EGU2020-5478, updated on 12 Jun 2020
https://doi.org/10.5194/egusphere-egu2020-5478
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Experimenting on simple and flexible top-down approaches for hydrological modelling

Sotirios Moustakas1 and Patrick Willems1,2
Sotirios Moustakas and Patrick Willems
  • 1KU Leuven, Hydraulics division, Department of Civil Engineering, Kasteelpark Arenberg 40, BE-3001 Leuven, Belgium
  • 2Vrije Universiteit Brussel (VUB), Department of Hydrology and Hydraulic Engineering, Boulevard de la Plaine 2, 1050 Brussels, Belgium

Nowadays, a plethora of modelling software on rainfall-runoff and groundwater dynamics are available. Considering the complexity and heterogeneity of natural processes governing the water cycle, many of those models involve physically-based formulations. Inevitably, a large amount of data is also required. However, the available data are often insufficient, while their quality questionable. At the same time, an increasing model complexity also gives rise to high computational requirements. In order to mitigate some of the aforementioned issues, during the past years a simple and flexible top-down approach for distributed rainfall-runoff modelling has been developed (Tran et al., 2018). Essentially, the distributed rainfall-runoff model is built starting from a simple lumped model, whose parameters are then spatially disaggregated. Disaggregation is carried out using conceptual links between model parameters and natural catchment characteristics.

We now test an extended version of this methodology involving disaggregation relationships for more model parameters. Moreover, we evaluate modelling performance for 2 different configurations. The first starts from the parameters of a lumped conceptual model and is essentially the original approach. The second one starts from the parameters of a uniform distributed conceptual model. The motivation behind the new approach is that it allows a better-integrated routing scheme with less model parameters. In turn, this can further reduce equifinality (denoting the “phenomenon” that largely different parameter-sets can often result to largely similar model outcomes). The two approaches are inter-compared and evaluated against flow observations.

With the disaggregated models as basis, we also experiment on the potential of simple methods for modelling groundwater levels. We approach this challenge by trying to identify links between a) the variations and b) the reference levels of the modelled groundwater storages and observed groundwater levels. For example, we hypothesize that modelled storages can be scaled to the actual level variations via the specific yield, which expresses the amount of interconnected pores in the soil. The modelling methodology is evaluated against groundwater level measurements.

 

Tran, Q.Q., De Niel, J., Willems, P., 2018. Spatially Distributed Conceptual Hydrological Model Building: A Generic Top-Down Approach Starting From Lumped Models. Water Resour. Res. 54, 8064–8085.

How to cite: Moustakas, S. and Willems, P.: Experimenting on simple and flexible top-down approaches for hydrological modelling, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-5478, https://doi.org/10.5194/egusphere-egu2020-5478, 2020

Display materials

Display file

Comments on the display material

AC: Author Comment | CC: Community Comment | Report abuse

Display material version 2 – uploaded on 22 Apr 2020
Version description: Enhanced boxplots for presenting results and replacement of logarithmic Kling-Gutpa Efficiency metric with[...]
  • CC1: Parameter disaggregation, Ioannis Sofokleous, 04 May 2020

    Hello,

    For slide 5 of your presentation could you please give some examples with pairs of model parameters (C) and physical properties (ρ) that can be used in the equation? I mean which physical property would give a parameter?

    Thank you,

    Yiannis

     

    • CC2: Reply to CC1, Sotirios Moustakas, 04 May 2020

      Hello,

      For example, you could link the overland flow time constant (CKOF, in NAM model) with the time of concentration. In turn, the time of concentration might be expressed using Manning coefficient and slope via the kinematic wave equation. The baseflow time constant (CKBF) could be linked with the reciprocal of hydraulic conductivity.

      This kind of relationships were used for all NAM model parameters. In fact, some of those relationships are already published. I provide the reference in the presentation. 

      Regards,
      Sotiris

      • CC3: Reply to CC2, Ioannis Sofokleous, 04 May 2020

        Thank you very much Sotiri!

Display material version 1 – uploaded on 21 Apr 2020, no comments