open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi Chi-Tuan,

I'm not sure if I understand you correctly.
I ran a simulation as you mentioned, with no prescribed elevation nor a LIQUID BOUNDARIES FILE and set the OPTION FOR LIQUID BOUNDARIES = 1. I checked and received the water depth =0 at the open boundary.

When I understand correctly, the domain is emptied completely in this set-up, which is the reason why there is no water depth, the free surface equals the bathymetry respectively.

Regarding your questions:
- In that area is almost no tidal influence. We observe tidal ranges of 5-10cm. The movement of surface elevation is mainly wind-driven.
-Bottom elevation is referred to as NHN which is a synonym for "standard elevation zero" which is equal to MSL in Germany.
-Unfortunately, I don't have atmospheric pressure measurements.

Time step seems ok as the simulation forced by the hydrograph at Kiel-LH is fine.
If also checked the CORIOLIS COEFFICIENT which was too small indeed, I had a mistake in calculating it.

However, I now have tried a run with a PRESCRIBED ELEVATION = 0.0 but received again only slight variations in surface elevation. This seems reasonable as surface elevation should be calibrated somewhat to 0.0m.
I'll now try to set up TPXO as an initial boundary condition as I imagine this might represent the changes in surface elevation excluding the wind forcing in real nature (which is included in the hydrograph at Kiel-LH). Do you have any other ideas?

Cheers Tim

Hello Tim,

Can you confirm that the problem of emptying of the domain has disappeared now and that what remains as issue is low variations of free surface elevation due to wind when using COEFFICIENT OF WIND INFLUENCE VARYING WITH WIND SPEED = YES but you get rather bigger variations when using COEFFICIENT OF WIND INFLUENCE VARYING WITH WIND SPEED = NO with COEFFICIENT OF WIND INFLUENCE VARYING WITH WIND SPEED = 1.2615E-3?

My first comment for this last computation is that this coefficient is too big, it should be around 1,000 times less (see the default value = 1.55E-6 and the explanation in the TELEMAC-2D user manual (it is not the value of a_wind but a_wind*rho_air/rho_water). That should be a reason why you got results so different. Can you tell me if when you change this value, you get more similar results between COEFFICIENT OF WIND INFLUENCE VARYING WITH WIND SPEED = YES or NO?

Thanks,

Chi-Tuan

Hello Chi-Tuan,

I'm stuck at the moment. I'm trying all sorts of setups but don't get it somewhat close to the measurements. I've attached a graph of having a prescribed elevation of 0.0m. The model reacts reasonably in my understanding, as it approaches the 0.0m after some time. But I can't see any wind effect. In the case of the graph, I adapted already your mentioned coefficient and had COEFFICIENT OF WIND INFLUENCE VARYING WITH WIND SPEED = NO

When I tried to use the TPXO the code handled the keyword inexplicable. When I wrote the:
BINARY DATABASE 1 FOR TIDE ='/Users/Shared/TELEMAC/TOPEX/DATA/hf.AO_2008.out'
BINARY DATABASE 2 FOR TIDE ='/Users/Shared/TELEMAC/TOPEX/DATA/uv.AO_2008.out'
in one line, it wasn't able to read it and the calculation stoped by an error. However, when I wrote it in two lines:
/BINARY DATABASE 1 FOR TIDE
='/Users/Shared/TELEMAC/TOPEX/DATA/hf.AO_2008.out'
/BINARY DATABASE 2 FOR TIDE
='/Users/Shared/TELEMAC/TOPEX/DATA/uv.AO_2008.out'
It seemed to read the tidal data.

In terms of the surface elevation, there was, however, not much difference as the surface elevation showed rather low deviations from 0.0m.

Regarding your first paragraph, I reckon you mean COEFFICIENT OF WIND INFLUENCE =1.2615E-3? As far as I'm aware COEFFICIENT OF WIND INFLUENCE VARYING WITH WIND SPEED has only the values 'YES' and 'NO', I haven't read about an option of numerical inputs for this keyword.

Regarding your second paragraph, you write my coefficient is too big. I got the value 1.2615E-3 from the examples. I did not understand the manual and wondering what a_wind seems to be? There is no sound explanation in the manual. The only idea I got from the manual is, that a_wind isn't U and also not considered as drag force. I've read about the factor 1000, though. Following the reference manual, the default is '0'; I can't comprehend the mentioned default value of 1.55E-6.

If I understand you correctly, I should try to run the model with the new value and test it with the COEFFICIENT OF WIND INFLUENCE VARYING WITH WIND SPEED = YES or NO. Should I use the TPXO data as well, or would you recommend a prescribed elevation of 0.0?

Regards, Tim

Hello Tim,

There is a limit of number of characters per line for steering files (72) as written in the user manual. If too long, you can write in 2 lines as you have done (the only condition is not to split the keyword name in 2 lines). You can also put the hf and uv AL 2008 out files not far from the directory where you run your simulation and use relative path (e.g.: ../../name_directory or ./name_directory).

I mistook when I wrote the end of the 1st paragraph of my previous answer, I should have written COEFFICIENT OF WIND INFLUENCE = 1.2615E-3? you are right (written too fast without double check).

I wonder if for the TELEMAC-2D wind example, the use of the value 1.2615E-3 was to amplify the wind effect.
a_wind is the coefficient that is multiplied by the magnitude of velocity times one component of velocity divided by water depth. The product is the source term of the dynamic equation. For TELEMAC-2D equations, it is then necessary to give this coefficient times air density divided by water density to fill in the keyword COEFFICIENT OF WIND INFLUENCE. This is how this coefficient is used when implemented in TELEMAC-2D. You can have a look at the PROSOU subroutine in $HOMETEL/sources/telemac2d folder if you want to exactly know.
Sorry, I gave you the new default value since release 8.2 (the latest release). I changed it to give users a good order of magnitude of what value to give for COEFFICIENT OF WIND INFLUENCE. If you use a release before 8.2, the default value is still 0.

If you use the regional solution Atlantic Ocean (or global solution TPXO), be sure that your open boundary is far enough so that tidal constituents can be interpolated with nodes of the AO regional solution. You should not have any writing like:

WARNING!!! THE POINT WITH X/Y IN THE MESH =
CONVERTED TO LONGITUDE/LATITUDE =
SEEMS TO BE OUTSIDE OF THE EXTENT OF THE TIDAL SOLUTION'
WHOSE MIN/MAX FOR LONGITUDE ARE:

at the 1st time step of your simulation.

If the open boundary is far enough from coast with respect of the Atlantic Ocean tidal solution extent, I would use it rather than a simple prescribing of water elevation but if the wind effect is the main physical driving force for free surface, the choice may not be so important.

I have not looked at your mesh yet but it can be improved I think:
- you have overconstrained triangles at some locations along solid boundary (the 3 nodes are located on this solid boundary, at least one of them should be liquid, inside the fluid domain,
- the transition between the area where you have fine elements and the one where elements begin to be bigger should be more progressive,
- you have some triangles with obtus angles.

I suggest you change your mesh a little bit to improve the results.

Hope this helps,

Chi-Tuan

Hello Chi-Tuan,

thanks again for your feedback and recommendations.
I've now tried several set-ups using the AO tidal data and played around with different values for COEFFICIENT OF WIND INFLUENCE.
The tidal data acquisition seems ok, at least I'm not experiencing the warning you mentioned.

Being clear now in terms of boundary conditions and using the tidal data, using the keyword COEFFICIENT OF WIND INFLUENCE VARYING WITH WIND SPEED seems to bring no reasonable improvements in terms of the surface elevation curve. The elevation oscillates around 0.0m.

As you mentioned, I'm currently rethinking the model domain. The model was able to perform quite well for the case I used some fairly local hydrograph to force it. Even though the mesh could be better, or at least has the potential to be improved, I'm now thinking I might miss some broader/regional effects. There could be some reflection from opposite coasts in the Kiel Bight, which the model is missing at the moment.

I'll have a look and will keep you informed.

Cheers, Tim

Hi Chi-Tuan,

I've now run some simulations. The broader domain didn't change the outcome though. Therefore, I think the wider regional effects don't really affect the local hydrodynamics within the fjord; at least not for the wind direction in my test case.

I've now updated to v8p2 as I saw the format of the ASCII ATMOSPHERIC DATA FILE has changed in a way, that it's less prone to user failures. I've now also changed the wind data from the station of the airport, close to the inner fjord, to data from the lighthouse where I received the hydrograph data from; some distance offshore. I also found pressure data from that location, which I adapted into the new ATMOSPHERIC DATA FILE.
However, also including pressure data seems to have no major impact.

From my understanding, the biggest influence is due to the COEFFICIENT OF WIND INFLUENCE. I will now try to calibrate the model by putting some further effort into different coefficients and see how it goes.
Any further recommendations are much appreciated:)

Cheers, Tim

Hello Chi-Tuan,

I've experienced some major improvements today. I did change my test case to a more recent one and prepared time series for Jan. 2021.

I kept my model domain and did not improve it, as I first wanted to get the general trend at least in the right direction. As you can see in the attachment, the model matches the general trend.
The hydrograph Time-Series is red. Green and yellow are modeled. The yellow line is a reduced roughness simulation (Manning 0.02).

However, the model seems to be improvable. It looks to me that around the late 10'th/11'th of Jan. when there is this a little peak, the model "overshoots" the small high water peak. After that point, the model seems phase-shifted, as the two major peaks on the ≈ 13'th and 14'th are a bit earlier than measured. Also, the magnitude doesn't exactly fit the measurements.

I've further set up the model without TPXO but an initial constant elevation which seems to reduce the phase shift.
Do you have any ideas on how to get rid of the shift?

Cheers, Tim