PISCES runtime error

Hi all,

I previously have run CROCO-PISCES on my current domain of interest, but it was a while back, firstly in climatological mode, and then with Mercator IBI initial and boundary conditions.

Now, I’m in the situation where I am encountering two errors.
Firstly, I was getting an error in init_scalars_.f (line 1170) where the vname array was smaller than required. This was in relation to reading point source tracers. I have switched off psources to postpone this issue.

Now, secondly, I’m getting an issue with reading boundary conditions (forrtl error 61, and netcdf error -49), where it would seem that at get_bry_bio_.f, line 1706, the code can’t read a variable that it’s looking for, even though the screen output indicates that it has read plenty of variables. I am forcing the model at the boundary with all 24 variables and they’re all included in the files.

Even though I’m debugging, I can’t see what variable the code is trying to read and missing. Forrtl error 61 indicates trying to read a real that is an integer, or vice versa.

It is as if the number of tracers is one more than reality, and thus not available in the BCs (a 25th bgc tracer for example).

The default option of using an analytical concentration isn’t kicking in either, which is strange.

The following is sample output in my file croco_Y2017M01.out
ERROR in nf_read_bry: nf_get_vara netCDF error code = -49
Cause of error: NetCDF: Variable not found mynode = 245
ERROR in get_bry_bio: cannot read variable ‘trc_south’

while the following is in run_croco_inter.out
forrtl: severe (61): format/variable-type mismatch, unit 6, file /proc/62228/fd/1
Image PC Routine Line Source
croco1 0000000001199E0B Unknown Unknown Unknown
croco1 00000000011E44E9 Unknown Unknown Unknown
croco1 000000000111C45F get_bry_bio_ 1706 get_bry_bio_.f
croco1 0000000000B781F4 MAIN__ 596 main_.f
croco1 000000000040BD22 Unknown Unknown Unknown
libc-2.17.so 00002B95DA360495 __libc_start_main Unknown Unknown
croco1 000000000040BC29 Unknown Unknown Unknown

Any suggestions would be greatly appreciated.

No clear idea, but can you attach:

  • your full log
  • the namelist you’re using for pisces


Hi Rachid,

The national HPC system here has just gone offline this morning until Thursday or Friday for maintenance so I can’t send on either to you right now. I will get it back to you as soon as I can.


Hi Rachid,

Just coming back to this now. I’ve attached the full log (Y2017M01.out and run_croco_inter.out). Having given this some thought, I remember that the difference between my most recent boundary files and previous attempt is that previously, for each month, I had daily physics and BGC.

Now, I have daily physics and three data points for BGC (previous month 15th day, current month 15th day, and 15th day coming month). Could the difference in the number of data points for PHY/BGC boundary conditions cause this? I modified the time array for each BGC param so that it’s only 3 long

and not 33/34 points long (providing a few days of overlap for interpolation).

The time array for some of the variables came from 1 source, with the time defaulting to integer because it was a whole number, whilst the time for the remaining variables was floating point. As a result, at runtime the timestamp which was assumed to be floating was met with an integer value which threw a runtime error.
When I forced the timestamps/time arrays to float64 when writing the bry files in python, the issue was resolved