Surface boundary condition

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
logvinov
Posts: 23
Joined: Wed Feb 28, 2007 8:00 pm
Location: Massachusetts Institute of Technology

Surface boundary condition

#1 Unread post by logvinov »

Hi all,

I have a question about surface boundary condition.

I run application with uniform wind in x-direction sustr=5e-8 and with a constant value of Akv=1e-3.

When I checked the output the stress condition (sustr=Akv*du/dz) didn't work! Value in the output (Akv*du/dz) is ~15% less.

What is especially tricky du/dz seems to go to desirable value(big enough to satisfy stress condition) but then drops sharply in the top 2 grid points!

Here is the link to du/dz(z) plot and his.nc file

http://web.mit.edu/~logvinov/Public/ROMS/


Thanks a lot,
Evgeny

kurapov
Posts: 29
Joined: Fri Sep 05, 2003 4:49 pm
Location: COAS/OSU

#2 Unread post by kurapov »

The Ekman layer thickness is delta=sqrt(2*Akv/f) (see Pedlosky'87, 4.3.16)

Akv=1.e-3 m2/s
f=1.e-4 1/s

=> delta= 4m

So the sharp change you see in u,v in the upper 4 m is Ekman veering.

Besides, sustr=Akv du/dz at the surface. You are probably trying to estimate
it at the points below the surface. That's why you get 15% diff.

logvinov
Posts: 23
Joined: Wed Feb 28, 2007 8:00 pm
Location: Massachusetts Institute of Technology

#3 Unread post by logvinov »

Hi Alexander,

Thanks a lot for your reply!

Yep, Ekman layer is there, but as you said it is 4.5m. And indeed there is a significant correction on the field in the Ekman layer, which actually tends to go to the right value of du/dz (that would satisfy a stress condition).

But We see a very sharp change in the top 1 meter that actually goes to a smaller value of du/dz at the very surface! (see the fig above). I am not sure why it happens.

I am wondering if I am not aware of some tricky ROMS boundary condition, that it tries to satisfy. (Not just a stress condition)

kurapov
Posts: 29
Joined: Fri Sep 05, 2003 4:49 pm
Location: COAS/OSU

#4 Unread post by kurapov »

Use a better resolution. Try to resolve the boundary layer. Or, for the sake of experiment,
increase Akv by an order of magnitude.

logvinov
Posts: 23
Joined: Wed Feb 28, 2007 8:00 pm
Location: Massachusetts Institute of Technology

#5 Unread post by logvinov »

The resolution is pretty hight already. It has 5-6 points in Ekman layer, so I don't think there should be a problem with resolution. I'll try the bigger values of Vertical mixing

logvinov
Posts: 23
Joined: Wed Feb 28, 2007 8:00 pm
Location: Massachusetts Institute of Technology

#6 Unread post by logvinov »

The problem was in using

#define SPLINES
#define UV_SADVECTION

I terned them off and now everything is working

Post Reply