Welcome to the new DiscussCESM forum!
We are still working on the website migration, so you may experience downtime during this process.

Existing users, please reset your password before logging in here: https://xenforo.cgd.ucar.edu/cesm/index.php?lost-password/

Relationship between the bc_a4 files and num_a4 file.

dbailey

CSEG and Liaisons
Staff member
So, I've been talking a bit with Simone about this, but I thought I would post it here for others. The bc_a4 file for srf_emis_specifier is the surface emissions of black carbon. This has units of molecules/cm2/s. The num_a4 file is the particle number emissions file for the surface black carbon. This has units of particles/cm2/s * molecules/mole * g/kg. So, these are combined somehow to produce a particle size in the modal aerosol scheme. Does anyone understand how the particle size distribution is derived here? What I need to do is run a sensitivity experiment where I increase the emissions of black carbon along the Northern Sea Route in the Arctic. So, say they emissions themselves go up by 10%. Do I have to do the same to the particle number emissions?

Dave
 

tilmes

Member
Hi Dave, thanks for the post. For the calculation of the particle number emissions we use this formula:

; number=mass * MW / MP
; MP=density*(pi/6)*Diameter^3
The density and diameter depends on the specifics of the aerosols and the diameter on the the aerosol mode.

For BC and POM we use:
emis_num = emissions*mw /mass_particle
units: (particles/cm2/s)(molecules/mole)(g/kg)
with:
mass_particle = density *(PI/6.) *(diam)^3
mw = mass weight of the specie in the chemical mechanism, (12 for BC and POM)>
diam = 0.134e-6
rho = 1700. (for BC)
rho = 1000. (for POM)

Since the number is proportional to the aerosol emissions, you can multiply the emissions and the numbers by the same factor.

Simone
 

ppolonik

ppolonik
New Member
Hi Dave and Simone,

Thank you for sharing this information, it's really helpful.
I have spent a fair amount of time trying to replicate these emission datasets since I ultimately want to modify the inputs. I have a couple questions that I'm hoping one of you can help me with (see 1-3 below) about the BC, POM, SO2, and SO4 anthropogenic emissions.

I started directly from the emissions datasets from the input4mips database (kg/m2/s).
All my tests are with SSP3 files (the existing CESM inputs are here: https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/atm/cam/chem/emis/emissions_ssp370/)

I conservatively regrid the SSP emissions (kg/m2/s) and converted from kg to molecules via the molar mass. SO2 emissions need an additional factor of 0.975, since 0.025 goes to SO4 (note: kg SO2 to mol SO4 conversion uses SO2 molar mass). SO4 energy and industry emissions have a vertical component and are in molecules/cm3/s (as opposed to cm2) - it is divided into eight 50m vertical layers (=5000cm, which is used to convert cm3 to cm2), but only four of them (4-7) are non-zero and they each contain exactly a quarter of the emissions. Doing all this, I was able to exactly replicate a lot of the 'mol' emissions and using Simone's description above, also many the 'num' emissions.

I am able to exactly replicate the 'mol' and 'num' emissions for BC, which is great.

Here are my questions - any help would be greatly appreciated:
1. Is there an additional conversion factor from OC to POM? The ratio of my replicated POM emissions (calculated assuming OC=POM) to the existing CESM emissions is a constant ~0.714. Since the ratio is 1 when I try the same thing for BC, I assume it's something specific to OC/POM.

2. Is there an additional conversion factor from SO2 to SO4? For all SO4 emissions, I am able to replicate the 'mol' variable (using MW=64), but not the 'num' variable. I am again off by a constant ratio, this time ~0.55, or ~0.83 if I apply a factor 96/64 to convert from SO2 to SO4. I used density=1770.

3. I'm reasonably sure that the SO4 surface (non-energy/industry) emissions are mislabeled - is this a known issue? Here is the mapping from the labeled to the actual sectors that I've had to use to get results to align (ag and sol are zero):
'emiss_ag_sol_was' -> shipping
'emiss_res_tran' -> transportation + waste
'emiss_shipping' -> residential
Besides causing a little confusion, I don't think it makes any difference for the 'mol' datasets. But it's possible if I'm right that the incorrect diameters were used to convert from 'mol' to 'num'. For example, I get the same constant ratio (described in #2) when converting 'emiss_ag_sol_was' (actually shipping) using the diameter for solvents as I do when I convert 'emiss_shipping' (actually residential) with the diameter for shipping. It's the same ratio as for energy/industry, which is not mislabeled.

I hope this all made sense. Sorry this is so long and thanks in advance for your help!
 

CESM researcher

HW doctor
New Member
Hi Dave and Simone,

Thank you for sharing this information, it's really helpful.
I have spent a fair amount of time trying to replicate these emission datasets since I ultimately want to modify the inputs. I have a couple questions that I'm hoping one of you can help me with (see 1-3 below) about the BC, POM, SO2, and SO4 anthropogenic emissions.

I started directly from the emissions datasets from the input4mips database (kg/m2/s).
All my tests are with SSP3 files (the existing CESM inputs are here: https://svn-ccsm-inputdata.cgd.ucar.edu/trunk/inputdata/atm/cam/chem/emis/emissions_ssp370/)

I conservatively regrid the SSP emissions (kg/m2/s) and converted from kg to molecules via the molar mass. SO2 emissions need an additional factor of 0.975, since 0.025 goes to SO4 (note: kg SO2 to mol SO4 conversion uses SO2 molar mass). SO4 energy and industry emissions have a vertical component and are in molecules/cm3/s (as opposed to cm2) - it is divided into eight 50m vertical layers (=5000cm, which is used to convert cm3 to cm2), but only four of them (4-7) are non-zero and they each contain exactly a quarter of the emissions. Doing all this, I was able to exactly replicate a lot of the 'mol' emissions and using Simone's description above, also many the 'num' emissions.

I am able to exactly replicate the 'mol' and 'num' emissions for BC, which is great.

Here are my questions - any help would be greatly appreciated:
1. Is there an additional conversion factor from OC to POM? The ratio of my replicated POM emissions (calculated assuming OC=POM) to the existing CESM emissions is a constant ~0.714. Since the ratio is 1 when I try the same thing for BC, I assume it's something specific to OC/POM.

2. Is there an additional conversion factor from SO2 to SO4? For all SO4 emissions, I am able to replicate the 'mol' variable (using MW=64), but not the 'num' variable. I am again off by a constant ratio, this time ~0.55, or ~0.83 if I apply a factor 96/64 to convert from SO2 to SO4. I used density=1770.

3. I'm reasonably sure that the SO4 surface (non-energy/industry) emissions are mislabeled - is this a known issue? Here is the mapping from the labeled to the actual sectors that I've had to use to get results to align (ag and sol are zero):
'emiss_ag_sol_was' -> shipping
'emiss_res_tran' -> transportation + waste
'emiss_shipping' -> residential
Besides causing a little confusion, I don't think it makes any difference for the 'mol' datasets. But it's possible if I'm right that the incorrect diameters were used to convert from 'mol' to 'num'. For example, I get the same constant ratio (described in #2) when converting 'emiss_ag_sol_was' (actually shipping) using the diameter for solvents as I do when I convert 'emiss_shipping' (actually residential) with the diameter for shipping. It's the same ratio as for energy/industry, which is not mislabeled.

I hope this all made sense. Sorry this is so long and thanks in advance for your help!
Hi,ppolonik,Have you solved your problem?recently, I have been doing the same work . I designed a case compsets BSTRATMAM7.
I found that the input emissions data srf_emis_specifier for ISOP = / ATM/CAM/chem/emis / 1992-2010 / emissions. ISOP. Surface. 1.9 x2.5 _c110426. Nc
srf_emis_specifier for bc_a3 =atm/cam/chem/trop_mozart_aero/emis/ar5_mam3_bc_surf_2000_c120315.nc.
I have three questions:
1: Does this case represent the current (2000) year cycle simulation? How can you implement a historical simulation? (which compsets?)
2: What is the difference between emission data 1 (Chem/Emis / 1992-2010) and emission data 2 (Chem/Trop_mozart_aero/Emis)
3:I replaced the data AR5_MAM3_BC_SURF_2000_C120315. nc (Trop_mozart_aero /emis/) with emissions- CMIP6_BC_A4_Anthro_Surface_1750-2015_0.9X1.25_C20170608.nc (CAM /chem/emis/CMIP6_emissions_1750_2015). Is that feasible?

I hope this all made sense. Sorry this is so long and thanks in advance for your help!​

 

ppolonik

ppolonik
New Member
I'm not sure I understand all of your questions and I'm also new to CESM, so I'm probably not the best person to ask.
But here are a few thoughts from what I understand - maybe it will help a little.

I'm still unclear about my questions 2 and 3. As far as I know my calculations are correct. For question 1 (about OC), it turns out that a factor of 1.4 is commonly used (and 1/1.4 ~= 0.71).

For your question 1, I think there are many compsets that can be used for historical simulations, but I don't know which one would work best for your situation. But the emissions you provide do not determine whether you are doing a historical simulation or simulating a cyclical year. Editing the namelist variables srf_emis_cycle_yr and srf_emis_type determine which year's emissions are used. In my case I used srf_emis_cycle_yr=2030 and srf_emis_type='CYCLICAL', which repeats 2030 emissions for every year.

I don't know the answer to your second question directly, but there are different emissions sources. I was using data from CEDS, which I think was used for CMIP6 and was the default for the compset I was using in CESM2. It might be easiest to download some emissions from the two paths that you provided and look for differences directly - either by comparing the data or by looking at the netcdf metadata (attributes), which often has helpful information about how the data was created.

For your third question: I believe that for BC, all emissions go to the A4 size bin, so as far as I know making the change you described should be fine. I am not familiar with the AR5 dataset, but one potential difference is whether non-anthropogenic emissions are included. The CMIP6 dataset name implies it is limited to anthropogenic, so biomass burning is probably not included. But from the technical aspect, based on my limited experience I think the model would run with that change.

I hope this helped!
 
Top