Detecting ML-based convective clouds using 3D observational data

This repository accompanies the article "A ML-based perspective on deep convective clouds and their organisation in 3D. Part I: Influence of deep convective cores on the cloud life-cycle" by Sarah Brüning and Holger Tost, which is to be submitted to ACP.

It contains the code for detecting and tracking convective clouds using ML-based 3D radar reflectivities. The code creates contiguous trajectories for convective clouds while separating DCCs and cloud anvil regions over tropical West Africa (30° N - 30° S, 30° E -  30° W). The result is a dataset of convective cloud trajectories and their cloud physical properties over the lifetime of each object.

The code for predicting 3D radar reflectivities from 2D satellite data using the ML model is available from https://doi.org/10.5281/zen…. The satellite data used in this study are free available at the CloudSat Data Processing Center (cloudsat.cira.colostate.edu) and the EUMETSAT Data Store (navigator.eumetsat.int).