Druckansicht der Internetadresse:

Faculty for Biology, Chemistry, and Earth Sciences

Department of Ecological Modelling - Prof. Dr. Michael Hauhs

print page
Lischeid, G; Büttcher, H; Hauck, A: Combining data-based and process-based approaches to minimize the complexity of a reactive sulfate transport model., IAHS , 277, 402-408 (2002)
Abstract In this study, advanced methods of nonlinear data analysis and mechanistic models are combined to investigate transport and turnover of airborne sulphate in the aquifer of a forested watershed. The objective is to understand why adjacent sites differ substantially with respect to short-term and long-term sulphate concentration time series. First, a groundwater model was parameterized as a basis for a coupled model. A detailed analysis revealed substantial uncertainties with respect to, e.g. residence times and the spatial pattern of groundwater discharge. Second, a time series of sulphate concentration in the catchment runoff was analysed using artificial neural networks. The chemical hydrograph provides spatially integrated information about the groundwater of the watershed. In addition, the pronounced short-term dynamics reflect the varying contribution of shallow and deep groundwater on stream discharge. The neural network revealed a substantial change in these dynamics during the last 14 years. Based on these results, only a few of a variety of candidate processes were selected for a mechanistic model of reactive sulphate transport in the aquifer. Sulphate transport and interaction with the matrix is described as an equilibrium sorption process. A very simplified model is used for water transport. Although none of the parameters was fitted by inverse modelling, the model matched the long-term dynamics of sulphate concentration. In addition, it succeeded in giving the envelope of observed short-term variability of sulphate concentration due to the varying contributions of different flow paths. It is concluded that using different process-based and data-based modelling approaches in an iterative way can considerably help the optimization of hydrologic models with respect to the available data, thus avoiding some of the problems associated with over-parameterized mechanistic models.
FacebookTwitterYoutube-KanalBlogKontakt aufnehmen
This site makes use of cookies More information