The Problem

On this page we'll decribe the problems that form the basis for our research project. The main idea is that 0D Rainfall-Runoff models are outdated. An example of a typical lumped Rainfall-Runoff model, (coupled to a 1D hydrodynamics schematization) can be seen below. Notice that the RR modeling objects (green and red squares) are only connected to the 1D schematization of the canals and have no mutual connections.


Below we'll point out the downsides of this modeling concept.

1. Unncessary steps in the working methodology

During the last few decennia, an enormous amount of spatially distributed data has become available in GIS-formats. Considering this fact, the standard working methodology for building Lumped Rainfall-Runoff models becomes increasingly redundant. We usually build and work with RR models in the following way:

  1. Define a spatial discretization for the RR-model
  2. Aggregate spatial data such as levels, soil type and landuse into parameter values for all 0D-model objects
  3. Calibrate the model
  4. Run simulations
  5. Plot the results in 2D on a map with levels

When we look at this process closer, it actually seems a bit silly: first we translate 2D data into 0D parameter values, then we calibrate these rather obscure parameters and finally we'll have to plot the simulation results in 2D again! However, since computational speed has increased considerably the last few decades, there is no real reason any longer to keep building model schematizations in 0D. In theory it should be much easier to build, calibrate & use 2D Rainfall Runoff schematizations.

2. Hard to calibrate

Because the parameters of a 0D model are always somewhat obscure (it's hard to really grasp what they stand for) it has always been hard to calibrate model schematizations of this type. Besides, it is rather unusual that lessons learned in one region pay off in another region. We therefore hope that models of the Distributed type perform better at this point.

3. No Groundwater Interaction

Another downside to "lumped" modeling is that the 0D-model objects do not have mutual interaction. Although a phreatic groundwater table is being computed on each 0D-object, the modeling objects do not exchange water. The only connection there is, is towards the river it discharges into. As a consequence, using RR-models of the "lumped" type cannot be used to compute the effects of e.g. Natural River Restoration on the phreatic groundwater table in the environment.