I am running a simulation on derecho using the compset BWSSP245cmip6 at f09_g17 resolution from 2015 to 2035 with MARBL in a 4p2z configuration. The case directory for this is at /glade/work/jcoupe/cases/ARISE-SAI/4p2z_spinup/b.e21.BWSSP245cmip6.f09_g17.CMIP6-SSP2-4.5-WACCM.001_4p2z_derecho
The simulation starts as a hybrid run with restart files from another simulation previously run on cheyenne, which ended at the year 2015 under the compset BHISTcmip6 and used CAM6 at the same horizontal resolution, whereas the 2015-2035 simulation is using WACCM6. The case directory for the BHISTcmip6 simulation is /glade/work/jcoupe/cases/ARISE-SAI/4p2z_spinup/b.e21.BHIST.f09_g17.CMIP6-historical.001_CAM6_POP2_4p2z
I noticed anomalies in pH and pCO2 at year 2015 compared to another BWSSP245cmip6 simulation run on cheyenne that I am trying to mimic (see green line below). pH suddenly stops declining at year 2015 at the transition from my BHISTcmip6 simulation (blue line) to my BWSSP245cmip6 simulation (orange line). In the previously run BWSSP245cmip6 simulation (green line), ocean acidification continues through 2035.
Surface pCO2(ocn) continues to increase and surface temperatures are increasing in both simulations as expected from the SSP2-4.5 scenario. ATM_CO2 is identical and increasing in both BWSSP245cmip6 simulations, so the ocean appears to have the same CO2 gas forcing. Next, I looked at FG_CO2. From here, it would appear that through the 2015-2035 period, the atmosphere is attempting to send more and more CO2 to the ocean. However, the ocean is not receiving it (?) as surface pCO2 is unchanged, but FG_CO2 continues to increase as the difference between pCO2(atm) and pCO2(ocn) grows. DIC stops increasing, alkalinity is unchanging in both simulations.
OCN_TRANSIENT=ssp245 and OCN_CO2_TYPE=diagnostic for both BWSSP245cmip6 simulations. The intentional differences are: MARBL-4p2z vs MARBL-3p1z, derecho vs cheyenne, and WACCM6 vs CAM6. Unintentionally, I realized that there are differences in the compset longnames that seem to arise from using CAM6 vs WACCM6. My historical simulation is:
HIST_CAM60_CLM50%BGC-CROP-CMIP6DECK_CICE%CMIP6_POP2%ECO%ABIO-DIC_MOSART_CISM2%NOEVOLVE_WW3_BGC%BDRD
whereas the longname for the SSP2-4.5 simulation (both on cheyenne and derecho) is
SSP245_CAM60%WCTS_CLM50%BGC-CROP-CMIP6WACCMDECK_CICE%CMIP6_POP2%ECO%NDEP_MOSART_CISM2%NOEVOLVE_WW3
Could POP2%ECO%ABIO-DIC...._BGC%BDRD vs POP2%ECO%NDEP lead to a step change in how the ocean receives carbon? Are there known differences in atm->ocn CO2 flux when switching from CAM6 to WACCM6? I have been trying to find documentation on the ABIO-DIC part but it seems to be sparse.
Any ideas for helping to diagnose this would be appreciated, thanks!
The simulation starts as a hybrid run with restart files from another simulation previously run on cheyenne, which ended at the year 2015 under the compset BHISTcmip6 and used CAM6 at the same horizontal resolution, whereas the 2015-2035 simulation is using WACCM6. The case directory for the BHISTcmip6 simulation is /glade/work/jcoupe/cases/ARISE-SAI/4p2z_spinup/b.e21.BHIST.f09_g17.CMIP6-historical.001_CAM6_POP2_4p2z
I noticed anomalies in pH and pCO2 at year 2015 compared to another BWSSP245cmip6 simulation run on cheyenne that I am trying to mimic (see green line below). pH suddenly stops declining at year 2015 at the transition from my BHISTcmip6 simulation (blue line) to my BWSSP245cmip6 simulation (orange line). In the previously run BWSSP245cmip6 simulation (green line), ocean acidification continues through 2035.
Surface pCO2(ocn) continues to increase and surface temperatures are increasing in both simulations as expected from the SSP2-4.5 scenario. ATM_CO2 is identical and increasing in both BWSSP245cmip6 simulations, so the ocean appears to have the same CO2 gas forcing. Next, I looked at FG_CO2. From here, it would appear that through the 2015-2035 period, the atmosphere is attempting to send more and more CO2 to the ocean. However, the ocean is not receiving it (?) as surface pCO2 is unchanged, but FG_CO2 continues to increase as the difference between pCO2(atm) and pCO2(ocn) grows. DIC stops increasing, alkalinity is unchanging in both simulations.
OCN_TRANSIENT=ssp245 and OCN_CO2_TYPE=diagnostic for both BWSSP245cmip6 simulations. The intentional differences are: MARBL-4p2z vs MARBL-3p1z, derecho vs cheyenne, and WACCM6 vs CAM6. Unintentionally, I realized that there are differences in the compset longnames that seem to arise from using CAM6 vs WACCM6. My historical simulation is:
HIST_CAM60_CLM50%BGC-CROP-CMIP6DECK_CICE%CMIP6_POP2%ECO%ABIO-DIC_MOSART_CISM2%NOEVOLVE_WW3_BGC%BDRD
whereas the longname for the SSP2-4.5 simulation (both on cheyenne and derecho) is
SSP245_CAM60%WCTS_CLM50%BGC-CROP-CMIP6WACCMDECK_CICE%CMIP6_POP2%ECO%NDEP_MOSART_CISM2%NOEVOLVE_WW3
Could POP2%ECO%ABIO-DIC...._BGC%BDRD vs POP2%ECO%NDEP lead to a step change in how the ocean receives carbon? Are there known differences in atm->ocn CO2 flux when switching from CAM6 to WACCM6? I have been trying to find documentation on the ABIO-DIC part but it seems to be sparse.
Any ideas for helping to diagnose this would be appreciated, thanks!