Articles | Volume 18, issue 13
https://doi.org/10.5194/gmd-18-4045-2025
https://doi.org/10.5194/gmd-18-4045-2025
Model description paper
 | 
03 Jul 2025
Model description paper |  | 03 Jul 2025

Computationally efficient subglacial drainage modelling using Gaussian process emulators: GlaDS-GP v1.0

Tim Hill, Derek Bingham, Gwenn E. Flowers, and Matthew J. Hoffman

Data sets

PROMICE and GC-Net automated weather station data in Greenland P. How et al. https://doi.org/10.22008/FK2/IW73UU

Model code and software

GladsGP: v1.0.0 T. Hill et al. https://doi.org/10.5281/zenodo.15053542

lanl/SEPIA: v1.1 J. Gattiker et al. https://doi.org/10.5281/zenodo.4048801

Download
Short summary
Subglacial drainage models represent water flow beneath glaciers and ice sheets. Here, we train fast statistical models called Gaussian process (GP) emulators to accelerate subglacial drainage modelling by ~ 1000 times. We use the fast emulator predictions to show that three of the model parameters are responsible for > 90 % of the variance in model outputs. The fast GP emulators will enable future uncertainty quantification and calibration of these models.
Share