[Paper Review] Cell2Fire: A Cell Based Forest Fire Growth Model

Cristóbal Pais, Jaime Carrasco+3 more|arXiv (Cornell University)|May 22, 2019

Fire effects on ecosystems10 references3 citations

TL;DR

Cell2Fire is an open-source, parallelized, cell-based forest fire growth simulator that uses the Canadian FBP fire spread model to predict fire dynamics across heterogeneous landscapes. Validated against the state-of-the-art Prometheus simulator, it achieves 91.82% accuracy (1-MSE) and 87.91% structural similarity over 22 hours of simulated fire growth, demonstrating strong performance and scalability for large-scale fire modeling in strategic forest and fuel management planning.

ABSTRACT

Cell2Fire is a new cell-based forest and wildland landscape fire growth simulator that is open-source and exploits parallelism to support the modelling of fire growth cross large spatial and temporal scales in a timely manner. The fire environment is characterized by partitioning the landscape into a large number of cells each of which has specified fuel, weather, fuel moisture and topography attributes. Fire spread within each cell is assumed to be elliptical and governed by spread rates predicted by a fire spread model such as the Canadian Forest Fire Behavior Prediction (FBP) System. The simulator includes powerful statistical and graphical output and spatial analysis features to facilitate the display and analysis of projected fire growth. We validated Cell2Fire by using it to predict the growth of real and realistic hypothetical fires, comparing our fire growth predictions with those produced by the state-of-the-art Prometheus fire growth simulator. Cell2Fire is structured to facilitate its use for predicting the growth of individual fires or embedding it in landscape management simulation models. It can be used to produce probabilistic fire scar predictions by allowing for uncertainty concerning the basic spread rate predictions and uncertain weather scenarios that might drive their growth.

Motivation & Objective

• To develop a high-performance, open-source fire growth simulator capable of modeling large-scale wildfire dynamics across heterogeneous landscapes.
• To enable integration of fire growth simulation into long-term strategic forest and fuel management planning frameworks.
• To support probabilistic fire scar predictions by incorporating uncertainty in spread rates and weather scenarios.
• To validate performance and accuracy against the state-of-the-art Prometheus simulator using real and hypothetical fire cases.
• To provide a modular, extensible tool for researchers and practitioners to simulate and analyze fire growth with transparent, reproducible methods.

Proposed method

• The landscape is partitioned into 100×100 m cells, each assigned fuel type, topography, fuel moisture, and weather attributes.
• Fire spread within each cell is modeled as elliptical, with spread rates derived from the Canadian Forest Fire Behavior Prediction (FBP) System.
• The simulator uses a cellular automata approach with parallel computation to scale efficiently across large spatial and temporal domains.
• It supports stochastic ignition, uncertain weather inputs, and uncertainty in rate-of-spread predictions to enable probabilistic fire scar modeling.
• Spatial and statistical outputs, including fire perimeter evolution and structural similarity metrics (SSIM), are generated for analysis and validation.
• Performance is validated by comparing Cell2Fire outputs with Prometheus across a real-world fire case (Dogrib fire, 2001) over 22 hours.

Experimental results

Research questions

• RQ1 How accurately can Cell2Fire predict the growth of real and hypothetical wildfires compared to the state-of-the-art Prometheus simulator?
• RQ2 To what extent does Cell2Fire maintain structural and statistical similarity with observed fire scars over time?
• RQ3 How does Cell2Fire perform in terms of computational scalability and parallelization efficiency for large-scale fire simulations?
• RQ4 Can Cell2Fire effectively support probabilistic fire scar prediction by incorporating uncertainty in spread rates and weather conditions?
• RQ5 How well does Cell2Fire integrate into strategic forest and fuel management planning frameworks?

Key findings

• Cell2Fire achieved a global average accuracy of 91.82% (1-MSE) and 87.91% structural similarity (SSIM) over 22 hours of simulated fire growth, closely matching Prometheus.
• The performance metrics remained stable during the first 4 hours, then declined between hours 4–11 due to extreme weather conditions amplifying differences in fire spread approximation.
• After hour 11, both accuracy and SSIM stabilized, indicating consistent performance despite persistent structural differences in fire perimeters.
• The simulator demonstrated strong agreement with real satellite-observed fire scars from the 2002 Dogrib fire, as confirmed by visual and quantitative comparison.
• Cell2Fire's results were within 20% of Prometheus in both MSE and SSIM across all 22 hours, confirming its reliability and validity for large-scale simulations.
• The model's open-source and modular design enables customization, integration into management frameworks, and support for uncertainty quantification in fire spread predictions.