open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi,

I am using telemac2d for rainfall-runoff modelling. The hydrograph I am getting as the result has very sharp peaks, so I tried to use various roughness coefficients. The expected hydrograph should have the same amount of water in the system but with flatter peaks, but increasing the roughness is decreasing the volume of water in the system which is very strange.

Can someone please help me to understand this strange behavior? and how to make the resulting hydrograph peaks flatter and a bit delayed in time instead of sharp peaks?

Best Regards!

Hi,

Could you please share your steering file? It is unclear how you set your friction law and coefficients based on the screenshot you shared.

Which rainfall-runoff model are you using for this? If you are using the Curve Number model (RAINFALL-RUNOFF MODEL = 1) keep in mind that this model is only valid for single storm events as certain hydrological processes are not taken into account (groundwater recharge and recession for example, which might have a non negligible impact on the flow regime of your example by looking at the shape of the observed data). I think it can be hard to obtain good comparison without more advanced hydrological model.

Finally, another important aspect is the rainfall definition vs the size of your study area. Is your precipitation data spatially distributed or not, etc...?

Kind regards
PL

Hi,

Attached is the steering file but the roughness values are changed in this file for the various simulations as shown in the screenshot shared in the first post.
The rainfall-runoff model used is the Curve Number model (as this is the only available RR model in t2d). I understand that groundwater recharge and recession are not taken into account. Also no discharge is added in the river as the initial condition (struggling to add that) as there is always some water available in the river. If I am able to add that, it may improve the results a bit. But I still don't understand the strange effect of roughness.
The precipitation data used is spatially distributed over the catchment area of 10.5 sqkm.

Hi,

Stricker coefficients lower than 5 are extremely low, I don't think it's even physical. Anyhow, the friction generated for a coefficient of 1 is extremely large and might block the water to flow in some areas, which might explain the trend you see in your results.

It would be nice to have more information on the geometry and mesh.

Kind regards
PL

Hi,

Attached is another image showing the effect of manning's roughness values (not extreme values as 1). It is a steep catchment, so all the water should come out instead of just getting lost somewhere.
The geometry file of the catchment and mesh is attached too.

I have another question. How can I add a constant discharge (lets say 1.5 m3/s) in the river because there is always some water in the river? it's never zero as it is in the simulation results.

Best Regards!

Hi,

I don't think your mesh is adapted for 2D runoff modelling. Your default mesh size is 500 m in the highest parts of the catchment. It seems like you want to perform something hyrid between pure hydrological modelling (on side slopes, with large cell sizes) and hydrodynamic modelling in the valley. I would not trust the results in the areas where your mesh size is 500 m.

I suspect that your results are more dependent on your mesh definition than anything else, so maybe try to refine your mesh. It looks like 95% of your cells are used to describe the valley, you should be able to refine the coarse areas by a large amount while still keeping a reasonable number of cells overall.

A simple way to introduce water discharge in such a model is to use a point source.

Kind regards
PL

Hi,

Yes, you understood correctly. The hydraulic modelling in the rest of catchment except the river is not very important for my case. Another reason for using such a coarse mesh size in the rest of the catchment to decrease the computation time.

If I make the mesh finer, is it possible to introduce some permanent water flow in the river? If yes, It will be the best.
If not, how to introduce it using point source?

Regards,
Nitesh

Hi,

I don't think you should do it this way. When performing 2D runoff modelling, one should make sure that the model geometry is compatible with the numerical problem to be solved, in this case overland flow using shallow water equations. Trying to emulate a hydrological model using large mesh size and extremely low Strickler coefficients is outside the reasonable range of application of a 2D hydraulic model for production work I would say.

When performing runoff modelling, the praxis is to use a rather high spatial definition (small cell size) in order to resolve the flow paths correctly. It is very common to work with meshes having more than one million elements.

This will make you model very slow, yes, but once again, I don't think it is reasonable to perform runoff modelling over a period covering more than one storm event being given the limitations of the CN model already mentionned.

Introducing water flow at point sources is very simple, please check the manual and the relevant examples provided with the installation.

Kind regards
PL

Thank you so much for the advice. I will make the mesh a bit finer.
Sorry for asking again but is it not possible to introduce water flow in the river while using spatially distributed rainfall even if I use a finer mesh?
Because I want to use spatially distributed rainfall as it is giving better results than the point rainfall.

Regards,
Nitesh

If you would like to have water flowing in the river prior to the rainfall event you could run a so called hotstart simulation in which you let river flow establish and reach steady state. If your river discharge is increasing in downstream direction due to tributuaries you could define several source points along the river. Once you obtained the desired steady state, you can use this result as a PREVIOUS COMPUTATION FILE and continue with your runoff simulation.

You can also use classical inflow BC but those might be tricky as you seem to have very low discharges (small water depth which can crash if some nodes get dry along the boundary). Hence, I find point sources a handy, albeit simplified, way to define inflow within such a model.

I am not sure to understand what this has to do with the type of rainfall (spatially distributed or not), so I can't really comment your last sentence.

Kind regards