open TELEMAC-MASCARET The mathematically superior suite of solvers

I have recently ported some older models I had validated in V6P3 accross to V7P2. The models are running in 3D (non-hydrostatic) and are forced at the boundaries using time series of water levels and depth averaged currents.

Time series of tidal water levels match perfectly at the boundaries of the domain as I would expect with them being imposed. The issue is that I am getting quite significant differences in the water levels between the 2 version with no changes in the setup of the 2 models. While both model are reproducing the tidal phase and amplitudes well there seems to be an overall reduction in water levels of about 20 cm in v7p2.

I'm not sure if it has to do with the fact that I am modelling with a salinity tracer initialized to around 40 ppt across the domain but it was also performed in the older v6p3 version.

I should also say that that the v6p3 model water levels have the better match between the imposed boundary conditions. The v7p2 model seems to be reducing the water levels within 1 node distance from the open sea boundaries of my model...

Is anyone aware of any major changes between the version that may have led to this?

So after some further testing this issue is definitely related to the inclusion of the salinity and temperature tracers.

Is this related to the Boussinesq approximation and is there any reason to expect such differences between the versions?

Hello,

To my knowledge there was no modification that could explain this. A question is whether your problem is well posed, given your boundary conditions. I also do not really understand how you can impose depth-averaged currents in 3D. The fact that salinity will trigger differences along the vertical could be a conflict with imposing depth-averaged currents as a constant on the vertical.

With best regards,

Jean-Michel Hervouet

Hi Jean,

Thanks for the input. The currents are imposed as you say, as a constant current across all layers of the boundary. This is manually performed within the BORD3D subroutine by forcing current values on UBORL and VBORL along with the water level. While this may not be the most correct approach for forcing a local model from a larger model domain I am welcome for suggestions on a better approach.

There is definitely a difference between the versions though. Either something must have changed within the more recent versions or something wasn't implemented in the older v6p3 version.

I have set the AVERAGE WATER DENSITY value to 1029.5 kg/m3, similar to what is given by the density law formula for a salinity of 42.4 ppt and water temperature of 20 degree Celsius.

I will extract some tide signals from both of the models for comparison over the next couple of days.

Hello
Even though prescribed values match well at the boucheries verify that:
1- you have updates your fortran user subroutines
2- between 6.9 and 7.2 the key word OPTION FOR LIQUID BOUNDARY (not sure about spelling) vas changed. So verify

Kind regards from Langkawi island
Riadh