Simulation of organic aerosol formation during the CalNex study: updated mobile emissions and secondary organic aerosol parameterization for intermediate-volatility organic compounds

Abstract

We describe simulations using an updated version of the Community Multiscale Air Quality model version 5.3 (CMAQ v5.3) to investigate the contribution of intermediate-volatility organic compounds (IVOCs) to secondary organic aerosol (SOA) formation in southern California during the CalNex study. We first derive a model-ready parameterization for SOA formation from IVOC emissions from mobile sources. To account for SOA formation from both diesel and gasoline sources, the parameterization has six lumped precursor species that resolve both volatility and molecular structure (aromatic versus aliphatic). We also implement new mobile-source emission profiles that quantify all IVOCs based on direct measurements. The profiles have been released in SPECIATE 5.0. By incorporating both comprehensive mobile-source emission profiles for semivolatile organic compounds (SVOCs) and IVOCs and experimentally constrained SOA yields, this CMAQ configuration best represents the contribution of mobile sources to urban and regional ambient organic aerosol (OA). In the Los Angeles region, gasoline sources emit 4 times more non-methane organic gases (NMOGs) than diesel sources, but diesel emits roughly 3 times more IVOCs on an absolute basis. The revised model predicts all mobile sources (including on- and off-road gasoline, aircraft, and on- and off-road diesel) contribute ~ 1 μgm^-3 to the daily peak SOA concentration in Pasadena. This represents a ~ 70% increase in predicted daily peak SOA formation compared to the base version of CMAQ. Therefore, IVOCs in mobile-source emissions contribute almost as much SOA as traditional precursors such as single-ring aromatics. However, accounting for these emissions in CMAQ does not reproduce measurements of either ambient SOA or IVOCs. To investigate the potential contribution of other IVOC sources, we performed two exploratory simulations with varying amounts of IVOC emissions from nonmobile sources. To close the mass balance of primary hydrocarbon IVOCs, IVOCs would need to account for 12% of NMOG emissions from nonmobile sources (or equivalently 30.7 t d^-1 in the Los Angeles-Pasadena region), a value that is well within the reported range of IVOC content from volatile chemical products. To close the SOA mass balance and also explain the mildly oxygenated IVOCs in Pasadena, an additional 14.8% of nonmobile-source NMOG emissions would need to be IVOCs (assuming SOA yields from the mobile IVOCs apply to nonmobile IVOCs). However, an IVOC-to-NMOG ratio of 26.8% (or equivalently 68.5 t d^-1 in the Los Angeles-Pasadena region) for nonmobile sources is likely unrealistically high. Our results highlight the important contribution of IVOCs to SOA production in the Los Angeles region but underscore that other uncertainties must be addressed (multigenerational aging, aqueous chemistry and vapor wall losses) to close the SOA mass balance. This research also highlights the effectiveness of regulations to reduce mobile-source emissions, which have in turn increased the relative importance of other sources, such as volatile chemical products.

Figure 1.

Scatter plot of first-generation mass-based SOA yields versus volatility (log C*, μgm⁻³) in the detailed parameterization (dots are colored by OH reaction rates).

Figure 2.

(a) Comparison of predicted SOA formation per unit of mass mobile IVOC emissions of new parameterizations and model of Zhao et al. (2015, 2016) at OA = 5 μgm⁻³ (average [OH] = 3 × 10⁶ cm⁻³). (b) Relative error in SOA formed between new and Zhao et al. (2015, 2016) parameterizations (solid line is the relative error at OA = 5 μgm⁻³; shaded area corresponds to OA from 1 to 50 μgm⁻³).

Figure 3.

(a) Modeled NMOG and IVOC emissions by source for the four simulation cases. (b) Measured and modeled IVOC mass concentrations in Pasadena, CA, during CalNex for the four simulation cases. Measured data in (b) from Zhao et al. (2014).

Figure 4.

Comparison of measured (boxplot: solid box denotes 25th to 75th percentiles and whiskers denote 10th to 90th percentiles) and modeled (line: shaded area denotes 25th to 75th percentiles) diurnal patterns in Pasadena, CA, during CalNex for the following species: (a) benzene, k_OH = 1.22 × 10⁻¹² cm³ molec⁻¹ s⁻¹; (b) toluene, k_OH = 5.63 × 10⁻¹² cm³ molec⁻¹ s⁻¹; (c) xylene, k_OH = 1.36 – 1.87 × 10⁻¹¹ cm³ molec⁻¹ s⁻¹; and (d) hydrocarbon IVOCs (blue: Case 2, red: Case 3), k_OH = 1.55 – 7.56 × 10⁻¹¹ cm³ molec⁻¹ s⁻¹. Measured data from Borbon et al. (2013).

Figure 5.

(a) PM₁-OA component hourly-averaged time series of measured data and model output in Pasadena, CA, during the CalNex campaign. (b, c) Diurnal pattern of measured and modeled SOA and POA mass concentration in Pasadena, CA, during CalNex. Measured data from Hayes et al. (2013).

Figure 6.

(a) Campaign-average NMOG emissions (td⁻¹) in the emission inventory. (b) Modeled campaign-averaged SOA concentration in Case 4. (c) Location of CSN sites used for model evaluation. (d) Comparison of modeled OA to measured OA (OC·1.8) at CSN sites in California.