CloudViT - Method code and data for the article "CloudViT: exploring cloud type classification with vision transformers in global satellite data."

Method code and data relating to the manuscript: CloudViT: exploring cloud type classification with vision transformers in global satellite data. Lenhardt, J. (1), Quaas, J. (1, 2), Sejdinovic, D. (3), Klocke, D. (4) 1 Leipzig Institute for Meteorology, Leipzig University, Leipzig, Germany 2 ScaDS.AI - Center for Scalable Data Analytics and Artificial Intelligence, Leipzig University, Humboldtstraße 25, 04105 Leipzig, Germany 3 School of Computer and Mathematical Sciences & Australian Institute for Machine Learning, University of Adelaide, Adelaide, Australia 4 Max Planck Institute for Meteorology (MPI-M), Hamburg, Germany Lenhardt, J., Quaas, J., Sejdinovic, D., and Klocke, D.: CloudViT: exploring cloud type classification with vision transformers in global satellite data, EGUSphere [ACP-preprint], https://doi.org/10.5194/egusphere-2024-2724, 2024. Data files available from /data/global_2016/ Global aggregated monthly cloud type predictions (count, time, ..) in netCDF files at xr_global_month_{month}_2016.nc Table of cloud type predictions (along with latitude, longitude, time, corresponding MODIS swath file, ...) in csv files at df_global_month_{month}_2016.csv Data files available from /data/climate_model/ Global aggregated cloud type predictions (count, time, ..) in netCDF file at xr_Apollo17_predictions.nc Method files available from /method/ Method class and functions available at /utils/utils.py Model PyTorch architecture available at /models/models.py Models' weights available from /models/models_save/ Example notebook using method's class and functions available at CloudViT_prediction_example.ipynb Archive structure: Notebook with method example CloudViT_prediction_example.ipynb Folder data with global predictions for the year 2016, test set predictions, climate model predictions, example data and data needed for the manuscript's plots Folder method with models files, corresponding PyTorch class, subfolder utils with code files for running the method Folder plots with saved figures for the manuscript and corresponding notebook plots_CloudViT.ipynb File requirements.txt for Conda environment specifications