Airborne measurements of sub-micron aerosol and trace gases downwind of Los Angeles are used to investigate the influence of aging on aerosol properties relevant to climate forcing and visibility. The analysis focuses on the Los Angeles plume, which in addition to strong urban emissions is influenced by local agricultural emissions. Secondary organic aerosol formation and repartitioning of semi-volatile ammonium nitrate were identified as key factors controlling the optical behavior observed. For one case study, ammonium nitrate contributed up to 50% of total dry extinction. At 85% relative humidity, extinction in the fresh plume was enhanced by a factor of ∼1.7, and 60–80% of this was from water associated with ammonium nitrate. On this day, loss of ammonium nitrate resulted in decreasing aerosol hygroscopicity with aging. Failing to account for loss of ammonium nitrate led to overestimation of the radiative cooling exerted by the most aged aerosol by ∼35% under dry conditions. These results show that changes to aerosol behavior with aging can impact visibility and climate forcing significantly. The importance of ammonium nitrate and water also highlight the need to improve the current representation of semi-volatile aerosol species in large-scale climate models.
 The interaction of aerosol particles with solar radiation is of key importance for atmospheric radiative transfer. The most recent consensus estimate of global average climate forcing from direct aerosol effects is −0.5 ± 0.4 Wm−2, which is comparable in magnitude (but opposite in sign) to that of methane, the second most important greenhouse gas [Intergovernmental Panel on Climate Change (IPCC), 2007]. On regional scales the aerosol direct climate forcing can be significantly larger [Ramanathan et al., 2005]. In addition, light extinction by aerosol particles can degrade visibility, both during extreme events such as dust storms, and more widely in the vicinity of urban regions [Hyslop, 2009; Malm et al., 1994]. It follows that understanding aerosol optical behavior and associated spatial and temporal variability is a necessary prerequisite to understanding its role in climate and visibility.
 The extent to which aerosol particles scatter and absorb radiation depends primarily on three factors: complex refractive index (and thus composition), size and morphology [Bohren and Huffman, 1983]. The aerosol source (e.g., combustion, biomass burning, condensation from the gas phase, mechanical generation etc.) initially determines these properties; however a wide range of atmospheric processing mechanisms can subsequently act to modify them, leading to the evolution of aerosol optical behavior with time. Among the most important processing mechanisms are those involving mass transfer between the particulate and gas phases. For example black carbon (BC), which is the most strongly absorbing aerosol in the atmosphere [Ramanathan and Carmichael, 2008], is initially fractal in nature when produced from efficient combustion and characteristically ages through the accumulation of a coating of secondary inorganic and/or organic material [Oshima et al., 2009]. The presence of coating material affects the optical behavior of BC in a variety of ways: scattering can increase due to larger particle size, absorption can increase due to coating-induced lensing effects [Bond et al., 2006; Lack et al., 2009], the coating material can cause the absorbing fractal BC to collapse, potentially changing optical behavior [Lewis et al., 2009; Ramachandran and Reist, 1995], and the initially hydrophobic BC particle can become associated with hygroscopic materials, leading to increased scattering under humid conditions [Weingartner et al., 1997; Zhang et al., 2008]. To further complicate this picture, the effects of aerosol processing are not necessarily irreversible. For example, semi-volatile aerosol components such as ammonium nitrate, some organics and water thermodynamically partition between the aerosol and gas phases [Myhre et al., 2006]. Capturing these processes in atmospheric models is challenging given the complex dependence of equilibria on aerosol composition and temperature, and as a result they have generally not been included in large-scale climate models [Myhre et al., 2006]. It follows that understanding the complexity of aerosol aging processes and their impacts on optical behavior represents a major challenge from both observational and modeling perspectives [Jacobson, 2001; Myhre et al., 2006].
 A large body of work has examined the evolution of aerosol composition with age, with particular attention in recent years being paid to the formation of secondary organic aerosol (SOA) [e.g., Bahreini et al., 2009; de Gouw and Jimenez, 2009; de Gouw et al., 2008; DeCarlo et al., 2010; Jimenez et al., 2009; Kleinman et al., 2008]. Aerosol mass can, in principle, be used to infer optical behavior indirectly using documented mass absorption and scattering efficiencies [Seinfeld and Pandis, 2006], however this approach fails to capture important complexity associated with the aerosol mixing state and microphysics [Jacobson, 2001; McMurry et al., 1996]. Therefore, direct observations of the impacts of aerosol processing on optical properties are also needed. Recent examples of such work include that of Kleinman et al. , who observed a doubling in the ratio of aerosol light absorption to CO (a conserved dilution tracer) in aged air masses compared to fresh emissions over the eastern U.S. during 2002. This observation was attributed to an increase in the absorption efficiency of BC due to the accumulation of organic and sulphate coatings. Similar increases in absorption by BC-containing aerosol have been reported from ground site measurements performed at a series of locations downwind of Mexico City [Doran et al., 2007]. In a different study, Morgan et al.  used ground and airborne measurements from northwestern Europe to highlight the importance of semi-volatile aerosol components for optical behavior. Partitioning of semi-volatiles to the aerosol phase, in particular ammonium nitrate and water, led to increased aerosol scattering which enhanced the direct radiative forcing by aerosols by over 100%.
 In this work we use airborne observations collected aboard the National Oceanic and Atmospheric Administration (NOAA) WP-3D research aircraft during the California Air Quality and Climate Nexus campaign (CalNex) to investigate the evolution of aerosol optical behavior downwind of Los Angeles (LA). This region provides an interesting environment in which to perform such a study as the LA basin is characterized by both dense urban emissions (e.g., NOx, volatile organic compounds, BC, particulate matter etc.) and significant agricultural and urban ammonia emissions. This enables aging of an urban aerosol influenced by ammonium nitrate, a semi-volatile species, to be examined in detail.
2.1. Study Region
 The 2010 CalNex project was designed to provide the scientific underpinning needed to develop regulatory strategies targeting air quality and climate change remediation concurrently in the state of California. With this aim, a large-scale, multi-institute field mission was conducted during May and June 2010 in which state-of-art instruments were deployed to measure atmospheric constituents and processes from the land, air and sea around California. As part of this activity, the NOAA WP-3D research aircraft conducted 18 research flights based from Ontario, located in the California South Coast Air Basin approximately 55 km east of LA (www.esrl.noaa.gov/csd/calnex). The scientific objectives of the WP-3D flights were wide ranging, spanning from emission inventory measurements to the characterization of cloud-aerosol interactions. To achieve these goals the aircraft was operated over large areas that included California's Central Valley and Mojave Desert. Here, the focus concerns measurements made in and around the LA basin area, shown in Figure 1.
 The population of the LA basin exceeds thirteen million people, with close to four million centered in LA itself (U.S. Census 2010). Significant agricultural activity is also present in the basin, as shown in Figure 1 by the black markers indicating the locations of confined animal feedlot operations associated with the dairy industry [Salas et al., 2009]. Ammonia emitted from dairy cattle adds to urban emissions yielding a chemically complex environment, well suited to the formation of ammonium nitrate aerosol. In addition to chemical complexity, pollutant transport within the basin is also complex. During summer months, synoptic scale circulation is largely dominated by the presence of a semi-permanent high-pressure cell that prevents frontal passage. Therefore, local meteorological processes, such as land-sea breeze circulation and mountain up- and downslope flows dominate [Targino and Noone, 2006 and references therein]. In simplified terms, during the daytime the sea breeze acts to push emissions from the west of the basin northward toward the San Gabriel Mountains and eastward toward Cajon and Banning passes. At night, the land breeze is less efficient at moving pollutants out of the basin due to the heat-island effect and pollutants can therefore accumulate in the shallow nocturnal boundary layer. Day-to-day ventilation of the basin is often relatively inefficient, which has historically contributed to air quality problems in the region.
 The NOAA WP-3D aircraft carried a comprehensive suite of fast-response instruments for the characterization of gaseous and aerosol-phase atmospheric constituents. Only those of relevance to the current study are described here.
 All fuselage-mounted aerosol instruments sampled from a common low turbulence inlet [Wilson et al., 2004] and, with the exception of an optical particle counter for the measurement of coarse mode particle size, downstream of a 2.5 μm aerodynamic diameter impactor. Dry aerosol extinction (RH < 10%) and its relative humidity dependence (RH range 60–95%) were measured using a newly developed 8-channel cavity ringdown spectrometer (CRDS) which operated at wavelengths of 405, 532 and 662 nm [Langridge et al., 2011]. The uncertainty of 10-s average CRDS measurements was <15% for dry aerosol sampled in the polluted conditions experienced in this work. Aerosol absorption (RH < 10%) was measured using a three wavelength (467, 530, 660 nm) Particle Soot Absorption Photometer (PSAP, Radiance Research Inc.). The PSAP data were corrected for filter transmission, flow and sample area effects [Bond et al., 1999; Sheridan et al., 2005] prior to the application of scattering corrections [Virkkula et al., 2005]. The uncertainty of 10-s average PSAP measurements is estimated to be 20–30% [Bond et al., 1999; Lack et al., 2008; Virkkula et al., 2005]. PSAP measurements may also be biased in the presence of organic aerosol [Lack et al., 2008], however corrections to account for this poorly quantified phenomenon were not applied.
 Dry particle size distributions spanning 4–6300 nm were compiled from measurements made using three different instruments. A white light optical particle counter covered particles in the range 550–6300 nm, an ultra-high sensitivity aerosol spectrometer covered 80–1000 nm (Droplet Measurement Technologies, Boulder, Colorado) and an instrument containing five condensation particle counters detected ultrafine particles with diameters below 80 nm [Brock et al., 2008]. Non-refractory (NR) aerosol composition and concentration was measured using a compact time-of-flight aerosol mass spectrometer (AMS, Aerodyne, Billerica, Massachusetts) [Drewnick et al., 2005]. The AMS quantified aerosol ammonium, nitrate, sulphate and organic mass with uncertainties ranging from 34 to 38% [Bahreini et al., 2009]. The upper size cut for particles measured by the AMS was ∼650 nm. The importance of loss of ammonium nitrate aerosol mass due to evaporation in the aircraft inlet and sample lines, which were generally 5–15°C warmer and drier (RH < 30%) than ambient conditions, was evaluated using a condensation model based upon a lognormally distributed aerosol mass distribution [Seinfeld and Pandis, 2006]. A mass distribution representative of that observed during CalNex was used for these calculations (400 nm median diameter, 400 nm full width at half maximum). During the ∼4 s residence time within the aircraft sample lines and AMS pressure controlled inlet, the predicted ammonium nitrate mass loss was <10%. This value may also represent an upper limit due to the potential for slower mass transfer in ambient particles compared to pure compounds [Huffman et al., 2009]. Single particle measurements of refractory black carbon were made using a Single Particle Soot Photometer (SP2, Droplet Measurement Technologies, Boulder, Colorado) [Schwarz et al., 2006]. The SP2 measured black carbon in a mass range covering 70–90% of the accumulation-mode mass distribution and provided 1 s BC mass with a total uncertainty of 40%. In addition, it provided information on the coating state of individual BC particles [Gao et al., 2007; Schwarz et al., 2008].
 A range of gas phase measurements was performed; each is listed here with associated sampling method, sampling time, detection limits and accuracy: CO (vacuum ultraviolet fluorescence, 1 s, 1 ppbv, 5% [Holloway et al., 2000]), NO (chemiluminescence, 1 s, 10 pptv, 3% [Ryerson et al., 2000]), NO2 (chemiluminescence, 1 s, 30 pptv, 4% [Pollack et al., 2010]), total gas phase reactive nitrogen NOy (chemiluminescence, 1 s, 40 pptv, 12% [Ryerson et al., 2000]), HNO3 (chemical ionization mass spectrometry CIMS, 1 s, 12 pptv, 15% [Neuman et al., 2002]), NH3 (CIMS, 1 s, 80 pptv, 25% [Nowak et al., 2007]), peroxyacetylnitrate, PAN (CIMS, 2 s, 5 pptv, 20% [Slusher et al., 2004]), toluene and benzene (proton transfer mass spectrometry, 16 s, 20 pptv, 20% [de Gouw and Warneke, 2007]). Biasing of NOy measurements in the presence of strong ammonium nitrate aerosol loadings was observed during CalNex. For the data subset used in this analysis, NOy measurements were cross-checked by comparison to NOy values calculated from the sum of NO, NO2, HNO3 and PAN. The two data sets agreed to within 6% and therefore, on the grounds of slightly superior data coverage, analysis was performed using the directly measured NOy values. All gas and aerosol data were averaged to a common 10 s resolution prior to analysis.
3.1. Data Selection
 From all flights flown within the LA area, specific days were identified in which transport was dominated by the sea-breeze circulation that acted to transport pollutants eastward through the basin. Further, specific instances where the sea-breeze led to clear spatial trends in photochemical age within the LA plume were selected. These conditions were chosen to enable measurements to be analyzed in a Lagrangian framework. Two flights were performed under appropriate conditions on Sunday 16 May and Sunday 20 June 2010 (the representativeness of using flights performed on Sundays is discussed further in section 6.4). Geographical constraints were applied to the flight tracks to isolate measurements from regions dominated by the urban plume (see Figure 2). This included a narrow area extending from LA northward to the foot of the San Gabriel Mountains and eastward toward Cajon Pass, and an area north of Cajon Pass where emissions were expected to vent from the basin. The southward extent of the area was limited in order to minimize the influence of strong direct ammonia emissions from the dairy facilities located south of Ontario. While part of the motivation for this work is to examine evolution of an urban plume evolving in an ammonia-rich environment, direct ammonia emissions from this concentrated source are considered distinct from the urban plume and were therefore not included in this analysis. The effectiveness of constraints applied to isolate measurements from direct ammonia emissions was verified from observations of total (gas + aerosol) ammonia north of the dairy facilities, which showed little enhancement beyond values in the fresh LA plume (ammonia-rich plumes from these facilities were located further to the south on these days). Finally, only measurements from the boundary layer were used, as the free-troposphere was expected to contain aged emissions from previous days that would interfere with the current analysis. Data collected during profile ascents and descents were used to estimate boundary layer heights, which varied throughout the study region. Accordingly, data collected below heights of 400–700 m (dependent upon location) and 400–630 m above ground level (AGL) were included for 16 May and 20 June respectively. No measurements were made below 235 m AGL. Following the application of all constraints, data contributing to analyses on 16 May and 20 June constituted 30 and 50 min of sampling respectively.
3.2. Markers for Photochemical Aging
 The primary marker used here to track photochemical aging of the LA plume is the ratio NOz (=NOy − NOx) to NOy [e.g., DeCarlo et al., 2010; Kleinman et al., 2007, 2008; Slowik et al., 2011]. This relationship captures the gradual transformation of primary NOx to oxidized products such as HNO3 and PAN that occurs during aging. It yields a ratio close to zero for fresh emissions and close to one for heavily aged air masses. Common concerns regarding its use relate to difficulty associated with assigning it to an absolute time base and its susceptibility to bias from either strong NOx sources located far from the primary emission area or loss of HNO3 during transport. In this work, the NOz/NOy ratio was not used for quantitative purposes where these limitations are most significant, but rather it served as a tool with which to qualitatively track mean plume age, with the understanding that NOx was continually added from the suburban region. Figure 2 shows the flight data that met the selection criteria on 16 May and 20 June, color coded to the ratio NOz/NOy. On each day an increasing trend in NOz/NOy is observed moving eastward through the basin, consistent with the photochemical aging of pollutants expected to occur during transport.
 To provide further evidence for the validity of using NOz/NOy to qualitatively track aging of the LA plume, we compare it to an additional marker of photochemical age: the ratio of toluene to benzene, which characteristically decreases during aging [Roberts et al., 1984; Warneke et al., 2007]. Figure 3 shows the correlation between these two proxies for the flight of the 16 May (similar correlations observed on 20 June are not shown). Data are presented in the form of a box and whisker plot displaying the 10th, 25th, 50th, 75th and 90th percentile values of all data within bins of width NOz/NOy = 0.1. Each data subset contained at least 8, 10 s average measurements. The correlation seen between the two proxies (r2 coefficient of 0.6 for analysis of raw 10 s data) provides further evidence supporting the observation that LA emissions aged during transport from west-to-east in the basin.
3.3. Accounting for Dilution
 Plume dilution and mixing affected observed aerosol extensive properties and was accounted for using CO as a tracer for urban and suburban emissions. This approach is justified given that CO is quasi-conserved chemically over the observation timescales of this experiment (<10 h based upon study area size and wind speeds). To calculate dilution-corrected quantities Δ[X − Xbkg]/Δ[CO − CObkg], background concentrations were evaluated for each study day. The choice of backgrounds was complicated by the potential for mixing to proceed via the entrainment of both boundary layer (BL) and free-troposphere (FT) air, which did not have the same composition. Given the predominantly westerly prevailing wind conditions in the BL on both days, BL background values were derived from flight transects conducted upwind to the southwest of LA. FT values were calculated by averaging all measurements performed in the altitude range 1500–2000 m above ground over the full geographical extent of the study region. The FT was influenced by aged regional emissions on 16 May and long-range transport on 20 June [Neuman et al., 2012]. The resulting mean background values are given in Table 1, together with standard deviations (2σ) describing the spread of data during the averaging periods.
Table 1. Mean Free Troposphere (FT) and Boundary Layer (BL) Background Concentrations Used for the Calculation of Dilution Corrected Aerosol Propertiesa
16 May BL
16 May FT
20 June BL
20 June FT
The 2σ standard deviation values are given in parentheses.
aerosol organic (μgm−3)
aerosol nitrate (μgm−3)
total NR aerosol mass (μgm−3)
black carbon (ngm−3)
 The relative contribution of each background to total plume dilution was determined using water vapor as a tracer for entrainment of FT air into the BL. This approach exploited the observation that water vapor mixing ratios in the BL were significantly higher than those in the FT, and that both FT and background BL mixing ratios were approximately constant across the study region. The fractional contribution of entrainment of FT background air to total dilution was calculated using the mixing parameter α:
where VH2O is the measured water vapor mixing ratio and FTH2O and BLH2O are the constant FT and BL water vapor mixing ratios. For 16 May and 20 June values of FTH2O = 1.5 gkg−1 and BLH2O = 9.5 and 9 gkg−1 respectively were used, which were estimated from vertical profiles conducted throughout the study region. Figure 4 shows median values of α calculated as a function of photochemical age for each day. On 16 May, α is close to 1 for all but the oldest air masses (those beyond Cajon Pass) indicating minimal entrainment of FT air into the BL. Beyond Cajon Pass, α values close to 0.4 indicate significant mixing of FT air into the boundary layer. Similar observations are seen for 20 June, with significant influence from FT entrainment seen only for the oldest air masses beyond Cajon Pass. Trace gas and aerosol background concentrations were determined from calculated α values through:
where XFTbkg and XBLbkg are the FT and BL backgrounds for the species of interest, shown in Table 1. For the analysis of aerosol intensive properties, such as aerosol single scattering albedo (SSA) and relative humidity extinction enhancement, dilution corrections were not required.
Figure 5 shows the variation in CO concentration with photochemical aging during the two study periods. Given the correlation between photochemical age and distance from LA seen in Figure 2, one might have expected CO concentrations to drop off with aging due to the effects of dilution. This trend was observed on 20 June, but not on 16 May where CO increased with transport in the basin and fell only for the oldest samples encountered beyond Cajon Pass. The reason for this behavior is not entirely clear. Nevertheless, on both days peak CO concentrations reached approximately 350 ppbv and minimum concentrations of only 16–23 ppbv above background (solid markers) were measured for the most aged air masses.
4.1. Aerosol Composition
4.1.1. Non-refractory Mass
Figure 6 shows dilution-corrected aerosol organic, nitrate and total mass concentrations measured by the AMS on the two study days. Dilution-corrected quantities were determined from the ratios of enhancements above background. Median values are shown for each age bin together with error bars calculated by propagating 2σ uncertainty in background concentrations (Table 1). We note that despite operating behind a 2.5 μm impactor, the AMS measurements captured primarily sub-0.65 μm particle mass due to limitations associated with transmission of larger particles through the instrument's aerodynamic lens [Bahreini et al., 2008]. The importance of the missing fraction of mass for comparison of chemical data to optical property measurements is examined in section 4.2.1.
 On 16 May a number of general trends were observed; total mass fell, largely in response to loss of ammonium nitrate from the particle phase, and modest growth of secondary organic material was observed. In relative terms, the ammonium nitrate mass fraction fell from ∼44 to 10% and the organic fraction grew from ∼36 to 85%. Based upon observations in the eastern U.S., de Gouw et al.  estimated urban primary d[organic]/d[CO] to be ∼0.01 μgm−3 ppbv−1, which is more than three times below that observed here for the least aged aerosol. Despite the different study environments, this result suggests that a significant fraction of the organic mass measured in the youngest age bin was likely secondary in nature. On 20 June quite different behavior was observed. Total mass increased with aging driven by strong SOA formation, and little formation or loss of ammonium nitrate was observed. Organic mass approximately tripled during the aging period and for the oldest aerosol constituted approximately 81% of the total non-refractory mass. Differences between ammonium nitrate formation and persistence observed on the two study days are examined in section 5.
4.1.2. Black Carbon
 The atmospheric lifetime of BC is controlled by wet scavenging and dry deposition and is typically of the order of one week [Rodhe et al., 1972]. Given that conditions were clear and dry during the two study days, significant loss of BC mass was not expected. In support of this assertion, the measured dilution corrected BC mass, d[BC]/d[CO], did not change significantly. On both 16 May and 20 June, median d[BC]/d[CO] values were constant to within ±20% across all age bins.
 Microphysical properties of BC, namely the number fraction of BC cores identified as thickly coated (coating thicker than ∼60 nm) and the average coating thickness, were calculated from single particle data following the approach of Schwarz et al. . Core-shell Mie theory using refractive indices of 1.95–0.79i (BC cores) [Bond and Bergstrom, 2006] and 1.5–0i (coatings) was used to translate the primary SP2 data products (BC-core mass and total particle scattering cross section) into an estimate of the thickness of non-BC material on the core for the individual particles. For this analysis, a subset of BC-containing particles in the BC mass range 4–6 fg was used (covering ∼15% of the measured mass distribution and corresponding to spherical volume equivalent diameters of 162–185 nm assuming BC density of 1.8 gcm−3 [Bond and Bergstrom, 2006]). This range was chosen to minimize possible biases associated with BC-containing particles with optical sizes outside the range detectable with the SP2 (as operated here). In the BC mass range of 4–6 fg, ∼95% of the BC particles were optically sized. Figure 7 shows these quantities for 16 May (Figure 7, left) and 20 June (Figure 7, right). On both days the fraction of thickly coated particles and the average coating thickness increased with aging, indicative of the accumulation of non-refractory material on the BC cores. Despite the increasing coated particle fraction, over 50% of BC in this mass range remained thinly coated at the level distinguished by the SP2 (∼60 nm) for the most aged particles. We note that this result does not demonstrate bimodality in the BC mixing state per se, as it is determined by the SP2 coating classification criteria used. It does however show that the BC mixing state evolved with aging. Quantitatively similar trends were also seen for other mass ranges examined. The increase in coating thickness observed on 16 May was somewhat surprising given that the AMS showed a fall in total non-refractory mass on this day (see Figure 6). This result suggests that the partitioning of non-refractory material to the BC was not equal for all components measured by the AMS. In particular, since mass loss on 16 May was dominated by ammonium nitrate leaving the particle phase it is likely that ammonium nitrate did not form a significant fraction of the BC coating material (assuming ammonium nitrate partitioning behavior was not modified by the BC substrate). Organics were the only non-refractory components to increase in mass with aging, suggesting that they formed the bulk of the BC coatings.
4.2. Optical Properties
4.2.1. Importance of Coarse Versus Fine Mode
 During the CalNex mission, aerosol optical measurements were made behind a 2.5 μm aerodynamic diameter impactor. These measurements sampled a greater portion of the aerosol population than the AMS composition measurements, which predominantly sampled sub-0.65 μm particle mass. In this section, we examine the importance of this difference with respect to using measured chemical composition to interpret observed optical behavior. Our approach is to model the fraction of total extinction generated from particles with geometric diameters below 2 μm (equivalent to aerodynamic diameters below 2.5 μm assuming spherical particles of density 1.6 gcm−3) and 0.65 μm using Mie theory [Bohren and Huffman, 1983]. The calculations use the full (sub-6.3 μm) size distributions measured aboard the WP-3D and assume particles have the refractive index of ammonium sulphate. This is appropriate given that calibration of the aerosol sizing instrumentation was performed using size selected ammonium sulphate particles.
Figure 8 shows the results of model calculations performed at 532 nm for 16 May. For all age bins, extinction from sub-2 μm diameter particles contributed more than 90% of the total dry particle extinction. Extinction from sub-0.65 μm particles contributed 78–81%. Thus, both the AMS and optical property measurements captured the majority of dry particle extinction predicted by Mie theory for this day. Further, sub-0.65 μm particles contributed 83–90% of sub-2 μm extinction, indicating the validity of comparing AMS composition measurements to optical property data. One factor not taken into account is the potential influence of sea-salt particles, which are not detected by the AMS. A number of factors suggest that sea-salt was not contributing significantly to aerosol measurements on 16 May. First, one would expect a large sea-salt influence to yield significant scattering in the super-μm range, which was not observed. Second, single particle composition measurements made during this period at a ground site in Pasadena (∼50 km west of Ontario) identified only 5–10% of sub-1 μm particles as sea-salt (K. Froyd, personal communication, 2011).
 Similar analyses could not be performed for 20 June as full size distribution data were not available. Therefore we proceed by assuming that composition measurements on 20 June were representative of the aerosol population measured by the optical property instruments. In support of this assumption, extinction Ångström exponents, which are primarily determined by particle size, were broadly similar on 16 May and 20 June, varying from 1.1 to 1.5 throughout the study region. Extinction Ångström exponents for 16 May were calculated from the measured size distributions using Mie theory and were found to agree with the direct measurements to within 16%.
4.2.2. Single Scattering Albedo
 Single scattering albedo, the ratio of light scattering to total light extinction, is among the most important properties controlling aerosol direct climate forcing [McComiskey et al., 2008]. Figure 9 shows the evolution of SSA measured for dry particles at 532 nm during the two study days, which experienced different RH conditions. We present dry particle SSA to enable direct comparison between the two cases, however we note that radiative forcing calculations should consider SSA under ambient RH conditions, which is examined in further detail in section 6.3. On 16 May the dry-particle SSA ranged from 0.98 to 0.99 in the fresh plume and fell to around 0.95 for the oldest air masses. A fall in SSA with aging is somewhat counter-intuitive based upon conventional wisdom that scattering mass increases with aging [Cheng et al., 2009]. The observed trend was however driven by loss of scattering mass as ammonium nitrate partitioned away from the particle phase. On 20 June opposite behavior was observed; SSA increased with aging from ∼0.91 for the fresh plume to ∼0.97 for the most aged aerosol. This trend was driven by the formation of scattering secondary organic material, with little influence from formation or loss of ammonium nitrate.
4.2.3. Extinction Relative Humidity Enhancement
 The extent to which extinction increases as aerosols take up water in response to elevated relative humidity (RH) is an important factor impacting visibility and radiative forcing. A significant body of work has identified the fraction of organic material present as a key determinant of water uptake properties [e.g., Baynard et al., 2006; Garland et al., 2007; Quinn et al., 2005]. Figure 10 shows measurements of the extinction enhancement factor at 85% RH, fRHext(dry, 85%), for the two study days. Similar trends were observed for each day, with aerosol hygroscopicity decreasing with age from around 1.7–1.8 in the fresh plume to 1.4–1.5 for the oldest aerosol. The decreasing hygroscopicity can be well understood as resulting from an increase in aerosol organic mass fraction with age. The factors driving increased organic mass fraction were however different on the two days; on 16 May loss of ammonium nitrate from the particle phase was of primary importance, whereas on 20 June strong SOA formation dominated. While the aging of aerosol organics is often associated with increased hygroscopicity [Rudich et al., 2007], these results provide an example whereby changes to bulk composition by mass transfer to the aerosol phase dominated such effects.
5. Ammonium Nitrate Partitioning
 Of the two days examined in this work, ammonium nitrate partitioning was key to controlling aerosol optical behavior on one (16 May) yet had little influence on the other (20 June). In this section we investigate why differences in ammonium nitrate behavior were observed and identify key factors that affected changes in partitioning throughout the basin area. The comprehensive thermodynamic model Isorropia II [Fountoukis and Nenes, 2007] was used to model the ammonium nitrate system. The model was driven using measurements of total (gas + aerosol) ammonia and nitric acid, aerosol sulphate, temperature and relative humidity. Particles were constrained to exist in the metastable state and results were examined in terms of the fraction of ammonia and nitric acid predicted to exist in the gas phase. The sensitivity to assuming metastable behavior was examined by performing additional runs in which solid and liquid states were allowed to co-exist. The partitioning predicted under each assumption differed by less than 5%.
Figure 11 summarizes ambient measurements of ammonia partitioning observed on each day and correlates observations to temperature, which is an important variable impacting the ammonium nitrate equilibrium [Stelson and Seinfeld, 1982]. Points are color coded to the aging ratio NOz/NOy. On both days temperature varied significantly and systematically in the basin, ranging from around 15°C for the youngest air masses observed near LA to 22–24°C for aged air encountered inland beyond Cajon Pass (relative humidities showed similar trends, ranging from ∼85% in the fresh plume to ∼30% for aged air masses). We note that these trends were not determined solely by the time at which data were collected. For example, on 16 May all data south of Cajon Pass were collected within a 1-h period from 3:30–4:30 P.M. local time and data north of the pass were collected 1.5 h later. On 20 June, data were sampled during a longer period from ∼11 A.M. to 6 P.M., although individual transects through the study region conducted over shorter timescales (<30 min) showed the presence of temperature gradients. Despite similar temperature trends observed on the two days, partitioning behavior of ammonia was quite different. On 16 May temperature appears to have driven ammonia partitioning to the gas phase with transport, which increased from 20% to 80%. In contrast, on 20 June greater than 60% of ammonia was present in the gas phase even for the youngest aerosol present at cooler temperatures close to LA.
 We first address why significant ammonium nitrate formation was not observed on 20 June. Comparing chemical budgets measured on this day to 16 May showed that total (gas + aerosol) ammonia and aerosol sulphate concentrations in the urban plume were very similar, but total nitric acid concentrations were approximately one half. This suggests that ammonium nitrate formation on 20 June could have been limited by the availability of nitric acid. This conclusion was supported by model calculations performed using Isorropia. Figure 12 shows calculated isopleths of ammonium nitrate formation as a function of total (gas + aerosol) ammonia and nitric acid concentrations. Calculations were performed for conditions approximating those encountered during sampling of the youngest air masses on both study days: 288.15 K, 80% RH and a sulphate concentration of 2 μgm−3. Also shown are the total ammonia and nitric acid concentrations measured for the youngest air masses (NOz/NOy < 0.4). These results confirm that ammonium nitrate formation in the urban plume on 20 June was limited by the availability of nitric acid.
 Second we examine how well Isorropia captured the dynamic partitioning observed. Starting with the 16 May case, Figure 13 (colored points) compares modeled and measured partitioning of ammonia and nitric acid and shows good agreement both in terms of observed magnitudes and trends. In order to further investigate the dependence of partitioning on temperature (and indirectly relative humidity) seen in Figure 11, additional model runs were performed in which the temperature was fixed to values measured for the youngest air masses (gray points). In these runs, ammonium nitrate partitioning to the gas phase with aging was less pronounced, illustrating the sensitivity of the ammonium nitrate system to temperature. The partitioning trends that remained provide an indication of how dilution alone acted to drive ammonium nitrate from the particle phase. Also shown in Figure 13 are the results of similar analyses performed for 20 June. As for 16 May, Isorropia reproduced the observed partitioning well on this day. The results of fixed temperature simulations also showed markedly different partitioning behavior to that under ambient conditions, indicating that temperature gradients present throughout the study region contributed to observed partitioning behavior on this day.
 In summary, we conclude that the formation and partitioning of ammonium nitrate in the LA urban plume results from complex interplay between chemical availability and meteorological conditions. Dynamic evolution of ammonium nitrate with aging can be observed and can be represented well using a state-of-the-art thermodynamic partitioning model.
6.1. Importance for Health
 Among the most significant impacts of air pollution are those affecting human health. Long-term exposure to particulates, particularly those below 2.5 μm in diameter (PM2.5), has been linked to premature mortality and impaired development of lung functionality in children [Gauderman et al., 2002; Pope et al., 1995; Schwartz, 2004]. Elevated PM2.5 concentrations are also known to exacerbate asthma [Meng et al., 2010], with particle hygroscopicity being an important factor determining the deposition efficiency of particles in the respiratory system [Mitsakou et al., 2005]. Evaluating health effects requires careful consideration of population exposure to total particle loadings, which is beyond the scope of this work. However, the aerosol trends observed here do highlight some important considerations for those interested in modeling total particle mass concentrations for the purpose of examining health impacts. In section 4.1.1 the variation in dilution-corrected dry-aerosol mass with aging was presented. It showed a decrease in aerosol mass of 50% on one measurement day (16 May) driven by loss of ammonium nitrate from the particle phase and an increase of 117% on the second day (20 June) driven by strong SOA generation. In addition, strong trends in aerosol hygroscopicity with aging on both days were shown in section 4.2.3. All of these trends were readily understood in terms of fundamental aerosol processes driving mass transfer between the aerosol and gas phases. The sensitivity of these processes to detailed chemical and physical conditions, together with their observed importance, indicates the need for their inclusion in chemical transport models used to generate PM2.5 fields for evaluating local-scale particle health effects.
6.2. Importance for Visibility
 Visibility impairment due to aerosol light extinction is among the most striking impacts of air pollution. It is not only of concern in densely populated urban environments, but also impacts rural locations, as illustrated by the U.S. Environmental Protection Agency's Regional Haze Program (http://www.epa.gov/visibility/program.html) which specifically targets visibility improvements in national park and wilderness areas. In this section, we explore how changes in aerosol optical properties during aging of the LA plume impacted visibility. This analysis does not attempt to provide an overarching description of factors controlling visibility in this region, which would require increased spatial and temporal data sampling, but rather aims to illustrate how changes observed during aging modified the ability of aerosol to impact visibility. Given this purpose, data analysis is performed for 16 May only, which provides an interesting illustrative case of evolving aerosol composition linked to ammonium nitrate partitioning (section 5) impacting aerosol optical behavior.
 Aerosol extinction is a key factor controlling visibility [Koschmieder, 1926] and therefore the approach used here examines how optical extinction and its RH dependence changed with aging. Although direct measurements of extinction and fRH were made aboard the WP-3D during CalNex, this analysis instead uses Mie theory to model extinction based upon measured size distributions. This constrained model-based approach was adopted to enable the importance of individual aerosol components, such as ammonium nitrate, to be explored, which would not have been possible from the extinction measurements alone. The aerosol refractive indices used for these calculations were determined from volume-weighting assumed refractive indices [e.g., Chyýlek et al., 1988] for the speciated aerosol components (ammonium nitrate: 1.6–0i [Weast, 1985], ammonium sulphate: 1.53–0i [Toon et al., 1976], organics: 1.63–0.02i [Dinar et al., 2008], BC: 1.95–0.79i [Bond and Bergstrom, 2006] and water: 1.33–0i [Lide, 2001]) using the following densities: ammonium nitrate: 1.72 gcm−3 [Lide, 2001], ammonium sulphate: 1.77 gcm−3 [Lide, 2001], organics: 1.4 gcm−3 [Alfarra et al., 2006] and BC: 1.8 gcm−3 [Bond and Bergstrom, 2006]. Kappa Köhler theory [Petters and Kreidenweis, 2007] was used to model particle growth, under the assumption of continuous water uptake (i.e., that particles had deliquesced). The kappa parameter is directly relatable to the hygroscopic growth factor, which was the necessary input for extinction calculations at elevated relative humidity. Kappa Köhler theory has been shown to represent water uptake in multicomponent systems well at sub-saturated conditions [Kreidenweis et al., 2008]. Effective kappa values were determined by volume-weighting kappas assigned to each of the AMS components (ammonium nitrate: 0.59, ammonium sulphate: 0.53, organics 0.01) and BC (0). A number of assumptions were implicit to the calculations, including full internal mixtures and spherical particles. Although certainly not correct at a fundamental level, their use was justified by comparing modeled and measured dry particle extinction and fRHext(dry, 85%) values which agreed to within 17 ± 6% and 4 ± 3% respectively.
Figure 14a shows a series of extinction calculations, each referenced to the magnitude of extinction calculated for dry aerosol of ambient composition. Calculations were performed for a wavelength of 532 nm. One run shows the extinction predicted for dry aerosol following removal of all ammonium nitrate from the particle phase. These calculations included the effects of changes to the particle size distribution, which were simulated assuming that all particles experienced the same fractional volume loss upon removal of ammonium nitrate. The results show the extinction to fall by up to 50% in the fresh urban plume, illustrating the significant contribution of ammonium nitrate to dry particle extinction. Figure 14a also shows the extinction relative humidity enhancement calculated at a series of elevated relative humidities. As expected from measurements shown in section 4.2.3, the enhancement of extinction due to water uptake was most significant in the fresh plume where aerosol was most hygroscopic. It acted to more than double the dry particle extinction at 90% RH. To explore further the importance of ammonium nitrate for water uptake, Figure 14b shows the fraction of enhanced extinction at elevated humidity due to water associated directly with ammonium nitrate. These calculations were performed by differencing the results of two runs; one under ambient conditions and one in which ammonium nitrate water uptake was prohibited by setting its kappa value to zero. The fractional contribution was then calculated by normalizing these values to the difference in ambient aerosol extinction under wet and dry conditions. The results indicate that, regardless of specific relative humidity, approximately 60–80% of additional extinction at high RH in the young LA urban plume was due to water associated with ammonium nitrate. We note that these estimates also likely represent a lower limit given that the shift in ammonium nitrate partitioning to the particle phase with increasing RH was not considered.
 Collectively these results show the important contribution of semi-volatile species ammonium nitrate and water to aerosol extinction, and therefore visibility, in the LA plume. They also show that the contribution of these species to extinction can be highly variable over relatively small spatial and temporal scales. Thus, in addition to highlighting the need for inclusion of processes involving semi-volatiles in atmospheric models used for visibility analysis, these results also show that models must be run at sufficiently high spatial resolution.
6.3. Importance for Direct Radiative Forcing
 Quantifying the direct impacts of aerosol particles on the climate system has been a research topic at the forefront of climate science for a number of years. Despite significant progress, uncertainty linking aerosol optical properties to radiative forcing still limits the accuracy of climate prediction in large scale models [IPCC, 2007]. One challenge is connecting aerosol mass loadings to optical behavior and capturing changes that occur to each throughout an aerosol's lifecycle. In this section, simplified calculations of aerosol direct radiative forcing are performed to evaluate the climate implications of observed changes to aerosol optical properties in the LA plume. Given the limited spatial extent of the current study, this analysis is not intended to constrain absolute forcing magnitudes, but rather aims to indicate more generally how fundamental changes in aerosol properties can impact climate forcing.
 Calculations of instantaneous top of the atmosphere radiative forcing efficiency ΔFeff, which represents the absolute radiative forcing per unit aerosol optical depth, were performed using the simplified equation of Haywood and Shine  for an optically thin, globally uniform partially absorbing aerosol layer present in the lower troposphere:
 Assumed values were used for the solar constant S0 = 1366 Wm−2, atmospheric transmission T = 0.76, cloud fraction Ac = 0 (clear sky conditions) and surface albedo R = 0.14. This albedo represents the mean value derived for the California South Coast Air Basin from satellite observations by Taha . The upscatter fraction (β) was calculated from measured sub-2 μm size distributions by first calculating the hemispheric backscatter fraction (b) using Mie theory and then applying equation (4), a parameterized relationship linking backscatter to upscatter fractions based on the Henyey-Greenstein phase function [Anderson et al., 1999]:
 The single scattering albedo (SSA) was calculated by combining 532 nm absorption measurements with scattering calculations performed using Mie theory. A calculation-based approach was employed, as in section 6.2, to enable the contribution of individual aerosol components impacting scattering, such as ammonium nitrate, to be evaluated. For the purpose of calculating single scattering albedo, aerosol absorption was assumed to be independent of relative humidity. This assumption has been shown to introduce negligible error to SSA estimates, largely due to the significantly stronger dependence of scattering than absorption on RH [Nessler et al., 2005].
 A range of calculations was performed to evaluate the sensitivity of forcing efficiency to aerosol aging and relative humidity changes, with specific attention given to evaluating the importance of ammonium nitrate. Forcing efficiency was calculated in place of absolute radiative forcing to facilitate simple comparison between cases. However, changes in aerosol optical depth for different simulations were incorporated in a relative sense by normalizing all forcing efficiency estimates to optical depth calculated for common conditions, chosen to be dry aerosol of ambient composition. This was accomplished by scaling the product of equation (3) by the ratio of aerosol extinction calculated for the specific simulation to that calculated for dry aerosol of ambient composition. This approach is valid given that aerosol optical depth scales linearly with extinction for a fixed aerosol population. The net impact of the scaling term was to yield a forcing efficiency that provided a relative indication of the how changes to SSA, β and aerosol optical depth impacted radiative forcing under different scenarios of humidity and aerosol composition (e.g., with and without aerosol nitrate included).
 As in section 6.2, calculations were performed for 16 May only. All calculations were performed at 532 nm. Figure 15a shows the forcing efficiency for dry aerosol as a function of chemical age. For aerosol of ambient composition (red) relatively little change (<5%) was seen with aging. This result, which is rather surprising given the strong trends in SSA shown in section 4.2.2, arose from competition between two opposing factors; a decreasing SSA that acted to decrease the forcing magnitude and an increasing upscatter fraction (from ∼0.18 to ∼0.21) that acted to increase the forcing magnitude. Also shown in panel a) are the radiative forcing efficiencies calculated assuming (1) ammonium nitrate mass loadings were conserved with aging to values representative of the youngest aerosol (blue) and (2) no ammonium nitrate formation occurred within or out of the basin (black). These simulations show that ammonium nitrate accounted for approximately half of the radiative forcing measured in the fresh LA plume. In addition, they show that failing to account for loss of ammonium nitrate from the particle phase with transport could lead to overestimation of the radiative cooling imparted by the most aged aerosol by approximately 35%.
Figure 15b shows the results of similar calculations performed at 80% relative humidity. Extinction relative humidity enhancement factors were calculated as described in section 6.2 and did not include the effects of increased partitioning of ammonium nitrate to the aerosol phase with increasing RH. The magnitude of the forcing efficiency calculated for ambient composition aerosol was larger than under dry conditions. This was due to added scattering from liquid water associated with the particles, which increased the aerosol optical depth with respect to dry conditions. The forcing efficiency also showed a more pronounced aging trend due to the importance of ammonium nitrate for water uptake. In order to investigate this further, an additional simulation was performed in which no water uptake by ammonium nitrate was allowed (gray). It showed that approximately 1/4 of the forcing in the fresh LA urban plume was due to liquid water associated with ammonium nitrate.
 Despite the simplicity of these calculations and the wide range of assumptions inherent to their use, the results serve to illustrate the general importance of capturing aerosol evolutionary trends, and specifically the behavior of semi-volatiles, for accurate representation of direct radiative forcing effects.
 Results presented in this section have provided an isolated snapshot of the importance of capturing the evolution of aerosol properties when considering health, visibility and radiative impacts of aerosol in the LA urban plume. They have also highlighted the important role of ammonium nitrate. It should be noted however that these conclusions were drawn from comprehensive analysis of data collected over a small portion of the LA basin during one day only (16 May) and should therefore not be assumed generally representative of the LA plume. Indeed, analysis of a second day (20 June) showed a significantly reduced role of ammonium nitrate, together with enhanced importance of organic formation. Although not examined in detail, changes to aerosol optical and hygroscopic properties observed on this day would also have acted to modify visibility and radiative forcing impacts. Differences between these days serve to underscore the complexity associated with capturing the dynamic behavior of semi-volatile aerosol components, whereby subtle changes in chemical and meteorological conditions can be of critical importance and lead to significant spatial and temporal variability in observed behavior.
 We also note that both measurement days examined in this analysis were Sundays and differences in ammonium nitrate and/or SOA formation during weekdays may be expected. Diesel truck-derived NOx emissions in California fall significantly at weekends, which drives faster oxidation chemistry that is known to enhance ozone formation [Harley et al., 2005; Pollack et al., 2012]. Impacts of these changes on ammonium nitrate and SOA formation have not been so clearly established. During summer and fall of 1995, Harley et al.  found no evidence for a weekend effect in either organic aerosol or ammonium nitrate across 5 sites in Southern California. Similarly, during CalNex, Bahreini et al.  found similar levels of organic aerosol in air masses of similar photochemical age during both week and weekend days in the LA basin. In contrast, during a recent study focused specifically on weekly trends in ammonium nitrate, measurements at a site in Claremont (∼8 km northwest of Ontario, California) reported a minimum in ammonium nitrate mass on Mondays of approximately 29 ± 23% below the weekly average [Millstein et al., 2008]. Although spatial and temporal behaviors may be different on weekdays, these findings suggest that the importance attributed to ammonium nitrate on Sunday 16 May was not an artifact of weekend sampling.
 The evolution of aerosol properties with aging in the LA plume was shown to have important consequences for visibility and direct climate forcing. Analyses were conducted using data collected from the highly instrumented NOAA WP-3D research aircraft, which participated in the California climate and air quality nexus project CalNex during May and June 2010. Two case studies were examined for flights conducted over the LA basin on 16 May and 20 June. On 16 May strong aging trends were observed, driven primarily by loss of ammonium nitrate from the particle phase resulting from the combined effects of dilution and temperature gradients throughout the LA basin region. Aerosol became optically darker with aging due to the loss of scattering material, and also became increasingly hydrophobic as composition became increasingly dominated by organics. On 20 June ammonium nitrate formation was suppressed by lower nitric acid concentrations and thus it contributed less to the aerosol evolutionary behavior observed. This day was characterized by strong SOA formation, which led to an increase in single scattering albedo with aging due to the accumulation of scattering mass and a decrease in hygroscopicity due to the rising aerosol organic fraction. No evidence was seen on either day for net re-evaporation of organics from the aerosol phase, despite increasing temperatures that accompanied aging in the basin. While these data do not discount the possibility that evaporation of semi-volatile organics occurred, they do show that the rate of condensation exceeded that of evaporation during the study periods. On both days black carbon trends were characterized by increases in both the fraction of particles with coatings and the mean thickness of coatings. Comparison of these trends to those seen for non-refractory aerosol components indicated that organics were likely the main species contributing to BC coatings. Collectively, these results highlight the complexity and variability of processes acting to modify aerosol properties with aging and specifically show the importance of mechanisms leading to mass transfer between gas and aerosol phases.
 For the 16 May case, further calculations were performed to assess the importance of observed changes for local visibility and radiative forcing. From a visibility perspective, ammonium nitrate was responsible for up to 50% of dry extinction in the fresh LA plume, but was of little importance for the most aged aerosol measured. In addition, of the enhanced extinction experienced under elevated relative humidity conditions in the fresh plume, 60–80% was from water associated directly with ammonium nitrate. Ammonium nitrate was also of significant importance for radiative forcing. Under dry conditions it accounted for approximately half of the radiative cooling exerted by the least-aged LA aerosol measured. Failing to account for subsequent loss of ammonium nitrate during aging would lead to overestimation of the radiative cooling exerted by the most aged aerosol measured by approximately 35%.
 Despite their limited extent, these observations and calculations indicate the importance of capturing the evolution of aerosol optical behavior for accurate assessment of visibility and direct climate impacts. Within the LA urban plume, ammonium nitrate formation was identified as an important factor controlling aerosol optical behavior. These conclusions were largely inferred from data collected near the surface in the boundary layer, but similar phenomena regarding the partitioning of semi-volatiles occurs throughout the vertical extent of the atmosphere [Neuman et al., 2003] and has been shown to be important for radiative forcing [Morgan et al., 2010]. The importance of ammonium nitrate also extends beyond this region of California. Indeed, under polluted conditions it forms a dominant fraction of aerosol mass over large areas of northwestern Europe [Putaud et al., 2004]. These results provide a detailed example of the need for state-of-the-art atmospheric models, including those used for regulatory purposes, to capture both the formation of secondary organic and inorganic material and the gas-aerosol partitioning of semi-volatiles, such as ammonium nitrate and water. Inclusion of the latter of these processes has traditionally proven challenging from a large-scale modeling perspective, but represents an important area for future development.
 This work was supported by NOAA climate and air quality base funding. We thank the flight and ground crews of the NOAA WP-3D aircraft for successfully supporting operations during CalNex. J.M.L. thanks CIRES for the award of a visiting postdoctoral fellowship.