Frequency-intensity-distribution bias correction and trend analysis of high-resolution CMIP6 precipitation data over a tropical river basin

Advancements in computational power have enabled general circulation models (GCMs) to simulate climate variables at a higher resolution. However, GCM outputs often deviate from the observed climatological data and therefore need bias correction (BC) before they are used for impact studies. While there are several BC methods, BCs considering frequency, intensity and distribution of rainfall are few. This study proposes a BC method which focuses on separately correcting the frequency, intensity and distribution of precipitation. This BC was performed on high-resolution daily precipitation simulations of Meteorological Research Institute-Atmospheric General Circulation Model Version 3.2 with a 20-km grid size (MRI-AGCM3-2-S) model which is part of Coupled Model Intercomparison Project Phase 6 (CMIP6) on Netravati basin, a tropical river basin in India. The historical rain gauge station data was considered for testing the effectiveness of the BC method applied. The quantile-quantile (Q-Q) plot, Taylor diagram, Nash-Sutcliffe efficiency (NSE), coefficient of determination (R-2), root mean square error (RMSE), mean absolute error (MAE), percentage bias (PBIAS) and correlation coefficient (R) are employed for the evaluation of the BC method. Higher R and R-2 and lower RMSE, MAE and PBIAS values were observed for the bias-corrected GCM data than raw simulation. The PBIAS reduced from 15.6 to 6% when BC was applied. The analysis suggested that the proposed method effectively corrects the bias in rainfall over the basin. Furthermore, an attempt has been made to analyse the trend of historical and future rainfall in the basin. The analysis revealed a declining trend of precipitation in monsoon months with the magnitude of 12.44 mm and 56.7 mm in the historical and future periods respectively. This study demonstrates that BC should be applied before the use of GCM simulated precipitation for any analysis or impact studies to improve the predictions.