Driven geospatial artificial intelligence modeling for climate change impact assessment: a global perspective

Abstract The challenge of anthropogenic climate change requires a paradigm shift in analysis, moving beyond traditional modeling constraints toward the application of computational intelligence. Geospatial Artificial Intelligence (GeoAI), an emergent interdisciplinary domain synthesizing spatial science, artificial intelligence (AI), and statistical learning, can reshape the epistemology of climate change impact assessment. Although previous reviews have examined discrete elements of geospatial technologies or AI in environmental science, a comprehensive, critical synthesis that captures the methodological and application-based convergence of GeoAI and climate change is rare. This study analyzes GeoAI applications in climate change science, employing a dual scientometric and systematic review methodology. A scientometric analysis of 152 core publications, retrieved from Scopus, Web of Science, IEEE Xplore, and Google Scholar, was conducted to quantitatively elucidate publication trends, thematic networks, and the global collaborative landscape. This mapping reveals a rapidly accelerating, interdisciplinary field characterized by robust clusters around deep learning architectures, multi-modal remote sensing, and specific climate impact pathways. The review deconstructs the GeoAI technological stack from convolutional and recurrent neural networks to transformative transformer-based models and hybrid physics-AI frameworks. It also evaluates their roles in climate variable prediction, extreme event attribution, ecosystem vulnerability diagnostics, and socio-economic impact modeling. The review identifies profound epistemological and practical challenges, including data-centric limitations (heterogeneity, paucity), computational intractability, the “black box” conundrum, model transferability failures, and critical ethical imperatives. This review outlines a strategic prospective research agenda, suggesting that the evolution of scalable, interpretable, and context-aware GeoAI systems is crucial for constructing a resilient and equitable planetary future.