[MITgcm-support] using RBCS to generate flow
katsman
katsman at knmi.nl
Tue Dec 24 07:54:37 EST 2013
Dear MITgcm developers,
Over the past years, I used the RBCS package to generate a flow in an
(otherwise unforced) basin representing the Labrador Sea, by restoring
the flow to a prescibed 3D-temperature and velocity field in a corner of
the basin (T and velocity in geostrophic balance). This worked excellent
in a (admittedly old) checkpoint58-version in which we manually added a
restoring term on U and V analogous to the programmed T, S part.
When I got a new workstation I thought it time to upgrade to the most
recent MITgcm version, but I cannot get the package to work in the same
way (and I tried many things over the past months, so I am getting a bit
desperate...).
The new checkpoint64 is supposed to be able to restore T,U and V, but I
find that - when I prescribe my T,V fields again as before - in the
sponge region defined by the mask, T is restored fine, while V is
restored fine only during the first few timesteps of the simulation.
Then a purely baroclinic U develops as well, while V becomes barotropic
(despite RBCS acting on it). Intriguing, but not what I wanted.
A test that really makes me think there is a bug in the rbcs package
rather than me doing something stupid is that when I only restore T with
a gradient in x, I expect to see a purely baroclinic V develop in
geostrophic balance with that. In stead, nothing happens (I can run it
for a month, no flow develops at all).
One of the earlier versions (c62) has yet another version of the RBCS
package. When I do a test with T-restore only (U,V is not standard in
that one), a baroclinic flow does develop obeying geostrophy.
Any ideas what is wrong here? I checked - the model does frequent the
appropriate lines of code in rbcs_add_tendency in the c64 version to
change the gV parameter - I suspect somewhere in the code U and V are
mixed up. Notably, the baroclinic flow that develops in the sponge
region is roughly the same strength as the V I prescribe
Any ideas?? (if the description is unclear I can send pictures)
Thanks in advance & happy holidays
Caroline
More information about the MITgcm-support
mailing list