high u_hvisc near the surface at the masked regions

[sonaljit.m]

Hello,
My ROMS setup is showing unrealistic velocities near the surface at the locations right next to the masked areas. The diagnostic output shows that the viscous fluxes (u_hvisc, v_hvisc) are shooting up to very high levels, messing up the velocity profiles only at the sigma points near the surface. Below, is a plot showing the profile of u_hvisc at time steps 7,8,9,10 at a point right beside the masked region.

u_hvisc_evolve.png (26.3 KiB) Viewed 2084 times

As you can see, the u_hvisc deviates from timestep 8 to 9, and keeps deviating then onwards. As a result, the velocity changes drastically. This is happening only near the surface. Below, is a plot of the velocity profile at the same time steps:

The location of this point is right beside a masked out region, shown in the contour of u_hvisc below:

As shown, this anomaly is occurring at the corner of the masked region. Everywhere else, the value of u_hvisc is almost negligible (order of 10^-06). Can anyone shed some light on this? I am not sure what parameters should be modified to prevent this anomaly near the surface points.

Note: This simulation is unforced, with 0 fluxes for momentum and scalar. The numbers rx0 and rx1 are 0.19 and 2.0 respectively.

The .h and .in input files are attached.

Thank you.

[kate]

Try turning off MIX_GEO_UV? It's an option I don't typically use, not sure how much people have really pushed it. Also, you have it with UV_VIS4, not sure that combo is supported.

[wilkin]

Sonaljit,

It looks like you're using the CariCOOS model configuration but with several modifications to the set-up (like MIX_GEO_UV that Kate noted, and I see biharmonic friction also). The CariCOOS model was running quite reliably in its configuration as of a few months ago, so any changes should be made methodically one at a time to monitor how they affect stability.

Note that just because the horizontal mixing terms are large does not mean they are the culprit driving instability. They may be a symptom only, because the divergence of the velocity gradient has already grown due to some other cause, and hence del^2(u) is large.

Something that might affect these terms specifically adjacent to the land when biharmonic friction is active is the gamma slipperiness factor. There were posts about this on the forum a long long time ago. Just be aware this could be an issue. My recollection is that no-slip is advised for biharmonic friction.