Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.
The main objective of this research is to keep up existing work related to the hydrological modeling of the Birse River in Switzerland. The present studies were performed using the software package HydroRoute System, being developed by Richard Metzger. HydroRoute incorporates the mathematical model entitled GSM-Socont (Glacier and Snowmelt - SOIL CONTribution Model) developed by Schaefli et al. (2005). Although the efficiency of this model has been proved (Beck, 2006), these studies have not yielded satisfactory results for watersheds of small and medium size. Consequently, this study works on carrying out changes in the hydrological model built in HydroRoute System so that it delivers hydrological responses closer to observations. According to previous studies, the extreme flows during summer thunderstorms are undervalued by the model for small watersheds. Two source code adjustments were developped and tested during this work. The first was kept as simple as possible, and under the assumption that too much water infiltrates during these extreme rainfall events. Therefore the production function was adapted by inserting a rainfall threshold parameter. Beyond this threshold, it assigns any remaining water quantity to the runoff reservoir. This change requires an additional parameter to calibrate. Once the calibration performed, the simulations lead to satisfactory results for the extreme discharges generated values. However, the agreement between observed and simulated discharge values was deteriorated. The modification seems to be insufficient and inadequate to the purpose of improving the overall model performance at the study site. Hence a second and more significant change was tested. Based on the knowledge of other existing models, adding an additional reservoir could allow greater flexibility of the model to approximate the observed data. The results showed an improved agreement between simulated and observed discharges and extreme discharges. In addition, the model has given better results for simulated discharge values during extreme events. However, it has imposed the insertion of three additional parameters. A disadvantage of increasing the number of parameters lies in an increase in parameters combinations to calibration of the model and therefore a substantial computing time required for this phase.