Applicant
Dr. Xinyang Fan
Department of Geography and Geosciences
GeoZentrum Nordbayern
Friedrich-Alexander-Universität Erlangen-Nürnberg
Project Overview
The grid-based Water Flow and Balance Simulation Model [1] (WaSiM) is a well-established tool for investigating the spatial and temporal variability of hydrological processes in complex river basins, such as high alpine catchments. The hydrological model is a conceptual representation of our understanding of the water flow processes and storage in the study area, and it is governed by physical laws. The parameters of physically based models such as WaSiM are generally set according to the expert judgement. They are most commonly tuned to obtain the model simulations that match the measured discharge at the closure of the catchment area. This approach, however, can be biased and introduce uncertainties when applied to climate change studies. Introducing optimization algorithms for an unbiased parameter estimation is hence of fundamental importance.
In the KONWIHR project, we aim to develop an auto-calibration framework for the physics-based fully-distributed hydrological model WaSiM on the HPC at FAU. The workflow composes (i) coupling the WaSiM model with an auto-calibration algorithm, and (ii) optimize the automatic calibration process on the HPC system. The key task is to find an effective strategy that allows the exchange of information between the auto-calibration algorithm and WaSiM. We aim to develop a workflow to allow the parallel runs of thousands of model simulations required in the auto-calibration. Each simulation is generated with different parameter sets within the user-predefined range. By adopting a suitable auto-calibration algorithm, this workflow allows for identifying the optimal parameter set(s) that contribute to plausible hydrological forecasts.
References
[1] Schulla, J., Model description WaSiM, 2024.
http://www.wasim.ch/en/products/wasim_description.htm