Abstract
Model selection is a critical aspect of designing a tool for integrated river basin management. Without scientific guidelines models may be selected in an ad‐hoc manner based on practical considerations such as data availability and computational load. A more systematic approach for model selection can be achieved by examining the interaction of coupled models in terms of sensitivity and uncertainty propagation at an early stage of the design. Such a design approach is also referred to by the term ‘appropriate modeling’. The recently completed decision‐support system (dss) for the River Elbe in Germany includes a large‐scale floodplain vegetation model and a 1D river schematization to describe the hydraulic conditions at a regional (500 km) scale. The key factors that influence the vegetation patterns are the flooding frequency in the floodplains and the land‐use practice in the floodplain area. The choice between a 1D hydrostatic and 2D hydrodynamic schematization to describe the effect of dike shifting on the floodplain vegetation is used as case example to demonstrate the principle of appropriate modeling. Whether a 1D approach will be sufficient depends on the purpose and sensitivity of the models that are coupled. The differences between the vegetation patterns obtained with the 1D and 2D hydraulic approach are compared with the effects of climate change, land use, and data uncertainty to illustrate this.