Assimilating near-real-time mass balance stake readings into a model ensemble using a particle filter

Short-term glacier variations can be important for water supplies or hydropower production, and glaciers are important indicators of climate change. This is why the interest in near-real-time mass balance nowcasting is considerable. Here, we address this interest and provide an evaluation of continuous observations of point mass balance based on online cameras transmitting images every 20 min. The cameras were installed on three Swiss glaciers during summer 2019, provided 352 near-real-time point mass balances in total, and revealed melt rates of up to 0.12 m water equivalent per day (mw.e.d-1) and of more than 5 mw.e. in 81 d. By means of a particle filter, these observations are assimilated into an ensemble of three TI (temperature index) and one simplified energy-balance mass balance models. State augmentation with model parameters is used to assign temporally varying weights to individual models. We analyze model performance over the observation period and find that the probability for a given model to be preferred by our procedure is 39 % for an enhanced TI model, 24 % for a simple TI model, 23 %, for a simplified energy balance model, and 14 % for a model employing both air temperature and potential solar irradiation. When compared to reference forecasts produced with both mean model parameters and parameters tuned on single mass balance observations, the particle filter performs about equally well on the daily scale but outperforms predictions of cumulative mass balance by 95 %–96 %. A leave-one-out cross-validation on the individual glaciers shows that the particle filter is also able to reproduce point observations at locations not used for model calibration. Indeed, the predicted mass balances is always within 9 % of the observations. A comparison with glacier-wide annual mass balances involving additional measurements distributed over the entire glacier mostly shows very good agreement, with deviations of 0.02, 0.07, and 0.24 mw.e.

Abstract licence: CC BY

77 more references, whose wording the paper’s licence keeps private.

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.