Long-Term Glacier Mass Balance Reanalysis Using Data Assimilation: Case Studies from Scandinavia

World-wide glaciers are losing mass which affects global sea-level, river runoff, freshwater influx to the oceans, glacier-related hazards, and landscape changes, with implications for human livelihoods and ecosystems.

Robust glacier mass balance estimates at a high temporal and spatial resolution are hence essential to effective adaptation strategies. 

We outline a probabilistic formalism, based on a modified particle scheme, for using stake, glacier wide, and geodetic mass balance measurements to infer the parameters of a numerical glacier evolution model - the Python Glacier Evolution Model, PyGEM. The particle method iteratively estimates the posterior probability distribution of the dynamical glacier state vector  while successfully accommodating data gaps as well as model nonlinearity and non-Gaussianity.

Our method is tested on different glaciers representing a broad range of climatic conditions and glacial contexts across Scandinavia. 

The approach leverages the combined strengths of the numerical model’s glacier physics-based predictive capabilities with the observations’ direct representation of glacier conditions, providing a robust estimate of glacier mass balance and its associated uncertainties.

This study underscores the value of Bayesian data assimilation, offering a robust and computationally tractable tool for estimating past, current and future glacier changes with high spatiotemporal coverage.