Glacier catalogue for IGM physics-informed deep-learning emulator pretraining

This dataset was created with the iceflow glacier model CfsFlow to generate glacier extent and retreat in the Alps and New zealand with the goal to generate realistic and diverse glacier states for pretraining the physics-informed deep-learning emulator of IGM (https://github.com/jouvetg/igm). The data consists of distributed surface topography (usurf) and ice thickness (thk) of 8 snapshots of 37 glaciers in different stages (advance and retreat). The data is organized glacier-wise: each folder corresponds to one glacier, which contains a unique NetCDF file with 2D distributed raster data of surface elevation and ice thickness.本数据集依托冰流冰川模型CfsFlow构建，用于生成阿尔卑斯山脉与新西兰境内的冰川范围及进退演化过程，核心目标是生成逼真且多样化的冰川状态，以用于预训练IGM（igm）的物理信息驱动深度学习模拟器（https://github.com/jouvetg/igm）。数据集包含37座冰川在不同演化阶段（前进与退缩）的8组快照数据，涵盖分布式地表地形（usurf）与冰厚（thk）两类参数。该数据集按冰川维度进行组织：每个文件夹对应一座冰川，内含一份唯一的NetCDF文件，存储地表高程与冰厚的二维分布式栅格数据。