neg. dp (m) Error

[Branden Spooner]

Hello HYCOM Experts,

I have been getting the following error on specific days while running a 0.01 resolution HYCOM run in the Gulf of Paria.

157447185 i,j,k=  220  518 11 neg. dp (m) in loop 19       -22.     max
157447185 i,j,k=  220  518 11 neg. dp (m) in loop 19       -22.     fatal
157447185 i,j,k=  220  519 12 neg. dp (m) in loop 19       -3.0     max
157447185 i,j,k=  220  517 11 neg. dp (m) in loop 15       -4.2     max
157447185 i,j,k=  220  519 12 neg. dp (m) in loop 15       -2.7     max

error: neg. dp (m) <    -10.00


I have changed the barotropic and baroclinic variables to the following values based on the CFL but still get the same error.
24  'baclin' = baroclinic time step (seconds), int. divisor of 86400
3  'batrop' = barotropic time step (seconds), int. div. of baclin/2

cfl:  min,max =      4.442   111.116

I am attaching the log file and blkdat.input file. Can you advise me on how I can resolve this issue?

Regards,
Branden

[Alan Wallcraft]

Suggested improvements:

a) Layers 36-41 are zero thickness everywhere.  So you could run with 35 layers and save about 15% of the computer time.  In principle the extra layers do no harm, although I see some zero salinity artifacts in these layers occasionally.

b) There is negative salinity at the surface:

157441083 i,j,k =  268  110  1 neg. saln after advem call      -0.01
157441086 i,j,k =  276   78  1 neg. saln after advem call      -0.01

I think this is due to too-strong river forcing.  I suggest smoothing the forcing.rivers input fields more that you do now (this spreads out the effect of the river).  Five smoothing passes is the minimum and you could go as high as 100 passes, but I would experiment with 10 or 20 smoothing passes.  In addition you could change the relaxation to SSS at the river by either setting relax.sssrmx to 0.5 (instead of -0.5), the former skips the relaxation if climatology is 0.5 psu away from the model SSS (as it will be near a river) and the later clips the relaxation at 0.5 psu.  Either use 0.5 everywhere or just near rivers.

c) The neg. dp is near the boundary, so it is probably the open boundary that is the problem.  A ten grid point nesting relaxation zone is the minimum and was used for GOMb0.08 because the outer model was also 0.08 degrees.  The usual rule of thumb is 5-10 grid points wide *on the outer grid*.  So the relaxation zone may be too narrow.

Alan.

[Branden Spooner]

Hello Alan,

Thank you for the suggestions.

Unfortunately, I am still having issues with "neg. dp" near the boundary. I have extended my domain to accommodate for the new nesting relaxation zone making it 10 points on the outer grid (80 points on the inner grid)  but I am still getting the error close to the boundary. I have attached the most recent log file, an image of the nesting relaxation zone and information on the domain.

Branden

 011.log

 depth_GOPa0.01_07.a

 depth_GOPa0.01_07.b

 GOP0.01_nest_rmu.png

 regional.grid.a

 regional.grid.b

--
--
You received this message because you are a member of HYCOM.org
To ask a question, send an email to fo...@hycom.org

---
To unsubscribe from this group and stop receiving emails from it, send an email to forum+un...@hycom.org.

[Alan Wallcraft]

The nest mask looks OK.

Sometimes weaker relaxation can help, e.g. change points with 0.1-0.3 days to 0.3 day e-folding time.

Plot the SSH and the currents near the blow up location in the model run and in the nest archive.

Sometimes removing a section of the barotropic port at the blow up location helps, but this is usually when there is an island just outside the domain that effects the outer model currents.

Unless it contributes significantly to the flow, I would remove the 5th port (i.e. close the strait at j=271,284), and perhaps make it land west of its choke point.

Note that 8x nesting is extreme, we normally do 3x or 5x at the most.

Alan.