GeoAI-Supported Early Warning Systems for Extreme Weather Events and their Implications in India

Extreme weather events — cyclones, flash floods, heat waves, cloudbursts, and droughts — are intensifying across India in both frequency and magnitude under the influence of accelerating climate change. Timely and spatially precise early warning systems (EWS) are therefore indispensable for safeguarding lives and livelihoods. The convergence of Geospatial Artificial Intelligence (GeoAI) — encompassing machine learning (ML), deep learning (DL), remote sensing, and Geographic Information Systems (GIS) — is transforming how such warnings are generated, disseminated, and acted upon. This systematic review examines GeoAI-enabled EWS architectures applicable to India's diverse hydro-meteorological and socio-geographic landscape, synthesising 218 peer-reviewed publications, 34 technical reports, and 12 government policy documents published between 2015 and 2024. Key performance findings include: augmenting physics-based flood models with Long Short-Term Memory (LSTM) networks reduces forecasting root mean square error (RMSE) by 15–30% in the Ganga and Brahmaputra basins; ensemble machine learning achieves landslide susceptibility classification accuracy (AUC) of 0.88–0.93 across the Himalayan arc; and attention-based convolutional networks approach AUC 0.77–0.81 for rapid cyclone intensification prediction. The review traces theoretical foundations, technological components, case applications across Indian states, institutional frameworks, data challenges, ethical dimensions, and emerging research frontiers. It critically examines the gap between technological capability and on-the-ground implementation, with particular attention to last-mile connectivity, community vulnerability, and governance. The paper concludes with a forward-looking research agenda and policy recommendations tailored to India's federal disaster risk management architecture.