open TELEMAC-MASCARET The mathematically superior suite of solvers

Hello,

I'm doing tide modeling within an estuary(PIC: domain) using Telemac2D recenly.
But after several study cases, there is always a problem on the modeled tidal curve: outside the mouth of the estuary(PART B ), modeled tidal curve is good(tide_01) but inside(PART A), modeled tidal curve is a little bit strange(tide_02).

I believe the result is wrong, but I really don't know why this happens,
is it that:
(1) my numerical paramters in 'CAS' file are improper ?
(2) does it have anything to do with the local bathymetry ?

Thank you for your help.

with best regards.

Hello robinson2012,

What is the bottom elevation of the point where you are looking at the free surface elevation time evolution in Figure tide01.PNG? If it is around 0.5 m, your result is normal, as the water depth cannot be lower that 0 m (or a little bit depending on which option you choose when you have tidal flats). If it is less than 0.5 m, do you have any tidal flats near this point in the estuary?

I suppose that inside the estuary, there may be tidal flats, may not there? If yes, in addition to default options (search for TIDAL FLATS in the dictionnary), you can use the option TREATMENT OF NEGATIVE DEPTHS = 2 (+ SUPG = 0), then you should not have negative depths on tidal flats, it is more physical.

Anyway, I have quickly looked at your steering file, and I have a few remarks to give you (that should not change the physics of your results). It is not improper, but can be improved a little bit.
In the validation directory, you can look at 123_tide test case which is a coastal study where tides are modelled. Used options are advised for such kind of studies. Moreover, you can read the topic www.opentelemac.org/index.php?option=com...7&limitstart=10#6080 where some advice is given.
In your steering file, if you use the option TREATMENT OF LINEAR SYSTEM = 2 (wave equation), you should use SOLVER = 1 (conjugate gradient), it is more efficient (then, you can decrease SOLVER ACCURACY). If you had spurious oscillations for free surface or velocities, you could tune FREE SURFACE GRADIENT COMPATIBILITY (< 1.).

You seem to have written free surface elevation in the FORMATTED DATA FILE 1 (with the code 5 4 4 in the boundary conditions file I suppose), do you? If you also have velocities, you could use Thompson's conditions (OPTION FOR LIQUID BOUNDARIES = 2;1 if the 1st open boundary is marine) like other TELEMAC users (please see other posts), it is more physical. Moreover, if your domain is located inside the area of a local solution of tide from Oregon State University (TPXO-like), from version 6.2, you can use developments of TELEMAC. If you do not have points in your domain interpolated in tidal flats in the solution from OSU, it should work correctly.

Hope this helps,

Chi-Tuan

Hello, Chi-Tuan

Thank you very much indeed for your very detailed answer.

The bathymetry of my model is refered to mean sea level and those tidal curves are elevations measured from MSL, the still water depth of which is more than -2 m. So I think in tide02.PNG with elevations above 0.5 m, it's not a problem of tidal flats. But there are tidal flats inside the estuary indeed as you said, and I will try again with your very helpful suggestions.

About 'FORMATTED DATA FILE 1', I use this file to provide tidal constituents(without velocities)for each node at open boundaries(boundary type:5 4 4) then use 'BORD.f' to calculate time series elevations. All tidal constituents are extracted from TPXO manually, but I don't know if it's reasonable compared with that developed in TELEMAC v6.2(frankly I'm not familiar with this new feature but I will try to learn) ...

With best regards.

Hello robinson2012,

I guessed that your bathymetry was refered to MSL by looking at tide01.PNG (sorry, I switched Figure 1 and Figure 2 in my previous post). Anyway, can you send a map of the bathymetry showing in particular the estuary (with e.g. one different colour every 0.5 m or 0.25 m) + the location of the point where you look at FS time series + scale for lengthes, please?
If there are tidal flats in the estuary and that the point where you look at FS time series is not so far from them (or worse, enclosed by tidal flats), time series can be different from classical shapes like in Figure tide01.PNG.

If you only use 5 4 4 codes in the boundary conditions file, everything has already been implemented to use tidal solutions from OSU. In a next future, it would be interesting to compare your method to use harmonic constants and the one available since version 6.2 of TELEMAC (and that mostly comes from OTPS). Maybe some parts of your Fortran file are the same as the one in TELEMAC, just have a look at tpxo module + tidal_model_t2d subroutine in telemac2d/tel2d_v6p2/sources directory. FYI, some original formulae have been changed to correct a few bugs (marked in comments)

Best regards,

Chi-Tuan

Hello, Chi-Tuan

Thanks a lot, it's OK .
Attached please find 'domain.PNG' with an area of 180 km * 165 km, which is only one part of my model domain(located at eastern China)near the coast line. The grid size ranges from 300 m to 2000 m, as can be seen in 'nodes.PNG'.

I have had another case study with some modifications according to your suggestions as follow:
TREATMENT OF NEGATIVE DEPTHS = 2
SUPG = 0;0
SOLVER = 1
FREE SURFACE GRADIENT COMPATIBILITY = 0.9
Model efficiency has been improved a little but I think they do not bring any improvement to the model quality, the problem is still the same as before.

In fact, I find that the shape of tidal curves have already changed at the estuary mouth(2 dash lines),with low tidal elevations a little bit distorted, then distored more and more obviously as going upstream. Then I tried an idealized study case with a regular slope for the whole domain, the result proves to be quite regular generally. So I guess it's a problem of local bathymetry but the real bathymetry is complex indeed which can't be changed arbitrarily, I think .

Could you please give me some suggestions on this problem ...
Thanks again.

With best regards.

p.s. It's interesting to compare the results using both method to assign open boundary conditions as you suggested and I will try.

Hello robinson (new name?),

I do not have so much experience with modelling tide inside estuaries (maybe other users can give their opinion?), but personnally, I think that due to shallows and resonance inside estuaries, the shape of tide can be changed a lot compared to coastal and deeper locations.
In another coastal model of mine, when having big tidal ranges, low waters were not well reproduced but high waters did.
I suppose that you want to correctly reproduce low waters to compare the model with measurements, do you?

I have an numerical trick to suggest you if the bathymetry is chaotic in your area: increase the number of BOTTOM SMOOTHING (keyword), but it will smooth bathymetry all over the domain. Perhaps, you can try to look at CORFON subroutine and modify it to smooth only one part of the domain (in the estuary e.g.), by using e.g. inpoly subroutine.

If any other users have better ideas, do not hesitate to tell!

Hope this helps,

Chi-Tuan

Hello, Chi-Tuan

Thanks a lot. ‘Robinson’ is more formal I think

I agree with you because I cannot find any problem in numerical parameters. And you say it quite right, my problem is about low waters which should be validated against measurements.
Since my bathymetric data are digitized from the chart which is very detailed, maybe a relatively coarse bathymetry is helpful. I have tried BOTTOM SMOOTHING, but it does not bring any improvement. About CORFON, I think it's tough because I don't know to what extent the bathymetry is to be modified in advance...

With best regards

Hello, Chi-Tuan

After a number of numerical test cases, it seems that my problem in this tide modeling case lies in the value of friction coefficient. I know that it's quite convenient to define spatially varying friction coefficients in Telemac system. But I'm wondering if it is possible to define temporally varying friction coefficient(flood/ebb tide)? if so, how to ...

Thanks for any help.:)

With best regards.

Hello,

Yes sometimes the friction changes with time (estuaries). There is also a procedure to estimate the friction when coupling with Sisyphe, it then varies with time.

If you want to program a friction varying in time, you can also do it in subroutine corstr.f (library telemac2d). It is called at every time step.

Regards,

Jean-Michel Hervouet