Evaluation and Projection of Northwest China's Extreme Precipitation Using Statistically Downscaled <scp>CMIP6</scp> Models

ABSTRACT Global warming is intensifying hydrological cycles, causing significant spatiotemporal variations in extreme precipitation. Since the 1980s, Northwest China has shifted from a warm‐dry to a warm‐wet climate regime, raising widespread concern. Employing skill metrics, this study quantitatively evaluates 23 statistically downscaled CMIP6 models from the NASA NEX‐GDDP dataset in simulating historical (1961–2014) extreme precipitation over Northwest China and, based on their performance, projects future changes in the mid‐ (2031–2060) and late‐21st century (2071–2100) under two scenarios. Six extreme precipitation indices are analysed: total wet‐day precipitation (PRCPTOT), very wet‐day precipitation (R95pTOT), maximum 5‐day precipitation (Rx5day), heavy precipitation days (R10mm), consecutive dry days (CDD) and consecutive wet days (CWD). Results show that most models reasonably capture spatial patterns (pattern correlation: 0.4–0.9), yet exhibit systematic dry biases (positive biases for CDD; negative biases for other indices). Interannual variability is better simulated in western subregions than eastern, particularly for PRCPTOT, R95pTOT, Rx5day and R10mm. Four models (CESM2, CESM2‐WACCM, CMCC‐CM2‐SR5 and EC‐Earth3‐Veg‐LR) demonstrate superior skill in spatiotemporal simulations. Projections from these best‐performing models indicate a mitigation of aridity and an increase in the frequency of extreme precipitation under SSP2‐4.5 and SSP5‐8.5 scenarios.