Assimilation of snow depth data from Sentinel-1 to improve the hydrological model of the Borgne catchment (VS)
Since 2013, Crealp (Centre de Recherche sur l’Environnement Alpin) is operating a realtime flood forecasting system for the upper Rhone river basin. The scarcity of meteorological input data at high elevation leads to model uncertainties. In this project, possible improvements of the sub-model of the Borgne catchment were tested and compared to the operational model and a newly calibrated model using the discharge measurement of the Grande Dixence network. First, a new snowmelt model considering measured solar radiation was tested. Second, the Sentinel-1 derived snow depth product by Lievens et al. (2019) was assimilated in the hydrological model during three winters between 2017 and 2019. The integration of solar radiation showed a promising increase of performance, especially for high elevation catchments. The assimilation of snow data led to contrasting results: some configurations showing a good improvement and some others are less convincing. The differences between modelled and measured water volumes and the validation with MODIS snow cover data led to the conclusion that non-assimilated models underestimate the snowpack, while the assimilated models tend to an overestimation. In the end, some ideas for subsequent steps are given in the outlook.
HURNI_PDM AUTOMNE 2020.pdf
publisher
openaccess
copyright
7.82 MB
Adobe PDF
133114c61efe8cde5a01561e16b1d920