This repository contains a Jupyter Notebook (main.ipynb), where the Fire Weather Index (FWI) is emulated using deep learning techniques with basic climate variables as input features. The config_nb.yaml file provides the deep learning optimization parameters, which can be adjusted by the user as desired.
The required packages and dependencies to run the experiments are listed in environment.yaml. To set up the environment, follow these steps:
- Create the environment using Mamba for faster dependency resolution:
conda create -n deep-fwi -c conda-forge mamba
- Activate environment
conda activate deep-fwi
- Use Mamba to install the packages required
mamba env create -f environment.yaml
Once the environment is installed and activated, open the Jupyter notebook and enjoy emulating! :)
The data required to run the code is available on Zenodo:
- Mirones, Ó., Bedia Jiménez, J., & Baño-Medina, J. (2025). Toy Dataset for Emulating the Fire Weather Index (FWI) Using Deep Learning Techniques [Data set]. Zenodo. https://doi.org/10.5281/zenodo.15075367
Instructions for downloading it can be found within the notebook.