This module will develop key innovations along 3 complementary lines. New approaches aiming at the integration of spatially distributed data for calibration and uncertainty analysis of physically-based models will be developed. Based on these approaches, a fully coupled, physically based for the Nant-Anzeindaz catchment will be elaborated. This model will then provide the basis to assess changes in the hydrological response to changes of vegetation structure and climate,
to assess change in the ‘water provision’ ecosystem service in module 6.
Task 1 – Data gathering, knowledge update & synthesis – Data for model setup, calibration and climatic forcing will be compiled for the modelled catchment. Basic topographical data, as well as the different types of land use, vegetation cover and soil types made available through the GeoDataHub will be used to identify proxies and spatial trends for the physical parameterization of the model (e.g. hydraulic conductivity, soil moisture retention, interception, root depth). In terms of observation data, pointwise hydrogeological and hydrological data such as river discharge and hydraulic heads will be gathered using the available national and cantonal observation stations. The spatial distribution of evapotranspiration will be calculated with the SEBS algorithm using albedo, NDVI and surface temperatures. Data on soil moisture, permafrost, snow cover will likewise be made available through the GeoDataHub as well as the spatial distribution of climatic forcing (i.e. rainfall, potential evapotranspiration).
Task 2 – Non-stationary model calibration of hydrological models using spatially-distributed observation data – Classical calibration approaches are based on the assumption of stationary conditions even though all catchments comprise spatial geological and ecological trends. Here, we will address the problem of inverse groundwater model calibration under non-stationary conditions, which is critical in alpine environments that present strong elevation gradients. The key innovation of this module is to use spatially distributed data in both the formulation of the inverse
problem as well as in the calibration data set. This will be made possible by integrating multiple predictors to help determining trends and covariates. These predictors can include topography gradients, granulometry trends, soil type distributions, repartition of the vegetation types and the occurrence of permafrost. In practice, the consideration of nonstationary conditions for calibration will be implemented by accounting for multiple covariates when kriging between
pilot points. As demonstrated by Brunner et al. (2012) the data worth of remotely sensed data such as soil moisture and evapotranspiration can be considerable, even if their absolute values are uncertain. These calibration approaches will be tested using synthetic models constructed with HydroGeoSphere (www.aquanty.com) and the PEST code (www.pesthomepage.org). These synthetic reference models will be inspired by the variations and covariations of physical properties present in the Nant–Anzeindaz catchment. The reference models will be used to generate observation data that is subsequently employed to estimate the calibration parameters and their spatial distribution. This synthetic approach will allow to systematically identify the information content (“data worth”) of different observation types under different geological, ecological and hydrological conditions. These analyses provide the basis for the calibration and
uncertainty analysis of real-world simulation approaches under non-stationary conditions and will allow identifying the most informative data sets to be included as observation in the calibration process.
Task 3 – Model application: feedback mechanisms between hydrological processes, vegetation and soils – The numerical model HydroGeoSphere is capable of simulating all hydrologic processes relevant to this project, including snow dynamics, the influence of permafrost on the infiltration capacity of the soil, surface and subsurface flow as well as transpiration processes. The model will cover the Nant-Anzeindaz catchment and will be constructed using spatially distributed data such as soil types and geological structures of the simulated catchments. The model will be calibrated on the basis of all available observations, using the calibration approaches developed in. The calibrated model will provide and unprecedented level of detail in terms of the processes simulated and of integration of spatial data. Based on the calibrated model, different climatic and ecological scenarios under consideration of the future soil maps will be assessed. For example, the changes of the catchments hydrographs to a rise of the treeline or changes in the spatial patterns and dynamics snow cover can be simulated. Given the holistic nature of the model, numerous hydrological system services can be explored under different climatic and ecological forcing. The focus of this analysis will be on the quantity and temporal dynamics of river discharge. These aspects are at the core of the management of electrical hydropower