A question about BPRP and BDRD

[Jay Zhu]

Hi all,

I have some question about the BPRP and BDRD in compset 'BSSP585cmip6'. I learned that BPRP uses prognostic CO2 as the forcing of ocean and land and ocean and atmosphere have mutual influence on atmospheric CO2, while BDRD uses prescribed CO2. I take this to mean that under the BPRP setting, the ocean and land can affect atmospheric CO2 concentrations, but cannot under BPRP.

But when I was setting up the compset, I use BSSP585cmip6 for long name and change ‘BGC%BDRD’ to ‘BGC%BPRP’ (that is, SSP585_CAM60_CLM50% BGc-crop-cMIp6deck_CICE %CMIP6_POP2%ECO%ABIO-DIC_MOSART_CISM2%NOEVOLVE_WW3_BGC%BPRP). When I run it there will be a high ATM_CO2 in the output file of POP. The first month 'ATM_CO2' is 446.8838. But in SSP585, the CO2 concentration in 2015 should be around 399. Why does this setting produce an overestimate of 50ppm?

Another problem is that there is a parameter 'OCN_CO2_TYPE' in the env_run.xml file. Is this parameter controlled by BPRP in compset? If I set the default BSSP58cmip6 and modify 'OCN_CO2_TYPE=prognostic', is it the same as using 'BGC%BPRP'?

The CESM version I used is CESM2.1.3.

Any help would be greatly appreciated！

Jay

 

[klindsay]

Jay,

I'll address your points in sequence:

under the BPRP setting, the ocean and land can affect atmospheric CO2 concentrations, but cannot under BDRD.

(You wrote BPRP at the end, but I assume that you meant BDRD.) A subtle point is that In both BDRD and BPRP, the ocean and land affect the CO2 tracer in the atmosphere. In BPRP, this CO2 tracer is used in biogeochemical computations in the land and ocean models (BP) and in radiative computations in the atmosphere (RP). In BDRD, these biogeochemical and radiative computations use prescribed CO2, and the CO2 tracer is a passive diagnostic tracer. So the CO2 tracer is always affected, but in BDRD, the CO2 tracer doesn't feed back to the rest of the model.

The first month 'ATM_CO2' is 446.8838. But in SSP585, the CO2 concentration in 2015 should be around 399. Why does this setting produce an overestimate of 50ppm?

I recommend that BPRP SSP experiments be initialized from BPRP historical experiments, for consistency. Such initial values are available in the input data repository with the settings
RUN_REFCASE=b.e21.BHIST_BPRP.f09_g17.CMIP6-esm-hist.002
RUN_REFDATE=2015-01-01

There are 2 possibilities to consider:

1. You are initializing your BPRP SSP585 experiment from this BPRP historical experiment. The atmospheric CO2 tracer in the BPRP historical experiment is being prognostically computed from surface fluxes from the land and ocean models, and from fossil fuel emisions. This computed CO2 concentration does not exactly match the observed CO2 concentrations. So the initial state of the CO2 tracer at 2015-01-01 does differ from the prescribed values used in BDRD SSP experiments.
2. You are initializing your BPRP SSP585 experiment from a BDRD historical experiment. The CO2 tracer in your BPRP SSP experiment is getting initialized from the (passive diagnostic) CO2 tracer in the BDRD SSP experiment. As in 1., this tracer's concentration does not exactly match the observed CO2 concentration, so the initial state of the CO2 tracer at 2015-01-01 does differ from the prescribed values used in BDRD SSP experiments.

I suspect that the (passive diagnostic) CO2 concentrations in the BDRD historical experiments differ more from observations from the CO2 concentrations in the BPRP historical experiment, because the BD framework lacks the negative CO2 concentrations feedbacks that are present in the BP framework. In particular, if CO2 concentrations grow large in a BP experiment, the land and ocean models will take up more CO2, reducing the excess CO2 in the atmosphere. This feedback is not present in a BD experiment, because the land and ocean models are using prescribed CO2 concentrations. So the (passive diagnostic) CO2 concentrations in BDRD experiments tend to drift further away from observations.

there is a parameter 'OCN_CO2_TYPE' in the env_run.xml file. Is this parameter controlled by BPRP in compset? If I set the default BSSP58cmip6 and modify 'OCN_CO2_TYPE=prognostic', is it the same as using 'BGC%BPRP'?

The default value of `OCN_CO2_TYPE` is `diagnostic` in coupled BDRD experiments and `prognostic` in BPRP experiments. The value of `OCN_CO2_TYPE` then determines a POP namelist variable that POP uses to determine where it gets its CO2 values from. You can override the defaults determined by BDRD vs BPRP by explicitly setting `OCN_CO2_TYPE`. The same applies to `CLM_CO2_TYPE` for the land model.

Hope that this helps,
Keith

 

[Jay Zhu]

Hi Keith,

Thank you for your careful and clear reply!

(You wrote BPRP at the end, but I assume that you meant BDRD.) A subtle point is that In both BDRD and BPRP, the ocean and land affect the CO2 tracer in the atmosphere. In BPRP, this CO2 tracer is used in biogeochemical computations in the land and ocean models (BP) and in radiative computations in the atmosphere (RP). In BDRD, these biogeochemical and radiative computations use prescribed CO2, and the CO2 tracer is a passive diagnostic tracer. So the CO2 tracer is always affected, but in BDRD, the CO2 tracer doesn't feed back to the rest of the model.

Regarding the explanation of BPRP and BDRD, can I understand it like this: both BPRP and BDRD will generate feedback on atmospheric CO2, but BDRD will use the CO2 forcing of SSP585 when forcing ocean or land modules? And the variable "ATM_CO2" in the output file pop.h.nc means the CO2 forcing in the ocean modules?

You are initializing your BPRP SSP585 experiment from a BDRD historical experiment. The CO2 tracer in your BPRP SSP experiment is getting initialized from the (passive diagnostic) CO2 tracer in the BDRD SSP experiment. As in 1., this tracer's concentration does not exactly match the observed CO2 concentration, so the initial state of the CO2 tracer at 2015-01-01 does differ from the prescribed values used in BDRD SSP experiments.

I think my problem belongs to the second possibility. I initialized the experiment with the files in ftp.cgd.ucar.edu/cesm/inputdata/cesm2_init/b.e21.BHIST.f09_g17.CMIP6-historical.010_v2/. Nominally, this should be the historical initial file of BDRD. As you suggested, I should look for BPRP's BHIST restart files and try again.

The default value of `OCN_CO2_TYPE` is `diagnostic` in coupled BDRD experiments and `prognostic` in BPRP experiments. The value of `OCN_CO2_TYPE` then determines a POP namelist variable that POP uses to determine where it gets its CO2 values from. You can override the defaults determined by BDRD vs BPRP by explicitly setting `OCN_CO2_TYPE`. The same applies to `CLM_CO2_TYPE` for the land model.

For 'OCN_CO2_TYPE', thank you for your explanation, which is very helpful to me.

Thank you again for your earnest answer！

Jay

 

[klindsay]

Regarding the explanation of BPRP and BDRD, can I understand it like this: both BPRP and BDRD will generate feedback on atmospheric CO2, but BDRD will use the CO2 forcing of SSP585 when forcing ocean or land modules? And the variable "ATM_CO2" in the output file pop.h.nc means the CO2 forcing in the ocean modules?

This is correct.

 

[Jay Zhu]

This is correct.

Thank you for your answer.

Jay