Error while running hycom: negative depth and NaN values

[Danielle Manalaysay]

Hi all,

I encountered the following error while running expt_01.0

   1942     15363 i,j,k=  164   19 30 neg. dp (m) in loop 15      -0.20E+74
   1943     15363 i,j,k=  127  195 31 neg. dp (m) in loop 15      -0.40E+74
   1944     15363 i,j,k=  164   19 32 neg. dp (m) in loop 15      -0.35E+82
   1945
   1946 error: neg. dp (m) <    -10.00
   1947
   1948
   1949
   1950      timer statistics
   1951     ------------------
   1952
   1953      cnuity   calls =        2   time =    0.17360   time/call =    0.08680000
   1954      tsadvc   calls =        2   time =    0.23310   time/call =    0.11655000



Based from the previous posts here this is an indication of a CFL violation. I have tried smaller barotropic time step using baclin =2400 and batrop =30.

But after trying running with the settings above I encountered the ff error:

   1926     11556 30 min/max of s after advection:999.000*******   (range:-2.00E+03)
   1927     11556 31 min/max of s after advection:999.000*******   (range:-2.00E+03)
   1928     11556 32 min/max of s after advection:999.000*******   (range:-2.00E+03)
   1929     11556 (0001/322 00) mean      SSH (mm):     NaN  (Infinity to     -Inf)
   1930 
   1931  error - NaN or Inf detected
   1932 
   1933     11556 (0001/322 00) mean mixl thk. (m):     NaN  temp:    NaN saln:    NaN
   1934     11556 (0001/322 00) mean surf thk. (m):    3.00   sst:    NaN  sss:    NaN
   1935     11556 (0001/322 00) mean clim thk. (m):   19.20   sst: 28.238  sss: 33.715


I have tried hycom_NaN and hycom_sea_ok but there were no NaN files detected on my *a files on data/SCRATCH.

Has anyone encountered this before?

Best regards,
Dani

[alan.wa...@hycom.org]

This looks like a problem with forcing files, since 10^74 is larger than I would expect from a CFL violation in a mostly running case.

Attach the entire .log file and I'll see if it provides any clues.

Alan.

[Danielle Manalaysay]

Hi Alan,

I'll try to look into my forcing files again. But here's the log file with negative depth.

Dani

[alan.wa...@hycom.org]

I think the problem is with your relaxation (climatology) files. 

If you set itest,jtest to a sea point then there will be a printout of the iniitial vertical profile at this point extracted from the climatology in the .log file:

 sigma(k):  28.10  28.90  29.70  30.50  30.95  31.50  32.05  32.60  33.15
                 33.70  34.25  34.75  35.15  35.50  35.80  36.04  36.20  36.38
                 36.52  36.62  36.70  36.77  36.83  36.89  36.97  37.02  37.06
                 37.10  37.17  37.30  37.42  37.48
        0    1    1  istate:  temp    saln  thstar   thkns    dpth   montg
                       mxl                                     0.0
                         1    0.00    0.00   34.00     0.0     0.0   0.000
                         2    0.00    0.00   34.00     0.0     0.0   0.000
                         3    0.00    0.00   34.00     0.0     0.0   0.000

The above is all zeros because you have itest=jtest=1.

The global means are written to the .log file, and they look strange:

    15361 (0001/321 00) mean L  1 thk. (m):    3.00  temp: 28.253 saln: 33.713
    15361 (0001/321 00) mean L  2 thk. (m):    3.54  temp: 28.234 saln: 33.714
    15361 (0001/321 00) mean L  3 thk. (m):    7.41  temp: 17.688 saln: 34.065
    15361 (0001/321 00) mean L  4 thk. (m):    5.60  temp: 11.494 saln: 34.311
    15361 (0001/321 00) mean L  5 thk. (m): 2670.78  temp:  4.897 saln: 34.432
    15361 (0001/321 00) mean L  6 thk. (m):  370.66  temp:  4.685 saln: 34.490
    15361 (0001/321 00) mean L  7 thk. (m):   50.91  temp:  4.666 saln: 34.503
    15361 (0001/321 00) mean L  8 thk. (m):    6.89  temp:  4.643 saln: 34.505
    15361 (0001/321 00) mean L  9 thk. (m):    0.89  temp:  4.505 saln: 34.511
    15361 (0001/321 00) mean L 10 thk. (m):    0.03  temp:  3.583 saln: 34.533
    15361 (0001/321 00) mean L 11 thk. (m):    0.00  temp: 11.946 saln: 34.235
    15361 (0001/321 00) mean L 12 thk. (m):    0.00  temp: 10.488 saln: 34.302

All the mean thicknesses are zero below layer 11 and layer 5 seems to cover most of the water column.  This may partially be a poor choice of target densities, but I'm not sure that is the entire story.  Note that if a layer is zero everywhere in the climatology, it is unlikely to change when the model is running (and it is usual for a regional model to not need the largest target densities from a global model, i.e. to exclude a few of the global near-bottom layers).

See the attached example of how to use relaxi, which is now our recommended way to map from a climatologies z-levels to HYCOM layers for relax files.

Alan.