6.3 Recap: The general modeling process

Hydrological modeling is an iterative process (see Figure 6.5. At the beginning, the purpose of the model has to be defined. The model goal determines the spatial and temporal resolution of the hydrological model. E.g. for analyzing the impact of climate change on seasonal, regional flows we are interested in the seasonal flow volumes whereas a model used for early warnings from floods requires high temporal resolution and a high model performance for threshold overflows. See for example Blöschl and Sivapalan (1995) for a thorough discussion of scales in hydrological modeling. Please note that the definition of the model goal also determines the performance criteria the model has to fulfill in order to be judged suitable for its purpose. As a next step, data about the catchment is gathered (see for example basin characterization in the Chapter Case Studies) and a first conceptual model of the dominant flow processes in the basin is drafted (e.g. low peaks and significant base flow point to large storage capacity in the basin whereas high peaks and low base flow indicate small storage capacity in the basin). The conceptual model of the river catchment is then implemented with a mathematical model (e.g. the HBV model in RS Minerve) and parameterized as well as possible (i.e. using reasonable estimates for initial parameter values). The model is then calibrated and validated (i.e. tested with a previously unseen data set) as described further below If the performance criteria for the model calibration are satisfied, a conscientious modeler performs a sensitivity analysis to gain confidence in the model. From each of these steps, the modeler can (and generally has to) go back to a previous step to gather more data, modify the conceptual model, adapt the model implementation, re-calibrate the model and/or re-evaluate the model sensitivity. Also, the modeler is aware of the inherent uncertainties of the data and the model (see for example Refsgaard et al. (2007) for a discussion on how to include uncertainties in the modeling process). The described steps produce a model (or an ensemble of models) that satisfy the performance criteria, only now the modeler can start actually using the model for the purpose it was implemented for.

Figure 6.5: The modelling process is iterative.

Modeling is an involved process requiring highly specialized knowledge and skills. A modeler has the responsibility to clearly state the underlying assumptions, uncertainties and limitations of their model, especially if it is to be used in decision making. This book chapter will guide you through the modeling process with the example of the Nauvalisoy catchment.

E4.2

What is a good model?

Discuss with your colleague(s):

Imagine you work for a government and have to judge if the hydrological model you mandated is of good quality. How would you decide whether or not to accept the model?