Evaluating ISIMIP3b bias-corrected data for precipitation extremes in China during 1981–2100

The Inter-Sectoral Impact Model Intercomparison Project (ISIMIP) bias-corrected data have been extensively used for climate change impact assessments, despite being derived from only five global climate models (GCMs). This study aims to enhance confidence in its application for projecting future changes in China by evaluating the accuracy of ISIMIP3b bias-corrected data in representing historical precipitation extremes compared to observational records. The findings reveal that the ensemble mean of ISIMIP3b data aligns well with observed precipitation extremes during the historical period (1981‒2010), with spatial correlation coefficients ranging from 0.87 to 0.96 and relative errors kept below 5%, affirming its suitability for studying future precipitation patterns. Future projections indicate an increase in heavy precipitation intensity and frequency, along with fewer consecutive dry days (CDD). These changes are more pronounced under higher emission scenarios and over the far-future period (2071‒2100) when contrasted with lower emission scenarios and the near-future period (2031‒2060). Specifically, by the end of the 21st century under the SSP585 scenario, relative to the 1981–2010 period: CDD is projected to decline by 7.4 d (11.4%); days of heavy precipitation to increase by 2.5 d (15.2%); maximum 5-day precipitation to increase by 16.3 mm (21.7%); and simple daily intensity index to increase by 0.7 mm/d (11.6%). A comparative analysis with projections from a broader range of GCMs highlights that the ISIMIP3b data with a small GCM ensemble can effectively capture key aspects of changes in extreme precipitation projections, while acknowledging certain limitations. This study underscores the potential of ISIMIP bias-corrected data to reasonably represent change in precipitation extremes with a reduced GCM ensemble. These findings help to justify the use of ISIMIP3b data with a small GCM ensemble for climate change impact assessment.