Empowering LLM Agents with Geospatial Awareness: Toward Grounded Reasoning for Wildfire Response

2025-10-13

AI Summary

Current large language models (LLMs) lack geospatial awareness, hindering evidence-driven decision-making in disaster response. To address this, we propose the Geospatial Awareness Layer (GAL), a framework that automatically fuses heterogeneous Earth observation data—including infrastructure, population, terrain, and real-time meteorological feeds—to generate dynamic, unit-annotated perception scripts, thereby endowing LLMs with embodied understanding of geographic contexts. GAL integrates remote-sensing-based wildfire detection, geodatabase querying, historical case retrieval, and diurnal signal incremental updating, forming an end-to-end geospatially aware intelligent agent. Evaluated on real-world wildfire incidents, the agent significantly improves both accuracy and interpretability of resource allocation recommendations across multiple LLM backbones. Furthermore, it demonstrates strong generalizability to other hazard types—including floods and hurricanes—validating its cross-disaster applicability.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Agentic search

Abstract

Effective disaster response is essential for safeguarding lives and property. Existing statistical approaches often lack semantic context, generalize poorly across events, and offer limited interpretability. While Large language models (LLMs) provide few-shot generalization, they remain text-bound and blind to geography. To bridge this gap, we introduce a Geospatial Awareness Layer (GAL) that grounds LLM agents in structured earth data. Starting from raw wildfire detections, GAL automatically retrieves and integrates infrastructure, demographic, terrain, and weather information from external geodatabases, assembling them into a concise, unit-annotated perception script. This enriched context enables agents to produce evidence-based resource-allocation recommendations (e.g., personnel assignments, budget allocations), further reinforced by historical analogs and daily change signals for incremental updates. We evaluate the framework in real wildfire scenarios across multiple LLM models, showing that geospatially grounded agents can outperform baselines. The proposed framework can generalize to other hazards such as floods and hurricanes.

Problem

• Enhancing LLM agents with geospatial awareness for disaster response
• Integrating earth data to overcome text-bound limitations in wildfire management
• Generating evidence-based resource recommendations using multi-source geospatial information

Innovation

• Integrates geospatial data layer into LLM agents
• Automatically retrieves infrastructure, demographic, terrain data
• Generates evidence-based resource allocation recommendations