[MITgcm-support] advection scheme for T/S

[samar khatiwala]

Hello

I am comparing a calculation in which temperature is advected with
the standard ABII scheme and one in which the DST3 (non fluxlimited)
scheme is used. I get substantially different steady state solutions
and was wondering if this is not unreasonable or whether I should be
looking for a bug in my configuration.

What I have is a sector domain (box with a ridge in the middle) 10 levels,
T is relaxed to a linear in y profile at the surface. There is no salinity
in this calculation.
Wind forcing is a simple cosine to drive a gyre. For the two calculations
all I do is change tempAdvScheme from 2 to 30. There is implicit mixing.

It would be great if someone could help me track down the problem
(assuming there is one). I've staged the steady state solution here:
http://www.ldeo.columbia.edu/~spk/temperature.mat
(arrays are called T_AB and T_DST3)
Data file is here:
http://www.ldeo.columbia.edu/~spk/data

Thanks

Samar