*** Physically-informed super-resolution downscaling of Antarctic surface melt *** 
Authors: Sophie de Roda Husman, Zhongyang Hu, Maurice van Tiggelen, Rebecca Dell, Jordi Bolibar, Stef Lhermitte, Bert Wouters, Peter Kuipers Munneke
Corresponding author: Sophie de Roda Husman Contact Information: S.deRodaHusman@tudelft.nl 
Delft University of Technology - Faculty of Civil Engineering and Geosciences, Stevinweg 1, 2628 CN Delft, The Netherlands 

***General Introduction***
The dataset provided in this repository corresponds to the original data used in the publication by De Roda Husman et al. (2024) titled "Physically-informed super-resolution downscaling of Antarctic surface melt" (DOI to be announced!). 

The dataset, named SUPREME, contains a record of surface melt (in meters water equivalent per year) on an Antarctic-wide scale. It offers surface melt data at a high spatial resolution of 5.5 kilometers for each melt year (1 July - 30 June), covering the period from 2001 to 2019. Our method downscales Antarctic surface melt from the regional climate model Regional Atmospheric Climate Model (RACMO, v2.3p2), employing a physically-informed super-resolution architecture. The super-resolution model integrates physical information derived from remote sensing data, specifically surface albedo and elevation.


***Description of the data in this data set***
The following data are available:
1. SUPREME: The main end product of this research, providing downscaled surface melt using the entire super-resolution algorithm and all input features (RACMO 27 km, albedo, elevation).
2. SRResNet: A single-image surface melt product, using only RACMO 27 km to downscale surface melt.
3. Additionally, several versions of the SUPREME dataset are provided where specific input features or plugins have been excluded. The following subsets are available:
- SUPREME_onlyEncoder
- SUPREME_onlySpotlighter
- SUPREME_onlyPhysicalActivation
- SUPREME_NoAlbedo
- SUPREME_NoElevation

All files are in TIF format with eighteen bands, each corresponding to melt seasons from 2001-2002 to 2018-2019. The data are measured in meters water equivalent per year and projected in EPSG:3031.

