Predicting Next-Day Wildfire Spread with Time Series and Attention

2025-02-17

AI Summary

This work addresses the limited accuracy of next-day wildfire spread spatial distribution prediction. We propose the first SwinUnet variant specifically designed for remote sensing time-series modeling. Our method integrates multi-source environmental rasters—including vegetation, topography, and meteorology—with historical fire data, leveraging a Swin Transformer encoder, a U-Net decoder, and a novel multi-temporal feature fusion mechanism. Crucially, we provide the first empirical validation that ImageNet pretraining yields significant performance gains for small-sample remote sensing time-series modeling. Evaluated on the WildfireSpreadTS benchmark, our model achieves state-of-the-art (SOTA) performance under both single-day and five-day input settings; notably, the five-day configuration substantially outperforms the single-day one, demonstrating that long-term temporal context is essential for modeling dynamic fire behavior. This work establishes a new, interpretable, and high-accuracy spatial prediction paradigm for intelligent wildfire early warning systems.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Time-Series/Data Streams
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Recent research has demonstrated the potential of deep neural networks (DNNs) to accurately predict next-day wildfire spread, based upon the current extent of a fire and geospatial rasters of influential environmental covariates e.g., vegetation, topography, climate, and weather. In this work, we investigate a recent transformer-based model, termed the SwinUnet, for next-day wildfire prediction. We benchmark Swin-based models against several current state-of-the-art models on WildfireSpreadTS (WFTS), a large public benchmark dataset of historical wildfire events. We consider two next-day fire prediction scenarios: when the model is given input of (i) a single previous day of data, or (ii) five previous days of data. We find that, with the proper modifications, SwinUnet achieves state-of-the-art accuracy on next-day prediction for both the single-day and multi-day scenarios. SwinUnet's success depends heavily upon utilizing pre-trained weights from ImageNet. Consistent with prior work, we also found that models with multi-day-input always outperformed models with single-day input.

Problem

• Predict next-day wildfire spread
• Utilize transformer-based SwinUnet model
• Benchmark on WildfireSpreadTS dataset

Innovation

• Transformer-based SwinUnet model
• Utilizes pre-trained ImageNet weights
• Multi-day input enhances prediction accuracy