DISOBased Evaluation of NEX‐GDDP‐CMIP6 Dataset for Projecting Extreme Temperature Indices in China

Escalating near‐surface temperatures have intensified the frequency, duration, and magnitude of extreme climate events across China. This study evaluates high‐resolution climate projections from the NASA NEX‐GDDP dataset using the Distance between Indices of Simulation and Observation (DISO) metric, integrating multidimensional performance indicators. Key findings reveal divergent behaviours between temperature extremes: T min displays higher median DISO values (0.70 vs. 0.48–0.56, p < 0.001) and wider dispersion (Interquartile Range, IQR = 0.43 vs. 0.24–0.47) than extreme cold indices, while T max shows lower medians (0.74 vs. 0.84–0.88, p < 0.001) and reduced variability (IQR = 0.24 vs. 0.59–0.66) compared to heat extremes. This suggests that GCMs are less effective at simulating extreme cold indices compared to T min , but perform better for extreme heat indices than for T max . Models reasonably replicate temperature extremes in most regions but persistently underestimate cold extremes in high‐latitude (Northeast China) and high‐altitude zones (Tibetan Plateau), particularly in extreme cold indices, due to unresolved snow‐albedo feedbacks and boundary layer processes. Mechanistically, the Multi‐Model Ensemble (MME) captures heat extreme trends but overestimates post‐1990 cold extremes, reflecting inadequate Arctic amplification parameterisation. Future projections demonstrate asymmetric intensification of temperature extremes under contrasting emissions pathways (2031–2090): Cold extremes (TN10p) moderate reductions under SSP2‐4.5 (−0.7%; 28% relative decline) but steep declines under SSP5‐8.5 (−2.5%; 63% relative decline); while heat extremes (TX90p) show progressive intensification with +7% (27% relative rise) and +22% (79% relative rise) increases under SSP2‐4.5 and SSP5‐8.5, respectively, revealing 3.2–3.6‐fold amplified changes under high‐emission scenarios. Additionally, TN10p declines most rapidly in the Tibetan Plateau, while TX90p increases accelerate in southeastern coastlines through intensified land‐atmosphere feedbacks. The amplified heat extremes under high emissions highlight critical needs for climate adaptation in vulnerability hotspots. These findings provide actionable guidelines for model selection in regional extreme event studies and underscore the imperative of process‐oriented model improvements to address persistent cold biases in complex terrain.