Atmospheric corrections for Earth observation with deep learning - Supplementary material

This is supplementary material for the paper, "Atmospheric corrections for Earth observation with deep learning", published in the IGARSS 2024 proceedings through IEEE Xplore 05 September 2024. --------------------------- Contents: Source_code.zip: Contains the Python source code for training the R2U-Net for doing atmospheric correctionsusing Sentinel-2 L1C and L2A image pairs. Data.zip: Contains information related to the Sentinel-2 image data used, like the Sentinel-2 image file names and ids for downloading from the Copernicus Browser, for both training and testing with the deep learning model. There are two Python-files for downloading and preprocessing the raw images into appropriate TIFs. There is one image showing the R2U-Net architecture, and one image showingwhere the Sentinel-2 images are located. Results.zip: Contains some of the results achieved with the trained R2U-Net. This includes the saved best model after training for 1626 epochs on the training dataset. And two folders; one containing various logged metrics during training and validation, and an image of the network graph, and a folder containing various prediction images for several of the test image chips.