Training dataset for land surface temperature guided super-resolution This datset contains the training, validation and evaluation sets for training a guided super-resolution alogirthm in order to downscale AVHRR data to the spatial resolution of the MODIS LST dataset. The dataset is splited in three distinct folders: Train Val Eval All data were extracted from the ESA CCI Climate toolbox (https://climate.esa.int/en/data/toolbox/) and saved as separate GeoTiff files. The LST data were extracted from the ESA LST CCI products 'MODIS-AQUA' and 'IRCDR'. In addition auxiliary data such as land cover, digital elevation model and vegetation height data used as a guide for the guided super-resolution are provided. The following datasets were used to create the guide: The Copernicus digital elevation model (DEM) GLO-90, upscaled to 0.01° spatial resolution (https://doi.org/10.5270/ESA-c5d3d65). ESA CCI Land Cover for the year 2005 (https://www.esa-landcover-cci.org). The spatial resolution of the original data was upscaled to 0.01° to match the MODIS LST dataset. This upscaling was achieved by selecting the most frequent land cover class within each upscaled pixel, ensuring consistency with the original land cover distribution. A high-resolution canopy height model (https://langnico.github.io/globalcanopyheight/). All data has been aligned to match the extact MODIS gridded product at 1km spatial resolution. More information can be found in the dedicated paper (Preprint). Credit: To use this data please cite this dataset and the respective journal publication: Dupuis, S., Metzger, N., Schindler, K., Göttsche, F., & Wunderle, S. (2025). Four decades of circumpolar super-resolved satellite land surface temperature data. arXiv. https://doi.org/10.48550/arXiv.2511.17134 @article{Dupuis2025, author = {Sonia Dupuis and Nando Metzger and Konrad Schindler and Frank Göttsche and Stefan Wunderle}, title = {Four decades of circumpolar super-resolved satellite land surface temperature data}, year = {2025}, eprint = {arXiv:2511.17134}, archivePrefix= {arXiv}, primaryClass = {cs.LG}, doi = {10.48550/arXiv.2511.17134} }