An extremes-weighted empirical quantile mapping for global climate model data bias correction for improved emphasis on extremes

Accuracy in the global climate model (GCM) projections is essential for developing reliable impact mitigation strategies. The conventional bias correction methods used to improve this accuracy often fail to capture the extremes, specifically for precipitation, due to the generic correction application to whole data. Given the importance of understanding future extreme precipitation behavior for disaster mitigation, we propose Extremes-Weighted Empirical Quantile Mapping (EW-EQM) bias correction with a specific emphasis on extremes. The EW-EQM applies separate EQM correction to threshold-exceeded extremes and frequency-adjusted non-extreme precipitation occurrences. The bias correction results demonstrated using station-observed precipitation records at 945 locations in the Mid-Atlantic region of the United States, and five Coupled Model Intercomparison Project Phase 6 (CMIP6) GCMs demonstrate the strength of EW-EQM to improve the bias correction abilities of extreme precipitation occurrences. The spatial median of Root Mean Square error between observed and bias-corrected extreme precipitation was mostly less than 6 mm for EW-EQM across GCMs, while EQM and Power Transformation had a median higher than 12 mm. Further, future bias-corrected precipitation series for 2021–2050 under SSP245 indicate a 0–10% increase in total annual precipitation and a 10% decrease to 25% increase in mean annual maximum precipitation in the region. The improved bias correction of extremes could be significant in climate change impact mitigation decisions such as flood management.