AEROSOL_NOTES - 30 September 2008 CMAQ v4.7 features a new aerosol module called AE5, which contains substantial scientific improvements over the aerosol modules released in previous versions of CMAQ and is recommended as the default aerosol module in this public release. All of the notes in this file pertain to the AE5 module in CMAQ v4.7. The AE4 module remains unchanged relative to CMAQ v4.6 except for updates #3 -- #6 listed below. The AE3 module is no longer supported so it is not part of the CMAQ v4.7 release. 1. Secondary Organic Aerosol. Major updates: In CMAQ v4.7, several new pathways for secondary organic aerosol (SOA) formation have been implemented based on the recommendations of Edney et al. (2007) and the recent work of Carlton et al. (in review). New SOA precursors include isoprene, sesquiterpenes, benzene, glyoxal, and methylglyoxal. The enthalpies of vaporization and SOA/SOC ratios have been revised based on recent laboratory data. In previous versions of the CMAQ model, all SOA was treated as semi-volatile. In CMAQ v4.7, four types of non-volatile SOA are simulated: i. SOA formed by in-cloud oxidation from glyoxal and methylglyoxal ii. enhancement of isoprene-derived SOA under acidic conditions iii. SOA originating from aromatic oxidation under low-NOx conditions iv. oligomerization of particle-phase semi-volatile organic material Extensive detail about the new SOA treatment, including references for each yield and partitioning parameter, can be found in the orgaer5.f subprogram. In-cloud formation of SOA is treated in the aqchem.F subprogram. Special treatment of methylglyoxal (MGLY) is employed when the CB05 gas-phase mechanism is used. This was done to generate parity with SOA results obtained between simulations using either the SAPRC99 or CB05 gas-phase mechanism. GLY and MGLY are explicit species in SAPRC99, but CB05 has only MGLY. The lower water solubility of MGLY (vs. GLY) and the absence of GLY as a precursor in CB05 simulations resulted in substantially less cloud SOA than in the paired SAPRC99 runs. Thus for CB05 simulations, MGLY is assigned the Henry's Law value of GLY in aqchem.F. In conjunction with the SOA code updates, several changes were made to the CB05 and SAPRC99 gas-phase chemical mechanisms (see CHEMISTRY_NOTES.TXT) and the biogenic emissions inventory system (see /www.smoke-model.org). The default boundary concentration profile released with CMAQ v4.7 has been updated to include the SOA species needed for both AE4 and AE5. The science enhancements described above necessitated a number of simplifications to the SOA treatment in order to preserve computational efficiency. First, the Aitken mode SOA species were removed such that all SOA in CMAQ v4.7 is confined to the accumulation mode. This has a negligible impact on model predictions of fine-particle SOA. Second, SOA formation from cresol oxidation was removed. As pointed out by Dr. Greg Yarwood of ENVIRON Corporation, the SOA formed from cresol was already accounted for in the aromatic SOA yields because cresol is itself an aromatic oxidation product. Third, the olefin SOA species (SGTOT_OLI_*) were removed. That pathway had been zeroed out in CMAQ v4.3 -- v4.6, so removal of those species reduces the computational burden without impacting the model results. In CMAQ v4.3 -- v4.6, the gas and particulate fractions of each semi- volatile oxidation product were lumped together as a single species named SGTOT_* for transport and removal. This approach was designed to minimize the number of species required to treat reversible SOA partitioning, but it caused some complications in the dry deposition treatment due to substantial deposition velocity differences between the gas and particulate fractions of each SGTOT species. Moreover, Dr. Bonyoung Koo of ENVIRON Corporation noted that it introduced some errors and inconsistencies in the wet deposition treatment. In CMAQ v4.7, the SGTOT species have been eliminated. Instead, the model now separately tracks the gas- and particle-phase concentrations of each semi-volatile oxidation product. For example, SV_TRP1 represents the gas-phase portion of the first oxidation product of monoterpenes and ATRP1J represents the particle-phase portion of that oxidation product. The 'SV' prefix is an abbreviation for semi-volatile oxidation product and all SV_* species can be found in the non-reactive species list (NR_SPC.EXT). All of the AE5 mechanism include files in CMAQ v4.7 reflect these changes. As a consequence of the SGTOT separation, the aero_depv.F subprogram has been revised so that SV_* species are removed based on the gas-phase deposition velocities and SOA species are removed based on the particle- phase deposition velocities. An ancillary benefit of this separation is that model users can easily determine the absolute contributions of each SOA precursor to the modeled SOA concentrations. Rather than having individual sources of biogenic SOA grouped together within the AORGBI and AORGBJ species, for example, the CMAQ v4.7 outputs contain a separate model species for each SOA precursor and pathway. When interpreting the CMAQ v4.7 outputs, anthropogenic SOA should be computed as the sum of AALKJ (as in previous CMAQ releases, alkane SOA is computed only when running the SAPRC99 mechanism), ABNZ1J, ABNZ2J, ABNZ3J, ATOL1J, ATOL2J, ATOL3J, AXYL1J, AXYL2J, AXYL3J, and AOLGJ. Biogenic SOA should be computed as the sum of AISO1J, AISO2J, AISO3J, ATRP1J, ATRP2J, ASQTJ, and AOLGBJ. In addition, a fraction of AORGCJ could be considered as biogenic SOA (and the remainder as anthropogenic SOA) because both isoprene and a number of aromatic hydrocarbons are precursors of GLY and MGLY. Users are refered to EVALUATION_TOOLS.TXT for equations to compute organic carbon (OC) concentrations from the CMAQ v4.7 output. 2. Coarse PM. Dynamic gas/particle transfer of volatile inorganics: In previous versions of CMAQ, coarse-mode particles were assumed to be dry and inert, and components in the coarse mode could not evaporate or condense. This approach does not allow important aerosol processes, such as replacement of chloride by nitrate in mixed marine/urban air masses, to be simulated. In CMAQ v4.7, the code has been updated to allow semi-volatile aerosol components to condense and evaporate from the coarse mode and nonvolatile sulfate to condense on the coarse mode. Since coarse-particle components are often not in equilibrium with the gas phase, dynamic mass transfer is simulated for the coarse mode (whereas the fine modes are equilibrated instantaneously with the gas phase). This treatment can be found in the new VOLINORG subroutine, which is embedded in the aero_subs.f subprogram. In coastal regions, significant emissions of sea salt originate from wave-breaking in the surf zone. Previous versions of CMAQ did not account for this process. In CMAQ v4.7, surf-zone emissions are computed in the ssemis.F subprogram by assuming that the surf zone is covered entirely by whitecaps (i.e., WCAP = 1) and that the size distribution of sea salt emitted from the surf zone is identical to that emitted from the open ocean. The computational efficiency of the CMAQ aerosol module is governed largely by the number of calls to ISORROPIA (Bhave et al., 2004). In previous versions of CMAQ, only one ISORROPIA call per model time step was required to equilibrate the fine-particle composition with the gas phase. In CMAQ v4.7, additional ISORROPIA calls are required during each time step to transfer inorganic gases to and from the coarse mode. In an effort to preserve computational efficiency, coarse-mode mass transfer is skipped when either RH < 18% (i.e., the metastable aerosol assumption is questionable) or the summed concentration of coarse NH4, SO4, NO3, and Cl is less than 0.05 ug/m3 (i.e., coarse-mode mass transfer is of negligible importance). Alternatively, users may turn off the coarse-mode mass transfer altogether by setting HYBRID = .FALSE. in the VOLINORG subroutine (line 2827 of aero_subs.f). This is only recommended for model sensitivity studies or for applications in which the inorganic PM composition is of less importance than minimizing model runtime. In previous versions of CMAQ, ISORROPIA was always run in the 'forward' mode, which partitions total-system concentrations between the gas and particle phases. In CMAQ v4.7, the 'reverse' mode of ISORROPIA is used for the first time. In this mode, vapor pressures of the gas-phase components are predicted based on the particle-phase composition, temperature, and RH. These vapor pressures are needed to determine the chemical driving potential between the gas phase and the coarse-particle mode. During testing, we found that ISORROPIA in reverse-mode occasionally returns negative vapor pressures [i.e., variables GAS(1), GAS(2), and/or GAS(3) are less than zero]. A corrected version of ISORROPIA was not available in time for this CMAQ release so, as an interim measure, volatile inorganic species are not transferred to/from the coarse mode during the integration time steps when negative vapor pressures are returned by ISORROPIA and a descriptive warning is written to the log file. In previous versions of CMAQ, the standard deviation of the coarse-mode particle size distribution was fixed at 2.2, and sulfate was the only condensing component to influence fine-mode standard deviations. In CMAQ v4.7, the standard deviations of all three modes are variable. However, a constraint is imposed such that standard deviations of the accumulation and coarse modes cannot change during the condensation process (see LIMIT_Sg in the GETPAR subroutine, which is contained in the AERO_INFO.f subprogram). This is a temporary patch that was required to achieve a numerically stable solution during dynamic mass transfer, and we hope to relax this constraint in a future public release of CMAQ. In conjunction with the updated coarse-mode treatment, two new species were added to AE_SPC.EXT: ANH4K and SRFCOR. ANH4K represents the coarse-mode ammonium ion and SRFCOR represents the surface area of coarse-mode particles. All of the AE5 mechanism include files in CMAQ v4.7 include these new species. Coarse-mode water (AH2OK) and nitrate (ANO3K) were included in the AE4 mechanism files released with previous versions of CMAQ, but their concentrations were fixed at zero. In the AE5 module, these concentrations are nonzero. The default boundary concentration profile does not include species commonly associated with sea-salt (e.g., ANAJ, ACLJ, ANAK, ACLK, ASO4K, ANO3K, etc.). For model applications in which one or more of the domain boundaries lies over the ocean, users are urged to add these species to the boundary profile. 3. Heterogeneous chemistry. Updated N2O5 hydrolysis probability: In CMAQ v4.6, the N2O5 heterogeneous reaction probability (GAMMA_N2O5) was revised based on an equation published by Evans and Jacob (2005). After the release of v4.6, Dr. Jerry Davis from NC State University discovered a typographical error in the published equation which had been copied verbatim into the CMAQ code. Correction of that error decreased GAMMA_N2O5 in grid cells where the temperature exceeded 282K and increased GAMMA_N2O5 at temperatures below 282K. The changes at low temperatures resulted in large overpredictions of total gas+particulate nitrate during winter. This prompted investigations into a revised parameterization of GAMMA_N2O5. In CMAQ v4.7, GAMMA_N2O5 is computed using the parameterization developed recently by Davis et al. (2008) which is a function of temperature, RH, inorganic PM composition, and phase state. A number of structural changes to the aerosol code were made in conjunction with this update. In previous versions of CMAQ, heterogeneous reaction rates had been computed in the EQL3 subroutine which was embedded in aero_subs.f. In CMAQ v4.7, a new subprogram entitled hetchem.f was created to house all calculations relevant to heterogeneous chemistry. The parent subroutine, HETCHEM, calls the N2O5PROB function to compute GAMMA_N2O5. By changing the value of GPARAM (which is an input parameter to the N2O5PROB function), users can select among various parameterizations for computing GAMMA_N2O5. These options include each of the default parameterizations that had been used in previous public releases of the CMAQ model. The HETCHEM subroutine also treats production of HONO from heterogeneous reactions of NO2 on particle surfaces (see CHEMISTRY_NOTES.TXT for details). 4. PM Emissions. Compatibility with new speciation profiles: In previous versions of the CMAQ model, input emissions of PM2.5 had to be specified using the following variable names: PEC (primary elemental carbon), POA (primary organic aerosol), PSO4 (primary sulfate), PNO3 (primary nitrate), and PMFINE (other primary PM2.5). In the PM2.5 speciation profiles used to develop those emission estimates (e.g., Table S1 by Bhave et al., 2007), the POA term had been defined operationally as 1.2 times organic carbon (OC). This multiplicative scaling factor accounted approximately for non-carbon organic mass that is bound with OC at the point of emission. In profiles developed from SPECIATE v4.0 and subsequent versions of the SPECIATE database, the POA term has been replaced with POC. POC consists of organic carbon only, while the primary emissions of non-carbon organic mass are incorporated into the PMFINE term in these new profiles. In CMAQ v4.7, the AERO_EMIS.F subprogram has been modified to read emissions of either POC or POA. If the input file contains a variable named POC, those emissions are used without adjustment to augment the ambient concentrations of AORGPAI and AORGPAJ. If the emission file does not contain POC, the file is searched for a variable named POA. If found, the POA emissions are used without adjustment to augment the ambient concentrations of AORGPAI and AORGPAJ. If neither POC nor POA are found in the emission file, the CMAQ model will halt with an error message. As a result of this code change, the AORGPAI and AORGPAJ variables in a given CMAQ v4.7 simulation may represent either POC or POA, depending on the type of emission file that was used as model input. This has implications on the way the CMAQ model results are postprocessed for evaluation against ambient OC measurements. See Equation 1 by Appel et al. (2008) and refer to EVALUATION_TOOLS.TXT for further details. 5. Aerosol Thermodynamics. Minor updates to ISORROPIA: The solution to cubic equations was modified to account for cases when the cubic and quadratic terms are negligible compared to the linear and constant terms. Without this modification, the POLY3 subroutine was often returning without finding a root and the fine-particle chloride concentration was being incorrectly set to a very low value. 6. PM2.5 Size Cut. INLET25 subroutine: In CMAQ v4.5 and subsequent releases, users have the option to calculate the volume fraction of each mode that is composed of particles smaller than 2.5um aerodynamic diameter (see AEROSOL_NOTES.TXT in CMAQ v4.5 documentation). Roger Kwok from HKUST and Baoning Zhang from the University of Windsor independently reported an error in the DST25 calculation, which has been corrected in CMAQ v4.7. This code change has no impact on modeled concentrations, but it does affect the diagnostic output variables PM25AT, PM25AC, and PM25CO. REFERENCES: Appel, K. W., P. V. Bhave, A. B. Gilliland, G. Sarwar, and S. J. Roselle, 2008. Evaluation of the community multiscale air quality (CMAQ) model version 4.5: Sensitivities impacting model performance; Part II—particulate matter, Atmos. Environ., 42: 6057-6066. Bhave, P. V., S. J. Roselle, F. S. Binkowski, C. G. Nolte, S. Yu, G. L. Gipson, and K. L. Schere, 2004. CMAQ Aerosol Module Development: Recent Enhancements and Future Plans, In CMAS Models-3 Users' Conference, Chapel Hill, NC. www.cmascenter.org/conference/2004/archive.html Bhave, P. V., G. A. Pouliot, and M. Zheng, 2007. Diagnostic model evaluation for carbonaceous PM2.5 using organic markers measured in the southeastern U.S., Environ. Sci. Technol., 41: 1577-1583. Carlton, A.G., B. J. Turpin, K. Altieri, S. Seitzinger, R. Mathur, S. Roselle, R. J. Weber, in review. CMAQ model performance enhanced when in-cloud SOA is included: comparisons of OC predictions with measurements, Environ. Sci. Technol. Davis, J. M., P. V. Bhave, and K. M. Foley, 2008. Parameterization of N2O5 reaction probabilities on the surface of particles containing ammonium, sulfate, and nitrate, Atmos. Chem. Phys., 8: 5295-5311. Edney, E. O., T. E. Kleindienst, M. Lewandowski, and J. H. Offenberg, 2007. Updated SOA chemical mechanism for the Community Multi-Scale Air Quality model, EPA 600/X-07/025, U.S. EPA, Research Triangle Park, NC. Evans, M. J. and D. J. Jacob, 2005. Impact of new laboratory studies of N2O5 hydrolysis on global model budgets of tropospheric nitrogen oxides, ozone, and OH, Geophys. Res. Lett., 32, L09813, doi:10.1029/2005GL022469.