Characteristics of Summer Midday Low-Visibility Events in the Los Angeles Area Susan M. Larsont and Glen
R. Cass”
Environmental Engineering Science Department and Environmental Quality Laboratory, California Institute of Technology, Pasadena, Callfornia 9 1125 ~
A five-site air monitoring network provided data during the summer of 1984 on pollutants that contribute to the midday visibility problem in the Los Angeles area. The data were supplemented by visual range observations and by nephelometer measurements of total light scattering. The network collected data over an extended region, obtained a data set describing the frequency distribution of high- and low-visibility events over an extended time period, and obtained information needed to calculate the cause and effect relationship between pollutant properties and the extinction coefficient from theories of light scattering and absorption, rather than from statistical techniques. Computed scattering coefficient values at the Pasadena site are on average within 26% of the measured values. Frequency distributions of calculated extinction coefficients at Pasadena agree with the frequency distribution of extinction coefficients estimated from visual range values for all but the extreme extinction cases. These data can be used to investigate the effect that an emission control program would have on visibility in the Los Angeles area.
Introduction The suspended particulate matter and gaseous pollutants present in the Los Angeles area regional haze can decrease visual range to less than a few kilometers (1-3). In the Los Angeles area, photochemical smog is heaviest during July, August, and September, and severe visibility reduction is often most apparent to the general public during the midday periods during these months. The extinction coefficient, a measure of light scattered and absorbed in the atmosphere, can be expressed as a sum of several components: light scattering by particles (bmh), light absorption by particles (babsp), light absorption by gases and light scattering by air molecules (bRayleigh) b e i t = bscab + bebs, + babs, + hayleigh (1) The value of the extinction coefficient can be obtained in several ways. A nephelometer can be used to measure the scattering contribution to the extinction coefficient (4). If scattering is the main cause of light extinction in a region, then the nephelometer measurement provides an approximate value of the extinction coefficient (5). The extinction coefficient also can be measured by using te‘Present address: Dept. of Civil Engineering, University of Illinois, Urbana, IL 61801. 0013-936X/89/0923-0281$01.50/0
lephotometers or teleradiometers (6-8). Alternatively, visual range values obtained by human observers can be inserted into Koschmieder’s formula in order to estimate the extinction coefficient (9, 10). Koschmieder’s formula is expressed by VR
-In A / &
(2)
where VR, the visual range, is the distance a t which an average observer can just barely distinguish a black object silhouetted against the horizon sky (11). This meteorological visual range is routinely measured by trained observers a t controlled airports and is reported in terms of the visual range that prevails around a t least half of the horizon circle, but not necessarily in continuous sectors (12). Parameter A in eq 2 represents the limiting contrast threshold for the average human observer. Commonly proposed values for the parameter A vary from 0.02 to 0.05 (13). The applicability of Koschmieder’s formula depends on the accuracy of the threshold contrast, the availability of black target objects, and the uniformity of illumination and atmospheric properties between the observer and horizon (13). Given the availability of suitable targets, Horvath (14) reports that errors in applying Koschmieder’s formula to estimate the extinction coefficient should be less than 10%. To improve visibility, the specific pollutants causing the reduction in visibility must be identified. Several researchers (1, 3, 15, 16) have employed regression analysis between observed extinction Coefficients and pollutant concentrations to estimate the level to which certain pollutants affect visibility. Unrealistic empirically determined extinction efficiencies, however, can be obtained for some aerosol species when statistical inference is used as a visibility modeling method. An alternative to purely empirical analyses can be constructed in which the size distribution and chemical composition of the atmospheric aerosol and the NOz concentration are used to calculate the contributions to the extinction coefficient (17, 18). Such visibility models have yet to be used as an integral part of the design process for engineering improvements in regional visibility, most probably because the data requirements of such models are difficult to satisfy. The purpose of the present study is to develop an experimental protocol and a visibility modeling approach by which the causes of regional visibility problems can be characterized. The data collected can be used to calculate the extinction coefficient based on theories of light scat-
0 1989 American Chemical Society
-
Envlron. Sci. Technol., Vol. 23,
No. 3, 1989 281
Figure 1. Map of Southern California showing air monitoring site locations (0)and adjacent airports (0).
Table I. Distance between Site Locations and Nearest Airports airport
station
distance, km
Los Angeles Ontario Norton Air Force Base Burbank
Lennox Upland San Bernardino Pasadena
4.8 5.6 1.6 16.1
tering and absorption rather than on statistical techniques, and the resulting data set is representative of the distribution of visibility events occurring over an extended period of time. Methods to accomplish these objectives are developed and tested in the Los Angeles area, and thus the experiments provide additional information on the visibility problem in Southern California. In this paper, a five-site atmospheric sampling program conducted in the Los Angeles area is described. The program served to acquire a data base representative of the distribution of visibility events occurring in summer midday periods. From the resulting chemical and physical description of the aerosol, scattering and extinction coefficient values are calculated. Comparisons are made between the calculated scattering coefficients and the scattering coefficient values measured with nephelometers. The calculated extinction results are compared to estimated extinction coefficient values obtained from airport observations of visual range. The data and visibility modeling procedure provide a description of the relationship between the various air pollutants and visual range that can serve as a basis for evaluating the effect of an emission control program directed at improving visibility (19).
Experimental Program During the summer months of July, August, and September of 1984, a five-site sampling network was operated in the Los Angeles air basin. The site locations spanned a distance of over 95 km, from Lennox, CA (located 8 km from the coast) eastward to San Bernardino. The intermediate sites were located a t Pasadena, Azusa, and Upland, CA (Figure 1). Atmospheric aerosol samples were collected at each station from 1000-1400 hours (PST)every sixth day over the summer. Each of the particulate sampling stations was colocated with a South Coast Air Quality Management District (SCAQMD) air monitoring station. The SCAQMD data on NOz concentrations were used to supplement the network particulate data. Airport weather (including temperature and relative humidity information) and visibility reports were available near the Lennox, Pasadena, Upland, and San Bernardino sites (Table I). The Pasadena, Upland, and Azusa sites were equipped with integrating nephelometers (Meteorology Research, Inc. Model 1550). To characterize the chemical composition of the aerosol, 4-h-average filter samples were taken for both fine (dp 5 282
Environ. Sci. Technol., Vol. 23, No. 3, 1989
2.1 pm) and for total particulate matter. Coarse mode aerosol mass concentrations (dp1 2.1 pm) were determined by subtracting the fine from the total particulate matter concentrations. Fine particle samples were collected downstream of an AIHL-design cyclone separator (20) which removed the coarse particulate matter from the air stream. Two sets of three parallel filter holders were used for sample collection, each filter holder containing a filter substrate that was compatible with a particular analysis. Teflon (Membrana) filters (47-mm diameter) were weighed at low relative humidity before and after sampling (air flow 10 L/min) to determine the dry aerosol mass concentration. Samples collected on Teflon filters were analyzed by X-ray fluorescence (XRF) to quantify the concentration of 34 trace elements (ranging in atomic weight from aluminum to lead). Aerosol samples for ionic species determination were collected on Nuclepore filters (47-mm diameter, 0.4-pm pore size, air flow rate of 5 L/min). Information on the concentration of the water-soluble ions NO, and Sod2was obtained by ion chromatography. A significant fraction of the fine particle ammonium nitrate may be lost from such filters during sampling (ca. 40% loss; ref 21), but data presented later in this paper show that fine particle nitrate concentrations were low in any case during these particular midday time periods. Coarse particle NaN03 measurements are not affected by aerosol volatilization during sampling. The samples taken on Nuclepore filters also were analyzed by atomic absorption for the concentrations of Na+, K', Mg2+,and Ca2+. The concentration of sulfur, potassium, and calcium obtained by XRF was used for subsequent calculations. Aerosol samples collected on quartz fiber filters (Pallflex QAO, 47-mm diameter, air flow rate 10 L/min) were analyzed by the method of Johnson et al. (22) by Dr. James Huntzicker and co-workers at the Oregon Graduate Center to determine the organic and elemental carbon concentrations. These filters were prefired to 600 "C for 2 h prior to use in order to reduce their carbon blank. Although there is no absolute standard available by which to judge the accuracy of elemental carbon measurements, the results obtained by this method are internally consistent and include a realistic pyrolysis correction. At Pasadena, the filter-based sampling system was accompanied by a photographic record of visual conditions and by measurement of the aerosol size distribution, relative humidity, and solar radiation intensity. The aerosol size distribution was measured with a Thermal Systems Inc. electrical aerosol analyzer (EAA) over the particle diameter range from 0.0075 to 1.0 pm and with a Particle Measuring Systems Model CSASP-100-HV optical particle counter (OPC) in 36 particle diameter intervals over the range from 0.5 to 47 pm. Measurements of total solar radiation intensity were made with an Eppley Laboratory pyranometer (Model PSP). To document the visual conditions at Pasadena, photographs were taken of five standard vistas [same scenes and photographic conditions as in Larson et al. (IS)]at noon (PST) during each sampling period. Data Analysis The data set acquired during the 1984 summer experiments will be evaluated to describe the Los Angeles summer midday visibility problem in terms of the distribution of observed visual ranges and corresponding pollutant concentrations and composition. The data set will be used to calculate the extinction and scattering coefficients.
1
100 PERCENT 75
PERCENT 75 DAYS EST BEXT LESS
DAYS OF
i
OF 50
-1
1
EST BEXT
50
I
LFNNOX-LAX
PASADENA-BURBANK 0
I
0
2
4 6 8 ESTIMATED BEXT
1
.j"
0
0
2
(XlE'K')
(X10'K')
100
1
PERCENT 75
PERCENT OF DAYS EST. BEXT
OF DAYS EST BEXT LESS THAN SHOWN
50
-
25
-
0.
4 6 8 ESTIMATED BEXT
LESS THAN SHOWN
0
I(
SAN BERNARDINONORTON AFB
- 0 0
2
4 6 8 ESTIMATED BEXT ( X 1E' K' )
1
0
0
2
4 6 8 ESTIMATED BEXT
1
0
( X 1E'K1)
Flgure 2. Frequency distribution of summer midday extinction coefficient values estimated from airport visual range values. Days on which the experiments reported in this paper were performed are shown by circles (0).
Midday Extinction Coefficient Values. The frequency distribution of estimated summer extinction coefficient values at Los Angeles area airports over the midday period (1000-1400 hours PST) is shown in Figure 2. These values were obtained by conversion of human observer visual range data into estimated extinction coefficient values using Koschmieder's formula with the contrast threshold ( A in eq 2) set to 0.02. The solid curve is formed from data for all days of record during the months July, August, and September, while the circles mark the estimated extinction coefficient values on the experiment days. The aerosol sampling events are distributed over the range of visibility events that occurred that summer. The 50th percentile estimated midday extinction coefficient value increases from 2.3 X m-l (corresponding to 17.0-km visual range) at the coast near Los Angeles International Airport (LAX) to 3.4 X m-l (11.5-km visual range) at the inland site at Ontario. The lowest median visual range (highest median bext value) among the airport sites examined is at Ontario. These data are consistent with the spatial distribution of the long-term average airport visual range data for the Los Angeles area presented by Trijonis (23) which likewise show that the lowest average airport visibilities in this area occur near Ontario. The Los Angeles International Airport site, which experienced the lowest median extinction coefficient value during the summer of 1984, also experienced the highest single day worst case extinction coefficient value, as seen in Figure 2. Clearly, no single statistic adequately conveys the visibility differences between the sites studied, illustrating the merit of describing regional visibility problems in terms of the frequency distribution of visibility events. Aerosol Characteristics. To provide a description of aerosol properties to support scattering and absorption calculations, measurements of aerosol properties made during the summer of 1984 were used. To construct a material balance on the chemical composition of the aer0801, the procedure of Larson et al. (18) was followed. In
Table 11. Speciation of Ionic Material Nat was associated with C1NH4+was associated with S042NH4+remaining, if any, was associated with NO, Nat remaining, if any, was associated with NO< remaining, if any Na+ remaining, if any, was associated with Sod2remaining, if any
this procedure, trace metals measured were converted to their common oxides (24). The mass of organic carbonaceous material was taken to be 1.4 times the organic carbon mass measured (25). The mass of NH,+ in the aerosol was taken to be that needed to neutralize the aerosol. The speciation of ionic material was assigned according to the scheme in Table 11. A portion of the airborne particulate matter may not be identified by the chemical analyses performed. Using the nomenclature of Sloane (26),this mass of dry material that is measured gravimetrically, but not accounted for in the chemical analysis is referred to as the "residue". The aerosol water content was estimated by the semiempirical approach formulated by Sloane (26), where data on the ambient relative humidity and the particulate matter solubility are used to estimate the amount of water present in the aerosol. The results of the analysis of the filter samples are presented in Figures 3 and 4, where the composition of fine and coarse particulate matter for the 1000-1400-hour (PST) period on experiment days for each site is shown. Table I11 lists the average chemical composition for each site for the fine and coarse particle modes. Results indicate the presence of sulfate in the fine particle mode, with nitrates appearing mostly as coarse suspended particulate matter. Elemental carbon and organic material are important contributors to the fine aerosol component and also are present in the coarse particulate matter. Crustal material (included in the catagory of "other identified" material) makes, as would be expected, a greater contriEnviron. Sci. Technol., Vol. 23,
No. 3, 1989 283
160
120
160.
1
!
120.
U
G
'5 1'1 1'7 2 3 29 JULY
'4 1'0 1'6 2 2 28 RUGUST
LE"oX
'3 '9 1'5 2 1 27 SEPTEMBER
160.
1
1
I
JULY
AUGUST
SEPTEMBER
I
I
AZUSA
UPLAND
U
G
11 17 23 29
160
120,
1
JULY
4 18 16 22 28 AUGUST
9 15 2 1 27
3
11 17 23 29 JULY
SEPTEMBER
4 18 16 22 28 AUGUST
3
9 15 2 1 27 SEPTEMBER
1
SULFATES
SRN BERNARDINO
NITRATES
U
ELEMENTAL CRRBDN
G
ORCRNIC MRTTER OTHER I D E N T I F I E D
40
RESIDUE WATER
0 1'1 17 2 3 29 JULY
'4 1'0 16 22 28 AUGUST
'3 '9 1'5 21 27 SEPTEMBER
Figure 3. Chemical composition of the fine suspended partlculate matter at each site. Values shown are 4-h averages for the period 1000-1400 hours (PST).
Table 111. Average Chemical Composition of Aerosol Lennox
Pasadena
Azusa
Upland
San Bernardino
14.8 2.6 4.8 26.6 7.5 37.6 6.1 58.08
14.8 2.9 4.2 36.9 11.8 25.1 4.4 34.08
fine Suspended particulate matter, av % by mass sulfates nitrates elemental carbon organics other identified residue water fine mass, pg/m3
21.3 3.0 3.8 19.8 7.4 27.7 17.0 51.99
18.0 3.4 4.4 32.0 8.1 24.4 9.7 55.60
13.4 1.8 5.3 25.3 7.9 38.4 8.0 78.30
coarse suspended particulate matter, av % by mass sulfates nitrates elemental carbon organics other identified: -41203
SiOl Fe20, remaining residue
water coarse mass, 4 r n 3
5.8 20.1 1.8 13.7
3.5 21.4 2.5 9.1
2.8 20.7 0.8 8.0
1.7 21.5 1.2 13.7
4.8 11.6 1.3 11.2
3.5 14.0 2.4 9.5 18.6 10.7 63.58
7.8 27.3 3.3
8.5 31.1 5.0 7.4 11.4 4.4 90.67
8.5 30.4 4.9 8.8 6.9 2.4 52.53
6.3 21.3 2.8 7.6 29.1 4.1 94.83
1.5 12.7 4.9 38.85
bution to the coarse material. There is noticeably more coarse airborne crustal material at the eastern sites (Azusa, Upland, and San Bernardino) than at the more westerly 284
Environ. Sci. Technol., Vol. 23, No. 3, 1989
locations (Lennox and Pasadena). Aerosol water is estimated to constitute from 4.4% to 17.0% of the fine aerosol on average at midday during the summer. Waggoner et
200
200
LENNOX
PASRDENR U
G
h3
100 50
50
0
0
5 1 1 17 23 29 4 10
JULY
RUGUST
JULY
SEPTEMBER
AUGUST
SEPTEMBER
AUGUST
SEPTEMBER
200
200
UPLAND 150
:
U
150
G
G3 100
i
50
0
1'1 17 2 3 JULY 200
29
4 1'0 16 22 28
3
9 15 21 27
SEPTEMBER
RUGUST
,
SAN BERNRRDINO
.
JULY
SULFATES NITRATES
,..^
ELEMENTAL CARBON ORGANIC MRTTER OTHER I D E N T I F I E D RESIDUE WATER
1 1 1'7 2 3 2 9 JULY
4 10 16 2 2 28 AUGUST
3
9 15 21 27 SEPTEMBER
Figure 4. Chemical composition of the coarse suspended particulate matter for each site. Values shown are 4-h averages for the period 1000-1400 hours (PST)
al. (27) find a 14% increase in light scattering between a Pasadena aerosol at 30% relative humidity and the same aerosol at 60% relative humidity. The relative humidities a t midday during the 1984 summer were observed to average only 45.0% (range 14.6-67.2%), and the small amount of water estimated to be present in Table I11 is consistent with the data of Waggoner et al. Scattering Coefficient Calculations. At the Pasadena monitoring site, aerosol chemical characteristics and size distribution data were available. At that site, the atmospheric aerosol scattering coefficient and the total light extinction coefficient were computed by the data reduction procedure of Larson et al. (18). Using the measured masses of individual chemical constituents, the calculated amount of water present in the aerosol, and the densities of these components, the contributions to total aerosol volume were calculated. Assuming an internally mixed aerosol of uniform chemical composition in the fine particle and separately in the coarse particle modes, and using the volume contributions, the volume average refractive indexes for the fine and for the coarse modes of the aerosol size distribution were computed based on the refractive indexes for each component. The fine and coarse particle refractive indexes and aerosol size distribution data then were used to calculate the aerosol scattering coefficient, bSmgat Pasadena via a Mie scattering code (28).
By combining these scattering coefficient predictions with data on light absorption by suspended particulate matter, light absorption by NOz, and scattering by air molecules, midday values of the total atmospheric extinction coefficient were computed at Pasadena. The particle absorption coefficient, babs,, was obtained by multiplying the measured total elemental carbon concentration by the light absorption efficiency of Los Angeles elemental carbon of 11.9 mz g-l measured by Conklin et al. (29). The gaseous absorption coefficient, babs,, was determined from the product of the light absorption efficiency of NOz and the NOz concentration (16, 30, 31). Tabulated values of the Rayleigh scattering coefficient for the atmospheric gases were taken from Penndorf (32)and were corrected for ambient temperature on each experiment day. Results of the calculated contributions to the atmospheric extinction coefficient a t Pasadena are presented in the upper right-hand corner of Figure 5. Larson et al. (19) have shown that nearly identical scattering coefficient values would be calculated for this data set if an externally mixed aerosol had been assumed rather than an internally mixed aerosol. Hasan and Dzubay (33)also found little difference in extinction coefficient values for internal versus external mixtures. Model Evaluation at Pasadena. To test the validity of the calculations at Pasadena, modeled scattering coefficient values were compared to the scattering coefficients Environ. Sci. Technol., Vol. 23,
No. 3, 1989
285
12.
12
I
PASADENA
LENNOX 9.
EXTINCTION COEFFICIENT
EXTINCTION COEFFICIENT (xl0-w) 6
RT 556 NM
9T
550 NM
3
3 0
‘5 1’1 1’7 23 29 JULY
‘4 1’0 1’6 22 28
AUGUST
‘3 ‘9 1’5 2‘1 27
0
’
5 11 17 23 29 JULY
SEPTEMBER
4 10 16 22 28
AUGUST
9 15 21 27
3
SEPTEMBER
12
12 AZUSA
EXTINCTION COEFFICIENT
EXTINCTION COEFFICIENT
tx10-%’, 6 .
(xl0-k’) 6
RT 550 NM
RT 550 NM
3
i
t
UPLAND
0
5 11 17 23 29
‘5 1’1 1’7 23 2‘9 ‘4 1’0 1’6 22 2‘8 ‘3 ‘9 1’5 2’1 27 JULY AUGUST SEPTEMBER
JULY
4 10 16 22 28
AUGUST
9
3
9 15 2 1 27
SEPTEMBER
SCATTERING BY AIR MOLECULES
EXTINCTION COEFFICIENT tXlE-’k’> A T 550 NM
ABSORPTION BY NO2 CAS ABSORPTION BY PARTICLES
ea
3
SChTTERINC BY COARSE PARTICLES SCATTERING BY FINE PARTICLES
0
12.
9.
MODELED SCATTERING COEFFICIENT (X18‘ M-’ )6’ AT 466 NM
WITH
BEXT
50
APREDICTED VRLUES
- ESTlMRTED DISTRIBUTION PRSRDENR 0 3 6 9 12 MEASURED SCATTERING COEFFICIENT (X10-‘ M-’
>
0
{k- 82)
PASADENABURBANK 2 4 6 8 EXTINCTION COEFFICIENT ( X l@K’ >
0
Figure 8. Measured versus modeled scattering coefficient and extinction coefficient values for the Pasadena site. Line with unit slope added to frame (a) as a guide for purposes of visualization.
measured by a nephelometer located at that site. The modeled extinction coefficient values also were compared to the estimated extinction coefficients implied by the human observer visual range data acquired at the closest airport. The wavelength dependence of the components of the extinction coefficient must be taken into account when making these comparisons. The human eye has maximum sensitivity a t a wavelength of approximately 550 nm (13). This wavelength is the value used for defining the standard visual range used in Koschmieder’s formula (34). When comparing the estimated extinction coefficient values against modeled extinction coefficient values, the calculations were carried out a t a wavelength of 550 nm. 286
Environ. Sci. Technol., Voi. 23, No. 3, 1989
However, the light source in the MRI 1550 nephelometer has a broad spectrum centered more about the blue wavelengths. In this model nephelometer, the light source is a xenon flash tube equipped with an ultraviolet cutoff filter. The Model 1550 has been estimated to have a “peak wavelength” of 480 nm (5) and a “peak sensitivity” of between 460 and 490 nm ( 4 ) . Harrison (5) reports “effective wavelengths” in the range of 479-488 nm, and early instrument calibrations were based on an assumed effective wavelength of 460 nm (4,35,36). Since calibration procedures assumed a wavelength of 460 nm, this wavelength was used in the calculation of the scattering coefficients that are compared against nephelometer measurements.
12
12 AZUSFl
UPLOND
9 MODELED SCATTERING COEFFICIENT
(xis-4
9 MODELED SCATTERING COEFFICIENT
-
~ - 1)6
AT 460 NM
( x m 4 M-I
)6
-
3
-
AT 460 NM
0
3..
o"o
rnql""""
0
"t
I
0""
"
$0"
"0
I
I
MEASURED SCATTERING COEFFICIENT
0
I
I
I
2 MEASURED SCATTERING COEFFICIENT
( ~ 1 0 M-' -~) (X10-4 M-' ) Figure 7. Measured versus modeled (wavelength 460 nm) scattering coefficients for the cases where calculations are based on rescaling the Pasadena aerosol size distribution to match the aerosol volume measured at these sites.
Direct comparison of the measured and modeled scattering coefficient values (wavelength 460 nm) at Pasadena is shown in Figure 6a. The frequency distribution of the modeled extinction coefficients at Pasadena (wavelength 550 nm) plotted with the distribution of the approximate extinction coefficients derived from visual range observations at Burbank Airport (threshold contrast of both 0.02 and 0.05) is shown in Figure 6b. Measured and predicted scattering coefficient values at Pasadena are highly correlated (r = 0.83). On average, the predicted scattering coefficient values are lower than actual observations by 26%. That degree of agreement is comparable to that obtained by other recent investigations. Ouimette (17) reported an average ratio of scattering coefficients as calculated by Mie theory to scattering coefficients as measured by a nephelometer of 0.85 f 0.34 for Zilnez Mesa in Arizona. For a China Lake, CA, site the calculated values overestimated measured values by an average of 52 70. Agreement to within 7-11 % was found by Dzubay and Clubb (6)between telephotometer measurements of the extinction coefficient and the extinction coefficient calculated by the sum of nephelometer b, measurements, opal glass bab$ determinations, values of b&.,leigh, and bakg, obtained from NO2 concentrations. Sloane (37) used Mie theory to calculate scattering coefficients with agreement between predicted and measured values of 5-3670. Sloane and Wolff (38) present a physical model for calculation of the extinction coefficient and find close agreement between predicted and measured scattering coefficients. In previous work, Larson et al. (18) obtained agreement to within 20% between measured and modeled scattering coefficient values. Frequency distributions of calculated extinction coefficients and of extinction coefficients estimated from visual range values are not expected to agree exactly due to errors in applying Koschmieder's formula and due to experimental errors. The confidence in the extinction coefficient calculation is strengthened by the favorable comparison between calculated and measured scattering coefficients. The frequency distribution comparison gives a further indication of the degree to which the calculated extinction coefficients are realistic. Examination of Figure 6b shows that median extinction coefficient values predicted from pollutant data at Pasadena and derived from visual ranges observed at Burbank airport are in close agreement. The highest extinction coefficient event a t Burbank is underpredicted by data taken at Pasadena. Exact agreement is not expected in this case. Burbank and Pasadena are separated by a distance of 16 km, and the pollutant properties at these two sites would easily differ by enough
to account for the differences shown in Figure 6b, especially under high pollutant loading conditions when Burbank and Pasadena are separated by a distance greater than the distance to extinction. Light Extinction Estimates at Other Sites. At monitoring sites other than Pasadena, aerosol size distribution measurements are unavailable. Previous studies (39) indicate that while the fine aerosol concentration (fine mass in pg mb3)varies from site to site in the western portion of the Los Angeles basin, the long-term average fine aerosol chemical composition (e.g., percentage of sulfates and elemental and organic carbon) is similar a t most sites. If size distribution properties were relatively similar from site to site, differing mainly in the total number concentration of particles present, then the relative aerosol size distribution measured at Pasadena might be used to approximate the aerosol size distribution at other monitoring sites. To test that hypothesis, the aerosol size distributions for the fine and the coarse particle modes measured at Pasadena were rescaled separately to match the fine and coarse aerosol volumes implied by the filter-based measurements made a t Lennox, Azusa, Upland, and San Bernardino on the same experiment day. The volumeaverage refractive index for fine and coarse aerosol was computed on a site-by-site basis by the method described previously. Predicted scattering and extinction coefficient values were calculated for Lennox, Azusa, Upland, and San Bernardino. Average contributions to the extinction coefficient at all sites are shown in Table IV, and the differences between individual days are shown in Figure 5. Scattering coefficient predictions were compared to measured values at sites equipped with nephelometers as shown in Figure 7, and the distributions of predicted extinction coefficient values were compared to extinction coefficients estimated from airport observer data at those sites near airports, as shown in Figure 8. The average scattering coefficients calculated at Azusa and at Upland under this approximation are 29% and 30% below the measured values, respectively. The estimated and predicted extinction coefficient frequency distributions (Figure 8) match as well for the Lennox-LAX site as they did for the Pasadena-Burbank site at which all model inputs were obtained by direct measurements at Pasadena. Lennox is located directly upwind of Pasadena during typical summer afternoons (40), and the two sites may have a similar particle size distribution. The agreement is less favorable for the Upland-Ontario site and for the San BernardineNorton Air Force Base location. At these sites, the higher extinction events are not reproduced well. The Environ. Sci. Technol., Vol. 23, No. 3, 1989
287
Table IV. Contributions to the Modeled Extinction Coefficient av % of total extinctn coeff"
fine particle scattering coarse particle scattering particulate absorption NO2 absorption Rayleigh scattering b,, (x lo4 m-l) Wavelength 550 nm.
Lennox
Pasadena
Azusa
Upland
San Bernardino
67.0 4.7 16.3 5.6 6.4 2.23
70.0
70.0 4.6 16.6 4.4 4.2 3.21
71.1 3.5 14.6 4.7 5.8 2.37
60.0 9.2 16.1 6.1 8.7 1.66
100
2.6 15.1 6.4 6.0 2.33
10a.
PERCENT 75 OF DAYS WITH BEXT 5 0 LESS THAN SHOWN 2 5
I
A
I 01 0.
PERCENT OF 75 DAYS WITH BEXT 50 LESS THAN SHOWN 2 5 LENNOX-LAX
!b,
UPLANDONTARIO
I
2. 4. 6. 8. EXTINCTION COEFFICIENT
4
0
I
2 4 6 8 EXTINCTION COEFFICIENT
(XlB'K')
(Xl@K')
/-
PERCENT 75 OF
. . .. . 25.
4 SAN BERNARDINO 0
2 4 6 8 EXTINCTION COEFFICIENT tXl@K'>
Figure 8. Measured versus modeled (wavelength 550 nm) extinctlon coefficient frequency diagrams for airports where calculations are based on rescallng the Pasadena aerosol size distribution to match the aerosol volume measured near those sltes.
scaled Pasadena aerosol size distribution in all likelihood does not accurately reflect the actual distribution a t these two locations, perhaps due to aerosol aging and growth or because those sites are downwind of a different set of emission sources.
Conclusions Summer midday (1000-1400 hours PST average) extinction coefficient vafues in the Los Angeles area range from less than 0.5 X lo4 (visual range of more than 78 km) to more than 9 X lo4 m-l (visual range of less than 4.3 km). By sampling a t five sites at 6-day intervals over the summer of 1984, it has been possible to collect a data base describing the distribution of midday aerosol composition and visual range values. The cause and effect relationship between pollutant properties and the frequency distribution of summer midday extinction coefficient values at Pasadena, CA, has been investigated. The procedure used combines on-site aerosol size distribution and chemical composition measurements with a calculation procedure based on theories of light scattering and absorption rather than on statistical techniques. Calculated scattering coefficients are highly correlated with measured scattering coefficients at Pasadena (r = 0.83) and agree on average within 26%. Reasonable agreement also is obtained between model calculations and the distribution of extinction coefficient values 288
Environ. Sci. Technol., Voi. 23, No. 3, 1989
estimated from observations of visual range at the airport nearest to Pasadena. The concept of size distribution "fingerprinting" was tested for its usefulnem in supporting visibility calculations a t sites other than Pasadena. Filter-based aerosol data taken at Lennox, Azusa, Upland, and San Bernardino were distributed by size in proportion to the aerosol size distribution measured on the same day a t Pasadena. At Lennox, reasonable agreement between predictions and observations was obtained, while at Upland and San Bernardino the high extinction events were poorly reproduced by this method. Therefore, on-site size distribution data like those taken a t Pasadena are recommended to support such modeling calculations. Since the calculation procedure that was demonstrated at Pasadena is based on fundamental aerosol and gaseous pollutant properties, the new distribution of visual ranges can be computed that would prevail if specific changes were made in the composition or size of the Pasadena aerosol as the result of emission controls (19).
Acknowledgments We thank Barbara Turpin for her assistance with filter handling and for her help during the field experiment8 and thank Kenneth McCue, Frank Vasquez, and Philip Lin for their help in preparing some of the graphs and statistical analyses in this work. Aerosol carbon analyses were per-
formed under the direction of James Huntzicker of the Oregon Graduate Center, and John Cooper of NEA Inc. directed the X-ray fluorescence analysis of aerosol samples. The South Coast Air Quality Management District provided information on gaseous pollutant concentrations. Registry No. Al, 7429-90-5;Pb, 7439-92-1;",+NO
