Assessment of CMIP5 and CORDEX-SA experiments in representing multiscale temperature climatology over central India

Abstract

The central India region has been seriously affected by repeated droughts in recent decades due to climate change, which is the main reason for conducting this research. It is still uncertain how the numerous climate models could precisely estimate the future climate for central India. The study mainly focuses on the forcing global climate models (GCMs) and the regional climate models (RCMs). The models have been checked using the coefficient of correlation (r2), Nash-Sutcliffe efficiency (NSE) and an improved method, skill score (SS). The performance is also spatially checked on ArcGIS using the kriging interpolation. The bias-corrected GCMs performed more authentically than the CORDEX RCMs in signifying maximum and minimum temperatures for the Bundelkhand region in central India. Bias-corrected GCMs, EC-EARTH, CCSM4 and GFDL-ESM-2M affirmed the best models on multiple time scales for maximum and minimum temperature in the study region. Maximum NSE and r2 have been observed for seasonal minimum temperature. GCM-EC-EARTH has shown 97% to 98% accuracy, while GCM-GFDL-ESM-2M has demonstrated 84% to 97% accuracy among other selected models. The research outcomes will also assist policymakers in developing strategies and policies for the future climate of central India with the help of more precise projected climatic data. HIGHLIGHTS The study assesses the reliability of the latest generation climate models on multiscale climatology.; The multi-model ensemble mean has been used to deal with uncertainty associated with models.; GCMs outperformed RCMs in predicting temperature.; The models performance could be best judged on the mean monsoon scale.; The outcomes will help to formulate effective climate policies for central India region.;

Keywords

• bilinear interpolation
• cordex
• empirical quantile mapping
• general circulation model
• regional climate model
• skill score