Kessler pressure artifacts

[jtruesdal]

There is an issue with the Kessler test. Here is a 1 day comparison of PS with FV. I've narrowed the pressure range to emphasize the pressure banding.

@PeterHjortLauritzen I think that @sjsprecious has standalone HOMME running on Derecho and could provide a reference for theta-l Kessler. I could also use that version as a comparison to my Kessler port to help debug the problem. Jian if your code is ready to go could you try running a baroclinic case using namelist-r100 (maybe change dcmip16_prec_type to 0 for Kessler physics.) If it would take a day or two to get running maybe just point me to a run script and code directory that I can use for debug.

[sjsprecious]

Hi @jtruesdal , thanks for posting these results.

I just did a standalone HOMME run on Derecho with theta-l dycore and 30 vertical layers. I used the dcmip2016_test1_baroclinic_wave test suite and set the dcmip16_prec_type = 0 in the r100 namelist as suggested.

The output can be found on Derecho at /glade/derecho/scratch/sunjian/e3sm_test/homme/dcmip_tests/dcmip2016_test1_baroclinic_wave/theta-l.

The namelist used for the run is at /glade/derecho/scratch/sunjian/e3sm_test/homme/dcmip_tests/dcmip2016_test1_baroclinic_wave/theta-l/namelist-r100.nl

In case you want to run it by yourself, the code and run script for standalone HOMME on Derecho are available at:

• code directory: /glade/derecho/scratch/sunjian/E3SM/components/homme
• run script: /glade/u/home/sunjian/e3sm/run_script/derecho/homme/job_submission.sh (you may change some settings or directories inside the script as needed)

Thank you and let me know if you have any questions.

[jtruesdal]

Thanks so much. At 24 hours HOMME standalone theta-l looks very much like FV. Same range for ps as above.

[jtruesdal]

@sjsprecious @PeterHjortLauritzen I was using an older incorrect version of cube_mod.F90 and found a few other bugs. After correcting these, the theta-l_kokkos and theta-l versions of Kessler look much better. Here is a theta-l_kokkos pressure field after 24 hours. Note that this is ne16 so looks a little bit different than the above ne30 but the pressure error is gone!