AI-Driven Prediction of Compound Drought–Dust–Heatwave Events: Comparative Evidence from Iran and the Southwestern United States

Core ProblemCompound climate extremes—particularly the co-occurrence of drought, dust storms, and heatwaves—pose escalating risks to human health, infrastructure, and ecosystems in arid regions. These events are becoming more frequent and intense under accelerated climate change, yet traditional models often fail to capture their nonlinear interdependencies. Iran and the Southwestern United States (SWUS) represent two of the world’s most climate-vulnerable dryland systems, where prolonged droughts, expanding desertification, and urban heat amplification converge. Despite advances in environmental monitoring, predictive frameworks that integrate multi-hazard dynamics remain critically underdeveloped.ApproachThis study employs an AI-driven spatio-temporal prediction framework combining Convolutional Long Short-Term Memory (ConvLSTM) and Random Forest models to forecast compound drought–dust–heatwave (DDH) events. Using twenty-five years (2000–2024) of satellite, meteorological, and land-surface data, we developed a Compound Hazard Index (CHI) integrating standardized precipitation–evapotranspiration, aerosol optical depth, and temperature extremes. The model was trained and validated for five major arid cities in Iran and five in the SWUS to evaluate spatial transferability and predictive performance.Key Findings.Results reveal that DDH events exhibit strong spatio-temporal clustering and cross-regional similarity, with urban heat islands amplifying hazard intensity by 20–35%. The ConvLSTM model achieved R² > 0.80 for both regions, outperforming conventional regression-based methods. Iran’s eastern plateau and the Sonoran Desert corridor in the SWUS emerged as primary hotspots of future compound risk.Impact:The findings demonstrate the transformative potential of AI-based predictive systems for early warning and climate adaptation in arid cities. By identifying emergent hazard patterns before onset, the framework supports evidence-based urban planning and resource management across global drylands.