Artificial Intelligence in Earth Science: A GeoAI Perspective

• Published in: Journal of geophysical research. Machine learning and computation Vol. 2; no. 3
• Main Author: Li, Wenwen
• Format: Journal Article
• Language: English
• Published: 01.09.2025
• ISSN: 2993-5210; 2993-5210
• DOI: 10.1029/2025JH000691

GeoAI, or geospatial artificial intelligence (AI), has transformative potential for Earth science by integrating geospatial data with AI to enhance environmental monitoring, predictive modeling, and decision‐making. This commentary, based on the Greg Leptoukh Lecture at American Geophysical Union 2024, explores the evolving role of GeoAI in addressing pressing challenges—from environmental change in the Arctic to disaster response in hurricane‐prone tropical regions. It highlights advancements in GeoAI‐driven analysis of multimodal Earth observation data, ranging from structured remote sensing imagery to semi‐structured data, and natural language texts. The integration of knowledge graphs and generative AI further strengthens GeoAI by enabling seamless integration of cross‐domain data, semantic reasoning, and knowledge inference. By bridging informatics and domain expertise, GeoAI is shaping a more intelligent and actionable digital future for Earth science. In December 2024, I had the opportunity to give the Greg Leptoukh Lecture at the American Geophysical Union (AGU) Fall Meeting in Washington, D.C. In my talk, I discussed how artificial intelligence (AI), especially its geospatial branch known as GeoAI, is helping scientists process and analyze large, complex Earth observation data more efficiently. This commentary summarizes key points from that lecture, highlighting important research at the intersection of GeoAI and Earth science. It also explores ongoing challenges and future directions for using AI to build a smarter and more sustainable digital future for our planet. GeoAI supports environmental monitoring and decision making through artificial intelligence (AI) powered analysis of Earth observation data GeoAI advances by using multimodal Earth data and generative AI to improve understanding of complex Earth systems GeoAI needs scientific validation and reproducibility to support reliable and trustworthy Earth science research