Evaluation of Bias Correction Methods for Regional Climate Models: Downscaled Rainfall Analysis Over Diverse Agroclimatic Zones of India

Abstract

AbstractRegional climate models (RCMs) are routinely applied for regional climate assessments. The RCM simulated rainfall typically overpredict the light rain/drizzle events. To correct the typical errors noted in RCM rainfall simulations, in this study, three bias‐correction methods: linear scaling (SCL), local intensity scaling (LOCI), and empirical quantile mapping (EQM), have been employed. These methods are used to correct monsoon rainfall simulations from 7 RCMs across 14 agroclimatic zones (ACZs) in India from 1970 to 2005. The corrected rainfall data were compared to the observations obtained from India Meteorological Department. The performance of the three methods was assessed using: probability distribution function, consecutive dry day index (CDD), R95 (rainfall distribution at 95th percentile), and spatial correlation. The results vary spatially across the different ACZs. Overall, SCL method is more effective followed by EQM while LOCI was relatively less effective in correcting the errors. Spatial analysis of the rainfall indicates notable improvements over the Western Himalayan Region, which has a complex topography and land use. Error metricsreveal broad improvements across different ACZs, except for Central Plateau Hill Region, East Coast Plain Hill Region, and Southern Plateau Hill Region. The SCL and EQM perform well, and the results are particularly good for simulating wet days (R95), while no distinct variation was found among correction methods to reduce dry bias (CDD). The results provide ACZ and region‐specific utilization of an effective bias correction technique for impact assessment studies in the Indian monsoon region.