Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Fire-EnSF: Wildfire Spread Data Assimilation using Ensemble Score Filter

Hongzheng Shi, Yuhang Wang, Xiao Liu

TL;DR

The paper tackles real-time wildfire spread prediction by applying the Ensemble Score Filter (EnSF), a diffusion-model–based data assimilation method, to fuse observations with FARSITE forecasts. It leverages a score-based diffusion framework with forward and reverse SDEs in pseudo-time to sample the posterior over time-varying fire-perimeter states, addressing high-dimensional, nonlinear filtering challenges. The authors demonstrate that EnSF yields higher accuracy, greater stability, and substantially faster computation than EnKF, validated on both simulated data and real California wildfire events, with publicly available code. This approach offers a scalable, real-time tool for improving wildfire management decisions and resource allocation by providing more reliable perimeter estimates under uncertain observations. The work also outlines practical implementation details (re-interpolation and vertex normalization) that enable robust handling of evolving polygon state representations.

Abstract

As wildfires become increasingly destructive and expensive to control, effective management of active wildfires requires accurate, real-time fire spread predictions. To enhance the forecasting accuracy of active fires, data assimilation plays a vital role by integrating observations (such as remote-sensing data) and fire predictions generated from numerical models. This paper provides a comprehensive investigation on the application of a recently proposed diffusion-model-based filtering algorithm -- the Ensemble Score Filter (EnSF) -- to the data assimilation problem for real-time active wildfire spread predictions. Leveraging a score-based generative diffusion model, EnSF has been shown to have superior accuracy for high-dimensional nonlinear filtering problems, making it an ideal candidate for the filtering problems of wildfire spread models. Technical details are provided, and our numerical investigations demonstrate that EnSF provides superior accuracy, stability, and computational efficiency, establishing it as a robust and practical method for wildfire data assimilation. Our code has been made publicly available.

Fire-EnSF: Wildfire Spread Data Assimilation using Ensemble Score Filter

TL;DR

The paper tackles real-time wildfire spread prediction by applying the Ensemble Score Filter (EnSF), a diffusion-model–based data assimilation method, to fuse observations with FARSITE forecasts. It leverages a score-based diffusion framework with forward and reverse SDEs in pseudo-time to sample the posterior over time-varying fire-perimeter states, addressing high-dimensional, nonlinear filtering challenges. The authors demonstrate that EnSF yields higher accuracy, greater stability, and substantially faster computation than EnKF, validated on both simulated data and real California wildfire events, with publicly available code. This approach offers a scalable, real-time tool for improving wildfire management decisions and resource allocation by providing more reliable perimeter estimates under uncertain observations. The work also outlines practical implementation details (re-interpolation and vertex normalization) that enable robust handling of evolving polygon state representations.

Abstract

As wildfires become increasingly destructive and expensive to control, effective management of active wildfires requires accurate, real-time fire spread predictions. To enhance the forecasting accuracy of active fires, data assimilation plays a vital role by integrating observations (such as remote-sensing data) and fire predictions generated from numerical models. This paper provides a comprehensive investigation on the application of a recently proposed diffusion-model-based filtering algorithm -- the Ensemble Score Filter (EnSF) -- to the data assimilation problem for real-time active wildfire spread predictions. Leveraging a score-based generative diffusion model, EnSF has been shown to have superior accuracy for high-dimensional nonlinear filtering problems, making it an ideal candidate for the filtering problems of wildfire spread models. Technical details are provided, and our numerical investigations demonstrate that EnSF provides superior accuracy, stability, and computational efficiency, establishing it as a robust and practical method for wildfire data assimilation. Our code has been made publicly available.
Paper Structure (13 sections, 28 equations, 10 figures, 3 tables)

This paper contains 13 sections, 28 equations, 10 figures, 3 tables.

Table of Contents

  1. Introduction
  2. An Overview of Wildfire Spread Models
  3. Data Assimilation using EnSF
  4. Data for FARSITE
  5. The Fire-EnSF Data Assimilation Framework
  6. Additional Implementation Details
  7. Vertices Re-interpolation
  8. Polygon Vertex Normalization
  9. Numerical Experiments
  10. Haversine Distances
  11. Example I: A Simulation-Based Experiment
  12. Example II: Application to Real Wildfire Data
  13. Conclusions and Discussion

Figures (10)

  • Figure 1: The growth of the number of vertices (of the fire boundary polygons) in FARSITE wildfire simulations (the legend on the top right shows the fire ID for four testing fire use cases)
  • Figure 2: Data assimilation for FARSITE using the EnSF
  • Figure 3: Hyperparameter tuning results for EnSF with the best 3 options circled with a red square. Left: RMSE between ground truth and estimation; Right: RMSE between observation and estimation.
  • Figure 4: Hyperparameter tuning for EnSF showing the RMSE between the estimated states and the ground truth.
  • Figure 5: RMSE comparison between the estimated states and the ground truth.
  • ...and 5 more figures