Inferring glacier meteorology with physical modeling and remote sensing

Meteorology is crucial to understand the rapid response of mountain glaciers to climate warming, but is often challenging to observe and simulate due to site inaccessibility, instrument maintenance and the complex interactions between glaciers and their surroundings. Recent, high-resolution, globally-available remote sensing observations create an opportunity to exploit the observed changes in glacier volumes and surface properties to infer bias-corrected high-mountain meteorology from climate reanalysis. In this study, we develop a unified method for model inversion based on Monte Carlo simulation and Bayesian inference, and then evaluate it on four benchmark glaciers with extensive in-situ measurements of surface meteorology (Argentière Glacier and Aletsch Glacier in the European Alps, Abramov Glacier and Parlung No.4 Glacier in High Mountain Asia).

Our approach is a multiparameter optimization that uses a physical-based land-surface model (Tethys-Chloris) driven by an ensemble of statistically-downscaled ERA5-Land reanalysis datasets, with remote-sensing-derived glacier surface mass balance and glacier albedo as targets. With this method, we obtain the bias of air temperature, precipitation and incoming shortwave radiation to correct the reanalysis data during the period 2015-2019 at the four sites. The results show that the derived multiyear meteorology is spatially variable over the glaciers and agrees with independent in-situ observations at each site. The good performance of this method in different climatic conditions paves the way to derive multiyear glacier meteorology on the world’s mountain glaciers and constrain globally a key control on their response to climate change.