Thematic Modules: Environment


Climate analytics and data processing


Providing a set of Python modules for supporting processing and analysis of wildfires related data

The module addresses wildfire analysis and prediction by providing tools that allow users to pre-process data, choose model architecture, train the model, and post-process results (basic visualisation).

Release Notes

The development of the Python package is currently ongoing. In this first version, a Machine Learning pipeline has been developed for running the different thematic module components. The capabilities include subsetting SeasfireCube data in Training, Validation and Testing, after a preliminary selection of drivers and target variables. The ML model used for this version is a standard UNet++ network; users can customise the network depth. The module includes utilities for training models and visualising data.

Future Plans

The package migration in PyTorch is underway for enabling seamless integration with other components from the project. Multiple Machine Learning models will be trained, validated, and tested with optimised hyperparameters configuration. The final models will then be applied on climate projection data with different CMIP6 scenarios. Furthermore, the potential of combining predictions from multiple models on CMIP6 data for enhanced accuracy in an ensemble approach we’ll be explored. The documentation will also be extended with additional information.

Target Audience
  • DT Developers
  • Expert Scientists

Apache v2.0

Created by

Related Use Case

interTwin Use Case: A Digital Twin for projecting wildfire danger due to climate change

Defining deep neural networks (DNNs) for creating Digital Twins for predicting fire danger on climate...