Future rainfall projections for Rwanda using statistical downscaling

Abstract

Rainfall variability over Rwanda is of major concern as the people’s welfare always relies on water availability, food security, environmental sustainability, etc. Policy makers need a complete information on rainfall to plan accordingly. The objective of this research was to project future rainfalls over Rwanda using statistical downscaling method. To achieve this, data from Meteo Rwanda (station data) and CORDEX output historical data were used. Different models were evaluated and one of them showed a better performance compared to others. Graph and statistical methods were used in the process of model validation. The RMSE, MBE, Index of agreement and correlation coefficient are the statistical measures used. Model performance was checked on monthly and seasonal basis while projections were done on annual, seasonal, decadal and climatological basis. CLMcom RCM driven by CNRM-CM5, showed a better performance than other models and was used in projections. As the rainfall distribution over Rwanda follows a topographical pattern, climate zones over Rwanda have been defined and stations were selected so as to cover all the climate zones. Future rainfall projections focused mainly on MAM and OND as they are the two rainy seasons experienced by East African region countries including Rwanda. The findings of this research demonstrated the distribution and variability of rainfall where high amount rains at the western part of the country and reduces when moving to the east. In general, projections showed increasing rainfall patterns under all scenarios at five of the selected seven stations. The decreasing patterns were projected at Gicumbi and Nyamagabe stations. For the seasonal rains, there are fluctuations from a decrease to the increase and vice versa. A significant decrease was projected at Gicumbi during short and long rain seasons while Kamembe, Nyamagabe and Ngoma showed a decrease during MAM. Rubavu station is likely to experience extreme rainfall which may cause disasters like floods and landslides while Gicumbi is likely to experience droughts.