AlKa-DL: Alpine karst spring discharge prediction

Karst aquifers are of immense importance in the Alpine region. They cover about 56% of the area and supply a considerable part of the population with drinking water (including major cities such as Vienna, Graz, and Innsbruck). Furthermore, karst springs are connected to valuable ecosystems and contribute to hydropower generation. However, of all things these very systems are severely affected by the consequences of climate change. For example, it can be assumed that climate change will lead to a reduction in snow and ice masses (which in turn shifts the water balance). Responsible and sustainable water management in karst areas requires precise models. Modern modelling approaches for karst spring discharge are often highly specialised and, due to the need for manual tuning, and thus mostly limited to site- specific questions. The AlKA-DL project addresses this challenge by developing innovative Deep Learning based approaches for modelling Karst runoff. Deep Learning is a subfield of machine learning that has already been successfully applied in many areas. It enables the development of data-driven models that can deliver highly accurate results with relatively little prior knowledge of the underlying systems. We use data from Austria, Switzerland, Germany, France, Italy, and Slovenia (the focus is on the mountain area defined by the Alpine Convention, which is particularly affected by climate change) as the basis for our model construction and testing. The aim of the project is to investigate the suitability of new data-driven modelling approaches for the impact assessment of climate change. In this way, AlKA-DL will enable short- and long-term predictive modelling tasks and thus make an important contribution to sustainable water management in karst areas.