Propagator Examples (Cargèse 2025)

This repository contains the material used during the workshop at Cargèse 2025. Each script demonstrates how to drive the Propagator fire spread core directly from Python, either through a lightweight script (simple_example.py) or via interactive marimo apps (venafro.py, monti_pisani.py). The focus is on fire-simulation workflows that can be re-used in your own studies.

Repository layout

• simple_example.py – fully scripted synthetic scenario showcasing the Propagator API in just a few lines.
• venafro.py – marimo app driven by DEM/vegetation/scar rasters (venafro_*.tif) to replay the Venafro case study.
• monti_pisani.py – marimo app running ensembles of random ignitions over the Monti Pisani rasters.
• utils.py – shared plotting helpers (vegetation map, probability/contour overlays).
• pyproject.toml / uv.lock – dependency pins (Propagator git revision, marimo, rasterio, matplotlib).

Prerequisites

• Python ≥ 3.13
• uv or any virtual environment capable of installing packages listed in pyproject.toml

Install the environment

# create and populate the virtual environment using uv
uv sync

# alternatively, standard venv + pip
python -m venv .venv
source .venv/bin/activate
pip install -e .

ℹ️ The propagator dependency is pulled directly from the CIMA Foundation repository at commit 6674076, matching the version demonstrated during the workshop.

Run the scripted example

uv run python simple_example.py

This script builds a synthetic flat DEM, applies constant wind/moisture settings, and runs the simulation for up to six hours (simulation time). When it completes it displays the probability surface using matplotlib.

Launch the marimo apps

Each marimo file can be run independently; choose the one that matches your scenario:

# Venafro case study (uses venafro_* rasters)
uv run marimo run venafro.py

# Monti Pisani ensemble (ignition uncertainty)
uv run marimo run monti_pisani.py

While the app is running you can experiment with wind speed/direction, moisture, simulation time, number of realizations, and (for Monti Pisani) ignition-radius uncertainty. The UI shows progress via marimo’s status spinner, and plots are updated once the solver finishes.

More information

• Workshop page: https://forefireapi.github.io/cargese2025
• Propagator core: https://github.com/CIMAFoundation/propagator_sim

---

Presentation PDF