low vertical velocity at coast

[flcastej]

Dear all,

I am working on a coastal lagoon and I have been able to run model properly, but I have a problem and I will be really grateful if someone could give me some hints that allow me to resolve it. We deployed several ADCP in order to validate the model, the vertical velocities recorder are in the order of 0.01 m/s but I haven´t been able to get this values in the model in anyway. I have tried different vertical mixing scheme and different value for vertical mixing coefficients but I couldn´t get value closer to the measured one. The vertical velocities obtained with the model is about 0.0001 m/s.

It ´s seems like the current follow the coast to strong and don´t let the vertical structure to appeared near the coast. It's just my opinion so perhaps it is nonsense.

It could be great if someone who has faced a similar problem could tell me how resolve the problem or some advanced user could point me out what parameter could be related with this problem.

Thanks a lot in advance.

-Francisco

[kate]

Well, the model does not compute vertical velocity as a "basic" field. It instead computes it from horizontal convergences and divergences as part of the hydrostatic approximation. Perhaps increasing horizontal resolution will allow larger vertical velocities to emerge - but they might just look like noisy crap, I don't know.

If you are truly aiming to study non-hydrostatic processes, you will need a different model. I hear the French ROMS_Agrif team is developing a non-hydrostatic code.

[rduran]

Vertical velocity at the surface tends to be much smaller than vertical velocity in the interior. Did you check different layers? (Just making sure it's not that).

Vertical velocity in ROMS should be decent enough to, say, simulate upwelling and downwelling as long as they correspond to the processes you can resolve in the horizontal. Small-scale ageostrophic motion tends to have strong vertical velocity but is often not resolved by ROMS (depending on resolution) while the adcp would not have a problem so ... This might be the wrong metric to evaluate your model (btw it's impossible to validate a model see Oreskes et al 1994 for example). Remember you are comparing pointwise with cell-average ... not exactly apples with apples.

Just out of curiosity what is the error for adcp vertical velocity?

[flcastej]

Dear Kate and Rodrigo,

Thanks a lot for your answers. I hope to be able to find a solution using ROMS, I have worked really hard to have my model working properly including 3 nesting layers and it could be a pity that finally I wasn't able to use ROMS. Also I like ROMS and all the community behind it, so I will try to use it as hard as I can. Answering your questions:

Kate:

I am studying the vertical currents produced in the coast of a coastal lagoon due to the elevation produce by the wind blowing to the coast. Do you have any reference of someone who has been able to do so using ROMS?. The horizontal velocity field and elevation matched quite well with the data measured, but the problems appear as soon as I start to compare the model output with the vertical velocities. Perhaps I should forget about vertical velocities and assume that the model is working properly if the horizontal currents and elevation data matched the model. I will try to work with differents horizontal resolutions and check if my result are improved.

Rodrigo:

You are right and I am not comparing exactly the same, but at least I expected to get a similar range for the model and the measurement data. The accuracy of the ADCP used is 0.001+-0.005 m/s.

Best regards,

-Francisco