Long-term hydrological behavior of an Alpine glacier

Abstract

Alpine glaciers are shrinking at a relentless pace, as an effect of global warming. The impact of these changes in the European Alps is relevant, given the importance of this territory from both ecological and economic viewpoints. While the ubiquitous reduction of glaciers’ mass through the Alps has been reported in several studies, its effect on streamflow is less studied. In the present work, we analyzed the long streamflow time series, available since 1976, of the Careser Stream, a headwater stream emerging from the Careser Glacier, in northeastern Italy. A large amount of missing data characterizes the streamflow time series, which we filled with a high level of confidence by using a Feed Forward Deep Neural Network algorithm. We explored the influence of climatic drivers on streamflow at monthly and annual time scales. Our analysis highlighted significant changes in the timing of streamflow due to the combined effect of early snow melting and a progressive reduction of the glacier’s area. The hydrological regime changed significantly with glacier water contributing to streamflow proportionally less after a tipping point identified in 1996. Projections performed by using eight bias-corrected climate models of the EURO-CORDEX collections confirmed this tendency with the complete transition of the catchment from the glacial to the nival regime by 2045 at the latest. The Deep Neural Network algorithm was very effective in filling the missing streamflow data and this offers an encouraging premise for applications in other glaciers of the Alps suffering from the same transformation.