GlaUnTI: Glacier surface mass balance dataset

<strong>Glacier surface mass balance dataset for GlaUnTI (GLAcier-UNiversal Temperature Index model)</strong><br><strong>Contributors</strong>Konstantin A. Maslov (University of Twente)Romain Hugonnet (University of Washington)Florian Tolle (Université Marie &amp; Louis Pasteur, CNRS, French Polar Institute IPEV)Eric Bernard (Université Marie &amp; Louis Pasteur, CNRS, French Polar Institute IPEV)Jean-Michel Friedt (Université Marie &amp; Louis Pasteur, CNRS, French Polar Institute IPEV)Madeleine Griselin (Université Marie &amp; Louis Pasteur, CNRS, French Polar Institute IPEV)Jakub Małecki (Adam Mickiewicz University)Thomas Schellenberger (University of Oslo)Claudio Persello (University of Twente)Alfred Stein (University of Twente) <br><strong>Summary</strong>This dataset provides a unified, observation-driven archive of glacier surface mass balance (SMB), glacier geometry, climate forcing and (optionally) glacier facies for the 78 European glaciers used in GlaUnTI. It is designed to support physically informed and machine learning SMB modelling, including the training and evaluation of differentiable temperature index models, hybrid models with spatial correctors, and purely data-driven baselines. The dataset harmonises multi-source SMB observations (point and glacier-wide), annually resolved elevation models, time-varying glacier outlines, ERA5-Land/ERA5 temperature and precipitation time series and facies maps derived from Landsat/Sentinel-2 when available. All spatial data are registered to the same 100 m grid per glacier. Temporal coverage spans 1995-2024. The dataset is structured so that each glacier forms an independent, self-contained directory.<br><strong>Code and usage examples</strong>The official implementation of GlaUnTI and the dataset dataloader are available at https://github.com/konstantin-a-maslov/glaunti/<br>