Fix inconsistencies between CO2 simulation and carbon simulation with CO2 only #2510
Conversation
… CO2 - Add missing ship logical to HEMCO_Config.rc.carbon, otherwise ship emissions for CO2 are not included - Update HEMCO_Config.rc.CO2 to: 1. Include DiagnFile to properly save out emissions diagnostics 2. Use same Aviation_SurfCorr_SclFac.1x1.nc file as carbon simulation Signed-off-by: Melissa Sulprizio <[email protected]>
# Conflicts: # CHANGELOG.md
msulprizio
added
category: Bug Fix
…sed on species defined In carbon_gases_mod.F90 all chemistry input fields needed for CO2, CH4, and CO chemistry were obtained regardless of the species included in the carbon simulation. Therefore when running the carbon simulation with only one species (e.g. CO2) fields were being obtained that weren't necessary. These fields have been blocked by logical statements checking if the species that needs those fields is advected. Also in this module, the chemistry routine has been updated to only call KPP when CH4 and/or CO are defined as advected species. KPP is not needed for CO2-only simulations. Instead CO2-only simulations utilize the CO2_COPROD field obtained from HEMCO. This field was originally obtained and applied to the species concentration array in subroutine Emiss_Carbon_Gases. The routine name was misleading and has been renamed to CO2_Production. The call to this routine has also been moved from emissions_mod.F90 to Chem_Carbon_Gases in carbon_gases_mod.F90. Notes: When we retire the CO2 simulation, we should be able to also remove the CO2 menu from geoschem_config.yml since it is not needed when using CO2 in the carbon simulation. Signed-off-by: Melissa Sulprizio <[email protected]>
The call to routine CO2_Production (formerly Emiss_Carbon_Gases) has been moved back to emissions_mod.F90 to reproduce the CO2 simulation. We may want to consider moving the call to this routine back to chemistry (i.e. in Chem_Carbon_Gases) since the routine applies the production of CO2 from CO and therefore represents a chemical source, not emissions. I will follow up with the Carbon Gases Working Group for guidance. Signed-off-by: Melissa Sulprizio <[email protected]>
…ased on species defines Logical switches have been added to HEMCO_Config.rc.carbon to only read the data necessary based on the carbon species used. For example, GLOBAL_OH is not needed for CO2-only simulations and will therefore not be read anymore. This should speed up single-species carbon simulations. Signed-off-by: Melissa Sulprizio <[email protected]>
msulprizio
force-pushed
the
bugfix/carbon_co2
branch
from
cb25002
to
e29adfd
Compare
|
Additional updates and fixes now include:
We are now able to reproduce CO2 concentrations between the CO2 simulation and CO2-only carbon simulations. See plots in #2463. |
yantosca
added
the
topic: Configuration Files
There was a problem hiding this comment.
Thanks @msulprizio. I had a couple of minor comments but will defer to @lizziel.
…U and on first timestep Also fixed a typo in CHANGELOG.md to use past tense. Signed-off-by: Melissa Sulprizio <[email protected]>
|
All GCClassic and GCHP integrations pass |
|
@msulprizio When I try to recompile GCClassic with the updated codes, it shows the following error information. ================================= |
Correction to my previous statement: All GCClassic integration tests are passing. The GCHP carbon CO2 simulation failed, even though the integration test reported it as passing (separate issue reported in #2527). I am investigating the failure (in ExtData) and will push a fix to this PR before requesting reviews again. |
…2-only simulations The carbon simulation with CO2 only was failing integration tests due to a failed ExtData import for CEDS_CO2_SHIP. The sed statements in singleCarbonSpecies.sh removed any entries with "CEDS_CO" if the CO species was not included in the simulation, also inadvertently removing the CEDS_CO2_SHIP entry. This has now been fixed by adding an underscore after CEDS_CO2. Other lines in that sed statement have also been updated to include an underscore or trailing white space to avoid this issue in the future. Signed-off-by: Melissa Sulprizio <[email protected]>
Several entries for CO2 emissions and chemistry data were incorrect in ExtData.rc.carbon. Primarily the climatology option and refresh string were inconsistent with the entries in HEMCO_Config.rc. Signed-off-by: Melissa Sulprizio <[email protected]> # Conflicts: # CHANGELOG.md
|
GCClassic and GCHP integration tests are now indeed passing. I pushed additional fixes for CO2 to ExtData.rc to get the GCHP carbon_CO2 simulation to complete. GCClassic differences for CO2 now look reasonable: GCHP differences in CO2 or the carbon (all species) simulation vs carbon with CO2 only looks reasonable as well: There are still large differences between CO2 in GCClassic and GCHP carbon simulations (all species vs CO2-only): Tagging @lizziel for any insight on the GCClassic vs GCHP differences. |
|
It looks like there are hot spots of CO2 at high elevations in GCHP. I am not sure what would cause this but can look into it. |
|
I turned off all the settings in HEMCO_Config.rc and geoschem_config.yml. It still shows some hot spots, especially at 500 hPa. Also set CO2 value in restart file to 400 ppmv. |
…ons for consistency with carbon simulations The CO2 and tagCO run directories were created using old restart files whose origin is not clear. To simplify validating the carbon simulation, these simulations now use the carbon simulation restart file which has been spun up for 5 years with a recent model version (14.1.0). The default start date for these simulations has changed to 20190101 as well. Signed-off-by: Melissa Sulprizio <[email protected]>
Name and Institution (Required)
Name: Melissa Sulprizio
Institution: Harvard
Describe the update
This pull request will address differences that still remain in CO2 between the traditional CO2 simulation and the carbon simulation with CO2 only. Fixes so far include:
a. Include DiagnFile to properly save out emissions diagnostics
b. Use same Aviation_SurfCorr_SclFac.1x1.nc file as carbon simulation
Expected changes
This is a zero difference update with respect to the full chemistry benchmark. It will impact CO2 in the carbon simulations.
Reference(s)
If this is a science update, please provide a literature citation.
Related Github Issue
See also: