Evaluating the Impact of Statistical Bias Correction on Climate Change Signal and Extreme Indices in the Jemma Sub-Basin of Blue Nile Basin

Abstract

This study evaluates the effect of the statistical bias correction techniques of distribution mapping and linear scaling on climate change signals and extreme rainfall indices under different climate change scenarios in the Jemma sub-basin of the Upper Blue Nile Basin. The mean, cumulative distribution function (CDF), mean absolute error (MAE), probability of wet days (Prwet (%)), and 90th percentile (X90 (mm)) of observed rainfall and the regional climate model (RCM) simulations of rainfall with and without statistical bias correction were compared with the historical climate (1981–2005). For future (2071–2100) climate scenarios, the change in climate signal and extreme rainfall indices in the RCM simulations with and without bias correction were also evaluated using different statistical metrics. The result showed that the statistical bias correction techniques effectively adjusted the mean annual and monthly RCM simulations of rainfall to the observed rainfall. However, distribution mapping is effective and better than linear scaling for adjusting the probability of wet days and the 90th percentile of RCM simulations. In future climate scenarios, RCM simulations showed an increase in rainfall. However, the statistically bias-adjusted RCM outputs revealed a decrease in rainfall, which indicated that the statistical bias correction techniques triggered a change in climate signal. Statistical bias correction methods also result in changes in the extreme rainfall indices, such as frequency of wet days (R1mm), number of heavy precipitation days (R10mm), number of very heavy rainfall days (R20mm), and other intensity and frequency indices.