open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

Coming from a fairly extensive background in Delft3D, I was excited to try out TELEMAC as part of a new project. I'm modelling the Gulf of St. Lawrence, with a domain stretching from Ottawa to the continental shelf, and while I'm getting fantastic results in 2D, the switch to 3D is creating significant issues.

I've attached a PPT that goes over the problem in detail, along with the .cas/ grid/ bound files. In short, the energy coming in through the shallower parts of the domain isn't being propagated correctly, and is producing a water level phase offset.

Having just passed my 75th test run, I'm very open to any ideas or suggestions about what might be off with this setup, and I'm more than happy to provide any additional details that would be useful!

Hej Alexander,

Interesting question! I don't have much experience with this topic myself, but your presentation does raise some questions:

1. Do you need a 3D model if a 2D model solves it nicely?
2. How did you setup the Z layers in your 3D model? And is there thermohaline stratification?
3. What is PEI?

You mention a reduction in 'tidal energy', but shouldn't it be more correct to talk about merily the tidal phase. Is the tidal wave going too slow or too fast in the model? I think the tidal propogation is based on the depth of the tidal layer. Could it be that the setup of the Z/Sigma-layers is introducing an issue?

Only questions so far, but I hope that is also somewhat of help.

Thank you so much for the quick reply! To answer your questions:

1. Great question, and one I always try to ask (along with "do we even need a model here"). Since we're trying to simulate the transport of microplastics in this estuary, we need a 3D model in order to explicitly track the surface velocity of the water, which should provide more accurate results compared to empirical approaches layered on a depth averaged model.

2. Z layers are currently implemented using the sigma layer approach, and there's no thermocline stratification in the model.

3. Should have been more clear here, but PEI is the island at the southern edge of the estuary. In the observations, water levels at this point track those in the rest of the domain; however, in TELEMAC-3D, they're lagged due to a delay in propagating around the island.


I mentioned tidal energy, but I see you mean, and speed makes more sense. The 3D model is producing a delay in propagating the water levels over the shallower areas, which creates the incorrect signature.

Hopefully this clarifies things somewhat- I'll also take a look at a fixed z-layer approach and see if there's anything there

One other thing worth noting here is that I've already run a test using 1-layer in TELEMAC-3D, and I get the same propagation issue, which suggests that it might not be related to the layering approach

Hello Alexander,

Which release of TELEMAC do you use? Can you upload the Fortran file please?

I used to teach a training course where I show how to move from a TELEMAC-2D tidal model to a TELEMAC-3D tidal model. My feedback was not to use constant vertical turbulence model + 2 bottom smoothings and it was OK, the results for water level could be quite the same, even with only 2 planes (= 1 layer). It was for a site located in the English Channel where tide can be seen.

Anyway, can you try a few ideas please:
- using TREATMENT OF NEGATIVE DEPTHS = 2 + MASS-LUMPING FOR DEPTH = 1.
- converting your grid directly in Mercator projection or UTM and not using SPHERICAL COORDINATES : YES (in that case give a value for keyword CORIOLIS COEFFICIENT = what could be the value in the middle of the domain); anyway if you use SPHERICAL COORDINATES = YES, you should give LATITUDE OF ORIGIN POINT and LONGITUDE OF ORIGIN POINT of a point located around the middle of the extent of the area, e.g. 47 and -55,
- without using TIDE GENERATING FORCE = YES,
- using characteristics for velocities (SCHEME FOR ADVECTION OF VELOCITIES = 1 + SCHEME OPTION FOR ADVECTION OF VELOCITIES = 1).

Can you also upload one listing file to see how your computation converge.

How deep are your shallows where the results differ?

Hope this helps,

Chi-Tuan

Hi,

Thank you so much for this detailed reply- I ended up sorting this one out! Like all difficult numerical problems, this one was a combination of two factors.

It seems that something is wrong with spherical coordinates. When I converted the grid to a UTM projection, I got a different set of results, and then when I added in coriolis, the results converged with 2d!

So now this mystery is why the same grid would work in spherical in 2d, Mercator in 3d, but not spherical in 3d. I'm going to take a look at the source for the projection step to see if there are any differences between 2d and 3d, but this seems to be some sort of bug

Hello Alexander,

It seems that you have pointed out a bug for SPHERICAL COORDINATES = YES when used with TELEMAC-3D.

I have tried to have a look but nothing very obvious at the moment (my current suspicion is in the $HOMETEL/sources/utils/bief/inbief.f subroutine).
There is not a lot of feedback for this feature in 3D.

I will try to investigate when I have time.

Chi-Tuan

Hello,

Coming from a fairly extensive numerical modelling background in Delft3D, I was excited to try TELEMAC as part of a new project. While the initial setup and runs were great, I've run into an issue trying to model the tidal propagation around the Gulf of St. Lawrence in 3D.

The domain for this model runs from two upstream rivers through a large gulf to off the continental shelf, and what seems to be happening is that the energy isn't propagating through shallower parts of the domain, resulting in a water level signal that's offset by ~90 degrees in some places in the domain. What's unusual is that this grid worked in TELEMAC-2D, and the issue has remained remarkably consistent over the course of some 75 testing simulations.

I've attached an overview PPT, example .cas file, and grid/ bound. If anyone has any ideas or suggestions, I would greatly appreciate it!