Modelling of future mass balance changes of Norwegian glaciers by application of a dynamical–statistical model

Abstract

The long-term behaviour of Norwegian glaciers is reflected by the long mass-balance records provided by the Norwegian Water Resources and Energy Directorate. These show positive annual mass balances in the 1980s and 1990s at maritime glaciers followed by rapid mass loss since 2000. This study assesses the influence of various atmospheric variables on mass changes of selected Norwegian glaciers by correlation- and cross-validated stepwise multiple regression analyses. The atmospheric variables are constructed from reanalyses by the National Centers for Environmental Prediction and the European Centre for Medium-Range Weather Forecasts. Transfer functions determined by the multiple regression are applied to predictors derived from a multi-model ensemble of climate projections to estimate future mass-balance changes until 2100. The statistical relationship to the North Atlantic Oscillation (NAO), the strongest predictor, is highest for maritime glaciers and less for more continental ones. The mass surplus in the 1980s and 1990s can be attributed to a strong NAO phase and lower air temperatures during the ablation season. The mass loss since 2000 can be explained by an increase of summer air temperatures and a slight weakening of the NAO. From 2000 to 2100 the statistical model predicts predicts changes for glaciers in more continental settings of c. −20 m w.e. (water equivalent) or 0.2 m w.e./a. The corresponding range for their more maritime counterparts is −0.5 to +0.2 m w.e./a. Results from Bayesian classification of observed atmospheric states associated with high melt or high accumulation in the past into different simulated climates in the future suggest that climatic conditions towards the end of the twenty-first century favour less winterly accumulation and more ablation in summer. The posterior probabilities for high accumulation at the end of the twenty-first century are typically 1.5–3 times lower than in the twentieth century while the posterior probabilities for high melt are often 1.5–3 times higher at the end of the twenty-first century than in the twentieth and early twenty-first century.