@masterthesis{Grosse2021,
  type      = {Bachelor Thesis},
  author    = {Grosse, Peter},
  title     = {Modellierung der Sedimentdynamik im Rofental, {\"O}tztaler Alpen, unter Verwendung eines nichtparametrischen Modells},
  doi       = {10.25932/publishup-50669},
  url       = {http://nbn-resolving.de/urn:nbn:de:kobv:517-opus4-506695},
  school      = {Universit{\"a}t Potsdam},
  pages     = {49},
  year      = {2021},
  abstract  = {The continued melting of glaciers releases more sediment deposits, making them more susceptible to erosion. Increased sediment discharge rates endanger water quality and water supply due to reservoir siltation. In order to better understand these hazards and their evolvement, erosion processes need to be studied, especially in high alpine catchment areas. In this bachelor thesis, sediment concentrations as well as other environmental variables, such as discharge, precipitation and temperature, were measured in the Rofental catchment ({\"O}tztal Alps) and in a highly glaciated sub-catchment of the Rofental. The quantile regression forest machine learning technique was applied to determine the correlation between sediment concentration and ancillary variables. The variables were aggregated to different levels, which allowed past hydroclimatic conditions to be taken into account. Considering the knowledge regarding the influence of these environment variables, the sediment concentration could be retrospectively simulated using a Monte Carlo approach. The resulting continuous time series was used to calculate the annual sediment export rates. The turbidity, as an indicator for the sediment concentration, was measured as well. By determining the correlation between the modelled data and the measured turbidity, the significance of the model could be assessed. The results show that the quantile regression forest model is a suitable approach to reconstruct the sediment dynamics in the Rofental. It further emerged that the discharge has a great impact on the sediment concentration in both catchments, with the relevance of the other variables varying widely between both study areas. The correlation between the measured turbidity and the simulated sediment concentration is strongly positive, whereby debris flows, measurement errors and the rinse of new sediment reservoirs deteriorated the model's performance.},
  language  = {de}
}