Jaime Carrasco-Barra^a,c,*, José Ramón González-Olabarria^j, Camilo Matus-Olivares^a, Felipe Ulloa-Fierro^d,e, Felipe Soto^f, David Palacios^c, Rodrigo Mahaluf^c, Felipe de la Barra^c, Carolina Espinoza^h, Matías Vilches^c, Matilde Rivas^c, Fernando Badilla^c, Horacio Gilabert^g, Miguel Castillo^f, Jorge Saavedraⁱ, Fulgencio Lisón^k, Jordi Garcia-Gonzalo^j and Andrés Weintraub^b,c

^a Universidad Tecnológica Metropolitana, Departamento de Industria, Santiago, Chile
^b University of Chile, Industrial Engineering Department, Santiago, Chile
^c Complex Engineering System Institute - ISCI, Santiago, Chile
^d Department of Industrial Engineering, Faculty of Engineering, Universidad de Talca, Sede Curicó, Chile
^e Programa Doctorado en Ingeniería, Facultad de Ingeniería, Universidad de Talca, Sede Curicó, Chile
^f Forest Fire Laboratory, University of Chile, Box 9206, Santiago, Chile
^g Centro de Cambio Global, Pontificia Universidad Católica de Chile, Santiago, Chile
^h University of Chile, Department of Geophysics, Santiago, Chile
ⁱ Forest Fire Analysis and Prediction Section, Corporación Nacional Forestal, Santiago, Chile
^j Forest Science and Technology Centre of Catalonia, Solsona, Spain
^k Ecosistemas Mediterráneos (ECOMED), Departamento de Ecología e Hidrología, Facultad de Biología, Universidad de Murcia, Campus de Espinardo, 30100, Murcia, Spain

^* Corresponding author.

Disclaimer

This software is for research use only. There is no warranty of any kind; there is not even the implied warranty of fitness for use.

Introduction

Spatial simulation of wildfires is a valuable tool for fire management, supporting both prevention and operational decision-making. However, many existing tools are not publicly available or require restrictive licenses. This paper introduces C2FK, an open-source simulator designed to model fire growth across Chilean landscapes and generate burn probability maps through iterative, spatially explicit simulations. Built on the Cell2Fire engine, C2FK integrates an adaptation of KITRAL, Chile’s national fire behavior system. The implementation includes new equations to estimate rate of spread and length-to-breadth ratio as functions of wind speed, using an elliptical fire growth model. It also incorporates a crown fire behavior module, extending predictive capacity beyond surface fires. We evaluate C2FK by comparing its outputs against synthetic and real wildfire scenarios, demonstrating accuracy and scalability. Cross-platform and parallel-computing enabled, C2FK provides a flexible, high-performance solution to support risk assessment, scenario analysis, and wildfire planning in Chile and other fire-prone environments.

Please cite as: Carrasco-Barra, J., González-Olabarria, J. R., Matus-Olivares, C., Ulloa-Fierro, F., Soto, F., Palacios, D., Mahaluf, R., de la Barra, F., Espinoza, C., Vilches, M., Rivas, M., Badilla, F., Gilabert, H., Castillo, M., Saavedra, J., Lisón, F., Garcia-Gonzalo, J., & Weintraub, A. (2026). New extensions of Cell2Fire software for fire risk analysis and evaluation in Chilean forests. Environmental Modelling & Software.

Usage guide

Compiling

# dependencies
sudo apt install g++-12 libboost-all-dev libtiff-dev
# or brew
brew install gcc@12 libomp boost libtiff # llvm ?

# fork & clone 
git clone git@github.com:<YOU>/C2FK.git
cd C2FK/Cell2Fire

# compile
make 
# there area other makefiles for other platforms, e.g. makefile.macos

Other platform details at .github/workflows/build-.yml and makefile.

Running simulations

Cell2Fire --input-instance-folder /path/to/input-data --output-folder /path/to/output-results --nsims 100

The basic information the simulator needs is fuel and weather data. These must be located in the input instance folder, and should be named "fuels.asc" or "fuels.tif" and "Weathers.csv" respectively. Additional information such as terrain elevation, canopy bulk density, and canopy base height can also be provided. Please use the examples found under /data to guide your input file names.

Configuring simulations

There are many command line options available to configure the simulation.

Argument	Type	Function
nsims	int	Number of simulations to run.
cros	bool	True if crown fire should be included.
fmc	int	Used in Kitral to calculate critical intensity. Default is 100.
nthreads	int	Defines the number of threads to use for simulations. Default is 1.
seed	int	Seed for the random generator.
IgnitionRad	int	If greater than 0, picks a point on the circumference of radius = IgnitionRadius from ignition point.
nweathers	int	Number of weather files. Used to randomly select one.
CCFFactor	float	Weighs by the cell’s CCF when calculating crown ROS in S&B. Default is 0.
EFactor	float	Weighs radial distance from the ellipse center. Default is 1.0.
BFactor	float	Weighs back ROS and ellipse "c". Default is 1.0.
FFactor	float	Weighs flank ROS. Default is 1.0.
HFactor	float	Weighs head ROS. Default is 1.0.
ROS-Threshold	float	ROS threshold for fire to spread. Default is 0.1.
ROS-CV	float	Used to calculate a random ROSRV value. Default is 0.
FirebreakCells	string	Firebreak plan, sets cells as “NonBurnable”.
weather	string	Options: “rows”, “constants”, “distribution”, “random”. Default is “rows”.
output-folder	string	Directory where results will be saved.
input-instance-folder	string	Directory containing all required files.
bbo	bool	True to use Black Box Optimization.
final-grid	bool	Saves grid state at the end of the simulation.
grids	bool	Saves forest grids during the simulation.
ignitions	bool	Reads ignition points from file or uses random ones.
ignitionsLog	bool	Saves a log of attempted ignitions.
verbose	bool	Prints ultra-detailed log.
out-cfb	bool	Saves crown burned fraction. Only with --cros.
out-crown	bool	Saves layer with crown fire cells.
out-ros	bool	Saves ROS per cell.
out-intensity	bool	Saves Byram fire intensities.
out-fl	bool	Saves flame length results. In S&B, saves surface, crown, and max.
output-messages	bool	Saves file with messages (origin, destination, time, ROS).

Output examples

Most outputs are ASCII raster layers, of the same shape as the original fuels raster.

The following is the fireline intensity output for a small simulation. The cells that have 0 are cells that did not catch fire.

ncols 7
nrows 7
xllcorner 457900
yllcorner 5.7168e+06
cellsize 100
NODATA_value -9999
0 4389.03 0 0 0 0 0
0 0 17663.2 18731.9 0 4201.66 0
1711.45 0 1731.95 0 5487.64 9923.44 0
0 14655.2 6652.72 5804.41 4371.23 9348.01 0
0 9441.69 7127.24 0 0 0 0
0 0 7692.64 4389.03 3107.24 2032.65 0
0 8251.82 0 0 17663.2 18731.9 0

Illustrative examples

Simulated fire propagation scars and isochrones for: (A) Talcamávida, (B) Santa Elcira, and (C) Trupán II.

The red point indicates the ignition location. The first columns highlight the initial stages of fire spread from the ignition point, while the final column shows the complete temporal evolution of the fire. This layout enables visualization of both early-stage spread dynamics and overall fire extent.

About us

We are a research group that seeks solutions to complex problems arising from the terrestrial ecosystem and its natural and anthropogenic disturbances, such as wildfires.

Currently hosted at ISCI offices.

Contact us on Discord.

Visit our public webpage.