AMIP HadISST boundary conditions issue

[shivr]

Hi,

We are doing an AMIP run with HadISST lower boundary conditions in CAM6. However, a few minutes after submitting the case the run aborts. The log file says that something is going wrong with the time interpolation:

2: ERROR: (shr_tInterp_getFactors) itime2 < itime1

The error is coming from the data ocean model. You can see it in the log file:
/glade/scratch/shivr/f.e21.F1850_BGC.f09_f09_mg17.CFMIP-hadisst-piForcing/run/ocn.log.2799533.chadmin1.ib0.cheyenne.ucar.edu.230823-013231

It looks like it is trying to interpolate between two times that are both 0, so it makes sense that it would freak out. It's probably an issue with
/glade/work/brianpm/cesm_release/components/cam/tools/icesst/bcgen/sstice_ts.nc

We're not seeing anything wrong with the file's time or date variables. So we're a bit stuck. Any help is appreciated, thanks!

Priyam and Brian

 

[jiamengl]

Hello, I wonder that if you have figured out what's wrong with this and how to fix it? I may run into the same problem

 

[dbailey]

Moving this to the CAM forum.

 

[brianpm]

Hi all,
I believe that we resolved this issue. I think the problem that Priyam was reporting was a symptom of something else going wrong in the setup of the run (Priyam can correct me if I'm remembering incorrectly).

For those looking into using HadISST or other data, here are the steps we used.
- Downloaded HadISST monthly data on 1° grid: Met Office Hadley Centre observations datasets
- Preprocess the downloaded file:
+ rename dimensions to lat, lon
+ rename variables (to "SST" and "SEAICE")
+ make sure SEAICE is in percent
+ make sure lat is -90 to 90
+ make sure lon is 0 to 360
+ construct date and datesec fields
- We then follow the instructions in the CAM tools/icesst README file. Note that we had to modify the Makefiles in bcgen and regrid to use the intel fortran compiler (ifort).

 

[shivr]

Yes, we resolved the issue by 1) creating a new case and 2) after creating the boundary condition file as Brian outlines above, copy it to a more permanent folder and have your run point there. We think an issue may have been that the run was pointing to the boundary condition file in /bcgen after it was created.

 

[Asiya]

Hi all,
I believe that we resolved this issue. I think the problem that Priyam was reporting was a symptom of something else going wrong in the setup of the run (Priyam can correct me if I'm remembering incorrectly).

For those looking into using HadISST or other data, here are the steps we used.
- Downloaded HadISST monthly data on 1° grid: Met Office Hadley Centre observations datasets
- Preprocess the downloaded file:
+ rename dimensions to lat, lon
+ rename variables (to "SST" and "SEAICE")
+ make sure SEAICE is in percent
+ make sure lat is -90 to 90
+ make sure lon is 0 to 360
+ construct date and datesec fields
- We then follow the instructions in the CAM tools/icesst README file. Note that we had to modify the Makefiles in bcgen and regrid to use the intel fortran compiler (ifort).

Could you please tell me what modifications we made in the Makefiles for bcgen and regrid? I am also encountering some compiler errors when I run make -f Makefile inside the bcgen directory.

 

[brianpm]

I am pretty sure that we just needed to change the default fortran compiler to be compatible with the system we were using. Probably we changed from an old compiler (Portland group?) to gfortran.