 22.9 23.6 23.9 27.2 26.9 24.3 24.4 24.4 21.1 24.5 23.4 22.1 27.5 1960 23.9 23.9 23.0 23.6 23.8 27.3 26.9 24.3 24.4 24.4 21.2 24.5 23.5 22.2 27.8 1961 23.5 23.4 22.6 23.1 23.2 26.6 26.3 24.0 24.0 24.1 20.9 24.3 23.2 21.8 26.9 1962 23.6 23.5 22.8 23.2 23.2 27.0 26.6 24.0 24.3 24.2 21.1 24.4 23.3 22.1 27.4 1963 24.0 23.9 23.2 23.6 23.6 27.3 27.0 24.4 24.4 24.3 21.5 24.9 23.7 22.4 27.8 1964 23.6 23.4 22.8 23.1 23.1 26.8 26.4 24.0 24.2 24.0 21.0 24.4 23.3 22.0 27.4 1965 23.6 23.5 22.7 23.1 23.2 27.0 26.6 24.0 24.1 24.2 21.0 24.4 23.3 21.9 27.4 1966 23.9 23.7 23.0 23.4 23.5 27.0 26.7 24.3 24.3 24.3 21.3 24.7 23.6 22.2 27.8 1967 23.6 23.4 22.9 23.1 23.0 26.8 26.4 24.1 24.2 23.9 21.2 24.5 23.5 22.1 27.2 1968 23.8 23.6 23.0 23.2 23.3 27.0 26.6 24.2 24.2 24.1 21.3 24.7 23.5 22.1 27.6 1969 24.1 24.0 23.4 23.7 23.6 27.5 27.0 24.5 24.5 24.4 21.7 25.0 24.0 22.6 28.3 1970 24.0 23.8 23.1 23.5 23.6 27.3 26.9 24.4 24.4 24.4 21.4 24.8 23.7 22.2 27.9 1971 23.3 23.1 22.6 22.8 22.6 26.5 26.0 23.8 23.7 23.4 21.0 24.4 23.3 21.9 27.4 1972 23.7 23.5 23.0 23.2 23.1 27.0 26.6 24.1 23.9 23.9 21.4 24.8 23.7 22.3 28.0 1973 24.1 23.9 23.4 23.5 23.5 27.4 26.9 24.5 24.0 24.0 21.7 25.1 24.0 22.6 28.5 1974 23.3 23.0 22.6 22.7 22.5 26.3 25.8 23.8 23.4 23.3 20.9 24.4 23.2 21.8 27.6 1975 23.3 23.1 22.6 22.7 22.6 26.5 26.0 23.8 23.6 23.3 20.9 24.3 23.2 21.8 27.6 1976 23.2 22.9 22.4 22.6 22.5 26.3 25.8 23.6 23.2 23.3 20.8 24.3 23.1 21.8 27.7 1977 23.7 23.5 23.0 23.1 23.0 26.6 26.2 24.2 23.8 23.8 21.4 24.8 23.7 22.2 27.4 1978 23.6 23.3 22.9 23.0 22.8 26.7 26.2 24.1 23.6 23.5 21.3 24.7 23.6 22.2 27.8 1979 24.0 23.8 23.4 23.5 23.3 27.2 26.8 24.5 24.2 24.0 21.8 25.1 24.0 22.6 28.2 1980 23.9 23.7 23.3 23.3 23.2 27.0 26.6 24.4 24.2 23.9 21.7 25.1 24.0 22.6 28.1 1981 23.7 23.5 22.9 23.1 23.0 26.8 26.3 24.2 24.0 23.8 21.3 24.7 23.6 22.2 27.8 1982 23.5 23.4 22.8 23.1 23.0 26.9 26.5 23.9 23.6 23.7 21.1 24.5 23.4 22.0 28.0 1983 23.9 23.7 23.1 23.4 23.4 27.2 26.8 24.3 24.1 24.1 21.3 24.8 23.6 22.3 28.1 1984 23.6 23.5 23.0 23.2 23.0 27.3 26.8 24.0 23.8 23.7 21.2 24.7 23.5 22.2 28.4 1985 23.7 23.6 23.0 23.3 23.2 27.1 26.7 24.1 23.7 23.7 21.3 24.8 23.6 22.2 28.0 1986 23.8 23.6 23.0 23.4 23.3 27.4 26.9 24.1 23.7 23.9 21.3 24.6 23.6 22.2 28.4 1987 24.3 24.1 23.6 23.8 23.7 27.6 27.1 24.7 24.3 24.3 21.9 25.4 24.2 22.8 28.5 1988 23.9 23.8 23.2 23.5 23.5 27.2 26.8 24.4 24.0 24.1 21.5 24.9 23.8 22.4 28.1 1989 23.7 23.5 22.9 23.1 23.1 26.6 26.2 24.3 24.1 23.9 21.3 24.8 23.6 22.1 27.4 1990 24.3 24.2 23.5 23.9 23.9 27.7 27.3 24.7 24.4 24.5 21.8 25.1 24.1 22.7 28.6 1991 24.1 23.9 23.3 23.6 23.6 27.4 27.0 24.5 24.3 24.3 21.6 25.0 23.9 22.5 28.4 1992 23.8 23.6 23.0 23.2 23.2 26.9 26.5 24.3 24.1 24.0 21.4 24.7 23.7 22.2 27.7 1993 23.9 23.7 23.2 23.4 23.3 27.2 26.7 24.3 24.0 24.0 21.5 24.9 23.8 22.4 28.1 1994 23.9 23.7 23.2 23.4 23.3 27.1 26.6 24.4 24.1 24.1 21.5 24.8 23.8 22.3 27.9 1995 24.1 23.8 23.4 23.5 23.4 27.1 26.6 24.6 24.4 24.4 21.8 25.2 24.1 22.6 28.0 1996 24.0 23.7 23.3 23.4 23.3 27.3 26.8 24.4 24.3 24.0 21.7 25.1 24.0 22.6 28.3 1997 24.2 24.1 23.4 23.8 23.8 27.6 27.2 24.6 24.4 24.5 21.6 25.0 23.9 22.5 28.5 1998 24.7 24.5 23.9 24.2 24.2 28.0 27.5 25.1 24.9 24.8 22.2 25.6 24.5 23.2 28.9 1999 24.0 23.8 23.3 23.5 23.3 27.3 26.8 24.4 24.2 24.1 21.6 25.0 23.9 22.5 28.3 2000 23.8 23.6 23.1 23.2 23.2 27.0 26.5 24.4 24.2 24.1 21.5 24.9 23.8 22.4 28.0 2001 23.8 23.6 23.1 23.3 23.2 27.1 26.7 24.3 24.1 24.0 21.4 24.8 23.7 22.3 28.0 2002 24.0 23.8 23.2 23.5 23.4 27.4 27.0 24.4 24.3 24.2 21.6 25.0 23.9 22.5 28.4 2003 24.2 24.0 23.5 23.7 23.6 27.5 27.0 24.7 24.5 24.4 21.9 25.4 24.2 22.7 28.5 2004 24.2 24.0 23.5 23.7 23.6 27.6 27.1 24.6 24.2 24.3 21.8 25.1 24.0 22.7 28.9 2005 24.4 24.2 23.7 23.9 23.8 27.9 27.4 24.8 24.4 24.4 22.0 25.4 24.3 22.9 28.9 2006 24.1 23.9 23.3 23.5 23.5 27.5 27.0 24.5 24.3 24.2 21.7 25.1 24.0 22.6 28.5 2007 24.2 24.0 23.4 23.7 23.6 27.6 27.1 24.5 24.3 24.2 21.7 25.1 24.0 22.6 28.6 2008 23.9 23.8 23.2 23.4 23.4 27.3 26.8 24.4 24.1 24.0 21.5 25.0 23.8 22.4 28.4 2009 24.5 24.4 23.8 24.1 24.0 28.1 27.7 24.8 24.5 24.5 22.0 25.4 24.3 23.0 29.2 Surf. Area (km2) 131,500 76,000 42,300 20,200 11,100 64,000 30,650 21,600 18,100 38,600 2,280 2,420 8,026 2,240 27,470 Avg 1944-2009 23.9 23.7 23.1 23.4 23.4 27.1 26.7 24.3 24.1 24.1 21.4 24.8 23.7 22.3 28.0 St. dev 0.30 0.31 0.32 0.32 0.36 0.38 0.37 0.29 0.30 0.31 0.34 0.35 0.34 0.34 0.47 Coeff of Var. 0.013 0.013 0.014 0.014 0.015 0.014 0.014 0.012 0.013 0.013 0.016 0.014 0.014 0.015 0.017 Table 2: Annual temperature data for selected sub-catchments for the period 1944-2009 (data represent the average annual temperature for the catchments upstream of the indicated flow gauging stations) 137 Runoff Edea Nachtigal Goura Mbakaou Betare-Oya Garoua Riao Eseka Ngoazik Ngbala Melong Mundame Yabassi Gouri (mm/year) Niger - Year Sanaga Basin Niger Basin Nyong Basin Ntem Basin Congo Basin North-Coastal Basins South 1944 436 1945 390 145 1946 358 249 1947 434 174 1948 522 275 1949 578 140 1950 532 153 258 1951 535 512 587 504 155 232 390 925 1,152 1952 495 499 503 540 141 210 431 1,066 1,126 1953 436 422 482 482 135 207 328 910 2,263 1,118 1954 541 519 632 588 186 286 387 364 1,276 2,397 1,278 1955 584 559 636 616 259 390 356 401 1,140 2,818 1,334 1956 561 549 598 514 188 305 480 581 355 1,076 2,647 1,390 1957 549 533 653 472 194 284 399 505 501 1,247 2,488 1,507 1958 445 485 540 524 138 179 314 270 251 910 1,918 1,253 1959 466 418 551 646 494 158 210 383 474 368 1,036 2,402 1,251 1960 511 459 586 709 509 289 416 411 614 444 979 2,167 1,301 1961 449 401 499 587 440 212 284 303 397 254 831 2,018 712 1962 550 468 622 664 470 214 324 529 535 365 1,203 2,418 1,314 1963 455 441 442 596 476 218 285 457 599 447 859 1,707 992 1964 512 488 540 721 457 161 256 455 601 425 980 2,259 1,328 1,732 1965 482 447 539 667 420 181 224 426 566 1,091 2,385 1,269 1,761 1966 537 470 621 720 505 186 275 538 708 984 2,209 1,211 1967 459 392 550 574 514 138 181 354 495 1,040 2,432 1,290 1968 472 424 554 586 517 170 254 377 579 367 916 2,502 1,170 1969 645 609 660 813 703 261 369 493 569 450 1,075 2,440 1,494 1,753 1970 504 464 509 664 560 219 319 489 536 337 866 1,884 1,193 1,315 1971 375 349 418 499 451 143 212 403 495 291 939 1,881 1,248 1,303 1972 331 305 386 403 367 80 101 355 404 264 646 1,964 1,045 1,291 1973 358 360 384 409 376 136 181 391 389 283 599 1,441 774 1,007 1974 453 417 506 633 464 131 230 357 438 390 905 1,964 1,198 1,431 1975 417 385 419 530 469 223 355 346 412 280 812 1,686 1,110 1,360 1976 502 449 528 691 479 124 218 322 459 300 1,070 2,172 1,577 1,546 1977 375 338 451 538 444 130 220 387 1,353 1978 501 441 526 669 487 179 313 422 1,436 1979 437 412 496 629 556 83 120 409 1,453 1980 460 346 499 544 585 166 231 1,401 1981 430 417 449 539 647 1982 473 306 466 630 513 1983 274 234 310 409 354 1984 390 357 367 456 321 1985 449 346 392 556 443 1986 325 281 330 408 389 1987 309 296 294 422 359 1988 387 360 279 373 475 1989 412 380 378 532 527 1990 439 403 385 539 495 1991 456 416 353 643 506 1992 455 416 358 624 478.08 1993 455 416 406 620 466.13 1994 459 419 414 650 480.56 1995 343 324 374 464 337.63 1996 448 410 391 687 386.82 1997 359 336 433 556 344.68 1998 410 392 429 626 419.68 1999 483 439 463 611 407.26 2000 402 372 469 532 352.75 2001 342 324 373 474 303.63 2002 419 386 494 508 412.58 2003 441 404 460 568 462.52 2004 484 2005 525 2006 472 2007 558 2008 534 2009 Avg (Q-data) 450 411 471 576 469 176 256 403 485 354 976 2,186 1,217 1,439 St. dev (Q-data) 74 74 97 100 82 50 73 65 100 77 161 332 195 207 Coeff of Var. 0.16 0.18 0.21 0.17 0.17 0.29 0.28 0.16 0.21 0.22 0.16 0.15 0.16 0.14 Surf. Area (km2) 131,500 76,000 42,300 20,200 11,100 64,000 30,650 21,600 18,100 38,600 2,280 2,420 8,026 2,240 Period Q-data 1944-2003 1951-2003 1951-2003 1959-2008 1951-2003 1945-1980 1950-1980 1951-1976 1954-1979 1956-1976 1951-1976 1953-1976 1951-1976 1964-1980 Table 3: Annual runoff data for selected sub-catchments for the period 1944-2009 (data represent the annual runoff for the catchments upstream of the indicated flow gauging stations) 138 Betare- Station Edea Nachtigal Goura Mbakaou Garoua Riao Eseka Ngoazik Ngbala Melong Mundame Yabassi Gouri Lake Chad Oya Basin - Nyong Ntem Congo Niger - North Basin Sanaga Basin Niger Basin North-Coastal Basins Basin Basin Basin South Surf. Area (km2) 131,500 76,000 42,300 20,200 11,100 64,000 30,650 21,600 18,100 38,600 2,280 2,420 8,026 2,240 27,470 Precipitation (mm); source: CRU TS 3.10 January 12 7 7 3 6 0 0 25 58 40 15 27 16 7 0 February 26 18 21 10 13 1 0 51 81 57 36 53 36 30 1 March 102 80 104 65 68 12 16 140 177 140 133 148 132 112 2 April 151 134 159 128 119 50 63 179 173 164 188 199 182 173 12 May 197 183 203 179 167 111 124 215 209 190 220 252 221 213 52 June 189 180 210 191 175 149 160 166 122 139 243 305 241 249 81 July 194 195 243 236 206 212 224 106 42 85 302 434 296 323 167 August 226 219 271 247 237 254 259 146 56 117 341 439 345 335 228 September 292 267 324 268 251 198 215 278 217 233 371 411 374 361 114 October 264 233 263 213 208 79 97 307 310 268 289 322 306 250 22 November 71 52 60 35 41 3 5 120 193 136 74 108 87 51 0 December 13 9 7 3 8 0 0 29 77 50 13 28 16 5 0 Annual 1,738 1,578 1,872 1,579 1,500 1,067 1,164 1,761 1,715 1,619 2,227 2,725 2,251 2,110 678 Temperature ( 0C); source: CRU TS 3.10 January 23.9 23.3 23.2 22.6 23.2 25.1 25.0 24.7 24.3 23.9 21.8 25.0 24.1 22.4 23.9 February 25.1 24.8 24.4 24.3 24.5 27.6 27.4 25.5 25.0 25.0 22.7 25.9 25.0 23.6 26.7 March 25.5 25.7 24.8 25.6 25.4 30.6 30.2 25.2 24.8 25.1 22.5 25.9 24.9 24.1 30.5 April 25.4 25.4 24.8 25.4 25.0 31.0 30.3 25.3 24.8 25.1 23.0 26.0 25.0 24.6 32.7 May 24.3 24.3 23.5 24.1 23.8 29.3 28.5 24.7 24.7 24.7 21.7 25.2 24.1 22.8 31.9 June 23.4 23.3 22.6 23.0 22.9 27.1 26.4 23.9 23.9 23.7 21.0 24.5 23.3 21.8 29.7 July 22.6 22.5 21.8 22.2 22.1 25.6 25.1 23.0 22.9 22.9 20.1 23.5 22.3 20.8 27.3 August 22.6 22.5 21.9 22.2 21.9 25.1 24.7 23.1 22.9 23.0 20.2 23.4 22.4 20.7 26.1 September 23.0 22.8 22.1 22.5 22.3 25.7 25.3 23.6 23.8 23.7 20.3 23.9 22.7 21.1 27.2 October 23.5 23.4 22.6 23.1 22.9 27.0 26.6 23.8 23.9 23.8 20.9 24.3 23.2 21.9 28.4 November 23.8 23.5 22.9 23.2 23.2 26.6 26.3 24.3 24.0 24.1 21.2 25.0 23.7 22.1 26.8 December 23.4 22.8 22.7 22.2 22.7 25.1 24.9 24.2 24.0 23.6 21.4 24.9 23.7 21.9 24.5 Annual 23.9 23.7 23.1 23.4 23.3 27.1 26.7 24.3 24.1 24.1 21.4 24.8 23.7 22.3 28.0 Potential evapotranspiration (mm); source: Climate Wizard Sanaga Basin Niger Basin Nyong/Ntem Congo Lake Chad January 147 148 146 155 150 162 163 133 134 138 133 132 137 141 159 February 145 148 145 154 150 167 167 131 130 136 131 133 136 139 165 March 160 163 157 169 164 200 196 148 151 154 140 146 148 148 210 April 146 148 143 151 147 187 180 143 144 150 132 139 140 136 210 May 137 139 135 142 139 173 166 134 138 143 127 132 133 131 200 June 121 123 119 124 123 146 141 117 117 124 110 114 114 115 173 July 114 115 113 120 118 137 134 112 114 120 106 108 108 110 155 August 121 122 120 125 122 135 134 121 124 129 111 110 114 115 143 September 127 129 127 131 128 140 138 125 129 134 118 116 119 123 152 October 137 139 134 142 139 159 155 131 138 140 123 125 128 127 177 November 134 138 130 143 142 164 162 125 128 131 116 119 121 123 167 December 143 147 140 153 153 163 165 128 128 135 123 125 129 133 158 Annual 1,631 1,658 1,609 1,710 1,675 1,934 1,901 1,548 1,572 1,632 1,472 1,499 1,526 1,541 2,069 Catchment runoff (mm/yr) January 14 15 14 12 20 1 1 21 20 16 24 53 30 26 February 7 10 10 6 12 0 0 11 15 11 16 43 19 21 March 7 11 11 3 11 0 0 14 24 15 21 61 24 24 April 9 11 14 5 13 0 0 23 39 20 33 93 34 46 May 16 16 20 16 19 0 0 33 50 28 45 128 47 77 June 24 22 28 32 30 3 3 36 45 28 62 190 74 122 July 38 35 49 67 50 13 19 28 28 21 109 293 135 228 August 56 51 64 109 68 46 72 21 13 18 152 355 201 260 September 85 75 87 134 93 71 106 33 32 31 206 375 261 272 October 108 90 100 122 100 32 45 64 83 61 185 320 228 255 November 62 52 53 49 54 6 7 73 85 62 85 181 111 122 December 25 23 23 21 30 2 2 43 48 32 40 87 52 47 Annual 450 411 471 576 499 176 256 402 481 344 976 2,178 1,217 1,500 Table 4: Monthly precipitation, temperature, potential evapotranspiration and runoff data for selected sub-catchments for the period 1901-2009 (data represent averages over the catchments upstream of the indicated flow gauging stations) 139 Annex 6: Turc-Pike model for assessment of climate change impacts on annual runoff 1. Aridity index Arora (2002) uses the aridity index φ = E0/P, i.e. the ratio of annual potential evapotranspiration (E0) to precipitation (P), to assess climate change impacts on annual runoff. Simple analytic expressions based solely on the aridity index of a basin are used to estimate changes in runoff due to changes in precipitation and (temperature driven) changes in potential evapotranspiration, as a first order estimate of the effect of climate change on annual runoff. Precipitation and available energy (expressed in terms of potential evapotranspiration E0) largely determine the actual annual evapotranspiration (E) and runoff (Q) rates in a region. The aridity index φ has been shown to describe the evaporation ratio E/P and the runoff coefficient Q/P (= 1-E/P) of catchments for a range of climatic regimes, in a number of studies. Arora (2002) also uses the aridity index to obtain analytic equations for the relative changes in annual runoff due to relative changes in annual precipitation and available energy, i.e. the precipitation elasticity εP = [dQ/Q]/[dP/P] and the evaporation elasticity εE0 = [dQ/Q]/[dE0/E0] of runoff. The latter is then used to derive the temperature elasticity of runoff, εT = [dQ/Q]/[dT/T]. The author discusses five functional forms, which describe the actual evapotranspiration ratio E/P as a function of φ, and assesses how the results from the Canadian Centre for Climate Modeling and Analysis (CCCma) third-generation atmospheric GCM (AGCM3) compared with those estimated by these five functional forms (see Fig. 1). For low precipitation and a high aridity index (as in Northern Cameroon) actual evapotranspiration tends to be equal to precipitation (E/P = 1), while for the reverse (low φ and highly humid conditions) actual evapotranspiration tends to equal the potential evapotranspiration (E = E0 and E/P = φ). Seasonal rainfall is one of the reasons causing deviations from the theoretical values. Fig. 1: (a) Comparison of evaporation ratio curves predicted by five functional forms, and (b) comparison of data from the CCCma AGCM3 model with these forms (source: Arora, 2002) 140 2. The Turc-Pike rainfall runoff model Because of its mathematical convenience, and since the differences between several functional forms are small compared to the random noise in actual precipitation and runoff data, we will use here the form introduced by Turc (1954) and Pike (1964), as follows: E/P = [1+ φ-2]-0.5 (evaporation ratio) Q/P = 1 – E/P (runoff coefficient) which yields after differentiation and manipulation: dQ/Q = (1 + β) dP/P – β dE0/E0 εP = 1 + β; precipitation elasticity of runoff εE0 = -β; potential evapotranspiration elasticity of runoff β = [1 + φ2]-1 / {[1 + φ-2]0.5 – 1} = (1 + E/P) E/P = 2 – 3 Q/P + (Q/P)2 For convenience we express the climate elasticities of runoff in terms of the runoff coefficient Q/P: εP = 3 – 3 Q/P + (Q/P)2 εE0 = -2 + 3 Q/P – (Q/P)2 The annual deviation ratios σE/σP and σQ/σP are derived as: σE/σP = [1+ φ-2]-1.5 = (E/P)3 = (1 – Q/P)3 (inter-annual variability) σQ/σP = 1 - σE/σP (valid for high correlation coefficients in annual Q, P and E) Results obtained with the above equations are shown in Figures 2 and 3 and Table 1. Generally, higher rainfall decreases the aridity index φ as well as the climate elasticities of runoff, and increases the runoff coefficient Q/P. The reverse is true for increasing potential evapotranspiration E0 or decreasing rainfall, which increases the aridity index and the climate elasticities of runoff and decreases the runoff coefficient. For for large φ (infinite) we obtain: E/P = 1, Q/P = 0, β = 2, εP = 3 and εE0 = -2. However, for large values of the aridity index, such as for Northern Cameroon (Lake Chad Basin), where the aridity index is about 4 (E0 = 2,000 mm and P = 500 mm), runoff is only a few percents of precipitation and a relatively large climate elasticity has only minor impacts in absolute terms of runoff φ 0.0 0.25 0.5 0.75 1 1.25 1.5 1.75 2 2.5 3 4 5 E/P 0.0 0.243 0.447 0.600 0.707 0.781 0.832 0.868 0.894 0.928 0.949 0.970 0.981 Q/P 1.0 0.757 0.553 0.400 0.293 0.219 0.168 0.132 0.106 0.072 0.051 0.030 0.019 σE/σP 0.0 0.014 0.089 0.216 0.354 0.476 0.576 0.655 0.716 0.800 0.854 0.913 0.943 σQ/σP 1.0 0.986 0.911 0.784 0.646 0.524 0.424 0.345 0.284 0.200 0.146 0.087 0.057 β 0.0 0.301 0.647 0.960 1.207 1.391 1.524 1.622 1.694 1.791 1.849 1.911 1.942 εP 1.0 1.301 1.647 1.960 2.207 2.391 2.524 2.622 2.694 2.791 2.849 2.911 2.942 εEo 0.0 -0.301 -0.647 -0.960 -1.207 -1.391 -1.524 -1.622 -1.694 -1.791 -1.849 -1.911 -1.942 εT (T=24 0 C) 0.0 -0.173 -0.372 -0.551 -0.693 -0.798 -0.875 -0.931 -0.973 -1.028 -1.061 -1.097 -1.115 ST (T=24 0 C) 0.0 -0.7% -1.5% -2.3% -2.9% -3.3% -3.6% -3.9% -4.1% -4.3% -4.4% -4.6% -4.6% Table 1: Theoretical values of actual evapotranspiration, runoff coefficient and climate elasticities 141 Fig. 2: (a) Comparison of evaporation deviation ratio curves (σE/σP) predicted by five functional forms, and (b) comparison of evaporation deviation ratio from CCCma AGCM3 with these forms (source: Arora, 2002) Fig. 3: The sensitivity factor β plotted against the aridity index φ for five functional forms (Arora, 2002) For φ = 0 we obtain: E/P = 0, Q/P = 1, εE0 = 0 and εP = 1. Under cold Nordic conditions with mainly snowmelt as runoff and minimal evaporation the precipitation elasticity of runoff is thus about 1. Under such circumstances ST = εT = 0. 3. Temperature elasticity of potential evapotranspiration Precipitation (P) and available energy (expressed in terms of potential evapotranspiration E0) largely determine the actual annual evapotranspiration (E) and runoff (Q) rates in a region. Potential evapotranspiration, together with precipitation, are the inputs to most hydrological models. For determining the temperature elasticity of runoff, we therefore need to first determine the temperature elasticity of the potential evapotranspiration E0. Xu and Singh (2001) analyzed, compared and generalized the various popular evaporation equations that belong to the category of temperature- based methods for the estimation of E0. For monthly evaporation values, they found that the modified Blaney–Criddle (1950) method and the Hargreaves method, described in Hargreaves et al (1982, 1985, 1994, 2003), Allen et al. (1998) and Droogers and Allen (2002), produced the least errors. The Hargreaves method provides good estimates of the reference crop evaporation compared to estimates derived with the standard Penman-Monteith method (Monteith, 1965). The Hargreaves method as described by W.J. Shuttleworth (1993) reads: 142 E0 = 0.0023 S0 Tmm0.5 (T + 17.8) (mm/day) where: E0 = reference crop evapotranspiration (mm/day) Tmm= difference between mean monthly maximum and mean monthly minimum temperature T = mean monthly temperature (0C) S0 = water equivalent of extra-terrestrial radiation (mm/day) The temperature elasticity and sensitivity of E0 can be easily derived from the Hargreaves equation, respectively as: εE0 = [dE0/E0] / [dT/T] = T / (T + 17.8) SE0 = 1/(T + 17.8) The usual form of the Blaney–Criddle equation (1950), converted to metric units is written as: E0 = 0.46 k p (T + 17.8) where: p is percentage of total daytime hours for the period used (daily or monthly) out of total daytime hours of the year (365*12), and k is a monthly consumptive use coefficient, depending on vegetation type, location and season. For our purpose the Blaney-Criddle and Hargreaves equations are essentially the same, and yield the same values of the temperature elasticity and sensitivity of E0. In the Sanaga, Coastal and Congo basins the average annual temperatures are about 240C, yielding εE0 = 0.574 and SE0 = 2.4% per 10C. The E0 increases in the Hargreaves method are based on a symmetrical increase of minimum and maximum temperatures, although climate change projections show that minima would increase more than maxima, and hence Tmm and thus E0 could be slightly reduced. In addition, all other variables are assumed to remain constant. It is likely that this constitutes an acceptable approach. For instance, increasing minimum temperatures will in reality be accompanied by increasing cloudiness, while higher maximum temperatures would entail more sunshine and thus lower the cloudiness. The two factors probably compensate each other. The same reasoning is applied for the standard Penman-Monteith method, which also uses minimum and maximum monthly temperatures for the estimation of E 0, and a host of other physically based parameters, as described in the following. The standard FAO Penman-Monteith method (Monteith, 1965) for the estimation of potential evapotranspiration (E0) is described by Allen26 et al. (1998) and implemented by Delobel27. The Penman- Monteith E0 applies to vegetated areas, i.e. theoretical grassland. As an example, the current averages of relevant parameters for Niamey on the Niger (representative for the most northern part of 26 http://www.fao.org/docrep/X0490E/X0490E00.htm 27 ftp://ext-ftp.fao.org/SD/Reserved/Agromet/PET/delobel/PETCALC1.xls 143 Cameroon) were taken from New_LocClim28, as presented in Table 2 below. The values correspond to the second decade of the selected months. A blue background indicates that the data are taken directly from New_LocClim, while the yellow background identifies derived data required by Delobel's spreadsheet. Table 2: Average values of variables required to compute the Penman-Monteith E0 from New_LocClim data Table3: Daily E0 values for Niamey, corresponding to various temperature increases The following months were selected to provide “representativeâ€? conditions: ï‚· January, the coldest month ï‚· May: the warmest month and the one with the highest E0 ï‚· August: the wettest month, and the one with the lowest E0 ï‚· November ï‚· E0 increases from August to November to January to May Figure 4: Daily E0 values at Niamey, corresponding to various temperature increases 28 Software and included database are available at ftp://ext-ftp.fao.org/SD/Reserved/Agromet/New_LocClim/ 144 Table 3 shows the E0 values corresponding to various temperature increases (between 0°C and 3°C). For all practical purposes, the increases can be considered linear and vary between 0.12 mm/°C in August to 0.15 mm/°C mm in May. Figure 4 shows the same results as Table 3 in graphical form. The month of August is taken as representative for the rainy season, when most of the basin runoff is generated. E0 for Niamey increases then with 5% for a 20C variation in temperature, equivalent to a temperature sensitivity of E0 of 2.5%. For an average August temperature of 280C at Niamey, the theoretical temperature sensitivity of E0 is 1 / (T+17.8) or 2.2%/0C and the temperature elasticity of E0 is 0.61. 4. Climate elasticities of runoff Combining some of the above equations yields for the temperature elasticity and sensitivity of runoff: εT = [dQ/Q] / [dT/T] = [-2 + 3 Q/P – (Q/P)2]. T / (T + 17.8) SE0 = εT /T For the Sanaga, Congo and Coastal basins with an annual average temperature of 24 0C, the temperature-based Hargreaves (1982) method for the estimation of E0 yields a temperature elasticity of E0 equal to 24/(24+17.8)=0.57 and a temperature sensitivity of E0 = (24+17.8)-1 = 2.4% . Thus, a 20C increase in temperature by 2050 will increase potential evapotranspiration by nearly 5%. For an average annual temperature of 28 0C, the temperature sensitivity of potential evapotranspiration is 2.2% per 10C in the Northern part of the Lake Chad basin in Cameroon; thus, a 20C increase in temperature by 2050 would increase potential evapotranspiration in this semi-arid region by 4.4%. 145 Annex 7: Climate change projections for 2050 and 2080 from the Climatewizard Fig. 1: Changes in annual precipitation projected by 15 GCMs for 2050. Left panel shows results for the 20% dry models; upper right panel shows the average of all GCM projections; lower right panel shows trend in precipitation projections for 2035 - 2065 @ 0.05% per year (source: Climate Wizard, 2013) 146 Fig. 2: Changes in annual precipitation projected by 15 GCMs for 2080. Left panel shows results for the 20% dry models; right panel shows the average of all GCMs (source: Climate Wizard, 2013) 147 Fig. 3: Annual precipitation projected by 15 GCMs for 2050 and 2080 (source: Climate Wizard, 2013) 148 Fig. 4: Changes in annual temperatures projected by 15 GCMs for 2050. Left panel shows results for the 20% coolest models; upper right panel shows results for the 80% warmest models; lower right panel shows the trend in projected temperatures for 2035 - 2065 @ 0.033 0C/year (source: Climate Wizard, 2013) 149 Fig. 5: Changes in annual temperatures projected by 15 GCMs for 2080. Left panel shows results for the 20% coolest models; right panel shows results for the 80% warmest models (source: Climate Wizard, 2013) 150 Fig. 6: Annual temperatures projected by 15 GCMs for 2050 and 2080 (source: Climate Wizard, 2013) 151