Projections of future rainfall and temperature using statistical downscaling techniques in Tana Basin, Ethiopia

Abstract

Global climate changes are becoming main threats to hydrological cycle, which thus influence environmental, social, and economic systems. Climate change studies using global climate models (GCMs) are mostly used for mitigation and adaptation strategies regarding the changing climate. The current GCMs’ data are, however, too coarse to use directly at the regional and local scales for climate change studies. Two widely used statistical downscaling methods, namely LARS-WG and SDSM models, were used to study the current and future climate change of Tana Basin, Ethiopia. Four GCMs (GFCM21, HadCM3, MPEH5, and NCCCS) for LARS-WG and two GCMs (HadCM3 and CanESM2) for SDSM with different emission scenarios were evaluated. Overall results indicated an acceptable response of the models to simulate and forecast climatic variables under HadCM3 and CanESM2 GCMs. Rainfall results downscaled by LARS-WG from the four GCMs indicated high intermodal variabilities and non-consistence; Increasing trend of rainfall showed on three of the GCMs while one GCM showed a decreasing trend in the range of − 9.6% to 45.2%. The four GCMs rainfall average ensemble value showed an increasing trend ranging from 3.9% to 18.8%, which is also consistent with HadCM3 projections ranging from 4.1% to 19.2%. However, the downscaled results from all four models showed increasing maximum and minimum temperature for all time periods. The mean annual maximum and minimum temperature change increased from 0.9 °C to 2.9 °C and 0.6 °C to 2.5 °C, respectively, while annual mean relative change of rainfall ranged from 9.9% to 44.7%. Both SDSM and LARS-WG methods were obtained good monthly rainfall data series than daily rainfall data series in the study area. However, both models with the selected GCMs (HadCM3 and CanESM2) performed reasonably well to simulate temperature than rainfall.