convml-tt example triplet datasets (GOES-16, Nx256, 200 km tiles)

Example triplet-tile datasets for use with convml-tt (Convolutional Neural Network for study of convective organisation via triplet training). Each triplet has anchor, neighbor and distant 256x256 pixel RGB tiles (200 km across) rendered from GOES-16 ABI radiances over the tropical Atlantic, plus a YAML meta file giving tile locations and source files. Nx256_s200000.0_N500study_N2000train.tgz: 2000 training triplets and 500 study triplets (convml-tt ExampleData.LARGE2000S500) Nx256_s200000.0_N0study_N100train.tgz: the first 100 training triplets of the large dataset (ExampleData.SMALL100) Nx256_s200000.0_N0study_N10train.tgz: the first 10 training triplets of the large dataset (ExampleData.TINY10) The two smaller datasets are subsets of the large one. They replace earlier versions that were hosted at homepages.see.leeds.ac.uk, which is no longer available.