← Back to Projects

Bayesian Data Assimilation on Arctic Glaciers

Learning from Large Ensemble Twin Experiments

Overview

This research develops Bayesian data assimilation methods to improve predictions of Arctic glacier mass balance by optimally combining observational data with physical models.

Methodology Workflow

1

Prior Model

Glacier dynamics simulation with initial parameter estimates

→
2

Observations

Satellite & in-situ measurements from Svalbard

→
3

Data Assimilation

Ensemble Kalman Filter combines model & data

→
4

Posterior Analysis

Improved predictions with quantified uncertainty

Core Components

EnKF

Ensemble Kalman Filter

Optimal state estimation combining model predictions with observations

OSSE

Twin Experiments

Synthetic observations to validate assimilation schemes

UQ

Uncertainty Quantification

Rigorous estimation through ensemble spread analysis

LE

Large Ensemble

Robust statistical inference with computational resources

Data Integration

🛰 Satellite Remote Sensing
📍 In-situ Field Measurements
🌡 Climate Reanalysis Data
⟶
Bayesian
Data Assimilation
⟶
📊 Mass Balance Predictions
📈 Uncertainty Estimates

Publication

Bayesian data assimilation on an Arctic glacier: learning from large ensemble twin experiments

Cao W, Aalstad K, Schmidt LS, Westermann S, Schuler TV · Journal of Glaciology, 2025

Read Full Article