WildfireSpreadTS: A dataset of multi-modal time series for wildfire spread prediction

This repository contains the code corresponding to the paper with the name above.

Setup the environment

pip3 install -r requirements.txt

Preparing the dataset

The dataset is freely available at https://doi.org/10.5281/zenodo.8006177 under CC-BY-4.0. For training, you will need to convert them to HDF5 files, which take up twice as much space but allow for much faster training.

To convert the dataset to HDF5, run: python3 src/preprocess/CreateHDF5Dataset.py --data_dir YOUR_DATA_DIR --target_dir YOUR_TARGET_DIR substituting the path to your local dataset and where you want the HDF5 version of the dataset to be created.

If you want to skip this step, and simply pass --data.load_from_hdf5=False on the command line, but be aware that you won't be able to perform training at any reasonable speed.

Re-running the baseline experiments

We use wandb to log experimental results. This can be turned off by setting the environment variable WANDB_MODE=disabled. The results will then be logged to a local directory instead.

Experiments are parameterized via yaml files in the cfgs directory. Arguments are parsed via the LightningCLI.

Grid searches and repetitions of experiments were done via WandB sweeps. Those are parameterized via yaml files in the cfgs directory prefixed with wandb_. WandB configuration files ending in _repetition contain the configurations for the runs that were used in the paper. They only needed to be repeated five times with varying seeds to estimate the variance of results. We omit explanations of how to use wandb sweeps to run experiments and refer the readers to the original documentation. To run the same experiments without WandB, the parameters specified in the WandB sweep configuration file can simply be passed via the command line.

For example, to train the U-net architecture on one day of observations, which is specified in cfgs/unet/wandb_monotemporal_repetition.yaml, we could simply copy and paste the WandB parameters to the command line:

python3 train.py --config=cfgs/unet/res18_monotemporal.yaml --trainer=cfgs/trainer_single_gpu.yaml --data=cfgs/data_monotemporal_full_features.yaml --seed_everything=0 --trainer.max_epochs=200 --do_test=True --data.data_dir YOUR_DATA_DIR

where you replace YOUR_DATA_DIR with the path to your local HDF5 dataset. Alternatively, you can permanently set the data directory in the respective data config files. Later arguments overwrite previously given arguments, including parameters defined in config files.

Re-creating the dataset

The code to create the dataset using Google Earth Engine is available at https://github.com/SebastianGer/WildfireSpreadTSCreateDataset.

Using the dataset for your own experiments

To use the dataset outside of the baseline experiments, you can use the Lightning Datamodule at src/dataloader/FireSpreadDataModule.py which directly provides dataset loaders for train/val/test set. Alternatively, you can use the PyTorch dataset at src/dataloader/FireSpreadDataset.py.

Citation

Will be added upon acceptance of the paper.