open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi everyone,

I'm still doing my calibration part. and the results as shown in the graph for the water level (shape of the WL look fine and almost the same with the measured data). but it's shifted and I have tried with different friction coefficient and redo geo file with new bathymetry and the results still the same. I'm not sure why the result for 1st spring there are some spike results for the model. then for the next spring, the results seem ok.

could someone explain about this? maybe I'm missing something in my coding in cas file. also, attach my cas file for this results.

should I put the wind data?

regards,

shahirah

Hi Shahirah,

For the shift between your data and the model results, did you first check the time reference (UT, local) of the 2 datasets? It's strange that the model results are in advance until about the 20th august and late after this date... what happens around this date?

Your solver accuracy (1.e-12) is too low. Default value (1e-4) should be enough or you can decrease it until 1e-6 or 1e-8 potentially. Be aware that the ZERO of Telemac2D is 1e-10 by default, so you're asking for a precision exceeding the ZERO of the model, this precision should not be reachable... Still related to this point, the maximum iterations for the solver is too high (100000). Reduce it to about 100-500 and reduce your time step if necessary so as to still have a Courant number around 1.

Your modelled water level signals seems to be noisy (if you zoom in for a one day plot for example). If this noise is still present with the previous modifications, try to also subplot the whole water depth (not the free surface) and the associated current speed for the same point. 2D plots of water depth and velocities can also help you to understand what is wrong at some instants of your simulation.

Do your water level measurements include weather effects? If yes, you can try to introduce wind (and pressure) effects. Regarding the wind, you are not considering it as you set WIND = NO in your steering file.

I hope it helps.

Regards,
Laurent

hi Laurent,

thank you for the reply. I have already found the solutions for the shifted graph yesterday. it's due too the time of TPXO data was based on GMT which is an 8hrs difference in Malaysia. so 1 issue solved for now.


yup, I agree, there's noise for the model data. I will try with your suggestion and reduce the time step (but this will increased in time-consuming in running the program am I right?).

for the wind at the moment, the setting is NO, but I'll be used the wind data. should I put the wind data after I calibrate the hydrodynamic model or I need to include the wind data and then calibrate it?.



regards,

shahirah

Hi Shahirah,

First reduce the accuracy asked to the solver and its maximum number of iterations. Maybe you will not need to decrease the time step, but if it's needed, then yes it'll be more time-consuming.

As I told you, if your water level measurements include weather effects, you can try to introduce wind (and pressure) effects. It also depends on which processes you want to calibrate (tide, meteo, both...)

Regards,
Laurent