High-Resolution Live Fuel Moisture Content (LFMC) Maps for Wildfire Risk from Multimodal Earth Observation Data

2025-06-25

AI Summary

Ground-based live fuel moisture content (LFMC) observations are costly, spatially sparse, and temporally lagged, hindering dynamic wildfire risk assessment. To address this, we propose an end-to-end LFMC mapping framework leveraging a pre-trained multimodal Earth model. Our method fuses multi-temporal Sentinel-2 optical, Landsat thermal infrared, and high-resolution topographic data within a spatiotemporal deep learning architecture, supported by an automated data processing pipeline. Compared to a randomly initialized baseline, our approach reduces RMSE by 20%. It achieves, for the first time, daily-resolution, spatially complete, and physically interpretable LFMC mapping across the contiguous United States. Validation in the Eaton and Palisades wildfire zones demonstrates substantially improved spatial continuity and temporal responsiveness, enabling high-accuracy, real-time wildfire risk monitoring and supporting rapid emergency decision-making.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Wildfires are increasing in intensity and severity at an alarming rate. Recent advances in AI and publicly available satellite data enable monitoring critical wildfire risk factors globally, at high resolution and low latency. Live Fuel Moisture Content (LFMC) is a critical wildfire risk factor and is valuable for both wildfire research and operational response. However, ground-based LFMC samples are both labor intensive and costly to acquire, resulting in sparse and infrequent updates. In this work, we explore the use of a pretrained, highly-multimodal earth-observation model for generating large-scale spatially complete (wall-to-wall) LFMC maps. Our approach achieves significant improvements over previous methods using randomly initialized models (20 reduction in RMSE). We provide an automated pipeline that enables rapid generation of these LFMC maps across the United States, and demonstrate its effectiveness in two regions recently impacted by wildfire (Eaton and Palisades).

Problem

• Generating high-resolution LFMC maps for wildfire risk
• Overcoming sparse ground-based LFMC data with satellite AI
• Automating large-scale LFMC mapping for operational wildfire response

Innovation

• Uses multimodal earth-observation data
• Leverages pretrained AI model
• Automated high-resolution LFMC mapping