Depth-Varying Roughness

The selection of Manning’s n roughness values are much discussed within the modelling community. These values essentially replicate the ‘friction’ caused by ground surface cover that can slow the flow of water to a greater or lesser degree. Concrete, for example, is relatively smooth and would have a lower Manning’s n roughness coefficient ; a heavily wooded area would conversely have a higher value applied. 

Whilst there are some broadly accepted figures for certain ground covers, it is always good practice to sensitivity test these values as part of the model build. This allows the modeller to assess whether the selection of this parameter has a significant impact on model results. If the results of the study are largely unaffected by the sensitivity test, this increases our confidence that the outcomes are valid since a significant source of model uncertainty is shown not to matter.

Where the depth of water is shallow, the nature of the ground surface is much more influential; this is where depth-varying roughness comes into play. Imagine a field of short grass covered by flood water to a depth of 1m; that grass has minimal impact on the flow of water. If the floodwater is very shallow – say, 2-3 cm – then the same short grass will exert much more influence on that water. We consider that that influence should be included in our modelled representation of a catchment and allows us to reach a more robust answer.

In TUFLOW, Depth-varying roughness can be applied using the additional columns in the TUFLOW Material File, or TMF. This allows the modeller to apply a higher value at shallow depths and a lower value when the water exceeds a certain depth. We’ve sometimes been asked what reasonable values might be and this is a great question. There isn’t a straightforward answer to this and data is limited. CIRIA guidance (Coppin, N.J. et al. (2007)) suggests that for grass, a factor of 10 is appropriate, and this is often our starting point. As ever, if verification data is available then model parameters can be refined.

You can view the roughness being applied during a model simulation via TUFLOW’s ‘n’ map output type, an example of which is shown in Figure 1 below. Its is also possible to plot variation in roughness at a specific location; a comparison of a simulation with and without depth-varying roughness is shown in Figure 2. 

Figure 1: Example of spatial variation in Manning’s n during a model simulation. Different colours indicate different roughness values.

Figure 2: Plot showing change in Manning’ n during a simulation at a specifici location. The blue line applies depth-variable roughness, whereas the yellow line shows a simulation where the n value is constant throughout the simulation.