Compound extremes and cascading hazards represent complex manifestations of coupled Earth-system processes in which multiple drivers interact simultaneously or sequentially across spatial and temporal scales. Heatwaves may develop in association with drought and atmospheric stagnation; precipitation extremes may trigger floods and landslides; drought and heat can precondition landscapes for wildfire; and cyclones can generate interacting wind, rainfall, flood and storm-surge hazards. Such events may involve nonlinear amplification, feedbacks, thresholds and transitions that are difficult to identify through conventional single-hazard analyses. Earth observation provides a unique observational basis for investigating these processes through thermal infrared, optical, microwave/SAR and atmospheric observations of land-surface temperature, soil moisture, evapotranspiration, vegetation stress, precipitation, surface water, fire activity, aerosols and atmospheric composition. We particularly welcome studies that integrate multisource and multiscale Earth-observation datasets with in situ observations, reanalysis, numerical models and high-resolution simulations.
The session also invites contributions demonstrating how artificial intelligence and machine learning, including deep learning, spatiotemporal learning, explainable AI, hybrid physics–ML approaches and data-driven process discovery, can improve the detection, characterisation and prediction of compound extremes and hazard cascades while addressing uncertainty and generalisation. Topics of interest include, but are not limited to, heat–drought, heat–humidity, heat–air-pollution, drought–wildfire, precipitation–flood, flood–landslide, cyclone–flood and other compound or sequential hazards. Contributions addressing land–atmosphere coupling, surface-energy and water-balance feedbacks, nonlinear dependence, threshold behaviour, regime transitions, hazard propagation, precursor identification, attribution, predictability and early warning are particularly encouraged.
The session aims to advance understanding of how compound extremes emerge, intensify, transition and cascade, and how Earth observation and AI/ML can reveal the underlying physical mechanisms across scales.