A study on soil temperature and moisture in the Swiss Plateau, using the model SNOWPACK

The extension of the well-known surface process model SNOWPACK to the soil is relatively new. Studies have mostly been conducted in high-altitude alpine areas, characterized by coarse-grained permafrost. Investigating the role of soil at all altitudes is crucial for various fields, such as catchment hydrology, geotechnical engineering, biology and agriculture. A thorough understanding of the model behavior can provide a tool to assess water cycle, soil moisture and other phenomena in future variable climate change scenarios. The work focuses on quantities such as ground temperature, soil moisture, water loss by evapotranspiration and produced runoff, which, despite often being studied separately, are highly interconnected. SNOWPACK allows to evaluate all of them together. By applying the model to low-altitude sites with fine-grained soils, this thesis pursues a double purpose: on the one hand, a sensitivity analysis is carried out in Payerne (Vaud canton, Switzerland), to determine how various configurations and soil parameters affect the mentioned hydrologic quantities. On the other hand, the results obtained in Payerne, thanks to a diffused availability of ground temperature and soil moisture field data sets, allow to validate the model in ten other locations in the Swiss Plateau. A sensitivity analysis is initially performed: the results are divided based on the water infiltration model and the soil evaporation method. A simple Bucket model is found to reproduce summer and spring temperatures well and autumn and winter ones decently (underestimation of 1 °C), but only when associated to the Evaporation Resistance method. If Richards equation is implemented, temperatures are underestimated by about 1-2 °C in all seasons, whereas evapotranspiration and soil moisture are satisfactorily simulated. Ground temperature evolution with depth results largely affected by thermal conductivity and heat capacity, while surface summer temperatures appear highly sensitive to roughness length variations. The average soil grain size is the variable that influences the most evapotranspiration, runoff and soil moisture. It is noted that the introduction of canopy, together with every parameter variation which contributes to its increased effect, causes a ground temperature reduction. Evapotranspiration also responds with a general decrease, except for the case in which the Leaf Area Index is raised. SNOWPACK validation in the ten new sites is overall successful; ground temperatures are simulated similarly as in Payerne, both with Bucket and Richards models. Likewise, modelled evapotranspiration has the same bias as in Payerne, with the difference that RMSE values are much lower in Rietholzbach. Regarding soil moisture, this is well reproduced (RMSE < 0.05) in most of the sites; a satisfying precision is reached, even though accuracy is not uniform among the simulations. Moreover, it can be affirmed that no evident difference is manifested between the seven loamy soils and the other four (silt loam and sandy loam soils) when a new soil type is set, but the same soil parameters are maintained. Finally, Payerne runoff is assessed at a catchment scale using Alpine3D. The cumulative runoff is found to have the same order of magnitude of the cumulative river discharge. As expected, runoff exceeds discharge more with Bucket than with Richards (43% and 21% respectively), confirming Richards as the most accurate hydrological model.