IceBoost - a Gradient Boosted Tree global framework for glacier ice thickness retrieval ...

• Main Author: Maffezzoli, Niccolò
• Format: Article in Journal/Newspaper
• Language: English
• Published: Zenodo 2024
• Subjects: glacier modeling; Machine learning; Artificial intelligence; ice sheet; Antarc*; Antarctica; glacier; Greenland; Ice Sheet
• Online Access: https://dx.doi.org/10.5281/zenodo.13145836; https://zenodo.org/doi/10.5281/zenodo.13145836

The repository contains: IceBoost model ensemble (XGBoost mudule: .json file, CatBoost module: .cbm file). Training datasets (.csv), in 2 versions: RAW and spatially downscaled. Comparisons between IceBoost and other two models, on 10 selected glaciers for each one of the 19 regions of the Randolph Glacier Inventory v.6 (.zip archive), alongside a .csv file listing the ids of the compared glaciers. In all comparisons: Model 1 refers to Millan et al. (2022), or BedMachine Greenland/Antarctica (if applicable). Model 2 refers to Farinotti et al. (2019). BedMachine Greenland v5: Morlighem, M. et al (2022). BedMachine Antarctica v3: Morlighem, M. et al (2022). All colorbars are tied to the IceBoost output, therefore the other two models may appear in saturated colors. The big circles (if any) reflect the ice thickness measurements from GlaThiDa v. 3.1.0. IceBoost is actively developed on Github: iceboost repository. ...