Machine learning aided forecasting of subseasonal drought indicators in the European alps
Annie Yuan-Yuan ChangVice
2020 - 2024
Drought is a natural disaster that is not historically associated with the European Alps, however, studies have indicated an increasing risk of drought in central Europe and the Alpine Space in the future with climate change. This research aims to explore the predictability of drought at a monthly time horizon in the form of low flow in Switzerland and the European Alps by using machine learning techniques and linking large scale weather patterns with local hydrological events. Furthermore, the scope of this research will be extended to the impact of drought and explore forest fire prediction at similar lead times. The outcome of this research is expected to provide useful information on drought indicators and its impacts to decision makers in sectors such as agriculture, forest management, hydropower production, navigation and transportation to better mitigate losses associated with drought.
This PhD project is supervised by Massimiliano Zappa and Konrad Bogner from the Hydrological Forecasts group at WSL, as well as Daniela Domeisen from the Institute for Atmospheric Science at ETH Zurich. It is also an integrated part of the Extreames propram at WSL. External partners include MeteoSwiss for providing extended range meteorological forecasts and Christian Grams from KIT, Germany for providing weather regime forecasts.
- Sub-Project 1: The first sub-project aims to predict low flows and lake levels in large rivers and lakes in Switzerland with lead time up to one month. The concept is to achieve this goal by training a machine learning model with catchment level hydrological forecasts from the in-house PREVAH model as inputs. The outcome of this project can contribute to the drought.ch platform, which provides information on current and potential drought in Switzerland.
- Sub-Project 2: The second sub-project will extend the scope to the European Alps with the same goal of low flow and lake level prediction at a monthly time scale. This is part of the Alpine Drought Observatory (ADO) INTERREG project commissioned by the European Union, which is supported by Canton Ticino, Canton Thurgau and Bundesamt für Raumentwicklung (ARE).
- Sub-Project 3: In the last sub-project, the goal is to connect hydrological drought with forest fire in order to issue early warnings at a monthly horizon for better forest management in Switzerland. Boris Pezzatti at WSL Cadenazzo is a collaborator of this sub-project.