Urban expansion and warming jointly amplify drought and surface heat island effects

Urbanization and climate change are intensifying the co-occurrence of drought and surface urban heat island (SUHI) effects, yet their combined dynamics remain poorly understood at high spatial resolution. This study evaluates the compound risk of drought and UHI in South Korea's Saemangeum region using a hybrid KNN–LSTM model, alongside three comparative approaches (KNN, CNN, CNN–LSTM), incorporating CMIP6 climate projections and satellite-derived land metrics. The KNN–LSTM model demonstrated superior performance (R² = 0.94; MSE = 0.0053), accurately capturing an increasing drought–SUHI association, which rose from 0.84 (1991–2020) to 1.00 by 2100 under the SSP585 scenario. High-risk zones predominantly overlap with built-up and cropland areas, where UHI indices exceed 4.0 and drought probabilities surpass 0.9. Although land use explains over 80% of drought variability under baseline conditions, its predictive power diminishes under future warming, highlighting the growing role of climate feedbacks. These findings emphasize the emergence of synchronized thermal and hydrological stressors in urbanizing coastal regions and underscore the need for integrated land-climate adaptation strategies supported by machine learning-based early warning systems.