I am attempting to set-up a 1-D ROMS application to study vertical T/S stratification and I was wondering whether anyone else out there has already done something like this?
I set-up a 9 x 9 grid in the horizontal with 20 vertical levels and the horizontal and vertical extents of the grid were 0.1 deg x 0.1 deg and 10m (flat bathymetry) respectively. I cooked up an initial T-S profile (having the correct density balance) and the surface forcings (winds, air T, air P, RH, sw rad, lwrad) all came from 32-km NARR.
For the lateral BCs, I tried (i) pure radiation, (ii) closed (default), (iii) wall and (iv) periodic - as the domain is so small in the horizontal. For vertical eddy-mixing I tried MY2.5, GLS k-kl, k-e, k-omega.
As expected I found radiation, (i) to give a drift in the water elevation, etc. However, to my surprise, I found the periodic conditions, (iv) induced huge vertical mixing (even though the w-velocities were smaller than for the other cases) and destroyed the stratification soon. The vertical eddy-viscosity models MY2.5, k-kl, k-e gave similar results but the k-omega gave really weird stratification and generated a lot of vertical mixing (thereby killing off the stratification). I am now going to see what happens when I switch off the meteorological forcing.
Has anybody else seen similar trends in 1-D ROMS applications? Any experiences and advice you can share will be much appreciated.
Thanks.
1-D ROMS application to test stratification
Re: 1-D ROMS application to test stratification
The code is distributed with the "mixed_layer" test case. It is a simple "1-D" application that we used to test the different GLS schemes for vertical mixing. The results are presented in our 2005 Ocean Modeling paper (see figure 4). I have not run that test case in a while. I do not remember getting strange results with k-w. Let me know if you try that case and there are any troubles.
Re: 1-D ROMS application to test stratification
Thanks for the reply! It is somewhat similar to my configuration. I have two questions :
(1) What is the justification for using E-W periodic open boundary conditions in one direction and N-S wall open boundary conditions in the other direction? Should not all boundaries/directions treated equally as we assume that there is no horizontal bias in the flow (and we are only interested in the vertical flow/mixing/dynamics)? and,
(2) For the GLS turbulent closures, we have the following additional CPP options (from cppdefs.h) :
** CANUTO_A use if Canuto A-stability function formulation **
** CANUTO_B use if Canuto B-stability function formulation **
** CHARNOK use if Charnok surface roughness from wind stress **
** CRAIG_BANNER use if Craig and Banner wave breaking surface flux **
** KANTHA_CLAYSON use if Kantha and Clayson stability function **
** K_C2ADVECTION use if 2nd-order centered advection **
** K_C4ADVECTION use if 4th-order centered advection **
** N2S2_HORAVG use if horizontal smoothing of buoyancy/shear **
** ZOS_HSIG use if surface roughness from wave amplitude **
** TKE_WAVEDISS use if wave breaking surface flux from wave amplitude **
When using the GLS formulation, how do we know which of these options to include for a particular application? Is there some guidance/notes/advice written up somewhere? I tend use only N2S2_HORAVG and KANTHA_CLAYSON and is this sufficient? In the 1-D mixing example included with the ROMS (MIXED_LAYER) in addition to these two, several other CPP options are also employed. What is the justification for including them?
(1) What is the justification for using E-W periodic open boundary conditions in one direction and N-S wall open boundary conditions in the other direction? Should not all boundaries/directions treated equally as we assume that there is no horizontal bias in the flow (and we are only interested in the vertical flow/mixing/dynamics)? and,
(2) For the GLS turbulent closures, we have the following additional CPP options (from cppdefs.h) :
** CANUTO_A use if Canuto A-stability function formulation **
** CANUTO_B use if Canuto B-stability function formulation **
** CHARNOK use if Charnok surface roughness from wind stress **
** CRAIG_BANNER use if Craig and Banner wave breaking surface flux **
** KANTHA_CLAYSON use if Kantha and Clayson stability function **
** K_C2ADVECTION use if 2nd-order centered advection **
** K_C4ADVECTION use if 4th-order centered advection **
** N2S2_HORAVG use if horizontal smoothing of buoyancy/shear **
** ZOS_HSIG use if surface roughness from wave amplitude **
** TKE_WAVEDISS use if wave breaking surface flux from wave amplitude **
When using the GLS formulation, how do we know which of these options to include for a particular application? Is there some guidance/notes/advice written up somewhere? I tend use only N2S2_HORAVG and KANTHA_CLAYSON and is this sufficient? In the 1-D mixing example included with the ROMS (MIXED_LAYER) in addition to these two, several other CPP options are also employed. What is the justification for including them?
Re: 1-D ROMS application to test stratification
1) I think that application has a wind stress applied in the e-w direction, and that creates a current. There is an initial temp profile that mixes downward, and we compare that rate of mixing with an analytical expression. So there is a horizontal advection, although there are no gradients in the horizontal. It should work with N-S periodic as well.
2) GLS:
There are 4 stability functions:
GALPERIN (default),
KANTHA_CLAYSON,
CANUTO_A
CANUTO_B
Users need to select one of these.
See our 2005 Ocean Modeling paper for descriptions of these.
When you activate MY25 or GLS, i strongly suggest that you also use N2S2_HORAVG. This averages the shear and buoyancy in the horizontal, and reduces oscillations.
For :
CHARNOK
CRAIG_BANNER
Sandro and others have a paper that was just reviewed in Ocean Modeling that describes these options. That paper should be out soon. These have to deal with surface flux of tke due to wave breaking, and use surface wind stress as the mixing forcing.
For:
ZOS_HSIG
TKE_WAVEDISS
These are also methods to allow surface flux of tke due to wave breaking, but they need wave height information (not wind stress). These are coded in, but not completely tested. I am testing them, as we add many more nearshore capabilities. These are typically for surf zone applications
K_C2, K_C4 - these are for horzontal advection of tke. If you dont select these, the default is a 3rd order scheme. I usually use the default method.
Hope this helps.
-j
2) GLS:
There are 4 stability functions:
GALPERIN (default),
KANTHA_CLAYSON,
CANUTO_A
CANUTO_B
Users need to select one of these.
See our 2005 Ocean Modeling paper for descriptions of these.
When you activate MY25 or GLS, i strongly suggest that you also use N2S2_HORAVG. This averages the shear and buoyancy in the horizontal, and reduces oscillations.
For :
CHARNOK
CRAIG_BANNER
Sandro and others have a paper that was just reviewed in Ocean Modeling that describes these options. That paper should be out soon. These have to deal with surface flux of tke due to wave breaking, and use surface wind stress as the mixing forcing.
For:
ZOS_HSIG
TKE_WAVEDISS
These are also methods to allow surface flux of tke due to wave breaking, but they need wave height information (not wind stress). These are coded in, but not completely tested. I am testing them, as we add many more nearshore capabilities. These are typically for surf zone applications
K_C2, K_C4 - these are for horzontal advection of tke. If you dont select these, the default is a 3rd order scheme. I usually use the default method.
Hope this helps.
-j