Geophysical Research Letters Geophysics

Urbanization Doubles Future Compound Heatwave Risk in the Yangtze River Delta Urban Agglomeration Based on a Machine Learning‐Statistical Hybrid Framework

Sisi Chen, Peng Ji, Shanshui Yuan, Qibo Xu, Chunhui Lu, Junliang Jin
Wiley · 2026-11-20 · DOI 10.1029/2026gl123497

Abstract

Abstract Urbanization intensifies compound heatwaves (CoHot) involving concurrent hot days and nights, yet coarse‐resolution climate projections cannot resolve urban‐scale processes. Dynamical downscaling can bridge this gap but requires substantial computing for long‐term projections. Here we introduce a novel and efficient hybrid framework combining daily‐scale stacking ensemble machine learning with monthly‐scale quantile mapping to downscale CMIP6 projections to urban resolution. Applied to China's Yangtze River Delta (YRD), a highly developed urban area, this framework improves simulation of CoHot frequency, duration, and intensity, increasing spatial correlation from −0.26–0.50 to 0.29–0.77 and reducing RMSE by 18%. High‐resolution (5‐km) projections reveal cities as evolving CoHot hotspots, where urban expansion drives shifting spatial patterns. By 2050, city centers could experience about four CoHot events and over 20 CoHot days annually, with total duration twice that in surrounding rural areas. This scalable, efficient tool aids urban heat risk assessment with broad applicability to rapidly urbanizing regions worldwide.

Read at publisher →