CanadaFireSat: Toward high-resolution wildfire forecasting with multiple modalities

2025-06-10

AI Summary

Under climate change, increasing frequency and intensity of wildfires in Canada necessitate high-resolution spatial prediction—yet existing models operate at coarse resolutions (~0.1°), limiting precise risk localization. Method: We introduce CanadaFireSat, the first nationwide 100-meter-resolution wildfire probability forecasting benchmark dataset for Canada, integrating Sentinel-2 multispectral imagery, MODIS land surface products, and ERA5 meteorological reanalysis data. We propose a multimodal temporal deep learning framework that jointly models high-resolution remote sensing, medium-resolution satellite observations, and environmental variables to learn spatiotemporal wildfire occurrence patterns end-to-end. Contribution/Results: This work achieves the first continental-scale, generalizable 100-meter-resolution wildfire risk prediction. Evaluated on the extreme 2023 fire season, our model attains an F1-score of 60.3%, significantly outperforming unimodal baselines. Results demonstrate the efficacy and practicality of high-resolution, multimodal fusion for operational wildfire early warning.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multimodal Learning
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Canada experienced in 2023 one of the most severe wildfire seasons in recent history, causing damage across ecosystems, destroying communities, and emitting large quantities of CO2. This extreme wildfire season is symptomatic of a climate-change-induced increase in the length and severity of the fire season that affects the boreal ecosystem. Therefore, it is critical to empower wildfire management in boreal communities with better mitigation solutions. Wildfire probability maps represent an important tool for understanding the likelihood of wildfire occurrence and the potential severity of future wildfires. The massive increase in the availability of Earth observation data has enabled the development of deep learning-based wildfire forecasting models, aiming at providing precise wildfire probability maps at different spatial and temporal scales. A main limitation of such methods is their reliance on coarse-resolution environmental drivers and satellite products, leading to wildfire occurrence prediction of reduced resolution, typically around $sim 0.1${deg}. This paper presents a benchmark dataset: CanadaFireSat, and baseline methods for high-resolution: 100 m wildfire forecasting across Canada, leveraging multi-modal data from high-resolution multi-spectral satellite images (Sentinel-2 L1C), mid-resolution satellite products (MODIS), and environmental factors (ERA5 reanalysis data). Our experiments consider two major deep learning architectures. We observe that using multi-modal temporal inputs outperforms single-modal temporal inputs across all metrics, achieving a peak performance of 60.3% in F1 score for the 2023 wildfire season, a season never seen during model training. This demonstrates the potential of multi-modal deep learning models for wildfire forecasting at high-resolution and continental scale.

Problem

• High-resolution wildfire forecasting in Canada using multi-modal data
• Overcoming coarse-resolution limitations in current wildfire prediction models
• Leveraging deep learning for precise wildfire probability maps

Innovation

• High-resolution 100m wildfire forecasting
• Multi-modal deep learning models
• Leveraging Sentinel-2, MODIS, ERA5 data