GOFMs: Geospatial Foundation Models and the New AI for Earth

Geospatial Foundation Models (GOFMs) are large AI systems pre-trained on Earth-observation data—satellite imagery, maps, and time-series—designed to learn transferable geospatial knowledge. In this episode we survey who’s building them (NASA/IMPACT/IBM/Clark University; Google; Atlas AI), what they can do—from flood spread mapping and burn-scar detection to high-resolution land-cover tasks—and the challenges of real-world data: quality, temporality, trust, and explainability. We also explore multimodal reasoning and natural-language interfaces that orchestrate geospatial analysis at scale for disaster response, agriculture, and environmental monitoring.

Note: This podcast was AI-generated, and sometimes AI can make mistakes. Please double-check any critical information.

Sponsored by Embersilk LLC