airborne analysis of the los angeles aerosol159.226.119.84/collins2000.pdf · airborne analysis of...

Atmospheric Environment 34 (2000) 4155}4173

Airborne analysis of the Los Angeles aerosol

D.R. Collins!, H.H. Jonsson", H. Liao!, R.C. Flagan!, J.H. Seinfeld!,*,K.J. Noone#, S.V. Hering$

!Department of Chemical Engineering, California Institute of Technology, Mail code 210-41, 1200 E. Calfornia Blvd.,Pasadena, CA 91125, USA

"Naval Postgraduate School, Monterey, CA, USA#Stockholm University, Sweden

$Aerosol Dynamics, Inc., Berkeley CA, USA

Received 17 December 1999; received in revised form 22 March 2000; accepted 30 March 2000

Abstract

As part of the Southern California ozone study (SCOS), a research aircraft was employed during August andSeptember of 1997 to characterize the physical and chemical properties of the aerosol present over the Los Angeles Basin.Aerosol size distributions measured using a di!erential mobility analyzer and two optical particle counters werecombined with "lter-based composition measurements to derive a physicochemical description of the aerosol sampled.The accuracy of this description was evaluated through comparison of derived and directly measured aerosol propertiesincluding mass, absorption coe$cient, hemispherical backscattering coe$cient, and total scattering coe$cient at twodi!erent relative humidities. The sampled aerosol exhibited a complex vertical structure possessing multiple elevatedaerosol layers. The most pronounced of these layers were observed to form by injection of aerosol above the ground-levelmixed layer along the southern edge of the San Gabriel Mountains, which form the northern boundary of much of theLos Angeles Basin. Over multiple inland areas, additional layers were observed at about 2500 m above sea level (asl),while o! the coast of Santa Monica, thin but concentrated layers were detected about 500 m asl. In addition to the sharpvertical gradients in aerosol concentration observed, horizontal gradients at multiple locations were found to be su$cientto result in more than 50% variability within a 5]5 km computational grid cell commonly used in atmospheric models.Vertically resolved aerosol measurements made over one location during several #ights, as well as over several locationsduring a morning and afternoon #ight on the same day, were used to investigate the temporally and spatially resolvedimpact the aerosol had on gas-phase photolysis rates. These calculations predict that for a 103 zenith angle the sampledaerosol enhanced photolysis rates by up to about 5%, although a slight decrease was often observed near groundlevel. ( 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Aerosol sampling; Aircraft sampling; Southern California; Photolysis rates

1. Introduction

Southern California has long struggled to comply withstate and federal air quality standards. Faced witha steadily increasing population and tightening ozoneand particulate matter standards, further emissions re-ductions will be necessary. Extensive atmospheric

*Corresponding author. Fax: #1-626-568-8743.E-mail address: [email protected] (J.H. Seinfeld).

modeling e!orts have provided a means for linking speci-"c emission control scenarios with probable air qualityoutcomes (see, for example, Meng et al., 1997). However,the complex terrain and meteorology associated withSouthern California, coupled with inherent uncertaintiesin model input "elds, complicates these e!orts. Aircraft-based measurements (Blumenthal et al., 1978; Wakimotoand McElroy, 1986) have demonstrated that verticaltransport in the Los Angeles area is not consistent withthe simple representation of a mixed layer trapped belowa temperature inversion. Recently, three-dimensional

1352-2310/00/$ - see front matter ( 2000 Elsevier Science Ltd. All rights reserved.PII: S 1 3 5 2 - 2 3 1 0 ( 0 0 ) 0 0 2 2 5 - 9

meteorological models have provided further insight intomechanisms responsible for formation of distinct pollu-tion layers that exist above the Los Angeles Basin forextended periods of time (Lu and Turco, 1994, 1995).Successful prediction of ground-level concentrations canonly be accomplished if the behavior and nature of ma-terial aloft are adequately described. In addition to thepoorly characterized variation of species concentrationswith altitude, strong horizontal gradients pose a uniqueproblem for models; observed variations within a com-putational grid cell (typically, 5]5 km) call into questionthe assumption that concentrations are uniform insidea grid volume. Use of a single measurement within a gridcell to represent the entire cell can lead to uncertainties ashigh as 25}45%, depending on the species of interest(McNair et al., 1996).

During the summer and fall of 1987, the SouthernCalifornia Air Quality Study (SCAQS) was undertakento provide a su$ciently detailed data set to test thecapabilities of available meteorological and air qualitymodels. Data from SCAQS, particularly a small numberof multi-day episodes, have served as the foundation formost of the modeling work in the years since (Pandis etal., 1992, 1993; Harley et al., 1993; Jacobson, 1997;Lurmann et al., 1997; Lu et al., 1997; Meng et al., 1998).Ten years after SCAQS, during the summer and fall of1997, the Southern California Ozone Study (SCOS) wasconducted in order to supplement the data set acquiredduring SCAQS by employing a network of emissions,meteorological, and air quality measurements. The statedgoals of SCOS were:

1. Update and improve the existing aerometric and emis-sion databases and model applications for repres-enting urban-scale ozone episodes in SouthernCalifornia.

2. Quantify the contributions of ozone generated fromemissions in one Southern California air basin tofederal and state ozone standard exceedances in neigh-boring air basins.

3. Apply modeling and data analysis methods to designregional ozone attainment strategies.

To satisfy goal 2, the study covered an extensive regionthat was roughly bounded by the Channel Islands to thewest, the San Joaquin Valley to the north, and the Cali-fornia state border to the east and south. In all,the study encompassed approximately 53,000 square miles,although a large fraction of the measurements were madewithin the Los Angeles Basin. SCOS was conductedbetween mid-June and mid-October when the highestpollutant concentrations are usually observed in the re-gion. However, during this campaign unusually highsea-surface temperatures caused by a signi"cant El Nin8 oevent resulted in deeper marine layers and enhanced mix-ing relative to typical conditions in Southern California.

Hence, air pollutant concentrations tended to be some-what lower during this study than is common for theregion.

As its name implies, the Southern California OzoneStudy was conducted primarily to improve our under-standing of ozone and its gas-phase precursors. Asmaller, but still extensive, component of the study focusedon understanding the formation and evolution of the atmo-spheric aerosol. The expanded monitoring network inplace for the ozone study facilitated interpretation ofaerosol measurements. As part of the aerosol componentof SCOS, a research aircraft was utilized during Augustand September of 1997 to provide a three-dimensionalcharacterization of the Los Angeles aerosol. This paperpresents a description of the measurements obtained bythat aircraft during the study period, and an analysis ofthe impact the aerosol column may have on key gas-phase photolysis rates involved in ozone production.

2. Instrumentation and analysis

Between 27 August and 12 September 1997 the Centerfor Interdisciplinary Remotely-Piloted Aircraft Studies(CIRPAS) Pelican aircraft #ew 12 missions over the LosAngeles Basin. The Pelican is a Cessna Skymaster thathas been modi"ed by replacing the front engine with anextended nose cowling that serves as the primary instru-mentation bay. The resulting pusher con"guration isideal for aerosol measurements because it minimizes thedisturbance of the atmosphere prior to sampling. Opticalparticle counters were mounted on each of the wings,while the remaining instruments were located inside theaircraft. These latter instruments sampled from two par-allel inlets, each of which included a cyclone with a nom-inal cut size of 2.5 lm. Since the cyclones were positionedimmediately downstream of the inlet opening, little dry-ing of the particles was likely prior to removal of thecoarse mode. Table 1 contains a complete list of instru-ments on board the Pelican during SCOS.

2.1. Filter samples

The chemical composition of the aerosol was probedusing a "lter system employing three parallel samplingtrains, each having a #ow rate of 24 l min~1. Separatetrains were used for analysis of trace metals, elementaland organic carbon, and inorganic ions. Each samplingtrain consisted of three identical "lter cassettes, allowingsequential exposure of multiple "lters through the use ofcockpit-activated solenoid valves that were in-line witheach of the "lter cassettes. During SCOS, three sets of"lter samples were taken during each of the "rst six#ights, with two sets taken on the seventh #ight. Filterswere unloaded immediately following each #ight andremained refrigerated during storage and transport until

4156 D.R. Collins et al. / Atmospheric Environment 34 (2000) 4155}4173

Table 1Measurements made on board the Pelican

Property measured Size range Instrument Time

Aerosol measurementsInorganic ions, trace metals, EC/OC,

gravimetric mass(2.5 lm Filter samplers &1 h

Particle size 0.01}0.5 lm TSI 3071 DMA and TSI 3010 CPC 1 minParticle size 0.15}3.0 lm PMS PCASP 1 sParticle size 0.5}20 lm PMS FSSP 1 s3-color light scattering coe$cient (total

and hemispherical backscattering)(2.5 lm TSI 3563 nephelometer 1 s

Light scattering/RH relationship (2.5 lm UW Humidigraph (Radiance Research nephelometers) 6 s

Radiation measurementsTotal solar Eppley pyranometer 1 sUV Eppley radiometer 1 s

Meteorological measurementsPressure (static and dynamic), temperature,and dew point

Various 1 s

Table 2Detection limits for species analyzed

Species Analyticallimit(lg "lter~1)

Ambientconcentration(lg m~3)

SO2~4

, NO~3

, NH`4

0.8 0.4OC 2 1EC 0.5 0.3Fe, Si, K 0.1 0.05Na 0.01 0.005Mg, Ni, Cu, Zn, Al 0.001 0.0005Mass 10 5

analysis. Several "lter blanks were taken from eachsampling train and were used to correct for the samplesanalyzed.

The "rst sampling train used a Te#on "lter to collectsamples for trace metal analysis. Inductively coupledplasma mass spectrometry (ICP-MS) was used to quan-tify Na`, Al3`, K`, Ni2`, Zn2`, Mg2`, Si4`, Fe3`, andCu2` for each of the "lters. With the exception of Na`

and Mg2` that were assumed to be associated withsea-salt particles, each of the metals was assumed to bepresent as oxides (Al

2O

3, SiO

2, K

2O, Fe

2O

3, NiO,

CuO, and ZnO) for determination of aerosol mass. Un-certainty in total aerosol mass introduced as a result ofthis assumption is relatively low since the elements ana-lyzed typically constituted only a small fraction of theaerosol.

Elemental and organic carbon collected on quartz"lters in the second sampling train was analyzed throughthermal optical re#ectance (Chow et al., 1993) by re-searchers at the Desert Research Institute (DRI). Toaccount for the positive artifact resulting from organicvapor adsorption on the "lter, a quartz back-up "lter wasused, with the corrected organic aerosol mass assumed tobe the di!erence between that measured on the front andback "lters. Since oxygen and hydrogen are not detectedin this type of analysis, organic carbon loadings weremultiplied by 1.4 to estimate total organic aerosol mass(White, 1990).

The "nal sampling train employed an MgO denuder toremove nitric acid in order to minimize the positiveartifact that it might otherwise cause. Downstream of thedenuder, a Te#on "lter was used to collect aerosol forinorganic ion analysis. Researchers with DRI used ionchromatography to analyze SO2~

4and NO~

3, and

colorimetry to quantify NH`4. The NO~

3mass measured

on the nylon back-up "lter was assumed to be volatilizedammonium nitrate, and was, therefore, used to correctthe NH`

4and NO~

3mass measured on the Te#on "lter.

To assess the degree to which the aerosol mass wasaccounted for through the speciation analyses, the Te#on"lters collected were analyzed by gravimetry. These ana-lyses were conducted by researchers at DRI under con-trolled conditions in which the relative humidity wasmaintained at between 45 and 55%. The correction ap-plied to the NH`

4and NO~

3concentrations to account

for ammonium nitrate volatilization was also applied tothe gravimetric mass.

Detection limits for each aerosol species analyzed arelisted in Table 2. Aerosol mass for each of the 20 sets of"lters analyzed is shown in Fig. 1. With the exception ofthe #ights of 4 and 5 September, "lter samples representthe integration of the aerosol present over a relativelywide area and over about 2 km variation in altitude.

D.R. Collins et al. / Atmospheric Environment 34 (2000) 4155}4173 4157

Fig. 1. Fine (D1(2.5 lm) aerosol chemical composition during the seven #ights for which samples were collected. Also shown is

aerosol mass determined gravimetrically, and through integration of the size distributions.

These spatially unresolved "lter measurements wereused to gain a general understanding of the chemicalnature of the aerosol and to aid in the analysis oftime-resolved size distributions, but were not used todirectly assess variations in composition within the studyarea.

2.2. Optical measurements

Three integrating nephelometers and an absorptionphotometer provided details of the optical and hygro-scopic properties of the aerosol sampled during SCOS.The University of Washington Passive Humidigraph,which consists of two Radiance Research nephelometersoperated in parallel at di!erent relative humidities (RH),was used to investigate the relationship between humid-ity and aerosol light scattering. In general, humiditieswithin the dry and wet nephelometers bracketed theambient RH. A TSI 3563 nephelometer that measurestotal scattering and hemispherical backscattering at blue(450 nm), green (550 nm), and red (700 nm) wavelengths

was operated at a humidity slightly below that of thedrier of the two Radiance Research nephelometers. Eachof the nephelometers was calibrated prior to, or during,the study using gases with known scattering properties.

A Radiance Research particle soot absorption photo-meter (PSAP) was used to make time-resolved measure-ments of the absorption properties of the dry aerosol.The PSAP continuously records transmittance of light(567 nm) through two regions on a "lter, one of which hasan aerosol-laden #ow passing through it, while the otheris used as a reference. The aerosol absorption coe$cient,p!1

, is determined from the rate of change of the transmit-tance ratio between the two regions. PSAP calibrationresults described by Bond et al. (1999) were used toanalyze the data obtained. It has been observed thatvariations in transmittance ratio unrelated to increasedparticle loading occur as a result of pressure variationswithin the instrument. Only during constant altitude legswhen the observed change in transmittance due to par-ticle loading dominated over that caused by pressurevariations were data analyzed.


2.3. Size distribution measurements

By integrating measurements from a di!erential mobil-ity analyzer (DMA) and two optical particle counters(OPCs), aerosol size distributions spanning the rangefrom 10 nm to '20 lm in diameter were determinedwith 1 min time resolution. A TSI 3071 cylindrical DMAwas operated with a TSI 3010 condensation particlecounter to size particles with diameters between approx-imately 10 and 500 nm. Each of the DMA #ow rates wasactively controlled to minimize #uctuations induced bypressure changes within the instrument. The voltage ap-plied to the DMA was scanned over 45 s, and was auto-matically adjusted to maximize particle size range, whileremaining below the pressure- and temperature-dependent threshold for electrostatic breakdown.A wing-mounted Particle Measuring Systems (PMS)passive cavity aerosol spectrometer probe (PCASP-100]) sized particles ranging from approximately140 nm to over 3 lm. The PCASP was calibrated prior toSCOS with polystyrene latex (PSL) particles. Mie Theorycalculations that accounted for the angular con"gurationof the light receptor in the PCASP were used to adjustthis PSL calibration to that expected for particles of anyspeci"ed complex index of refraction. Mounted on thewing opposite to the PCASP was a PMS forward scatter-ing spectrometer probe (FSSP-100). By utilizing anexternal laser/detector, the FSSP causes only minimaldisturbance to large particles and cloud droplets. The 1 Hzmeasurements of both the PCASP and FSSP were aver-aged over the &1 min measurement time of the DMA.

The PCASP and DMA systems both change the rela-tive humidity of the aerosol during the act of sampling,while the FSSP has little e!ect on this important para-meter. To combine these measurements into a coherentdescription of the sampled aerosol, the ambient particlesize distribution must be estimated from the perturbeddata. An accurate description of the e!ect of humidity onparticle size requires knowledge of, or assumptionsabout, the aerosol composition, mixing state, and deli-quescence state. Filter samples taken during the majorityof the Pelican #ights were used to estimate the composi-tion of the sampled aerosol during the respective "lterintegration interval. For those #ights during whichno "lter samples were taken, the average compositionanalyzed during the #ights with "lter measurementswas used. Since "lter samples were not size-resolved, itwas assumed that size-dependent composition andmixing relationships were similar to those described byZhang et al. (1993) in their analysis of data taken duringSCAQS. External mixtures of salt particles were assumedso that available single-salt solution thermodynamicsdata (Tang and Munkelwitz, 1994; Tang, 1996; Tang etal., 1997) could be used for particle growth/evaporationcalculations. Tang (1996) showed that only minor di!er-ences exist in overall optical properties of internal and

external mixtures of common nitrates and sulfates. More-over, the similar densities of common nitrates and sul-fates limit the uncertainty in the calculated aerosol massthat might be caused by this assumption. Hygroscopicgrowth was considered only for the salt species, and notfor carbon or dust. The assumption that carbonaceousaerosol is non-hygroscopic is known to be incorrect forat least some organic species found in the aerosol phase(Saxena et al., 1995). However, in their interpretation ofSCAQS impactor data, Zhang et al. (1993) suggest thatexternally mixed carbonaceous particles were non-hy-groscopic, while carbon internally mixed with saltparticles was less hygroscopic than the salt itself. Therelatively low humidity encountered throughout thepresent study, and generally characteristic of the region,further limits uncertainties caused by this assumption.

The ambient relative humidity measured during SCOSoften lay between the crystallization and deliquescencepoints of the salt species known to be present in theaerosol. Whether the aerosol would be wet or dry underthese circumstances depends on the RH history it experi-enced. Humidity data taken during the #ights, coupledwith continuous ground-based measurements at "ve lo-cations in the Los Angeles area, were used to estimate themaximum humidity encountered by the aerosol that wassampled during a given #ight. As an estimate of theuncertainty that this assumption might introduce, it wasdetermined that, averaged over all of the #ights, PM2.5(dry aerosol mass )2.5 lm aerodynamic diameter) cal-culated using the assumption that the maximum humid-ity encountered was the ambient humidity at that time isonly 5.1% greater than that calculated assuming all saltshad previously deliquesced. After adjusting for relativehumidity di!erences, excellent agreement in measuredconcentration was observed in the overlapping size rangeof the PCASP and FSSP. In general, the size distributionrecovered from the DMA had a similar shape to thatmeasured by the PCASP, but was shifted by about 10%to larger size. The cause of the discrepancy is not known,but it generally had little impact on derived aerosol massand optical properties. Lacking de"nitive evidence of therelative accuracy of the di!erent measurements, indi-vidual distributions were simply averaged in overlappingsize ranges.

2.4. Optical and mass closure

Although composition and optical measurements af-ford valuable information about the properties of thesampled aerosol, only an accurate physicochemical de-scription can provide a full range of spatially and tem-porally resolved aerosol characteristics. In order to testthe validity of the size distribution-based physico-chemical description of the aerosol, a number of closurecomparisons were performed with the optical and com-position measurements. The utility of such comparisons


is that they provide con"dence in the over-determined setof measurements when closure is achieved, and indicatepotential sources of error when it is not.

To estimate aerosol mass collected on the "lters duringsampling intervals, aerosol size distributions were "rstadjusted to the relative humidity within the cyclone,which was slightly warmer and drier than outside. Thedensity of each aerosol type considered was then cal-culated and used to determine the appropriate cyclonepenetration e$ciency s-curve. For hygroscopic particles,density was calculated as a function of relative humidity,whereas constant densities of 2.6 g cm~3 for dust (Tegenand Fung, 1994), and 2.0 and 1.4 g cm~3 for elementaland organic carbon, respectively (Larson et al., 1988),were used. The aerosol mass of those particles thatpassed through the cyclone was then calculated throughintegration of the size distribution. Aerosol mass derivedin this way is included in Fig. 1. Although there are somedi!erences between the three measures (species sum,gravimetric, and size distribution integration) of aerosolmass during a given "lter interval, there are no pro-nounced systematic discrepancies.

As with aerosol mass calculations, modi"cation of theaerosol size distribution to account for removal in thecyclone is necessary for comparison with the variousoptical instruments on board. The remaining particlesare then adjusted to the relative humidity within eachinstrument. For those cases in which the relative humid-ity of the sample #ow is reduced below the crystallizationpoint of a given salt, only if the RH in the humidi"ednephelometer is above the deliquescence point of the saltis it assumed to be hydrated. For analysis of scatteringand absorption coe$cients, constant refractive indices of1.95}0.66i for elemental carbon (Bergstrom, 1972) and1.55}0.0i for organic carbon (Larson et al., 1988) wereassumed. The wavelength-dependent complex index ofrefraction of Saharan dust particles described by Patter-son et al. (1977) was assumed to be representative of dustpresent in Los Angeles. Optical properties of dust varywith location and even time, but few representative dataare available. Fortunately, the contribution of dust toscattering and absorption was generally small, minimiz-ing the uncertainty in derived optical properties resultingfrom this assumption. Refractive index of aqueous aero-sols was determined by the partial molal refractionapproach of Moelwyn-Hughes (1961). For internalmixtures of salt and carbon, the volume-weighted aver-age of the refractive indices of the components was used.Fuller et al. (1999) found that volume-weighted calcu-lations tend to overestimate absorption of a mixedsalt/elemental carbon particle relative to more physicallyrealistic scenarios such as a salt shell that contains anelemental carbon inclusion. However, because of uncer-tainty in the mixing state and size distribution of carbonin the aerosol, the magnitude of the error remainsunknown.

Mie Theory was used to predict the response of theoptical instruments to the sampled aerosol with speci"edsize-resolved concentration and refractive index. Forcomparison with the nephelometers, the angular sensitiv-ity of the TSI 3563 (Anderson et al., 1996) was incorpor-ated into the Mie Theory calculations. Fig. 2 showsclosure comparisons with the total scattering coe$cientmeasured by each of the Radiance Research nephelo-meters, the 550 nm hemispherical backscattering co-e$cient measured by the TSI nephelometer, and theabsorption coe$cient measured by the PSAP. Data fromeach of the seven #ights during which "lter samples weretaken are included in the comparisons with the threenephelometers, while data from only those #ights thatincluded extended constant altitude legs were included inthe PSAP comparison to minimize biases caused bypressure variations. Although each comparison is char-acterized by some discrepancy, that with the PSAP isclearly the most signi"cant. Potential causes of this dis-agreement include errors in the analysis of carbon or inthe distinction between organic and elemental carbon,errors in assumed mixing state and sphericity of theparticles, and errors in the absorption measurementitself, potentially due to the broadside enhancement e!ectdescribed by Fuller et al. (1999). It is unlikely that thisdiscrepancy results from use of volume-weighted refrac-tive indices in the size distribution-based calculations,since this would tend to increase the derived absorptioncoe$cient, thereby improving agreement.

3. Three-dimensional aerosol structure

A variety of #ight patterns were utilized during SCOSin order to explore the three-dimensional distribution ofthe aerosol in the Los Angeles Basin. The pattern usedduring the "rst four missions of SCOS consisted of a se-quence of ascending and descending spirals. Both spatialand temporal variations in the aerosol were explored byconducting one of these 4-h missions in the morning andanother in the afternoon on both 27 and 28 August.Meteorological measurements showed the temperatureinversion on 27 August had a height of approximately350 m until breaking up in the early afternoon. A weako!shore #ow was observed during the evening, followedby onshore #ow and an inversion having a height ofapproximately 500 m the following day. Peak 1-h ozoneconcentrations observed in the Southern California AirBasin were 116 and 132 ppb on 27 and 28 August,respectively.

To present the three-dimensional structure of the aero-sol during this period, data from spirals along two crosssections have been considered. An east}west cross sectionwas chosen that includes spirals over Santa Monica, ElMonte, Azusa, Pomona, and Rialto, while a north}southcross section includes spirals over Altadena, El Monte,


Fig. 2. Closure comparisons between derived and directly measured optical properties. Data presented are from those #ights duringwhich "lter samples were collected. Comparisons shown include total scattering coe$cients measured by the Radiance Researchnephelometers at varying relative humidity, hemispherical backscattering at 550 nm measured by the TSI nephelometer, and absorptionmeasured by the PSAP.

Fullerton, and Seal Beach. It is not possible to com-pletely deconvolute the spatial variations between spirallocations from variations that result from evolution oradvection of the aerosol during the course of a #ight. Nor

is it possible to predict with con"dence the aerosol prop-erties at a point along one of the cross sections throughsimple interpolation of measurements made. Neverthe-less, Figs. 3 and 4 present data along these cross sections


Fig. 3. Distribution of PM2.5 throughout the Los Angeles basin as sampled during morning and afternoon #ights on 27 August. Thetop plots show contours of near-ground-level aerosol mass as well as the east}west (E}W) and north}south (N}S) cross sections used inthe image plots shown below. As indicated in the 27 August contour plot, spiral locations are Altadena (AL), Azusa (AZ), Chino (CH), ElMonte (EM), Fullerton (FU), Pomona (PO), Rialto (RL), Riverside (RV), Santa Monica (SM), and Seal Beach (SB). Represented in theimage plots is PM2.5 interpolated from adjacent spirals. The solid gray area at the bottom of each of the image plots represents theapproximate ground level.

as if they were snapshots in time exhibiting smoothvariations between locations. PM2.5 has been consideredhere because of its regulatory relevance. Shown at the topof these "gures are contour plots of aerosol mass derivedfrom measurements made at the lowest point of eachspiral. Also presented are the east}west (E}W) andnorth}south (N}S) cross sections considered. Beginningon the morning of 27 August, the aerosol was mostconcentrated over the inland valleys between El Monteand Chino, where PM2.5 exceeded 40 lg m~3. Stronggradients in PM2.5 are apparent along both cross sec-tions between the coast and inland areas. Only a weakaerosol layer at approximately 1000 m above sea level(asl) is visible above the ground-level polluted layer.Along the east}west cross section, little change is ob-served in near-ground-level aerosol mass from morningto afternoon, although the columnar aerosol burden over

the inland areas increased, and a weak, but distinct, layerat &400 m asl developed that stretched to SantaMonica. At approximately 1200 m asl along thenorth}south cross section, a much more intense aerosollayer is observed, in which PM2.5 exceeds 30 lg m~3.These data suggest that aerosol was injected above themixed layer at the southern edge of the San GabrielMountains and was then transported away from themountains with the sea breeze return #ow. Aircraft-basedmeasurements have previously identi"ed such elevatedpollution layers both over Southern California (Blumen-thal et al., 1978; Wakimoto and McElroy, 1986; Li et al.,1997) and over similar coastal regions (Wakamatsu et al.,1983; Lalas et al., 1983; Ho! et al., 1997). A second, lesspronounced, layer is present approximately 200 m abovethe "rst. The two layers present along the east}west crosssection the following morning are likely remnants of this


Fig. 4. The same format as in Fig. 3 for the aerosol sampled during morning and afternoon #ights on 28 August.

(these) layer(s). As the ground-level mixed layer deepensin the afternoon of 28 August, the morning layers cease tobe distinct, suggesting that they mixed with the underly-ing layer. Along the north}south cross section, an elev-ated layer formed on 28 August in much the same way aswas observed the previous day.

The climb and descent rates used during this studywere typically 150 m min~1, leading to a vertical resolu-tion of &150 m in the size distribution measurements.Aliasing due to this limited vertical resolution becameexcessive during several spirals, in which case the fasternephelometer measurements were examined. Fig. 5shows aerosol scattering coe$cients recorded with thenon-humidi"ed Radiance Research nephelometer duringeach of the spirals #own o! the coast of Santa Monica onthe "rst four #ights. The morning spirals show relativelylittle vertical variation on either day. With the strength-ening of the sea breeze in the afternoon, pronounced, butthin (&100 m), aerosol layers developed. The scatteringcoe$cient of 0.00014 m~1 measured in the August 28layer was among the highest observed at any location

during the study. Like the layers over the inland areas,these coastal layers have been observed in previous stud-ies and are attributed to vertical transport along theheated coastal ranges, followed by horizontal transporttowards the coast as part of the sea breeze return #ow(Lu and Turco, 1994).

During one #ight on 9 September and two #ights thefollowing day three spirals over El Monte and two overLong Beach reached altitudes of over 3000 m. On both ofthese days upper level winds were from the south andwest, and temperature inversions were present until mid-afternoon. Maximum 1-h ozone concentrations were 105and 113 ppb on 9 and 10 September, respectively.Though the spirals were #own over a 32-h period at twodissimilar locations, each encountered an aerosol layer atapproximately 2500 m having a peak PM2.5 mass ofabout 2 lg m~3, as shown in Fig. 6a. The early morningspiral #own over El Monte on 9 September indicates thata more concentrated layer may have been present above3000 m, although no similar layers were observed during3400 m spirals on the morning of 10 September. The


Fig. 5. Vertical pro"les of ambient temperature and aerosol scattering coe$cient measured during spirals #own o! the coast of SantaMonica on four missions conducted on 27 and 28 August.

Fig. 6. (a) Vertical pro"les of PM2.5 measured during 3000#mspirals #own over El Monte and Long Beach, and (b) nor-malized mass size distributions averaged over each of the layersindicated by arrows in (a).

averaged mass distributions for the layers indicated byarrows in Fig. 6a are presented in Fig. 6b. There is a clearshift in the aerosol mass distribution within the 2500 mlayer relative to the aerosol within either the ground levelor 1500 m layers. This observation is consistent with anaerosol that has aged through condensation and coagu-lation.

Seven di!erent #ight patterns were employed for the 12missions #own during SCOS, making analysis of day-to-day variations di$cult. However, the majority of #ightpatterns included spirals over El Monte, Riverside, andFullerton. These repeated spirals provide the data neces-sary for a statistical analysis of the aerosol over theselocations. Day-to-day variations in the aerosol concen-tration and properties obscured any diurnal trends thatmight otherwise have been apparent. Therefore, in com-bining data from each set of spirals, no distinction wasmade between samples taken at di!erent times. Fig. 7presents the mean and standard deviation of severalintensive and extensive aerosol properties as a function ofaltitude for 10 spirals #own over both El Monte andRiverside, and 9 #own over Fullerton. The data havebeen smoothed to more clearly show vertical trends. Asindicated in these plots, the aerosol variability at eachlocation exceeds the mean variability between locations.Consistent with expectations, aerosol mass decreaseswith increasing altitude and with proximity to the coast.Aerosol number concentration was about 50% higher inEl Monte than in Riverside, while aerosol mass washigher in Riverside, indicating that much of this in-creased mass results from growth of existing particles.Steadily improving air quality in Los Angeles, coupledwith atypical meteorology associated with a strong ElNin8 o event, led to signi"cantly lower aerosol loadings


Fig. 7. Vertical pro"les representing the mean and standard deviation of a number of extensive and intensive aerosol properties. Datataken during the 10 #ights during which spirals were #own over El Monte and Riverside, and 9 #ights with spirals #own over Fullertonwere used in the analysis. For the derived ratio of scattering at 80% RH to scattering at 30% RH, only those data below the altitude atwhich random error began to dominate are shown.

during SCOS than during SCAQS 10 years earlier. How-ever, the relative contributions of "ne (PM2.5) and coarse(PM10}PM2.5) fractions remained much the same.Chow et al. (1994) found that, during SCAQS, PM2.5constituted between 50 and 67% of PM10, while near-ground-level measurements made on board the Pelicanduring SCOS indicate a corresponding range of 53}69%.At each of these locations, PM2.5 constituted an increas-ing percentage of PM10 with increasing altitude. Theratio of aerosol scattering at 80% RH to that at 30% RHis derived using the data from Radiance Research nephelo-meters along with an empirical relationship betweenscattering intensity and RH (Kasten, 1969). Only thosedata below the altitude at which random error began todominate this ratio are shown. The enhanced ratio ob-served for Riverside suggests that aerosol hygroscopicityis greater in the inland areas, although variations inaerosol size distribution could also account for thechange.

4. Sub-grid variability

The primary goal of SCOS was the acquisition ofa su$ciently detailed database to lead to e!ective evalu-ation of photochemical and aerosol modeling capabili-ties. Computational limitations, sparse meteorologicaland air quality measurements, and limited emissionsinventories all serve to limit the minimum useful com-putational grid size of three-dimensional models. Typical

grid cells are 5 km on a side with a vertical dimensionthat varies with height (Meng et al., 1997). McNair et al.(1996) discuss potential errors involved in assuming thatpoint measurements of a given species represent the gridcell volume average. In their analysis of data obtainedduring SCAQS, they found that for the one pair ofmonitoring sites that were separated by less than 5 km(4.8 km), variations in peak ozone concentration of up to50% were observed. Similar discrepancies were notedwhen point measurements were compared with interpo-lations between surrounding measurements.

Identical patterns were #own on 4 and 5 September aspart of an intensive measurement period during SCOSthat was designed to investigate nitrate formation alonga trajectory beginning east of downtown Los Angeles inDiamond Bar, passing over the ammonia-rich area nearMira Loma, and ending near the eastern edge of the LosAngeles Basin in Riverside. Meteorological measure-ments describe an initial o!shore #ow on 4 Septemberthat gave way to onshore winds and a 120}150 m deepmarine layer on 5 September. This transition was largelythe result of increasing in#uence of tropical storm Kevino! the tip of the Baja Peninsula. Peak 1-h ozone concen-trations diminished with this transition from 157 ppb on4 September to 113 ppb on 5 September. To ensurea su$cient amount of aerosol was collected for "lteranalyses, the Pelican circled &300 m above each samp-ling site for approximately 1 h. Fig. 8 shows #ight pat-terns for these missions. Inspection of data from these#ights during some of the constant altitude circular #ight


Fig. 8. Flight patterns for missions conducted on 4 and 5 September. Insets show vertical pro"les of PM2.5 and temperature at spirallocations. The heavy horizontal line in each of the pro"les represents the approximate altitude of the circles #own above Diamond Bar,Mira Loma, and Riverside. The solid arrows through the #ight track represent the direction of maximum aerosol gradient at each of thesampling locations that was used for the analysis presented in Figs. 9 and 10.

patterns showed persistent gradients in aerosol concen-tration. The diameter of the circles ranged from about 5.5to 6.5 km. This is comparable to the computational gridsize of conventional three-dimensional models. Unlikesimultaneous ground-based samples at multiple loca-tions, airborne analysis of relative changes in pollutantconcentrations is largely insensitive to instrumentalvariability. On the other hand, with aircraft sampling it ispossible that variations perceived as being attributable tohorizontal gradients could, in fact, result from samplingalong the boundary of a vertical layer that varies inheight. Because of #ight pattern restrictions, spirals werenot #own over any of the sampling locations, but were#own above adjacent areas. As shown in the insets in

Fig. 8, vertical pro"les of PM2.5 and temperature over ElMonte and Riverside do not indicate the presence ofsharp vertical gradients near the sampling altitude(&500 m), suggesting that any variations are horizontalin nature.

For each sampling location shown in Fig. 8, an arrowindicates the direction corresponding to the maximumgradient in aerosol concentration. Figs. 9 and 10 showvariations along the indicated direction of maximumgradient of several normalized extensive and intensiveaerosol properties, as well as size distributions. The dataalong the gradients were averaged to limit scatter. Somevariability was apparent at each location, but that ob-served at Diamond Bar and Mira Loma on 4 September


Fig. 9. Variation of extensive aerosol properties, intensive aerosol properties, and number size distributions as a function of distancealong the direction of maximum gradient indicated in Fig. 8.

and Riverside on 5 September is consistent with thepresence of strong gradients in aerosol concentration,resulting in PM2.5

.!9: PM2.5

.*/ratios of 1.61, 1.33, and

1.39, respectively. The lack of variability in either theintensive properties or the shape of the size distributionssuggests that the aerosol concentration varies with loca-tion, but its characteristics do not. As is true for most ofthe Los Angeles Basin, there are major freeways as wellas other emissions sources in the vicinity of each of theselocations, potentially causing the observed variability. Itis also possible that the variability is a result of changes inboundary layer depth over the area of the orbit. It isinteresting to note that none of the locations exhibitedthe same variability, or lack thereof, on both days. Addi-tional measurements are necessary to determine whetherthe presence of these strong gradients follows a diurnal

pattern, which might explain the di!erences observedover the two-day period since the 4 September missionwas conducted approximately 4 h earlier than the #ightconducted on 5 September.

5. E4ect of the aerosol on photolysis rates

Formation of ozone and a number of other gas andaerosol phase species is driven by photochemistry (Sein-feld and Pandis, 1998). Accurate prediction of gas-phaseconcentrations requires a detailed understanding of thearray of reactions responsible for their formation anddestruction, as well as the spectrally resolved actinic #uxthroughout the area of interest. Direct measurementof actinic #ux or key photolysis rates themselves can


Fig. 10. The same format as in Fig. 9 for the aerosol sampled on 5 September.

provide the necessary information, although such dataare typically available only at ground level. Alternatively,vertical aerosol pro"les derived from airborne measure-ments or vertically resolved models can be used to deter-mine the potential impact of scattering and absorptionon photolysis rates. Several investigations of this typehave been reported for di!erent regions with varyingcloud cover, surface albedo, and solar zenith angle(Demerjian et al., 1980; Ruggaber et al., 1994; Lantzet al., 1996; Castro et al., 1997; Dickerson et al., 1997;Jacobson, 1998; Liao et al., 1999). The extensive aerosolmeasurements made with the Pelican provide much ofthe data necessary for such an analysis, thereby reducinguncertainty in the role of the aerosol in atmosphericphotochemistry.

The approach taken here to determine photolysis rateshas been described in Liao et al. (1999), and will bedescribed only brie#y. Actinic #ux was calculated usingthe one-dimensional discrete ordinate radiative transfer(DISTORT) model (Stamnes et al., 1988). The calcu-lations employed 80 vertical layers from 0 to 70 km thatranged in thickness from 100 m in the lowest 3 km, to5 km between 50 and 70 km in altitude. Aerosol concen-tration was assumed to be zero above the maximummeasurement height. Clearly, this will result in someerror in derived photolysis rates. An analysis of the im-pact of the layers observed at about 2500 m will bediscussed below, but no consideration of even higherlayers is made, although it is likely that the aerosolpresent above this height contributes negligibly to the


Fig. 11. Fractional change in "ve gas-phase photolysis rates calculated to have resulted from the aerosol sampled over El Monte relativeto an aerosol-free atmosphere. For each reaction considered, the mean and standard deviation were determined from photolysis ratecalculations using measurements made during spirals #own over El Monte on 10 #ights. Unlike most of the other analyses presentedhere, altitude is relative to ground level and not sea level.

columnar aerosol burden. Actinic #ux was calculated forthe 290}700 nm wavelength range with spectral intervalsranging from 1 nm for wavelengths between 290 and330 nm, to 5 nm between 600 and 700 nm. Publisheddata for temperature-dependent ozone absorption crosssection (Malicet et al., 1995; WMO, 1985), solar irra-diance (Neckel and Labs, 1984; Woods et al., 1996),temperature pro"le (Nagatani and Rosen"eld, 1993),ozone pro"le (McPeters, 1993), and spectrally resolvedsurface albedo (Demerjian et al., 1980) were utilized forthese analyses. As is common in Southern Californiaduring the summer months, clouds were rare during thisstudy and were, therefore, not considered. To clearlydemonstrate the variability associated with aerosol con-centration and properties, and not time of day, a constantsolar zenith angle of 103 was considered.

Consistent with the approach taken to characterizephysical and chemical aerosol properties, photolysisrates were computed based on data taken along a crosssection during one #ight, and taken at one locationduring several #ights. Poor spectral resolution and size-dependent non-idealities prevented direct use of nephel-ometer and absorption photometer measurements.Instead, aerosol optical properties at each height interval

were determined through Mie Theory calculations usingthe size distributions. As discussed above, the size distri-bution measurements have vertical resolution of approx-imately 150 m, which is slightly greater than theminimum vertical layer thickness used in DISORT, solimited interpolation was necessary. The fractionalchange in several key photolysis rates resulting from theaerosol sampled over El Monte during 10 #ights is pre-sented in Fig. 11. As with the statistical analysis of verti-cally resolved aerosol properties shown in Fig. 7, themean and standard deviation were calculated and pre-sented as a solid line and shaded region, respectively.Little variation is observed in the impact the aerosol isexpected to have on photolysis rates of NO

2, HCHO,

CH3COCH

3, and O

3PO(1D), while the impact on

photolysis of O3

to form O(3P) is smaller in magnitudeand exhibits less vertical variability than is observed forthe other reactions considered. For each reaction ana-lyzed, the El Monte aerosol causes a slight decrease inphotolysis rates in the lowest 100 m, with a more pro-nounced enhancement observed above this height. Thepresence of a cross-over point at which the fractionalchange of a given photolysis rate is zero results froman o!set between increased di!use actinic #ux due to


Fig. 12. Vertically resolved fractional change in the NO2

photolysis rate calculated using data obtained during spirals #own on themorning and afternoon #ights of 28 August. The east}west cross section indicated in Fig. 4 was used for presentation of these data.

aerosol-enhanced scattering and decreased direct actinic#ux caused by upward scattering and absorption by theoverlying aerosol. These cross-overs have been observedin similar analyses with relatively non-absorbing aerosollike that encountered during SCOS (e.g., Liao et al.,1999). The maximum increase in photolysis rates abovethat expected if no aerosol were present is approximately2% for O

3PO(3P), and 5% for each of the other reac-

tions considered. Inclusion of the 2500 m aerosol layersmeasured during spirals over El Monte on 9 and 10September results in a further enhancement of approxim-ately 1% (0.3% for O

3PO(3P)). Similar results were

obtained when the fractional change in the photolysisrate of NO

2was considered along the east}west cross

section described in Fig. 4 using data from morning andafternoon #ights on 28 August. As shown in Fig. 12, only

slight variations in photolysis rate alteration are ob-served along the cross section, both in the morning andafternoon. The sharp gradient in fractional photolysisrate change with height in the lowest 500 m demonstratesthe potential error caused by extrapolations of ground-based radiative #ux measurements. The impact of aero-sols on photolysis rates presented here is similar to esti-mates of Jacobson (1998) using data obtained over LongBeach on 27 August 1987 during SCAQS, but di!erssigni"cantly from calculations performed using data ob-tained over Claremont on the same day. Speci"cally,Jacobson (1998) predicted that the aerosol present overClaremont resulted in a decrease of up to about 12% inthe NO

2photolysis rate, resulting from a more concen-

trated and highly absorbing aerosol than was sampledduring SCOS. The atypical meteorological conditions


during this campaign prevented a comparable severepollution episode, while improved air quality in SouthernCalifornia since 1987 would probably limit the likelihoodof aerosol loadings similar to those observed inClaremont during SCAQS. Therefore, the results fromSCOS and SCAQS probably bound the likely impactthat a signi"cant, present-day, pollution episode wouldhave on photolysis rates.

6. Summary

As part of the Southern California Ozone Study(SCOS), 12 missions were conducted with an aircraftinstrumented to characterize the chemical and physicalproperties of the Los Angeles aerosol. Instrumentationon board included three parallel "lter sampling systems,three nephelometers, an absorption photometer, a DMA,and two OPCs. Data from the variety of measurementsmade were combined to provide a time-resolvedphysicochemical description of the aerosol. Several clos-ure comparisons were performed between derived anddirectly measured properties. Best-"t lines through scat-ter plots containing these data for comparisons of "neaerosol mass, `drya aerosol scattering coe$cient, `wetaaerosol scattering coe$cient, hemispherical backscatter-ing coe$cient, and absorption coe$cient had derived/measured slopes of 0.91, 1.06, 1.13, 1.21, and 0.55, respec-tively. The exact cause of the large disagreement in theabsorption comparison is not known, although a numberof potential errors have been suggested.

The aerosol present over the Los Angeles Basin ex-hibits a complex three-dimensional structure. Verticallyresolved data suggest that pronounced elevated layerspresent over the inland areas were formed by injection ofaerosol above the ground-level polluted layer along theSouthern edge of the San Gabriel Mountains, followedby advection towards the coast through incorporationinto the sea breeze return #ow. Additional layers wereobserved about 500 m asl o! the coast of Santa Monica,and approximately 2500 m asl over El Monte and LongBeach. Data from spirals #own over El Monte, Fullerton,and Riverside on several #ights over the 3-week samplingperiod were used to provide a limited statistical descrip-tion of the vertically resolved aerosol at each location. Ingeneral, it was found that variability over time exceededvariability among locations. Constant altitude circles#own over Diamond Bar, Mira Loma, and Riversideyielded evidence of gradients in aerosol concentrationsu$cient to cause over 50% variability within a 5]5 kmcomputational grid cell commonly used in atmosphericmodels.

Data from spirals #own over El Monte during several#ights, as well as from spirals #own over several locationson 28 August, were used to analyze the impact of the aero-sol on important photolysis rates in the photochemical

generation of ozone. On average, the aerosol was pre-dicted to cause a slight decrease in photolysis rates in the"rst 100 m above ground level, but led to a more pro-nounced (up to &5%) increase above that height. Formost of the reactions considered, the 2500 m elevatedaerosol layers above El Monte caused an enhancement inphotolysis rates of about 1% above that resulting fromthe aerosol present below these layers. Collectively, thedata obtained in this sampling program provide furtherinsight into microphysical processes that govern the size,composition, and spatial and temporal behavior of theLos Angeles aerosol.

Acknowledgements

The authors gratefully acknowledge Prakash Bhaveand Glen Cass for providing back trajectories used in thisanalysis. We also thank Yanzeng Zhao for assistancerelated to the El Monte lidar measurements. This workwas supported by the California Air Resources Board.

References

Anderson, T.L., Covert, D.S., Marshall, S.F., Laucks, M.L.,Charlson, R.J., Waggoner, A.P., Ogren, J.A., Caldow, R.,Holm, R.L., Quant, F.R., Sem, G.J., Wiedensohler, A.,Ahlquist, N.A., Bates, T.S., 1996. Performance characteristicsof a high-sensitivity, three-wavelength, total scatter/back-scatter nephelometer. Journal of Atmospheric and OceanicTechnology 13 (5), 967}986.

Bergstrom, R.W., 1972. Predictions of the spectral absorptionand extinction co$cients of an urban air pollution model.Atmospheric Environment 6, 247}258.

Blumenthal, D.L., White, W.H., Smith, T.B., 1978. Anatomy ofa Los Angeles smog episode: pollutant transport in thedaytime sea breeze regime. Atmospheric Environment 12,893}907.

Bond, T.C., Anderson, T.L., Campbell, D., 1999. Calibration andintercomparison of "lter-based measurements of visible lightabsorption by aerosols. Aerosol Science and Technology 30(6), 582}600.

Castro, T., Ruizsuarez, L.G., Ruizsuarez, J.C., Molina, M.J.,Montero, M., 1997. Sensitivity analysis of a UV radiationtransfer model and experimental photolysis rates of NO

2in

the atmosphere of Mexico City. Atmospheric Environment31 (4), 609}620.

Chow, J.C., Watson, J.G., Fujita, E.M., Lu, Z.Q., Lawson, D.R.,Ashbaugh, L.L., 1994. Temporal and spatial variations ofPM(2.5) and PM(10) aerosol in the Southern Californiaair-quality study. Atmospheric Environment 28 (12),2061}2080.

Chow, J.C., Watson, J.G., Pritchett, L.C., Pierson, W.R., Frazier,C.A., Purcell, R.G., 1993. The dri thermal optical re#ectancecarbon analysis system } description, evaluation and ap-plications in United States air-quality studies. AtmosphericEnvironment Part A } General Topics 27 (8), 1185}1201.


Demerjian, K.L., Schere, K.L., Peterson, J.T., 1980. Theoreticalestimates of actinic (spherically integrated) #ux and photo-lytic rate constants of atmospheric species in the lowertroposphere. Advances in Environmental Science and Tech-nology 10, 369}459.

Dickerson, R.R., Kondragunta, S., Stenchikov, G., Civerolo,K.L., Doddridge, B.G., Holben, B.N., 1997. The impact ofaerosols on solar ultraviolet radiation and photochemicalsmog. Science 278 (5339), 827}830.

Fuller, K.A., Malm, W.C., Kreidenweis, S.M., 1999. E!ects ofmixing on extinction by carbonaceous particles. Journalof Geophysical Research } Atmospheres 104 (D13),15941}15954.

Harley, R.A., Russell, A.G., Mcrae, G.J., Cass, G.R., Seinfeld,J.H., 1993. Photochemical modeling of the Southern Califor-nia air-quality study. Environmental Science and Techno-logy 27 (2), 378}388.

Ho!, R.M., Harwood, M., Sheppard, A., Froude, F., Martin,J.B., Strapp, W., 1997. Use of airborne lidar to determineaerosol sources and movement in the Lower FraserValley (Lfv). Bc. Atmospheric Environment 31 (14),2123}2134.

Jacobson, M.Z., 1997. Development and application of a new airpollution modeling system. 3. Aerosol-phase simulations.Atmospheric Environment 31 (4), 587}608.

Jacobson, M.Z., 1998. Studying the e!ects of aerosols on verticalphotolysis rate coe$cient and temperature pro"les over anurban airshed. Journal of Geophysical Research } Atmo-spheres 103 (D9), 10593}10604.

Kasten, F., 1969. Visibility in the phase of pre-condensation.Tellus 21, 631}635.

Lalas, D.P., Asimakopoulos, D.N., Deligiorgi, D.G., Helmis,C.G., 1983. Sea-breeze circulation and photochemical pollu-tion in Athens, Greece. Atmospheric Environment 17,1621}1632.

Lantz, K.O., Shetter, R.E., Cantrell, C.A., Flocke, S.J., Calvert,J.G., Madronich, S., 1996. Theoretical, actinometric, andradiometric determinations of the photolysis rate coe$cientof NO

2during the Mauna Loa Observatory Photochemis-

try Experiment 2. Journal of Geophysical Research } Atmo-spheres 101 (D9), 14613}14629.

Larson, S.M., Cass, G.R., Hussey, K.J., Luce, F., 1988. Veri"ca-tion of image-processing based visibility models. Environ-mental Science and Technology 22 (6), 629}637.

Li, S.M., Macdonald, A.M., Strapp, J.W., Lee, Y.N., Zhou,X.L., 1997. Chemical and physical characterizations ofatmospheric aerosols over Southern California. Journal ofGeophysical Research } Atmospheres 102 (D17),21341}21353.

Liao, H., Yung, Y.L., Seinfeld, J.H., 1999. E!ects of aerosols ontropospheric photolysis rates in clear and cloudy atmo-spheres. Journal of Geophysical Research } Atmospheres104 (D19), 23697}23707.

Lu, R., Turco, R.P., 1994. Air pollutant transport in a coastalenvironment. 1. 2-Dimensional simulations of sea-breezeand mountain e!ects. Journal of the Atmospheric Sciences51 (15), 2285}2308.

Lu, R., Turco, R.P., 1995. Air pollutant transport in acoastal environment. 2. 3-Dimensional simulations overLos-Angeles Basin. Atmospheric Environment 29 (13),1499}1518.

Lu, R., Turco, R.P., Jacobson, M.Z., 1997. An integrated airpollution modeling system for urban and regional scales. 1.Structure and performance. Journal of Geophysical Re-search } Atmospheres 102 (D5), 6063}6079.

Lurmann, F.W., Wexler, A.S., Pandis, S.N., Musarra, S., Kumar,N., Seinfeld, J.H., 1997. Modelling urban and regional aero-sols. 2. Application to California's South coast air basin.Atmospheric Environment 31 (17), 2695}2715.

Malicet, J., Daumont, D., Charbonnier, J., Parisse, C., Chakir,A., Brion, J., 1995. Ozone UV spectroscopy. 2. Absorptioncross-sections and temperature-dependence. Journal ofAtmospheric Chemistry 21 (3), 263}273.

McNair, L.A., Harley, R.A., Russell, A.G., 1996. Spatialinhomogeneity in pollutant concentrations, and their im-plications for air quality model evaluation. AtmosphericEnvironment 30 (24), 4291}4301.

McPeters, R., 1993, Ozone pro"le comparisons. In: Remsberg,E.E., Prather, M.J. (eds.), The atmospheric E!ects of Stratos-pheric Aircraft: Report of the Models and MeasurementsWorkshop, NASA Ref. Publ. 1292, A1}A47.

Meng, Z., Dabdub, D., Seinfeld, J.H., 1997. Chemical couplingbetween atmospheric ozone and particulate matter. Science277 (5322), 116}119.

Meng, Z.Y., Dabdub, D., Seinfeld, J.H., 1998. Size-resolved andchemically resolved model of atmospheric aerosol dynamics.Journal of Geophysical Research } Atmospheres 103 (D3),3419}3435.

Moelwyn-Hughes, E.A., 1961. Physical Chemistry. Pergamon,Tarrytown, NY.

Nagatani, R.M., Rosen"eld, J.E., 1993. Temperature, heating,and circulation. In: Remsberg, E.E., Prather, M.J. (eds.), TheAtmospheric E!ects of Stratospheric Aircraft: Report of theModels and Measurements Workshop, NASA Ref. Publ.1292, A1}A47.

Neckel, H., Labs, D., 1984. The solar-radiation between 3300-aand 12500-a. Solar Physics 90 (2), 205}258.

Pandis, S.N., Harley, R.A., Cass, G.R., Seinfeld, J.H., 1992.Secondary organic aerosol formation and transport. Atmo-spheric Environment Part A } General Topics 26 (13),2269}2282.

Pandis, S.N., Wexler, A.S., Seinfeld, J.H., 1993. Secondary or-ganic aerosol formation and transport. 2. Predicting theambient secondary organic aerosol-size distribution. Atmo-spheric Environment Part A } General Topics 27 (15),2403}2416.

Patterson, E.M., Gillette, D.A., Stockton, B.H., 1977. Complexindex of refraction between 300 and 700 nm for Saharanaerosols. Journal of Geophysical Research 82, 3153}3160.

Ruggaber, A., Dlugi, R., Nakajima, T., 1994. Modeling radiationquantities and photolysis frequencies in the troposphere.Journal of Atmospheric Chemistry 18 (2), 171}210.

Saxena, P., Hildemann, L.M., McMurry, P.H., Seinfeld, J.H.,1995. Organics alter hygroscopic behavior of atmosphericparticles. Journal of Geophysical Research } Atmospheres100 (D9), 18755}18770.

Seinfeld, J.H., Pandis, S.N., 1998. Atmospheric Chemistry andPhysics. Wiley}Interscience, New York.

Stamnes, K., Tsay, S.C., Wiscombe, W., Jayaweera, K., 1988.Numerically stable algorithm for discrete-ordinate-methodradiative-transfer in multiple-scattering and emitting layeredmedia. Applied Optics 27 (12), 2502}2509.


Tang, I.N., 1996. Chemical and size e!ects of hygroscopic aero-sols on light scattering coe$cients. Journal of GeophysicalResearch } Atmospheres 101 (D14), 19245}19250.

Tang, I.N., Munkelwitz, H.R., 1994. Water activities, densities,and refractive-indexes of aqueous sulfates and sodium-ni-trate droplets of atmospheric importance. Journal of Geo-physical Research } Atmospheres 99 (D9), 18801}18808.

Tang, I.N., Tridico, A.C., Fung, K.H., 1997. Thermodynamicand optical properties of sea salt aerosols. Journal of Geo-physical Research } Atmospheres 102 (D19), 23269}23275.

Tegen, I., Fung, I., 1994. Modeling of mineral dust in the atmo-sphere } sources, transport, and optical-thickness. Journal ofGeophysical Research } Atmospheres 99 (D11), 22897}22914.

Wakamatsu, S., Ogawa, Y., Murano, K., Goi, K., Aburamoto,Y., 1983. Aircraft survey of the secondary photochemicalpollutants covering the Tokyo Metropolitan Area. Atmo-spheric Environment 17, 827}835.

Wakimoto, R.M., McElroy, J.L., 1986. Lidar observation ofelevated pollution layers over Los-Angeles. Journal of Cli-mate and Applied Meteorology 25 (11), 1583}1599.

White, W.H., 1990. Rep. No. 24, Natl. Acid Precip. Assess.Program, Washington, DC.

Woods, T.N., Prinz, D.K., Rottman, G.J., London, J., Crone,P.C., Cebula, R.P., Hilsenrath, E., Brueckner, G.E., Andrews,M.D., White, O.R., Van Hoosier, M.E., Floyd, L.E., Herring,L.C., Knapp, B.G., Pankratz, C.K., Reiser, P.A., 1996. Vali-dation of the VARS Solar irradiances: Comparison with theATLAS 1 and 2 measurements. Journal of Geophysical Re-search } Atmospheres 101 (D6), 9541}9569.

World Meteorological Organization (WMO), 1985. Rep. No. 16.Zhang, X.Q., McMurry, P.H., Hering, S.V., Casuccio, G.S., 1993.

Mixing characteristics and water content of submicron aero-sols measured in Los Angeles and at the Grand Canyon.Atmospheric Environment 27, 1593}1607.


airborne analysis of the los angeles aerosol159.226.119.84/collins2000.pdf · airborne analysis of...

Documents