ml-fuel: Predicting Fuel Load with Machine Learning on Climate Data

We use earth observation data with Machine Learning to predict Fuel Load, which embodies the content of burnable vegetation in an area.

Two models are developed, for the Mid-Latitudes and the Tropics using Gradient Boosted Decision Tree style Machine Learning methods. We can see below a visual comparison of the predictions made by the model and corresponding ground truth.

• CatBoost for Mid-Latitudes

  

• LightGBM for Tropics

  

We recommend you go through the Getting Started section to first set up your development environment. Once you have the input data (as .nc NetCDF files) and the dependencies installed, you can either refer to the example notebooks in notebooks/ or go through the further notes on preprocessing, training and testing. More details can be found in the Module API docs in the index below.

Contents:

• Getting Started
• Data Description
• Pre-processing
• Training
• Training CatBoost (Mid-Latitudes)
• Training LightGBM (Tropics)
• Inference
• CatBoost inference
• LightGBM inference
• Adding New Features
• Training and Inference using Docker

Indices and tables

• Index

• Module Index

• Search Page

Note

This repository was developed by Anurag Saha Roy (@lazyoracle) and Roshni Biswas (@roshni-b) for the ESA-SMOS-2020 project. Contact email: info@wikilimo.co. The repository is now maintained by the Wildfire Danger Forecasting team at the European Centre for Medium-range Weather Forecast.