Deliverable 2.4 Glacier mass balance data set

Report covering data description and methods has been added.This dataset provides annual glacier mass balance estimates derived using an uncertainty-aware machine learning framework based on Bayesian Neural Fields. The dataset covers all glacierized regions defined in RGIv7 except Regions 19 and 20 (19: Sub-Antarctic and Antarctic Islands, 20: Antarctic Mainland). Temporal coverage spans 1979–2019. This dataset is intended as a global, consistent Bayesian prior for glacier mass balance. This is a preliminary release of pretrained outputs; it is intended to work as a prior for Bayesian ML and has not been finetuned on observations. The dataset structure is subject to significant change after finetuning on observations.