Testcase WC13/PSAS ends with netCDF error (in posterior.F)

[TomLav]

My runs of test case WC13/PSAS (r824) end with a netCDF error:

READ_STATE - error while reading variable: zeta at time record = 50
in NetCDF file: wc13_hss.nc
ROMS/TOMS - Output error ............ exit_flag: 3
ERROR: Abnormal termination: NetCDF OUTPUT.
REASON: NetCDF: Start+count exceeds dimension bound

This happens right after Utilities/posterior.F nf_fread2d() of zeta (free-surface) with rec=50.

Taking an "ncdump -ch wc13_hss.nc" reveals that the dimension ocean_time is "ocean_time = UNLIMITED ; // (49 currently)". It does thus seem that we are attempting a read at rec=50 which is beyond the current time dimention in wc13_hss.nc.

When looking closer at the log file, and specifically the lines indicating output/write to the hessian file, I see that the last one to appear reads:

WRT_HESSIAN - wrote Hessian fields (Index=1,1) in record = 0000049

So we do seem to only write 49 time records in the hessian file indeed.

I am too new to ROMS and 4DVAR to understand if we should be writing 1 more, or if we are reading 1 too many.

[TomLav]

A further note on this:

WC13/PSAS runs fine at r820 (that is "svn update -r 820" both in ROM's trunk/ and in tests/WC13/PSAS).

With r820, "ncdump -ch wc13_hss.nc" reveals that 50 time stamps were written to the hessian file (instead of 49 when the test is run on r824 - see original post).

New edit:
I get the same netCDF error than pure r824 when I use the r820 version and edit build.bash to add "-DRPCG".