Quantitative assessment of precipitation changes under CMIP5 RCP scenarios over the northern sub-Himalayan region of Pakistan

Abstract

Modelling the probable effect of global warming on precipitation over the northern sub-Himalayan region is very important to ensure sustainable water supply for Pakistan. The aim of the study is to develop statistical downscaling models for the projection of precipitation using the outputs of Coupled Model Intercomparison Project Phase 5 global circulation models and using future scenarios. The models were developed considering the Global Precipitation Climatology Centre precipitation data as model predictands. The downscaling models were developed using non-local model output statistics approach based on support vector machine (SVM). Random Forest was applied to formulate multimodal ensemble (MME) for the projection of precipitation. The accuracy of models was judged using the percentage of bias, normalized root mean square error, and the modified index of agreement (md). Results showed that the SVM downscaling model simulated the temporal and spatial distributions of historical precipitation with high skills. The MME showed variations in the range of − 12.68% to 6.31%, − 9.61% to 3.45%, − 8.70% to 9.15%, and − 9.40% to 5.47% for RCP2.6, RCP4.5, RCP6.0, and RCP8.5 scenarios, respectively. The spatial pattern of annual mean rainfall of MME revealed an expansion of high rainfall area, especially in 2070–2099 under all scenarios.