Extreme precipitation simulation capability of CMIP6 models and their future spatiotemporal evolution: Case study in Poyang Lake Basin, China

Poyang Lake Basin, China. This research evaluates the spatiotemporal simulation ability of extreme precipitation (intensity, frequency, and duration) in the Poyang Lake Basin using global climate models from Phase 6 of the Coupled Model Intercomparison Project (CMIP6). We employ quantile mapping and a novel performance-based multi-model ensemble weighting approach to assess model performance. Subsequently, near-, medium-, and long-term spatiotemporal trends of extreme precipitation are predicted under four shared socioeconomic pathways. Our performance-based multi-model ensemble weighting method significantly improves the simulation accuracy of extreme precipitation in the Poyang Lake Basin, yielding a central root mean square error of 0.93, a spatial standard deviation ratio of 1.07, a correlation coefficient of 0.67, and a temporal interannual variability score of 0.11. This method demonstrably provides better simulations of interannual distribution and spatial pattern characteristics compared to single models and simple multi-model averages. Projections under future emission scenarios indicate that extreme precipitation is generally lower in the southwest and higher in the northeast of the study area. Higher emission pathways accelerate the upward trend of extreme precipitation occurrences, with the most substantial increases observed in long-term projections. Spatially, the most dramatic increases in precipitation intensity are concentrated in the south and central regions, while high growth in frequency and duration is predominantly observed in the north and central regions. • Introduce a multi-performance rank-weighting method for combining GCM projections. • Multi-model ensemble enhances extreme precipitation simulation accuracy. • Investigate future changes to extreme precipitation in the Poyang Lake Basin. • Increased emissions intensify extreme precipitation, long-term changes more obvious.