FloodML - Monitoring floods using Sentinel-1, Sentinel-2, Landsat 8, landsat 9 and TerraSAR-X data

This project aims at detecting and monitoring floods using a Machine Learning appraoch with Random Forest.

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development, demonstration and testing purposes.

Prerequisites

The program has dependencies on preconfigured pip environments, which is found in the requirements.txt.

Installing

The python environment is managed via pip, but we recommend creating a virtual environment using conda first:

cd floodml
conda create --name rapids-0.21.08 --file requirements-R02108.txt
conda activate rapids-0.21.08

Run the software

The following scripts are used for preprocessing, training and inference of both approaches.

Satellite imagery:

• RDF-1-preparation.py: Prepares the data in numpy format, creates the training database.
• RDF-2-training.py: Runs the training algorithm
• RDF-3-inference.py: Runs a prediction using the trained model and an image file

Trained models

Trained models based on Sentinel 1 and Sentinel 2 data are only available on request

Copyright

• Copyright 2020-2024, Centre National d'Etudes Spatiales (CNES)

License

This project is licensed under the ApacheV2 License - see the LICENSE.md file for details

Acknowledgements

Thanks to JPL and CLS group for their contribution on the development of FloodML.