The module adddresses tropical cyclones analysis by providing the tools for gathering and pre-processing data, training ML models, post-process results.
A version that uses GNNs instead of CNNs is available and written in PyTorch
The development of the Python packages is currently underway. In this version a set of data-driven models based on VGG have been developed and tested. An ML ensemble of VGG neural networks has also been implemented as well as an alternative version of the pipeline that makes use of GNNs rather than CNNs.The library includes a set of features for preparing the data for training the model including: capabilities for splitting the input gridded data into non-overlapping patches, patches selection and storage as TFRecords. Moreover, the code includes the trainable model for TC detection.
Next versions will also include features for gathering and managing climate change future projection data like from CMIP6 archive. However, since such data will be needed also by the wildfires and eddies-related ML models, the gathering capabilities could be developed as a separate thematic module. A multi-model thematic module will be also developed to support ensemble detection over multiple CMIP6 data. Finally, we will explore the possibility of implementing a data-driven approach for TC tracking.