On bias correction of summer monsoon precipitation over India from CORDEX‐SA simulations

Abstract

In the present study, four different bias correction methods are assessed in improving the precipitation output from a set of two regional climate simulations for 1971–2005 period. The simulations come from Coordinated Regional Climate Downscaling Experiment over South Asia (CORDEX‐SA). The bias correction methods applied here are linear scaling (Scaling), empirical quantile mapping (EQM), power transformation (PTR) and local intensity scaling (LOCI). All the bias correction techniques show skill in improving the model results, but there are clear differences in the degrees of success depending upon the characteristic of precipitation studies—like seasonal mean, seasonal extremes or inter‐annual variability. Most of the methods show equivalently high ability in correcting the seasonal mean precipitation values; however, there also exist residual biases which vary spatially over India. The largest reduction in bias was observed for the frequency of dry days and precipitation on very wet days, which are significantly altered after adjustment by EQM and PTR. It is found that distribution‐based EQM method is more skilful followed by PTR and LOCI in most of the cases including correction of frequency‐based indices (e.g., dry days number and inter‐annual variability).

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

“…The findings from the study are supported by the results from Choudhary and Dimri (2019), which show improved performance of rainfall simulations in India after bias correction. They concluded that the EQM method performed best relative to LOCI and SCL.…”

“…The findings are again complemented by the results in Choudhary and Dimri (2019). For inter‐annual variability (as standard deviation) of JJAS mean precipitation, the application of EQM on RCA4 data could successfully capture the spatial pattern of variability.…”

“…The choice of the model to study rainfall characteristics depends upon the performance of a model. Choudhary and Dimri (2019) showed that the GCM, after bias correction, could capture the rainfall over central, northern, and northeastern parts of India and along the foothills of Himalaya. They further mentioned that each bias correction method performed equally well in simulating mean rainfall for JJAS.…”

“…Three bias correction methodologies: SCL, LOCI, and EQM, that have been used in prior studies were considered (Acharya et al., 2013; Chaudhary et al., 2019; Fang et al., 2015; Prasanna, 2018; Singh et al., 2017; Teutschbein & Seibert, 2013). The use of these methods in the rainfall variables is well explored in Fang et al.…”

“…The results showed that EQM was more skillful than PTR and LOCI. The study further highlighted the variability in bias correction performance depending upon the rainfall characteristics such as seasonal mean, seasonal extremes or inter‐annual variability, and the spatial variability across India (Choudhary & Dimri, 2019). There is a high likelihood of significant rainfall bias over the northeast and Himalayan regions as well as over western ghats (Basha et al., 2017; Tiwari, Kar, Mohanty, Dey, Sinha, et al., 2016).…”