cecilehannay commented 7 months ago

Purpose: Rougher topography has shown indications of improving precip in the polar regions and marginal improvements in stratospheric jets: https://github.com/NCAR/amwg_dev/issues/473 Therefore we would like to science evaluate rougher topography in coupled setup.

Description:

Same as 78 (https://github.com/NCAR/amwg_dev/issues/485) but with

rougher topography.

bnd_topo = "/glade/campaign/cgd/amp/pel/topo/files/se/ne30pg3_gmted2010_modis_bedmachine_nc3000_Laplace0050_noleak_20240116.nc"

new pe_layout
```
./xmlchange NTASKS_OCN=512
```
Case directory:
Locally (if still available): /glade/p/cesmdata/cseg/runs/cesm2_0/b.e23_alpha16g.BLT1850.ne30_t232.079
On github: https://github.com/NCAR/amwg_dev/tree/b.e23_alpha16g.BLT1850.ne30_t232.079

Sandbox:
Locally (if still available): /glade/work/hannay/cesm_tags/cesm2_3_alpha16g
On github: https://github.com/ESCOMP/cesm/tree/cesm2_3_alpha16g

Diagnostics:
Diags (if available): https://webext.cgd.ucar.edu/BLT1850/b.e23_alpha16g.BLT1850.ne30_t232.079/

Output:
Raw output on derecho (if still available): /glade/derecho/scratch/hannay/archive/b.e23_alpha16g.BLT1850.ne30_t232.079/
CAM Climos: /glade/campaign/cgd/amp/amwg/climo/b.e23_alpha16g.BLT1850.ne30_t232.079

Contacts: @PeterHjortLauritzen @adamrher @cecilehannay @JulioTBacmeister @swrneale

Extra details:

cecilehannay commented 7 months ago

Timings

79 includes the faster dycore + new pe_layout.

It is a significant improvement (faster and cheaper) compared to where we were in 75 (with the slower dycore).
The pe_layout from 78 might be slightly better.
@jedwards4b: what do you recommend?

Timing for 79 (faster dycore + pe_layout2)

  Overall Metrics: 
    Model Cost:            5465.62   pe-hrs/simulated_year 
    Model Throughput:        10.68   simulated_years/day

Timing for 75 (old dycore)

  Overall Metrics: 
    Model Cost:            6436.15   pe-hrs/simulated_year 
    Model Throughput:         8.11   simulated_years/day

Timing for 78 (faster dycore + pe_layout1)

  Overall Metrics: 
    Model Cost:            5135.40   pe-hrs/simulated_year 
    Model Throughput:        10.77   simulated_years/day

jedwards4b commented 7 months ago

I agree, 78 looks like the winner.

cecilehannay commented 7 months ago

@PeterHjortLauritzen: diags for that run are here

adamrher commented 6 months ago

@whlipscomb @Katetc the BLT1850 runs do have NOEVOLVE on and there is CISM output:

/glade/derecho/scratch/hannay/archive/b.e23_alpha16g.BLT1850.ne30_t232.079/glc/hist

And so you could probably look at acab in this run relative to these baselines https://github.com/NCAR/amwg_dev/issues/497 and https://github.com/NCAR/amwg_dev/issues/485 to identify the impact of the rougher topo.

My understanding is that all our F-cases are using stub glacier, but these BLT1850 compsets have NOEVOLVE on, and therefore produce cism output.

Katetc commented 6 months ago

Hi Adam, just wanted to point out that the files in this directory are only initial_hist files from each of Cecile's restarts. They are not full CISM history output. After discussing with Bill Sacks and Mariana, we have discovered that even when CISM NOEVOLVE is on, the run loop is not called from the NUOPC coupler so it does not produce history files in a B case. There is an issue around this in CMEPS 426 and discussions and work going on to figure out how/what we want to address this going forward.

adamrher commented 6 months ago

Oh my, I guess I should've looked a level deeper. Thanks for the clarification, and I'm glad you guys caught this!

NCAR / amwg_dev

b.e23_alpha16g.BLT1850.ne30_t232.079 #489

Timings