Physics-guided spatiotemporal neural models for fuel density prediction

2026-07-08

AI Summary

This work proposes a physics-informed spatiotemporal neural network approach to enhance the accuracy and physical consistency of fuel moisture predictions for wildfire risk management and prescribed burning decisions. For the first time, physical priors—such as mass conservation and fire spread rate dynamics—are explicitly embedded into ConvLSTM, Adaptive Fourier Neural Operator (AFNONet), and Video Vision Transformer (ViViT) architectures, which are then trained end-to-end using a differentiable physics-informed loss function. Experimental results demonstrate that the proposed method significantly outperforms purely data-driven baselines across multiple independent test scenarios, achieving substantial improvements in prediction accuracy, numerical stability, and physical plausibility. This advancement establishes a new paradigm for efficient and reliable wildfire forecasting.

Abstract

This paper presents a physics-guided machine learning (PGML) framework for fuel density prediction, integrating physics constraints and domain knowledge into deep learning models to enhance model accuracy and stability. We explore three deep learning architectures -- ConvLSTM, Adaptive Fourier Neural Operator (AFNONet), and Video Vision Transformer (ViViT) -- to model the spatiotemporal evolution of fuel density. Our approach incorporates differentiable physics-informed terms in the loss function, including a mass-conserving fuel transport term and a rate-of-spread estimation. Experimental results, averaged across multiple independent trials, demonstrate that the proposed PGML framework outperforms purely data-driven baselines without physics constraints in both accuracy and stability. This framework enables computationally efficient, physically plausible fire forecasting to support adaptive prescribed burn management.

Problem

• fuel density prediction
• physics-guided machine learning
• spatiotemporal modeling
• fire forecasting
• mass conservation

Innovation

• physics-guided machine learning
• spatiotemporal modeling
• fuel density prediction
• differentiable physics constraints
• fire forecasting