Author: Alexis Neven and Valentin Dall'Alba
Publication: The Cryosphere, 2021
DOI: https://doi.org/10.5194/tc-15-5169-2021
This repository contains the data processing pipeline and simulation scripts used in the study "Ice volume and basal topography estimation using geostatistical methods and ground-penetrating radar measurements: application to the Tsanfleuron and Scex Rouge glaciers, Swiss Alps" from Neven and Dall'Alba 2021.
The goal is to interpolate glacier bedrock topography from ground-penetrating radar (GPR) measurements using multiple-point statistics (MPS), with a focus on capturing complex subglacial features and providing robust uncertainty estimates.
Ground-penetrating radar (GPR) is widely used to estimate glacier thickness, but measurements are typically acquired only along discrete profiles. This creates challenges when interpolating between profiles, especially in glaciated terrain with complex basal geomorphology.
Traditional interpolation methods like Ordinary kriging or Sequential Gaussian simulation (SGS) often fail to reproduce realistic basal structures and tend to underestimate associated uncertainty.
We apply the Direct Sampling (DS) algorithm from the Multiple-Point Statistics framework to interpolate bedrock topography from GPR measurements.
This approach allows :
- Reproduces complex geomorphology (e.g., karstic features)
- Provides a set of equiprobable realizations for uncertainty quantification
- Enhances ice volume estimation accuracy
We evaluate MPS against kriging and SGS on:
- A synthetic glacier dataset with known ground truth
- The real Tsanfleuron Glacier dataset (Switzerland)
Estimated ice volume for the Scex Rouge and Tsanfleuron glaciers : 113.9 ± 1.6 million m³ MPS accurately captures basal morphology and uncertainty, improving the quality of subglacial flow and hydrology modeling.
