Global Model BC/IC in CMAQ

I am trying to create BC/IC for CMAQ from a global model (AM3) which has a single SOA species to represent Secondary Organic Aerosol, but I am not sure how to divide the single global model SOA species into multiple CMAQ cb6r3_ae7_aq aerosol species in a CONUS domain.

We cannot provide specific mappings for all possible models, but here are some things to consider. Generally, you should map the same precursor system to the same precursor system. For example, if the SOA species is from monoterpene oxidation, then map to the monoterpene SOA set in CMAQ. Second, if there is an assumed volatility in one model, it should be mapped to the same volatility in CMAQ. C* is a measure of volatility with C*=0.1ug/m3 effectively nonvolatile if you need a nonvolatile mapping. To get a better description of aero7 species, see (Organic Aerosol tab) and Pye et al. 2017 ACP, Murphy et al. 2017 ACP, and the aero7 release notes should define all species. We also include species descriptions in the code (CCTM/src/MECHS/mechanism_information;


1 Like

On a related note, I was curious if any assumptions about OM/OC ratio is made in CMAQ cb6r3_ae7_aq mechanism or CMAQ postprocessing? I want to map Primary OM in the global model to a CMAQ species, and in my opinion, the lumped aerosol species APOC (Primary Organic Carbon; mol wt = 220.0) is the most suitable one but the global Primary OM will have to be converted to Primary OC first. I was wondering if could use something like OM/OC=1.5 for the entire US?

The OM/OC ratios assumed by CMAQ for all OM species can be found by looking at the definitions for the post-processed species APOCI, APOCJ, ASOCI, and ASOCJ in the mechanism-specific species definition file.

For cb6r3_ae7_aq, this shows the following:

APOCI ,ugC m-3 ,ALVPO1I[1]/1.39 + ASVPO1I[1]/1.32 + ASVPO2I[1]/1.26

APOCJ ,ugC m-3 ,ALVPO1J[1]/1.39 + ASVPO1J[1]/1.32 + ASVPO2J[1]/1.26
+ASVPO3J[1]/1.21 + AIVPO1J[1]/1.17 + APOCJ[1]

ASOCI ,ugC m-3 ,ALVOO1I[1]/2.27 + ALVOO2I[1]/2.06
+ASVOO1I[1]/1.88 + ASVOO2I[1]/1.73

ASOCJ ,ugC m-3 ,AISO1J[1]/2.20 + AISO2J[1]/2.23 + AISO3J[1]/2.80
+AMT1J[1]/1.67 + AMT2J[1]/1.67 + AMT3J[1]/1.72
+AMT4J[1]/1.53 + AMT5J[1]/1.57 + AMT6J[1]/1.40
+ AMTNO3J[1]/1.90 + AMTHYDJ[1]/1.54
+AGLYJ[1]/2.13 + ASQTJ[1]/1.52
+AORGCJ[1]/2.00 + AOLGBJ[1]/2.10 + AOLGAJ[1]/2.50
+ALVOO1J[1]/2.27 + ALVOO2J[1]/2.06 + ASVOO1J[1]/1.88
+ASVOO2J[1]/1.73 + ASVOO3J[1]/1.60 + APCSOJ[1] /2.00 \
+AAVB1J[1]/2.70 + AAVB2J[1]/2.35 + AAVB3J[1]/2.17

You are correct that if you want to map primary OM from the global model to the CMAQ species APOCIJ (note that APOCJ is both a CMAQ model species and, in the post-processing equation above, the sum of all primary OC species in the accumulation mode, including APOCJ), you’d need to decide on what kind of organic aerosols are represented by “OM” from the global model and, therefore, what OM/OC ratio you want to assume when mapping it to the CMAQ species APOCJ. This is more of a question for the global model from which you obtain OM than it is a CMAQ question.

From the CMAQ perspective, you could instead decide to map “OM” from the global model to one or more of the other primary (and/or secondary, unless you know that your OM is indeed all primary) CMAQ OM species, like ALVPO1J, ASVPO1J, etc., depending on what you know about the characteristics such as sources, OM/OC ratio, volatility, etc. of OM from the global model.

Thanks for clarifying the expressions. I have ‘Primary OM’ from the global model where OA = Primary OM + SOA. Now to map that Primary OM, I am looking at

APOMI ,ug m-3 ,ALVPO1I[1] + ASVPO1I[1] + ASVPO2I[1] + APOCI[1] + APNCOMI[1]

from the species definition file and thinking I could use global model OM/OC=3:2 to map 40% of Primary OM to APNCOMI[1]; similarly, I could set aside 60% of Primary OM to ALVPO1I[1] , ASVPO1I[1] , ASVPO2I[1] and APOCI[1] although I am not sure in what ratio I should distribute the mass to different volatility aerosols? Also, is APOCI[1] representing permanently aerosol POC that will never partition to gas phase unlike ALVPO1[1], ASVPO1[1], and ASVPO2[1] that can partition between gas and aerosol phases?

Others can weigh in, but I would suggest to split your primary OM either between APOC and APNCOM or between ALVPO1, ASVPO1, and ASVPO2.

The first option represents the traditional non-volatile primary organic aerosol approach while the second represents the semivolatile primary organic aerosol approach first implemented in CMAQv5.2 and described in the Murphy et al. reference listed by Havala above. In CMAQv5.2, the two approaches were implemented as separate aero6 mechanisms (aero6 vs. aero6nvPOA). You can see by comparing the AE namelist files for these two mechanisms in CMAQv5.2 here and here that the first one only contains APOC and APNCOM while the second one only contains ALVPO1, ASVPO1, and ASVPO2 for primary organic aerosols.

In CMAQv5.3 aero7, all of these five species are included and that is why the post-processor species definition file lists all of them when computing the post-processed species APOCI and APOCJ. A user can decide to map emissions of primary organic aerosols to either the non-volatile or non-volatile framework separately for each emissions stream through DESID depending on what is know about these emissions, and can also use either framework to represent boundary and initial conditions of primary organic aerosols.

If you decide to split OM from AM3 as 60% APOC and 40% APNCOM, you’re assuming an OM:OC ratio of 1.5:1 and selecting to map to the non-volatile primary organic aerosol framework since APOC will not partition to the gas phase.

If, on the other hand, you split OM between ALVPO1, ASVPO1, and ASVPO2, you’re selecting to map to the semivolatile primary organic aerosol framework. Each of these species has a different volatility as noted by Havala and documented in the Murphy et al. reference, so you need to decide what the most appropriate split would be for the type of primary OM represented by AM3. Note that this split will not require you to explicitly assume any OM:OC ratio since these three CMAQ species, like all CMAQ secondary organic aerosol species, already include the mass of the non-carbon species associated with them, so you’d just be splitting total primary OM mass between three primary OM species.

Thanks for explaining the Primary OA frameworks in CMAQ to me. Using the non-volatile framework seems easier since the OM:OC ratio in the global model is given to me. However, as a process, the representation of aerosols with different volatility in the semi-volatile framework seems more realistic (just an opinion). I have two questions here:

  1. I believe using a specific NEI emissions inventory (say 2014) should not affect my decision to use the non-volatile or semi-volatile framework, but please correct me if I am wrong here. NEI emission files will probably have primary OM emissions but then CMAQ will assign those emissions according to the framework chosen.
  2. I am just curious how lumped species molecular weight is calculated (or estimated)? For example, APOC and APNCOM both have molecular weights of 220 in CMAQ species definition. I remember molecular weight of air = 29 g/mol calculated averaging three of four main component gases with different weights, but it is probably a different process for large organic compounds in a chemical mechanism inside a chemical transport model.

For your first question, you are correct, in CMAQv5.3 the choice whether to use the non-volatile or semivolatile POA framework for mapping BC and IC can be made independently of which framework you use to represent emissions. The choice for mapping POA emissions is made through the DESID module, specifically the EmissCtrl namelist file, and can be done on a stream-by-stream basis. By default, POA emissions (i.e. emission species POC and PNCOM) are mapped to the semivolatile framework, see the section “! Fine-Mode Primary Organic Aerosol Scaling” in the EmissCtrl namelist file.

I do not know the answer to your second question.

Thanks - I had some more related questions:

  1. What is the lumped species AOTHR (Other Particulate Mass; mol. weight 200) representing? Is it organic, inorganic or unidentified?
  2. Is it possible to map to only the Coarse mode (or some other mode like Aitken) species in CMAQ BC/IC? Say instead of mapping to APOC, can I map to APOCI or APOCK even though they don’t appear on Species list of the chemical mechanism?

AOTHR represents primary unspeciated fine PM. It was first introduced in CMAQv5.0. Please see the corresponding release notes and the Reff et al. (2009) reference provided in these release notes for further details on speciation of primary fine PM.

I am not sure I understand the second question correctly. You can only map to species listed in the first column of the mechanism-specific GC, AE, and NR namelist files. APOCK does not appear in the AE namelist file and is therefore not a CMAQ species.

In the conversation above, mapping to “APOC” and “APNCOM” was shorthand for mapping to both the Aitken (I) and accumulation (J) mode for which these species are defined and appear in the AE namelist file, i.e. APOCI, APOCJ, APNCOMI, and APNCOMJ. The split between I and J mode depends on what you know about the size distribution of the aerosol you’re mapping from, but you likely want to assign most of your mass to the J and only a small fraction to the I mode. Default diameters and geometric standard deviations for the three aerosol modes used by CMAQ can be found in AERO_DATA.F

I am thinking of using the semi-volatile treatment in CMAQ for my simulations just because it is a better representation of the process.

First, about SMOKE Emissions - the default DESID table for is:

Question 1) Is it OK to just use this default table for splitting all primary OM emission surrogates from SMOKE for a 12km US domain simulation in CMAQ?

Second, about BC/IC from global model for the same CMAQ simulation - I also want to map global model species to CMAQ BC/IC for the same CMAQ simulation using semi-volatile framework.
Question 2) If I mapped the primary OM from the global model to SMOKE emission surrogates POC and PNCOM rather than CMAQ species of varying volatilities, would CMAQ still make the conversion as per the semi-volatile framework table above in DESID? Or is DESID strictly for emissions surrogate species only?

I cannot provide a definitive answer to question 1), but if your emissions were produced with a fairly standard NEI / SMOKE setup, using the DESID defaults is probably a good starting point.

DESID is for emissions only, mapping emitted species to model mechanism species. If you map your global model OM concentrations to CMAQ APOCI, APOCJ, APNCOMI, and APNCOMJ IC and BC concentrations, these non-volatile species will not be converted into any other species. Put differently, DESID controls the mapping of emission species to CMAQ model mechanism species, while processing IC and BC from a non-CMAQ model involves mapping species from that model’s mechanism to CMAQ model mechanism species.

Thanks @hogrefe.christian for responding - I will start with the default DESID table for splitting SMOKE emission surrogates to semi-volatile framework CMAQ species.

Sorry I was not clear about the question on global model to CMAQ mapping. Now that I am using the semivolatile framework (default DESID table) for primary OA from SMOKE emissions, I could still use OM:OC = 1.5 to map global model primary OM aerosol to CMAQ’s non-volatile framework species APOC and APNCOM for BC, based on cb6r3_ae7 species definition,

APOCI ,ugC m-3 ,ALVPO1I[1]/1.39 + ASVPO1I[1]/1.32 + ASVPO2I[1]/1.26 + APOCI[1]
APOCJ ,ugC m-3 ,ALVPO1J[1]/1.39 + ASVPO1J[1]/1.32 + ASVPO2J[1]/1.26 + ASVPO3J[1]/1.21 + AIVPO1J[1]/1.17 + APOCJ[1]

CMAQ postprocesor will sum both APOC from BC and ALVPO1,ASVPOn from emissions (but processed in CMAQ) to calculate the final PM2.5? I wanted to check with you if this approach of mine would be fine to use.

Yes, the species definition files for the ‘combine’ postprocessor will consider both the semi-volatile and non-volatile OM species when computing APOCI, APOCJ, etc.

For many applications, only one group (semi-volatile and non-volatile OM species) would have non-zero concentration values, but in your case, the effects of emissions would be reflected in the semi-volatile species and the effects of boundary conditions would be reflected in the non-volatile species, and the post-processor would add them when computing APOCI, APOCJ, etc.

aero7 was deliberately set up to accommodate such an approach, while in aero6 you had to decide upfront which framework to use and then provide both emissions and boundary conditions using that framework.

Is there a way to check what boundary conditions were used or the BC contribution to a pollutant concentration in CMAQ output, log or diagnostics file? This is with regards to any pollutant of interest, not just OM.

If you use CMAQ ISAM you can determine the contribution to a pollutant from the boundary conditions, initial conditions, and from specified regions of interest.

We also have a model feature in development now that will output the boundary contributions (in and out) for any desired species for each domain face of the rectangular grid. This should be released with the next version of CMAQ.