Added value of distributed information in rainfall-runoff models for the Meuse catchment

Background

Modelling in Meuse catchment

The Meuse is a dominantly rain fed river with quick hydrological responses. The Meuse basin is a densely populated area and many people use the river for different purposes. Therefore, it is important to make good predictions for the discharge in the Meuse. Although the Meuse is dominantly rain fed, snow also occurs on an irregular basis. The subsequent snow melt can have a large influence on the hydrological regime [de Wit et al., 2007]. The currently used rainfall-runoff model for the Meuse has some shortcomings. The most important ones are poor model performances during snow or rain-on-snow events. In addition, the conceptualisation of the wetting process is not right, resulting in poor  predictions for the first flood peaks.

Realism in models

One of the most suggested solutions to improve the current hydrological models is by putting more 'realism' in this models [Kirchner, 2006; Fenicia et al., 2008; Hrachowitz et al., 2013a], and thus representing the occurring runoff processes in the catchment better. However, there are a lot of different ways to possibly achieve this. Most of them are based on using more knowledge from (within) individual catchments in a smart way, which generally leads to more distribution in models. There is still a large discussion going on about the optimal level of distribution. A higher degree of realism does not necessary mean a better performance for specific criteria; however, it generally means a more comparable result for the performance in the calibration and validation period and therefore a smaller predictive uncertainty [Kirchner, 2006; Fenicia et al., 2007]. In addition, it often also means that more evaluation criteria can be met by using the same set of parameter values [Euser et al., 2013].

More realism in model structures can for example being achieved by using distributed instead of lumped forcing data [Fenicia et al., 2008]. Another way to put more realism in models is by using different model structures for different hydrological response units (HRU's). These model structures mainly function in parallel and are often connected via the groundwater reservoir [Uhlenbrook et al., 2004; Savenije, 2010].

Use of hydrological signatures

It is increasingly acknowledged that model evaluation based on single objective optimisation, often performed with standard least squares optimisation, is insufficient to appropriately identify dominant processes. The use of a multi-objective optimisation offers more insight into the processes underlying the observed catchment response. The use of specific characteristics of the hydrograph, or hydrological signatures, for the (multi-objective) evaluation of the performance of hydrological models can give even more information about the hydrological behaviour of the modelled catchments.

Main research questions

The main research question of this PhD research is:

'What can be the added value of distributed information for the improvement of the rainfall-runoff model of the Meuse?'

This research question can be divided in the following questions.

1. How can the added value of distribution or additional complexity be tested?
2. What is the added value of using different model structures within one catchment?
3. What is the added value of using distributed data?
4. How can the results from distributed model experiments be validated?
5. Which scale and transfer issues play a role in distributed models?

Study area

The main research question refers to the rainfall-runoff model of the Meuse catchment. However, the Meuse catchment is large and diverse, and therefore not really suitable to use for model experiments. The Meuse has 8 tributaries which supply each more than 5% of the total discharge in the Meuse (de Wit [2008]). Especially the runoff in the catchments originating in the Ardennes Massif have a large influence on the discharge of the Meuse at St. Pieter (de Wit et al. [2007]). Of these, the Vesdre and Amblève both have artificial reservoirs and therefore a less natural discharge regime. Therefore, the Ourthe will be used for the first modelling experiments.

References

Euser, T., Winsemius, H. C., Hrachowitz, M., Fenicia, F., Uhlenbrook, S., Savenije, H. H. G., 2013. A framework to assess the realism of model structures using hydrological signatures. Hydrology and Earth System Sciences 17 (5), 1893-1912.

Fenicia, F., Savenije, H. H. G., Pfister, P. M. L., 2008. Understanding catchment behavior through stepwise model concept improvement. Water Resources Research 44.

Fenicia, F., Savenije, H. H. G., Pfister, P. M. L., 2007. A comparison of alternative multiobjective calibration strategies for hydrological modeling. Water Resources Research 43.

Hrachowitz, M., Savenije, H., Blöschl, G., McDonnell, J., Sivapalan, M., Pomeroy, J., Arheimer, B., Blume, T., Clark, M., Ehret, U., Fenicia, F., Freer, J., Gelfan, A., Gupta, H., Hughes, D., Hut, R., Montanari, A., Pande, S., Tetzlaff, D., Troch, P., Uhlenbrook, S., Wagener, T., Winsemius, H., Woods, R., Zehe, E., Cudennec, C., 2013a. A decade of predictions in ungauged basins (pub) - a review. Hydrological Sciences Journal.

Kirchner, J. W., 2006. Getting the right answers for the right reasons: Linking measurements, analyses, and models to advance the science of hydrology. Water Resources Research 42.

Savenije, H. H. G., 2010. Topography driven conceptual modelling (Flex-topo). Hydrology and Earth System Sciences 14, 2681-2692.

Uhlenbrook, S., Roser, S., Tilch, N., 2004. Hydrological process representation at the meso-scale: the potential of a distributed, conceptual catchment model. Journal of Hydrology 291, 278-296.

de Wit, M. J. M., Peeters, H. A., Gastaud, P. H., Dewil, P., Maeghe, K., Baumgart, J., 2007. Floods in the meuse basin: Event descriptions and an international view on ongoing measures. International Journal of River Basin Management 5, 279-292.