Added Value of Bias Adjusted and statistical downscaled ISIMIP Models in Simulating Extreme Precipitation Characteristics over West Africa

Abstract

Abstract This paper evaluates the skill of five Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) models in simulating regional variability pattern of precipitation extreme characteristics over West Africa by comparing their simulation outputs with CHIRPS, which is a high resolution observation dataset from a 34 years reference period (1981 to 2014). Based on the findings, it was established that the bias-adjusted ISIMIP models were able to accurately replicate the geographical extent and annual year to year variability pattern of extreme precipitation characteristics over the study region. Moreover, the GCMs were able to captured extreme precipitation patterns over mountainous areas, indicating an added value skill of ISIMIP models over topographical features. The performance of ISIMIP models and their multimodel ensemble in capturing extreme precipitation characteristics was further evaluated using the normalized mean bias error, normalized root mean square error, Taylor skill metric and interannual variability skill score. Our result revealed that MPI-ESM1-2-HR and multi model ensemble mean were judged to be the most promising models for reproducing the extreme precipitation pattern over the two delineated sub regions (Guinea and Sahel) of west Africa and can therefore be used as a basis for selection of models for hydrological impact studies and adaption strategy.