Extreme event assessment and model diagnostic for attribution of the record-breaking July 2025 heat in Shandong, China

Abstract In July 2025, Shandong Province, China, experienced a record-breaking regional extreme heat event. Based on observational data from 77 meteorological stations (1961–2025), this study assessed the event’s extremity and systematically examines the feasibility of its attribution using Coupled Model Intercomparison Project Phase 6 (CMIP6) models. The event reached a magnitude close to that of a 1-in-50-year event in the current climate, with 84.4% of stations recording high-temperature days (HTD, Tmax ≥ 35 °C) ranking among their top three historical values and 50.6% setting new records. This study employs a rigorous process-oriented evaluation framework that combines statistical performance metrics with hierarchical physical diagnostics to assess 31 CMIP6 models. The results reveal that none passed our stringent validation criteria. A hierarchical physical diagnosis uncovered a coherent weakening of high-pressure systems across timescales: models not only systematically understated the Western Pacific Subtropical High (WPSH) in the climatology but, more critically, failed to generate the intense continental ridge observed during extreme years. This dual failure is compounded by our finding that even a best-performing model subset could not simulate the key driver, and individual models fail discordantly, producing a spectrum of physically implausible anomaly patterns. Consequently, the current CMIP6 ensemble does not adequately represent the physical processes driving the extreme heat, rendering reliable probabilistic attribution analysis unfeasible. This “non-conclusion” is critical because it reveals systematic deficiencies in current models for capturing East Asian monsoon heat extremes, thereby outlining clear priorities for future model development and cautioning against the unconditional application of current models for attribution in this region.