Sulfate air quality control strategy design

~.~981/81~71227-23 Q 1981 Rrgamon

Amosphcric Environmenr,Vol. 1.5, No. 7,pp. 1227-1249. 1981. Printed in Great Britain.

SOZ.OO/O Press Ltd.

SULFATE AIR QUALITY CONTROL STRATEGY DESIGN GLEN R. CASS Environmental Engineering Science Department and Environmental Quality Laboratory, California Institute of Technology, Pasadena, California 91125, U.S.A. (First received 5 May 1980, and infinalform

31 October 1980)

Abstr~%--An approach to the design of emission control strategies for sulfate air quality improvement is described. Methods developed are tested within a case study of the nature and causes of the high sulfate levek observed in the Los Angeles area. An air quality model for sulfate formation and transport is developed which computes long-term average sulfate concentrations using a Lagrangian marked particle technique. The air quality model is verified by application to the Los Angeles air basin during each month of the years 1972 to 1974. The time sequence of observed sulfate air quality is reproduced closely in spite of the fact that emission peaks are six months out of phase with observed peak sulfate concentrations. An important finding is that there is a seasonal trend in the rate of SO, oxidation to form sulfates in the Los Angeles atmosphere with conversion rateaveraging about 6% per hour in the late spring, summer and eariy autumn, and declining to between 3 % and 0.5 % per hour in winter months. The problem of identifying the least costly combination of emission controls needed to achieve a major sulfate concentration reduction is addressed using the air quality model results. Example calculations show that close to a 50 %improvement in sulfate air quality could be achievedin downtown Los Angeles at a cost of circa SlOOmillion annually. Since the effect on visibility of such a sulfate concentration reduction has been estimated previously, a partial remedy for the Los Angeles visibility problem is described.

1. INTRODUCTION

Particulate sulfates accounting for a few percent of the sulfur content of fuel are emitted directly from most combustion processes. Additional sulfates form from atmospheric oxidation of SO2 downwind from sulfur oxides sources. These sulfate particles contribute to visibility deterioration (Eggleton, 1969; Charlson et al., 1974; Waggoner d al., 1976; Weiss et al., 1977; White and Roberts, 1977; Trijonis and Yuan, 1978ab; CBS, 1979) and to the acidification of rain water (Cogbili and Likens, 1974; Likens, 1976) throughout the United States and Europe. Visibility reduction due to light scattering by submicron aerosols, including sulfates, is so pronounced that emission control strategies designed for sulfate abatement are likely to be considered by governmental air pollution control agencies. The costs of such a control program will be high. In the Los Angeles area, for example, $100 million in annual emission control expenditures would be required to obtain about a 50 “i, reduction in sulfate concentrations (Trijonis et al., 1975; Cass, 1978; South Coast Air Quality Management District, 1978). With such a large committment of resources at stake in those portions of the United States and Europe which are affected by similar problems, it is important to develop methods which will identify the least costly means for sulfate air quality control. The purpose of this paper is to formulate an approach to the design of cost-effective emission control strategies for sulfate air quality improvement. A regional sulfate air quality model will be described 1227

and validated against historical air quality data. The modeling study will be structured so that two important objectives can be met. First, an explanation for the origin of pollutants cont~buting to Los Angeles sulfate air quality will be provided. Secondly, this will be done using a model which is designed to assist a later study of sulfate abatement alternatives. Total concentration estimates will be built-up by superposition such that the effect of alterations in some source types but not others can be evaluated without repeating expensive air quality modeling calculations. Information on the air quality impact of different emission source types will be combined with data on emission control measures and their costs. Emission control strategies capable of achieving close to a 50 y0 reduction in sulfate concentrations in Los Angeles will be described. Since the visibility consequences of such a sulfate concentration reduction have been assessed previously (Cass, 1979), the net result is a procedure for engineering a specified improvement in visibility.

t. CHARACTERISTICSOF LOS ANGELES SULFATE QUALITY

AIR

Two areas of the United States are affected by high sulfate concentrations: the entire eastern United States and portions of southern California (National Research Council, 1975). The Los Angeles sulfate problem was chosen as the testing ground for the approach developed in this study because it is selfcontained: it spans a small enough geographic area that one can draw a box around it and analyse it in great

1228

detail. In the process, insights will be acquired that should help those charged with control of sulfate problems of great geographic extent in the eastern United States and in Europe. But what are the characteristic features of the Los Angeles sulfate problem which could be used to guide air quality model formulation and to check the physical consistency of such a model? That question will be addressed by analysis of the extensive historical air quality data base accumulated in the Los Angeles area. Data from 27 monitoring sites at the locations shown in Fig. 1 have been anaiysed as part of this study. Data acquisition and preparation are described by Cass (1978). Particular attention is paid to the years 1972 to 1974. During those years the Los Angeles Air Pollution Control District (LAAPCD), the U.S. Environmental Protection Agency’s CHESS program and the National Air Surveillance Network (NASN) monitoring programs operated concurrently, yielding the widest geographic coverage of sulfate air quality data available at the start of this investigation. Sulfate concentrations measured in remote areas of the Pacific Ocean and the Mojave Desert average about 1 pgm- 3 (Junge, 1957; Gillette and Blifford, 197l), rising to 3-5 pg m _ 3 at near-basin locations like San Nicolas Island and Lancaster (Hidy et al., 1974; MacPhee and Wadley, 1975a, 1975b). A sharp contrast exists between the sulfate concentration in incoming marine or desert air vs. that observed at the central Los Angeles Basin monitoring stations shown in Fig. 1. In contrast to the problem of long distance transport of sulfates in the eastern United States, a sulfate air quality model can be validated in the Los Angeles area while employing only local emissions data plus a small increment from background sulfates. Sulfate concentrations observed at the downtown Los Angeles station of the Los Angeles Air Pollution Control District during the period 1965 to 1977 are shown in time series in Fig. 2a. Concentration fluctuations are quite large, with high values occurring at least occasionally in all seasons of the year. However, the data can be smoothed to reveal seasonal trends, as shown in Fig. 2b.i When this is done, a broad summer seasonal peak in sulfate levels is apparent in all years of record. During the winters of 197@71 and 1971~72, isolated days of very high sulfate concentrations occurred which led to elevated annual averages for those years. A successful air quality modeling study

* The graph in Fig. 2b was generated by passing the time sequence of 24 h average sulfate readings over the period of interest through a linear digital filter. The effect of this urocessing is to reveal long-term air quality trends by suppressing fluctuations with frequency greater than four cycles per year, leaving seasonal variations intact. The filter’s characteristics are such that it returns the low frequency signal with unit gain, half power cutoff set to remove disturbances with period shorter than three months, and roll oB at the half power point of 20dB per octave. For a discussion of digital Altering methods see Bendat and Piersol (1971) Chapter 9.2.

should attempt to account for the origin of both the summertime and wintertime peak sulfate levels observed in Los Angeles. In Fig. 3, the same filtering process has been applied to the LAAPCD and CHESS sulfate data at all monitoring stations active during the period 1972 to 1974. NASN data are not presented because their infrequent monitoring schedule provides insufficient data for this sort of treatment. The similarity of seasonal pollutant patterns at all monitoring stations is quite striking. The timing and relative magnitude of seasonal concentration peaks and troughs is apparently related from Thousand Oaks in the northwest to Vista on the south, a distance of nearly 200km. This near equality of sulfate concentration trends at widely separated locations leads to long-term average sulfate concentrations of the same magnitude over most portions of the basin as shown in Fig. 1. The tendency toward spatial uniformity of longterm average sulfate concentrations within the Metropolitan Los Angeles area is in marked contrast to the highly localized nature of the major sources of precursor sulfur oxides shown in Fig. 6. Emissions inventory data for the South Coast Air Basin (Hunter and Helgeson, 1976; Cass, 1978) show that over 75 7; of the SO, emissions from major point sources in 1974 originated from sources located adjacent to the coast. A daily sea breeze/land breeze reversal in wind direction is typical of Los Angeles with net transport inland along streamlines roughly normal to the coast (see DeMarrais el al., 1965). For a conserved or slowly decaying pollutant emitted from major coastal point sources, one expects pollutant concentrations to drop greatly with distance downwind as atmospheric dispersion and removal processes come into effect. Sulfur dioxide concentrations do decline with distance inland and crosswind from the largest grouping of emissions sources located in the Long Beach Harbor area. The contrasting constant sulfate levels with distance inland from major sulfur oxides sources as illustrated by the data of Figs 1 and 3 require explanation. Can the demonstrated sulfate enrichment above background in the Los Angeles metropolitan area be explained reasonably in terms of local sulfur oxides emission sources? And if so, how do sulfur oxides emissions which are concentrated at a relative handful of locations along the coast become mapped into a longterm average air quality pattern with comparable sulfate concentrations observed over such a large geographic area?

3. A MODEL FOR LONG-TERM

AVERAGE SULFATE AIR

QUALITY UNDER UNSTEADY METEOROLOGICAL CONDITIONS

Long-term average sulfate air quality modeling studies have been reported for the eastern United States and for western Europe. Eliassen (1978) explored a mul-

kEO

;x

BACKGROUND

CUESS

A

NONIlORINO

STATION

AMNCY

Fig. 1. The South Coast Air Basin which surrounds Los Angeles showing 1972-1974 mean sulfate concentrations at the air quality monitoring sites studied.

m

1230

GLEN

R.

CASS

0

1965

64

1966

1997

1968

1989

1970

1971 YEAR

1972

1973

I9N

197s

1976

1977

Fig. 2a. LAAPCD sulfate data at Los Angeles.

YO-

IO’t (W

1 , 18551969

I 1967

I 1968

1969

1970

1971 YERR

1 1972

I 1973

I 1974

1 1975

197%

I 1977

Fig. 2b. Filtered sulfate data at Los Angeles. tiple source model for sulfate formation in Europe based on tracking the fate of Lagrangian particles within a mixed layer of assumed constant 1OOOm thickness above the ground. Lagrangian puff models recently have been applied to sulfate air pollution problems in western Europe by Johnson et 01. (1978) and to the eastern United States by Sheih (1977). among others. The air quality model employed in the present study computes pollutant concentrations from long-term average source to receptor transport and reaction probabilities. These. transport and reaction probabilities in turn are estimated from Lagrangian marked particle statistics based on the time sequence of historical measured wind speed, wind direction, and inversion base height motion within the airshed of interest. First order chemical reactions and pollutant dry deposition are incorporated. The model is oriented toward identification of the emission source types contributing to air quality predictions at each receptor site in a way that facilitates later selection of emission control equipment.

In Lagrangian air quality models, dispersion is described by the trajectories of representative fluid particles. Pollutant concentrations at downwind receptors are determined by the probability that a pollutant-laden fluid parcel emitted at a known time, t’. and place x’, will occupy a given location, x, in the airshed at later time t. Thus most Lagrangian air quality models in current use can be expressed in the form:

x

S(x’, r’)R(r, r’) dt’ dx’

(1)

where (c(x, t) ) is the ensemble mean pollutant concentration at point x at time t; Q(x, tlx’, t’) is the transition probability density that a fluid particle at location x’ at time r’ will tmdergo a displacement to location x at time t; R(t, I’) is the probability that a fluid particle will retain its chemical identity at time t’ until time r, and S(x’, t’) is the spatial-temporal distribution of emission sources (Lamb and

Sulfate air quality control

1231

1232

GLEN R. C‘ASS

Neiburger, 1971; Lamb, 1971; Lamb and Seinfeld, 1973). Equation (1) is valid only for first-order chemistry because each fluid particle’s probability of undergoing chemical reaction is represented as being independent of the chemical conversion of other components of that species elsewhere in the system. Long-term average air quality models based on Equation (1) directly calculate long-term average poilutant concentration as a function of location in the airshed: (c(x)> =-

1

T

7 so

where T is the averaging

(c(x,t))dt

(2)

time. For a conserved

pollutant, (c(x) ) is efficiently obtained by directly integrating over a long-term average source to receptor pro~bility density function, & rather than by averaging over the results of a long time series of fully determined calculations of ins~n~n~us pollutant concentrations, (c(x, r) >. The form of 3 is easily obtained if meteorological behavior can be viewed as a sequence of steady state conditions. Steady meteorological conditions exist when wind speed, wind direction and atmospheric stability persist for time periods longer than the characteristic time for pollutant advection to beyond the air basin boundaries. In that case c is often approximated by superposition of Gaussian plume formulae weighted by the frequency with which combi~tions of wind speed, wind direction and atmospheric stability fall into defined classes. In Los Angeles, for example, the characteristic times for wind reversal, inversion base change, and air stagnation periods are shorter than transport times out of the air basin. An air parcel may wander in the basin for more than a day until a high speed wind event clears it from the airshed. A long-term average air quality model for sulfate formation in such situations will have to cope with unsteady meteorolo~~l conditions. Lamb and Seinfeid ( 1973) suggest that one feasible way of determining source to receptor transport probabilities under such conditions is by means of a numerical simulation. Analytical solutions like the Gaussian plume formula are unavailable. Additional complication is provided in cases where pollutants temporarily located within the stable layer above the inversion base are not affected by ground level removal processes, while pollutants located within the mixed layer adjacent to the ground are subject to depletion by surface removal reactions. The result is then that air parcel transport and chemical status no longer can be calculated independently over very long times as was the case for situations which could be modeled by Equation (1). However, these problems can be approached as follows. A large number of mass points representing pollutant emissions can be inserted into a mathematical representation of atmospheric fluid flow. Those mass points are propagated by a stochastic chain through a simulation of unsteady transport and re-

action processes. By averaging over the fate of a large number of fluid particles, a source to receptor transport and reaction probability density can be determined by inspection. Depending on the rules written for propagating particle status, that density function may contain embedded within it the effect of simultaneous pollutant reaction and deposition, wind reversal, inversion base motion, and source initial conditions such as effective stack height and the partition of sulfur emitted into SO, and primary sulfates. Formal derivation of the model developed for use in this study is given by Cass (1978); a summary of the calculation approach follows. Single particles marked with the initial chemical composition of sulfur oxides emissions from a source (fraction SO,; fraction SO:or SO,) are inserted at measured time intervals into the atmosphere above the location of their point of origin. binding on the plume rise characteristics of each source and meteorological conditions at the time of release, a pollutant parcel may be inserted either above or below the base of the temperature inversion which separates a well mixed layer next to the ground from a stable air mass aloft. Each particle carries with it six probabilities which describe its chemical status over time. They are PsoJr 1lo, 0, Pso~,@ 1to, ib Pso,(r ( co, 0, Pso,,

it j go, 0, Pso~,

(t 1t,,

4 and

Pso,~@

/ to, 9% where

the

subscripts denote the probability that the fluid particle is present as SO2 or sulfates (SO,) either above (a) or below (b) the inversion base, or has deposited (d) at the ground by time t, given that it was emitted from source class i at time to. At the time of pollutant release, these probabilities are initialized such that either PSOl,(fO,0 +

PSO,,(~O~

4 =

1

(3)

Psorb(ro, 0 f

PsO,,(tO,

i) =

1

(4)

or while all other chemical descriptors are initially set equal to zero. Whether the parcel is inserted above or below the inversion base is determined by meteorological conditions at tbe time of release and by source stack parameters keyed to index i. The partition of source emissions at time of release into SO2 and sulfates is based on source test data for each source type i. Each fluid particle also carries with it a weighting function, o(t,, i), proportional to the diurnal variation of emissions source strength. As these sulfur oxides laden air parcels are transported downwind, chemical reactions and surface removal processes act to alter the probable chemical status of each particle. Sulfur oxides residing within the mixed layer next to the ground are affected both by ground level dry deposition and by atmospheric oxidation of SO, to form additional sulfates: (5)

Sulfate air quality control

Fig. 4. Hypothetical time history of interaction between the in\rersion base and a fluid particle released at time to (O’sindicate a single particle stabilized within the inversion; zone between arrows ( 5) indicates that vertical mixing has occurred while the particle resided below the inversion base.) OBJECTIVE

Stgl:

To compute theaverage poWant concentration observed resulting from a unit source located at x0.

Supwimpse slreaklms cmbW dl fluidpoiicks ObserWdjOne streokline for each tow of the mmth

ste92:

Locate particles within the cells of 0 receptor grid

stsp3:

Aaunubte the rmqnlbdes ossociaM wth the particles falling wilMlecdlgridcall

step 4:

Divide ttm accumubted FlolMult mmbodingbytiskeofthe WXJplWcsllandfh8MlllbaOf “hart’~supYhpoaed

ROWIt:

The-tOrOCOptOrrelotanrtdp beencalcubtad~maps

l-m

erdeei3tlefmnoMtscumot bcaiten point0 avemgepol~tont -hotion

obeewed

Fig. 5

1233

GLEN

1234

where Sso,, and Sso_ denote sulfur atoms present as SO2 and sulfates below the inversion base, V, and VP are the deposition velocities for SOZ and sulfates respectively, h is inversion base height and k is the pseudo-first order rate of SO2 oxidation due to all competing chemical processes. Pollutant parcels located within the elevated stable layer are isolated from surface removal processes but still can participate in chemical reaction processes. Exchange of air parcels between the mixed layer next to the ground and the stable layer aloft occurs as inversion base height changes over time, as depicted in Fig. 4. Equations used to propagate the status of particles between time steps, At, have the property that Pso,. + PSOlb+ Pso,* + Pso,, + Pso*, + Pso,l = 1 (7) at all times. Transport calculations in the horizontal plane are described schematically in Fig. 5. The trajectories of successive particles released from a source form streaklines downwind from that source. Streaklines present at each hour of the month are computed and superimposed. The horizontal displacement of each particle located below the inversion base is paired with the particle’s relative diurnal source strength and probable chemical status and divided by the depth of the mixed layer at the time that the parent streakline of interest was computed. The resulting magnitudes are assigned to a matrix of receptor cells by summing the contribution for all particles falling within the same receptor cell. Totals are accumulated separately for SOZ and for sulfates. The accumulated totals are divided by the dimensions of a receptor cell and the number of streaklines being superimposed in order to directly obtain a spatially resolved estimate of the long-term average source to receptor transport and reaction probability density functions, &o, and Eso,, for SO1 and sulfates due to the source type of interest. That result can be scaled by the absolute magnitude,X of the source of interest in order to obtain the incremental pollutant concentration due to that source throughout the airshed. By repeating that process for each source in the airshed and superimposing the results onto an estimate of sulfate background air quality, a multiple source regional air quality model for sulfates is obtained. Superposition is permitted because all chemical processes are modeled in a form which is linear in emissions. The final governing equation for sulfate concentration becomes: < cso.(x; T, r,) > N

Q

K’

=

&,,(x CJ i=l

+

-m


where (c,,~(x;

(x,,i;

T

t,)~(x,,

4)

>

T, t,) ) is long-term

CASS

concentration as a function of horizontal location x in the airshed over a long time period T which began at time t,; ~SO,(~(~O, i; T, t,) is the long-term average source to receptor transport and reaction probability density function which maps total sulfur oxides emissions originating at location x0 from source class i into air quality impact at any location x over the averaging time of interest; X(x,, i) is long-term average sulfur oxides emission source strength at location x0 from source class i; and &,Jx; T, r,,> is an estimate of sulfate background air quality. The long-term average Lagrangian marked particle air quality model has several particular merits. The order of integration over air parcel release and transport may be arranged to minimize computing time and intermediate data storage requirements. The calculations conserve mass, are simple, numerically stable, and free of computational diffusion. The model builds its own initial conditions by integrating backward to connect all locally emitted air parcels in the airshed at a given time to their source. Unlike Eulerian grid models, air parcels advected beyond the edges of the receptor region are not lost to the system. Their position is remembered but their magnitude is not accumulated to a receptor cell unless the air parcel is advected back into the region of interest. Receptor cells may be specified only over those areas where concerntration estimates are desired.

4. APPLICATION

OF THE MODEL TO LOS ANGELES

The air quality model was used to estimate the sulfate concentration patterns observed in the Los Angeles Basin during each month of the years 1972 to 1974. When adapting the model for use in Los Angeles, three approximations were made as a practical consideration aimed at conserving available computing resources. First, inversion base height above ground level over the central Los Angeles Basin was treated as being spatially homogeneous at any given time.t Secondly, it was assumed that inversion base motion could be represented by a stylized diurnal cycle which passes through the known daily maximum and minimum inversion base heights (see Fig. 4). Finally, at any single time, the wind field over the flatlands of the Los Angeles coastal plain (see the area shown in Fig. 6) was approximated as a uniform parallel flow (single carefully chosen wind station used to specify flow). The first and third approximations above result in a huge savings in computing time by permitting the separation of trajectory and chemical calculations from detailed dependence on a given starting location in the airshed. Model validation results indicate that these

W,

J -a

T

R.

(8)

average sulfate

t The reader may evaluate that approximation against the long-term average inversion base profile data given by Cass (1978), and the short-term data of Edinger (1973) and Blumenthal et al. (1978). Inversion base height above sea level rises as one goes inland from the coast, but so does ground elevation.

Sulfate air quality control approximations typically do not lead to errors in sulfate concentration predictions in this case which exceed the error bounds on the field air quality observations. The lack of appropriate measurements on winds aloft prevents a more detailed incorporation of the effects of wind shear and vertical transport within the application illustrated later in this paper. Input data and data preparation steps needed for model application are outlined in Table 1. An important feature of this study is that the input data used are based almost entirely on values reduced from experiments performed within the airshed of interest. The attempt is to insure to the extent possible that the modeling approach is being tested rather than assumptions in the data base. An exception to this rule is that SO, deposition velocity had to be estimated based on a concensus drawn from experiments performed elsewhere. The spatially

resolved inventory of sulfur oxides emissions required was compiled from a source by source accounting of pollutant emissions. Virtually all

1235

emissions data used are based on individual source tests, process sulfur balances, or fuel consumption and fuel quality reports. A grid system was laid down over the central portion of the South Coast Air Basin as shown in Fig. 6. Emissions estimates for both sulfur dioxide and primary sulfates resolved over that grid system were obtained for the 27 classes of mobile and stationary sources listed in Table 2 for each month of the years 1972 to 1974. The spatial distribution of average daily total sulfur oxides emissions during 1973 illustrated in Fig. 6 was obtained by overlaying similar maps developed for each source class of interest. Major off-grid sources located within the region shown in Fig. 1 also were surveyed for inclusion in the air quality model calculations. Energy and sulfur balance calculations performed on commerce data for fuels in the Los Angeles area were used to confirm independently that the emission inventory was accurate to within about 10 %. Emission inventory compilation and verification is described elsewhere (Cass, 1978).

Table 1. Data resources assembled for air quality model validation in Los Angeles Approach

Data requirement

(1) Spatially resolved inventory of total sulfur oxides emissions consolidated classes.

into i = 1, 2, .

, N source

(2) Specification of effective stack height for each source class as a function of time of pollutant release. (3)

Specification emissions.

(4)

Specification of the initial fraction of the SO, emissions which originated as SOs or sulfates, f,(i), from each source class.

(5)

An estimate of the maximum retention time for an air parcel in the modeling region, rc. A continuous record of hourly wind speed and direction data for each month of interest, plus the ~~ hours preceding the start of each month. A sequence of hourly estimates for inversion base height for each month of interest plus the r, hours preceding the start of each month. A function for estimating the rate of horizontal eddy diffusion in the form (at (t - to), u2 (t - to)). Estimate of SOa deposition velocity, V, Particulate sulfate deposition velocity, V, Atmospheric pseudo-first order SO, oxidation rate, k.

(6) (7)

(8) (9) (10) (11)

of the diurnal

variation

(12) An estimate of the seasonal variation background concentrations.

in source

in sulfate

(13) A choice of receptor call size in the horizontal plane (Axr. Ax,) and a time step, AI. (14) Ambiit monitoring data for comparison to model calculations.

Compiled by Cass (1978) for the 27 classes of mobile and stationary sources shown in Table 2 within each square of the grid system of Fig. 6 (plus major off-grid sources) for each month of the years 1972-1974 (see Fig. 7). Brings (1971) plume rise formula appl& to stack parameters obtained from the Federal Power Commission (Thomas, 1976), Hunter and Helgeson (1976), and Hunter (1975). Diurnal variation of electrical utility emissions from Sjovold (1973); diurnal variation of highway emissions from Nordsieck (1974); other source classes taken as operating without a pronounced diurnal cycle. Obtained from source tests by Hunter and Helgeson (1976) for stationary sources and from Pierson and Brachaczek (1976) and Pierson (1977) for mobile sources. Minor source types with no data assumed to have & (i) = 3 %. Maximum retention time of 48 hours for 95 % removal of trajectory end points from the 80 by 80km grid (see text). Data obtained from the Los Angeles Air Pollution Control District (1975a). Daily morning (4-7 a.m.) and afternoon mixing depth data

from the Los Angeles Air Pollution Control District (1975b) interpolated to form diurnal cycles. From experimental data of Fig. 8 in text. 0.7 cm s- r, from literature. 0.03cm s- *, from Davidson (1977) and Garland (1974). July 1973 average value of k = 8 % h-r supplied to model from Los Angeles area experimeits by ‘Roberts (1975). Average value of k for other months obtained by iteration on over-determined system of equations. Average value at San Nicolas Island from Hidy et al. (1974) scaled to the seasonal variation observed at Vista (the most remote near-coastal sampler with data for entire 1972-1974 period). Receptor cell size set coincident with source cell size (i.e. 3.2 km by 3.2 km); time step set equal to 1 h. Monthly and annual mean sulfate and total sulfur oxidea data for the years 1972-1974 from the Los Angeles Air PoUution Control District, EPA’s NASN Network, and EPA’s CHESS Network.

1236

GLEN

R.

CASS

Table 2. Sulfur oxides emissions summary Total sulfur oxides emissions* 1973 annual mean (tons/day as SO2 equivalents) On-grid sources + Stationary sources Fuel combustion Electric utilities Refinery fuel burning Other interruptible gas customers Firm natural customers

181.71 9.42 2.29 0.29

Major off-grid sources $

58.20

Primary sulfate emissions 1973 annual mean (tons/day as SO:-) On-grid sources +

8.18 0.42 0.10 0.01

Major off-grid sources $

2.62

Chemical plants Sulfur recovery plants Sulfuric acid plants Other chemicals Petroleum refining and production Fluid catalytic crackers Sour water strippers Delayed cokers Miscellaneous refinery processes Oil field production Miscellaneous stationary sources Petroleum coke calcining kilns Glass furnaces Secondary metals industries Primary metals (off-grid steel mill) Mineral products Sewage treatment Other industrial processes Permitted incinerators Mobile sources Autos and light trucks-surface streets Autos and light trucks-freeway Heavy trucks and busses-surface streets Heavy trucks and busses -freeway Airport operations Shipping operations Railroad operations Total

60.40 20.00 0.09

0.40 0.27 0.00

52.07 0.13 2.28 1.02 4.50

2.19 0.00 0.15 0.02 0.12

25.52 2.00 8.78 0.00 0.02 0.07

3.22 0.54 0.13 0.00 0.00 0.03 0.00 0.00

15.71 10.05 10.64 6.80 1.06 10.13 3.32

0.07 0.05 0.37 0.23 0.05 0.46 0.11

0.00 0.64

428.94

0.23 41.46 1.90

101.79

17.12

0.06 1.55 0.08

4.31

Total sulfur oxides emissions shown include both SOz and primary sulfates. t On-grid sources are those located within the 80 km by 80 km gridded area shown in Fig. 6. $ Large stationary sources locatedoutsideof the 80 km by 80 km grid were inveutoried if they emitted over 25 tons of So, per year. Emissions from small stationary sources and mobile sources in the o&grid area were @e&d after determination that their contribution was small. l

Figure 7 shows the time history of sulfur oxides emissions from sources located within the 80 by 80 km grid. An underlying increment to sulfur oxides emissions from mobile sources is observed which shows little seasonal variation. Added to that is a nearly constant contribution from miscellaneous stationary sources (principally from petroleum coke calcining kilns). Petroleum refinery process emissions are shown, mostly from refinery fluid catalytic cracking units. Emissions from chemical plants (which constituted the largest single emissions source class during 1972) decline sharply during our three-year period of interest as local sulfur recovery and sulfuric acid +t.s added new emission control equipment. A strong seasonal variation in emissions from fuel burning sources is noted as interruptible natural gas customers

switch to fuel oil in response to increased wintertime natural gas demand by high priority home heating customers. A severe test for air quality model performance has been identified. The air quality model must be able to track large changes in SO,Vemissions which are typically six months out of phase with the summertime peak sulfate concentrations observed. Hourly records of wind speed and wind direction at Los Angeles were obtained on magnetic tapa from the Los Angeles Air Pollution Control District (1975a). Daily data on the early morning (e.g., 47 am.) inversion base height above Los Angeks International Airport aad tbe afternoon maximum mixing depth above downtown Los Angeks likewise were obtained on magnetic tape (Los Angeles Air Pollution Control District, 1975b). These inversion base motion data

Sulfate air quality control

0.i

0.5

X 0.1

0.0

0.0

0.0 0.0

0.0 0.l

0.0

U.0

0.0

0.0

0.0

Fig. 6. Spatial distribution

oxidesemissions for ttK year 1973, in short tons per day stated as SO2 equivalents, Non-zero values in ocean reflect the presence of ship traffic.

of tcxai sulfur

were interpolated to form a stylized inversion base motion estimate for each hour of each day of the form illustrated in Fig. 4. An estimate of the maximum air parcel retention time within the modeling region is needed in order to determine how long each air parcel must be tracked. Trajectories starting at three major point source locations at the coast were calculated~for air parcels released at each hour of the years 1972 to 1974. It was found that a trajectory integration time, sq5, of 48 h was necessary in order to insure that over 95 % of the end points of trajectories of age r9s originating at major sources will be .outside of the 80 by 80km gridded area shown in Fig 6. Data on the rate of homonym d~~sion’~urring during fang distance (or long time) transport over

urban areas are rare. Therefore an estimate of long distance transport properties over the Los Angeles urban area was prepared from raw data on dispersion of SF, releases obtained from Drivas and Shair (1975) and Shair (1977). The Los Angeles data are plotted in Fig. 8 together with data from the urban area of St. Louis reported by McElroy and Pooler (1968). These limited data on dispersion rates in Los Angeles defy useful separation into distinctly different stability classes. What can be said is that the Los Angeles data fall within the band formed by the extrapolation of McElroy and Pooler’s observations under their least stable and most stable atmospheric conditions. The heavy solid line shown in Fig. 8 represtnts an estimate of the rate of horizontal dispersion drawn through all of the Los Angeles data with a slope the same as the

1238

GLEN

R. CASS

Fig. 7. Sulfur oxides emissions within the 50 by 50 mile square.

5

103

5 "b i i

102

5

ST

LaJlS

..+..ts .*..0...

lOlY ’ ’ ’ 1”“’ 10'

5

I 102

‘1’1’1’1

DISPEASWNSTUW

18 -22 -20 8 _,3

I

5

103

TRAVEL

TIME. seconds

I I’11111 5

I l@

“‘l”lJ 5

105

Fig. 8. Cross-wind standud deviation of tracer mnterkl aa P function of travel tim for St. Louis from McElroy

and Pooler (1968) and for L+osAn@ufrom Drives and St& (1975) and from sbrir (1977). HUvy solid+ine shows least squares fit for u,(r) fkom the Los Ad@s

band formed by McElroy and Poola’s results under unstabic and stable atmospheric combtions. That function has been used to calculate horizontal dispersion throu#out this model a&kation. The equation of that line is (a,(mA r(s)): cry= 1.73t*?

(9)

&m aRa having ttrarformed

the time axie BJ I’.*.

Typical values for So2 dry deposition vdocitics recommmdcd for mass bnlurce cakulations over gms&ndandwaterbyGarla&(1974)aadbyOwers and Powell (1974) m from 0.7 to 0.9cms-‘. The only attempt to integrate @to&ion v* ovor an entire urban area that we hoHe fouaM is due to Chambertin (i!%O).He estimated a deposition k&city

1239

Sulfate air quality control of 0.7cm s- l for SO1 removal during a London pollution episode. A value of 0.7cm s- ’ has been adopted for use in this study. Size distribution data for Los Angeles sulfate areosols show that sulfate mass in the Los Angeles atmosphere typically is strongly peaked in the accumulation mode with a mass median diameter of about 0.5pm. Particles of that size should have a dry deposition velocity of about 0.02 to 0.03 cm s- ’ over a high grassy field under typical Los Angeles

dry deposition velocities lie in the range 0.01 to 0.1 ems- l. A value of O.O3cms-’ has been used for sulfate dry deposition velocity calculations in this study. Liljestrand (1980) has shown that wet deposition of sulfur containing species over the whole surface area of the South Coast Air Basin accounts for about 2 % of the magnitude of the sulfur emitted to the atmosphere locally and is less than 10 % of the magnitude of the dry deposition of sulfur containing species occurring on an annual basis. Wet removal has been neglected in our calculations. SO1 oxidation rates have been measured in the Los Angeles atmosphere and found to vary from 1 to 15 % h- under both

measurements of pseudo-first order of SO, oxidation along cross-basin trajectories terminating in Pasadena. The average value of k from those experiments was just under 8% h- ‘. A value of SO, oxidation rate k equal to 8 y0 h- ’ will be used to test the air quality model during the month of July 1973. For the remaining months of interest the SOz oxidation rate is unmeasured. Some clues to the nature of the underlying process do exist. A summertime peak in the ratio of sulfates to total sulfur oxides is observed in the Los Angeles atmosphere (see Cass, 1978). This suggests that there is a seasonal dependence in SO, oxidation rate in the Los Angeles atmosphere which can be explored by use of the air quality model. Estimation of a seasonal trend in SO, oxidation rate is a tightly constrained process in spite of the fact that values of k must be obtained by iteration. That is because the system ofequations involved is highly overdetermined. The value of k which solves the chemical

reaction equations must result in sulfate air quality predictions which match observations at a large number of air monitoring sites. Certain relationships between sulfate and total sulfur concentrations must be satisfied within a model which conserves sulfur and which does not provide any means for adjustment of total sulfur concentration predictions. A prior estimate of the likely range of values for k exists. The only parameter in the model with any freedom of adjustment, k, must simultaneously satisfy a huge system of equations or inequalities.

5.

maximum SO2 rate of % h-’ to purely processes, which clearly too to explain of the made in Angeles. Heterogeneous of SO1 or within particles is important in the rapid oxidation Empirical studies Cass (1975) by Hidy al. (1978) that Los sulfate concentration not only days of photochemical activity, also track of high humidity and total suspended loadings which facilitate heterogeneous oxidation processes. (1975) showed fog or evidence of atmospheric liquid content present during out of days of high sulfate observed in Angeles over ten year Data on concentrations, ammonia and strong concentrations (e.g., sufficient to heterogeneous processes a practical quality model lacking at Therefore it decided that conversion of to form due to competing processes be modeled a slow order reaction the rate by local experiments. During 1973, Roberts made 11

RESULTSOF THE AIR

QUALITY

MODEL VALIDATION

Model results closely reproduced observed sulfate concentration patterns within the central portion of the Los Angeles Basin, particularly during the years 1972 and 1973. Figures 9-11 outline air quality model results in time series at several widely separated air monitoring sites. In the upper graph of each pair, the sulfate air quality mode1 results are represented by a continuous horizontal line. The small circles indicate the monthly means of sulfate observations at each monitoring site. The error bars represent a 95% confidence interval on the ambient air quality observations. Approximately 80% of the sulfate concentration predictions at Los Angeles Air Pollution Control District (APCD) air monitoring stations are within the error bounds on the ambient monitoring results. The results are particularly good at Pasadena during July 1973 (the site and time of Roberts (1975) reaction rate experiments) with all parameters in the model fixed by experimental data. Model predictions follow observed sulfate levels dosely at the critical CHESS stations in the eastern San Gabriel Valley at Glendora and West Covina. A tendency to underpredict the summer peaks observed near the up-coast and down-coast edges of our study area at Santa Monica and at Garden Grove and Anaheim during 1973 and 1974 was noted. A

GLEN R. VASS

1240

tJ0

I ‘@SElWEO

70-

-HaJa_

t

1

IIERN RESUTS

f

01

L

L

i

Fig. 9a. Monthly

arithmetic mean sulfate concentrations at downtown air quality model results vs. observed values.

Fig. 9b. Source class contribution

to sulfate concentrations Angeles.

statistical comparison of the relationship between observed and predicted sulfate concentrations over all APCDand CHESS monitoring sites combined is given in Table 3. Observed and predicted monthly average sulfate concentrations are presented graphically in Fig. 12. The data shown in Fig. 12 give typical results rather than the best or the worst year for model performance. Source class increments to predicted sulfate air quality were examined in time series at each air monitoring station, as shown in the lower graphs of Figs 9-11. It was found that three to five source

1

187%

SF?2 1973 JFIlR!tJJASONOJFHRtIJJRSONOJF~R~JJffSOND

Los Angeles (APCD)

observed at downtown

Los

classes with significant impact, plus background sulfates, must be considered simultaneously in order to come close to explaining sulfate levels observed at most locations. For example, during the year 1973 at downtown Los Angeles, contributors to the annual mean sulfate concentration observed were estimated to be: (1) Background sulfates-28 “/, (2) Electric utility generating stations-23 “/,. (3) Heavy duty mobile sources-1 5 7;. (4) Sulfur recovery and sulfuric acid plants-l 2 “/b. (5) Petroleum refining and production-l I “4,;.

1241

Sulfate air quality control

Fig. 1Oa. Monthly arithmetic mean sulfate concentrations at Pasadena (APCD) air quality model results vs. observed vatues.

Fig. lob. Source class ~ntribution

to sulfate con~ntmtions

(6) Autos and light trucks-t%. (7) Petroleum coke calcining kilns-3 %. (8) Ail remaining sources-4 %. The rdative importance of particular source classes varies from one monitoring site to another. Within the last few years, a number of investigators of Los Angeles sulfate air quality have attempted to dedu@e~source to receptor influence relationships for sulfates by regression analysis of air quality data on emissions data (He&r, 1976; Joint Project, 1977; White et d., 1978). Great di~uity was encountered in fortnukttihg and validating empirical emissions-air

observed at Pasadena (APCD).

quality relationships. The reason is now fairly clear. Referring to Fig. 7, it can be seen that most of the source classes responsible for observed sulfate levels have so little variance in their emissions patterns over time that it would be impossible to explain the wide variance in observed air quality on the basis of variance in emissions alone. The only sources with large well-defined short-term fluctuations in emissions are power plants, and they do not account for enough of the total sulfate burden to dominate the variance in ambient sulfate air quality. However, using statistical techniques White et al.

1242

GLEN

m t

-ma_

R. CASS

ftmJ_TS

Fig. 1la. Monthly arithmetic mean sulfate concentrations at Glendora (CHESS) air quality model results vs. observed values.

Fig. 1I b. Source class contribution to sulfate concentrations observed at Glendora (CHESS).

(1978) derived an upper limit estimate for the contri-

bution of power plant emissions to sulfate air quality at West Covina. They determined that power plant emissions accounted for no more than 17% of the sulfate burden at that location. Our model results at West Covina support this finding and indicate that power plants accounted for 17.4 ‘x of the total sulfate loading at West Covina, averaged over the years 1972-1974. Power plant impact was proportionately higher elsewhere however during those years and would be lower at present due to more recent reductions in the sulfur content of utility fuel.

A seasonal variation in the overall rate of SO, oxidation in the Los Angeles atmosphere was computed from simultaneous comparison of observations and model predictions at a large number of monitoring sites. As shown in Table 4, monthly mean SO2 oxidation rates of between 0.5% h- ’ and 3% h- ’ prevail from October through February of our test years. During late spring, summer, and early autumn, SO, oxidation rates were estknated to riac to an avemge of about 6 % h - ‘, with individuai months ranging f 2% h-t about that mean value. Those numerical results must be qualified since a better

Sulfate air quality control

1243

Table 3. Statistical comparisoo of monthly average sulfate concentrations: observations vs. air quality model results at all 11 LAAPCD and CHESS air monitoring sites located within the receptor zooe of Fig. 13

Year 1972 1972* 1973 1974

Predictions

Observations

Number of pairs of monthly average observations and predictions*

Meant pgm-’

108 106 120 128

12.18 12.31 13.12 10.45

Standard deviation pgme3

Meant Ccgm-’

Standard deviation pgmm3

Correlation: observed vs. predicted ms

5.87 5.84 7.38 5.65

12.51 12.32 12.52 10.80

5.92 5.81 6.80 4.86

0.82 0.89 0.82 0.70

* Biased observations at two CHESS stations discarded for January 1972.These stations did not begin operation until the last four days of that month and thus missed the early January 1972 sulfate episodes. Days sampled thus were not representative of that month. t Uoweighted average of available data pairs; because of changing number of days per month and absence of observations at some locations in some months this does not represent an annual mean value. t NASN monitoring sites report insufficient data for comparison to model performance on a monthly basis. For comparison to NASN data on an annual mean basis, see Cass (1978).

7-----l W-

&_

+++= + + ++*; + ++

P 8 La

++ +*+t, 3, +++*+*+++ + 8 :+ *++ + + 10&g.*~~ l

OL 0

+ a + I

10

l

I

I

#

UO

PFEOf&EO sou zH,f4*

Fig. 12. Sulfate air quality model results-1973 means at ten air monitoring stations.

A so

monthly

understanding of seasonal trends in background sulfate concentrations or SO, deposition velocity could alter the outcome. One striking feature of Los Angeles sulfate air quality is that long term average upwind/downwind concentration gradients observed between monitoring sites are rather uniform in spite of the highly localized nature of major SO, emission sources. Annual average sulfate air quality model predictions shown in Fig. 13 confirm that observation: most monitoring sites lie within the 10 and 15 pgrn - 3 isopleths in all years of interest. The air quality model results of Fig. 14 showing individual source class contributions to observed sulfate air quality help to explain this phenomenon. In winter months with a pronounced daily sea breeze/land breeze wind reversal, air parcel trajectories wander widely over the basin. Sulfur oxides emitted from all source classes are dispersed

widely within the airshed by the rotation of the wind vectors. In contrast, during mid-summer, onshore flow persists for most of the day. However, the sequential siting of major SO, sources along the coast means that the central portion of the air basin is downwind of one major source group or another at most times. Lateral dispersion of emissions is just about sufficient to balance sulfate formation, with the result that upwind/ downwind pollutant gradients are rather small in spite of the direct inland transport from sources to receptors. Annual mean sulfate concentrations are further smoothed by seasonal transport cycles in which peak sulfate concentrations appeared far inland during the summer and near the coast during the winter, as is most easily seen from the automotive impact graphs of Figs 14c and 14f. In January 1972, extreme resultant wind stagnation occurred during a period of high SO, emissions. The highest localized sulfate concentration predictions for Table 4. Calculated rate of SO, oxidation to form sulfates in the Los Angeles atmosphere, in % h- ’ (overall average values of k for the month shown) Month

1972

January February March April May June July August September October November December

3% k$ 3%0 6% 6% 4% 5% ;; :$

1973 1% 1.5% 1% ;F 72 8%. 5% 5% & 1’%O

1974 0.5% 1% :$ 52 6% ;: 82 3”/ 1+ 0.5;

3-year mean 1.5% 1.2% 3.3% 3.0% 6.3 % 6.3 % 5.7 % 6.0% 5.7 % 2.0% 0.9 % 0.8 %

* Average value from 11 field experiments in that month (Roberts, 1975).

GLEN

1244

any month of our three year period occurred at that time. While such extended stagnation is unusual, the fact that it can occur means that sulfate air quality control strategy design should consider avoidance of wintertime as well as summertime pollution episodes in Los Angeles.

6. SULFATE AIR QUALITY ~O~ROL~~AT~Y

DESIGN

Figures 9-11 show that an emission control strategy diversified over a large number of source types Average

(4

sulfate

concentfa~~s,

Catenda

year

R. CASS will be needed if significant sulfate air quality improvement is to be achieved in this locale. Intuitive application of emission control regulations is likely to lead to costly errors because the air quality impact of sources is not in direct proportion to their relative share of basin-wide SO, emissions. Economically efficient emission control strategies can be defined if data on control technologies and their costs are combined with the air quality model results. Points representing the least costly combination of emission controls needed to attain a variety of altered levels of sulfate air quality can be computed. This sequence

pg zw3

(W

1972

coiendar

Fig. 13a

yecr

Fig. 13b

Average

s&fate

concentrations,

pg me3

GpQ” A&O &West covmo

Fig. 13c.

1973

1245

Sulfate air quality control

FEERJWY

Fig- 14a. Sulfate air quality increment due to electric utility boilers (pgmm3)

1972

Fig. 14d. Sulfate air quality increment due to electric boilers (pg m- ‘)

FEBRRRY 1972

Fig. 14b. Sulfate air quality increment due to chemical plants (Mgm- 3,

Fig. 14e. Sulfate air quality increment due to chemical plants (pg mm3)

‘._

‘X. k...

ALY

‘L ‘. ..,...,,

-..

1873

Fig. 14c. Sulfate air quality increment due to autos and light trucks (pg m- ‘)

FEm.mY

c

1972

Fig. 14f. Sulfate air quality increment due to autos and light trucks @gme3)

1246

GLEN

of points forms a curve of steepest descent, that is the least costly path toward sulfateair quality improvement. Consider the following example. The effect of a 50% reduction in sulfate concentrations on visibility in Los Angeles has been estimated by Cass (1979). The number of days per year of prevailing visibility less than three miles (4.8 km) would be cut approx. in half. We wish to define attractive control strategy paths that would approach such an air quality improvement. The number and location of major SO, emissions sources in the Los Angeles area has not changed in recent years. Hence the sources to be discussed are those present during the air quality model validation study. The South Coast Air Basin as it existed in 1973 will provide the starting point for our calculations. To confine the example to manageable proportions, analysis will be limited to those stationary source emissions control technologies and costs previously identified by Hunter and Helgeson (1976). Following Hunter and Helgeson, it is assumed that each control measure listed provides proportionately the same removal efficiency for both SO, and primary SOi- emissions as is expected for total SO,. It is further assumed that each emission control technique listed will be applied equally to all members of the source class for which it is specified. Each emission control option will be used continuously if adopted (i.e., we are not interested in emergency episode abatement but rather a proportionate reduction in SO, emissions on each day of record). Emission control techniques were associated with the source classes for which they could be prescribed. Annual average SO, emissions for the year 1973 to which each control measure would apply were estimated from the on-grid plus off-grid portions of the emission inventory of Table 2. The source to receptor transfer function&O? (x/x,, i; T, tJ from the 1973 air quality model validation effort provides an estimate of the effect on air quality of a unit change in emissions from each source type i. That function was used to map the potential emission reduction, Ax&,, i) from use of a given control technology into expected annual mean sulfate air quality improvement at downtown Los Angeles. The incremental air quality improvement obtained from each technology was divided by the total annual cost’ of the control measure of interest in order to arrive at an index of control measure effectiveness, in pg m - 3 of sulfate air quality improvement per dollar per year spent on emission control. If the control measures identified by Hunter and Helgeson (1976)

$ Costs given are for both recovery of capital investment and operation on an annual basis, stated in 1975-76dollars. Control cost estimates given by Hunter and Helgeson (1976) were intended to represent actual “manufacturing costs” for each control technology in those cases where market price data were unavailable. These cost data are limited to Hunter and Helgeson’s work because that study forms a single consistent set of cost estimates across a variety of South Coast Air basin emission sources.They should not be interpreted as the only control measures available.

R. CASS were to have been implemented in 1973 in order of declining cost-effectiveness index, the progression of air quality improvement versus cumulative control cost would have been as shown in Fig. 15. The points along the curve of Fig. 15 form one possible emission control strategy. Up to $97 x lo6 per year could have been spent on these control measures, with up to 43 7; improvement expected in average sulfate air quality at downtown Los Angeles. Maximum obtainable visibility improvement would have been somewhat less than that illustrated in Fig. 5 of Cass (1979). A decision could have been made to operate at any point along the control strategy curve of Fig. 15 without spending more than necessary to obtain the level of air quality sought. A further application of the emission control strategy design approach developed in the preceding example was completed recently by the South Coast Air Quality Management District (1978). Air quality modeling results obtained from Cass (1978) at a large number of air monitoring sites were combined with forecast emissions estimates and additional data on emission control technologies. The results of that study are similar to Fig. 15 and were used by the government to revise the sulfur oxides emission control regulations in Los Angeles.

7. SUMMARY

AND

CONCLUSIONS

An air quality model has been described which calculates long-term average sulfate air quality levels in a multiple source regional setting. The model computes pollutant concentrations by a long-term average Lagrangian marked particle technique. First-order chemical reactions and pollutant ground level dry deposition are incorporated. The model is capable of handling unsteady meteorological conditions in a region characterized as having an idealized persistent temperature inversion. This atmospheric temperature structure results in a well-mixed layer of time varying depth near ground level capped by a stable air mass aloft. The model is adapted to assist emission control strategy design in a way that the effect of each source type contributing to observed air quality is identified. The air quality dispersion model was applied to Los Angeles sulfate air quality over each month of the years 1972 to 1974. It was found that the model can track major changes in SO, emissions while closely reproducing observed sulfate concentration patterns within the central portion of the Los Angeles Basin. The correlation between observed and predicted sulfate levels was 0.82 in both 1972 and 1973 and rises to 0.89 in 1972 if two biased air monitoring observations are discarded. A seasonal variation in the overall rate of SO, oxidation to form sulfates in the Los Angeles atmosphere was apparent from the results of the air quality modeling study. Monthly mean SO2 oxidation rates of

1247

Sulfate air quality control

0

10

XI

ANNUAL THAT

3c

40

COST

30

60

OF

EMISSION

ACTUALLY 118 DOLLARs,“EAR

70

so

90

CONTROL

100

110

120

BEYOND

1973 ACHIEVED IN ws-ni COST Elksis

Fig. 15. Stationary source emission controls identified by Hunter and Helgeson (1976) applied to SO,Kemissions sourceslocated in the South Coast Air Basin as they existed in 1973.

between 0.5 % h- ’ and 3 % h-r prevail from October to February of our teat years. During the late

REFERENCES

spring, summer and early autumn, SO, oxidation rates were estimated to jump to anaverage of about 6 % h- I, with individual months ranging + 2 y0 h-r about that mean value. Source class increments to predicted sulfate air quality were examined in time series at each air monitoring station. It was found that three to five source classes of roughly equal impact, plus background sulfates, must be considered simultaneously in order to come close to explaining sulfate levels observed at most locations. The implication is that a mixed strategy targeted at a combination of source types will be needed if significant sulfate air quality improvements are to be achieved in this airshed through precursor SO, control. Design of economically efficient emission control strategies in a complex urban setting was discussed. Air quality model results were combined with data on emission control measures and costs. Example calculations showed that close to a 50% reduction in Los Angeles sulfate concentrations could be achieved at a cost of circa %100 million annually (1976 cost basis). Since the impact on visibility of such a sulfate concentration reduction has been estimated previously (Cass, 1979), a partial remedy for the Los Angeles visibility problem has been described.

Blumenthal D. L., White W. H. and Smith T. B. (1978) Anatomy of a Los Angeles smog episode: pollutant transport in the daytime sea breeze regime. Atmospheric Environment 12, 893-907. Bendat J. S. and Piersol A. G. (1971) Random Data: Analysis and Measurement Procedures, John Wiley, New York. Briggs G. A. (1971) Plume rise.:a recent critical review. Nucl. Saf: 12(l). C&s G. R. (1975) Dimensions of the Los Angeles SO,/Sulfate Problem, California Institute of Technology, Environmental Quality Laboratory Memorandum No. 15. Cass G. R. (1978) Methods for sulfate air quality management with applications to Los Angeles. Ph.D. Thesis, California Institute of Technology. Available from University Microfilms, Ann Arbor, Michigan, or the Environmental Quality Laboratory, California Institute of Technology, Pasadena, California. C&s G. R. (1979) On the relationship between sulfate air quality and visibility with examples in Los Angeles. Atmospheric Environment 13, 1069-1084. Cass G. R. and Shair F. H. (1980) Transport of sulfur oxides within the Los Angeles sea breeze/land breeze circulation system. Proceedings of the Second Joint Conference on Applications of Air Pollution Meteorology, American Meteorological Society, New Orleans, 24-27 March 1980. Chamberlin A. C. (1960) Aspects of the deposition of radioactive and other gases and particles. Jnt. J. Air Pollut. 3, 63-88. Cogbill C. V. and Likens G E. (1974) Acid precipitation in the northeastern United States. Water Resour. Res. 10, 1133-1137. Charlson R. J., Vanderpol A. H., Covert D. S., Waggoner A. P. and Ahlquist N. C. (1974) HsO,/(NH,),SO, background aerosol: optical detection in St. Louis region. Atmospheric Environment 8, 1257-1267. Davidson C. I. (1977) Deposition of trace metal-containing aeroaolr on smooth, Bat surfaces and on wild oat grass (Avena fat@. Ph.D. Thesis, California Institute of Technology, Pasadena, California. DeMarrais G. A., Holzworth G. C. and Ho&r C. R. (1965)

Acknowledgements-This work has been supported by the California Air Resources Board (Contract No. A6-061-87). Thanks are due to the staff of the South Coast Air Quality Management District (formerly the Los Angeles Air Pollution Control Diet&t) and to the staff of the California Air Resources Board for their cooperation and assistance. *E 15:7- J

1248

GLEN R. CASS

Meteorological Summaries Perfinenr fo Afmospheru Transport and Dispersion over Southern Calijornia. Washington, D.C., U.S. Weather Bureau Technical Paper No. 54. Drivas P. J. and Shair F. H. (1975) The Chemistry. Dispersron and Transporf of Air Pollufanfs Emitted from Fossil Fuel Power Plants in Calijornia-Transport of Plumes Associated wirh fheComplexCoasra/ Mefeorology. Division of Chemistry and Chemical Engineering, California Institute of Technology, performed under California Air Resources Board Contract No. ARE915. Edinger J. G. (1973) Vertical distribution of photochemical smog in Los Angeles basin. Enrir. Sci. Technol. 7,247 252. Eggleton A.E.J. (1969) The chemical composition of atmospheric aerosols on Tees-side and its relation IO visibility. Armospheric Environmenr 3, 355.372. Eliassen A. (1978) The OECD Study of long range transport of air pollutants: long range transport modeling. Afmospheric Encfronmenr 12, 479 487. Garland J. A. (1974) Dry deposition of SO, and other gases. In Afmosphrre-Surjirce Exchanyr o/ Parricularr and Gaseous Palluronfs. Procerdfng.\ oj a Symposium held UI Richland. Wushinyron. 4-6. Sept. 1974. Technical Information Center. Office of Pubhc Affairs. U.S. Energy Research and Development Administration, 1976. Gillette D. A. and Blilford 1. H. (1971) Composition of tropospheric aerosols as a function of altitude. J. ufmos. Sci. 28. 1199-1210. Heisler S. L. (1976) A.~.w.t\menf o/ rhe Influence q/ Eltv fret Generaffny Sfurions on Daily Su//afe Lrcels IU fhr Soufh C‘oasf Air Basin. Westlake Village. Califorma. Environmental Research and Technology Document P5088R4. Hidy G. M. et al. (1974) Observations of aerosols over Southern California coastal waters. J. appl. Mef. 13. 96107. Hidy G. M., Mueller P. K., Tong E. Y. (1978) Spatial and temporal distributions of airborne sulfate in parts of the United States. Atmospheric Enrironmenf 12, 735.-752. Hunter S. C. (1975) KVB Incorporated, personal communication. 19 November 1975. Forwarded stack data taken during source tests reported by Hunter and Helgeson (1976). Hunter S. C. and Helgeson N. L. (1976) Control @‘Oxide.\ o/ Sulfurfrom Sfarionary Sources in the Soufh Coasr Air Basfn o/ Calijornia. Tustin, California, KVB Incorporated. Document No. KVB-5802432. Johnson W. B.. Wolf D. E. and Mancuso R. L. (1978) Long term regional patterns and transfrontier exchanges of Armospherfc airborne sulfur pollution in Europe. Encironmenr 12, 511-527. Joint Pro&t (1977) Joinr Projecf /or Relating Emissions oj SO, f; Air. Quulity - -Inrer;m Reporf. Westlake Village. California, Environmental Research and Technology Inc.. Document P-5088. Junge C. E. (1957) Chemical analysis of aerosol particles and aas traces on the island of Hawaii. Tellus 9. 528.-537. La& R. G. (1971) The application of a generalized Lagrangian diffusion model to air pollution simulation studies. In Proceedings ojrhe Symposium on Air Pollufion, Turbulence and Diflusion, (Edited by H. W. Church and R. E. Luna) New Mexico State University, December 7 10. 1971, Las Cruces, New Mexico. Lamb R. G. and Neiburger hi. (1971) An interim version of a generalized urban air pollution model. Atmospheric Enrironmenr 5. 239-264. Lamb R. G. and Seinfeld J. H. (1973) Mathematical modeling of urban air pollution-general theory. Em:ir. Sci. Technal. 7. 253-261. Likens G. E. (1976) Acid precipitation. Chem. Enyny News 54,

W), B-44.

Liljestrand H. M. (1980) Afmospheric transport of acidify in Southern Calijornia by wef and dry mechanisms, Ph.D.

Thesis. California Institute of Technology. Pasadena. California. Los Angeles Air Pollution Control District (1975a) Wmd data file. DSNAME WDMAST. Technical Services Diviston. Los Angeles Air Pollution Control District. magnetic tape of the District’s wind data. Los Angeles Air Pollution Control District (1975b) Master VAR99 disk file. DSNAME SYSAG. VAR99. Technical Services Division. Los Angeles Air Pollution Control District. magnetic tape copy of disk file containing District’s inversion base height data. MacPhee R. D. and Wadley M. W. (1975a) Atrborne particulate matter in Los Angeles County 1974. Technical Services Division, Los Angeles Air Pollution Control District. Air Quality Report No. 79. MacPhee R D. and Wadley M. W. (lY75b) Airborne particulate matter in Los Angeles County 197.5. Technical Services Division. Los Angeles Air Pollution Control District. Air Quality Report No. 81. McElroy J. 1.. and Pooler F Jr. I 1968) Sr. LOU.\ DrApersion Srudb. C’olumv II .4nu/j,si.\. Arlington, Virgmia, U.S. f’ubiic Health Service. National Air Pollution Control Admnnstration Publication No. AP-53. National Research Counctl I lY751 Air quality and stationary source emission control. Washmgron D.C., U.S. Government Printing Office. Senate Document No. 94-4. Nordsieck R. A. (1974).4ir Qualify Impacfs ofE:lecrrfc Curs in /.o.\ Atfgc*/~ Appendfx A. Po//uronr Emissions Esfimafes and Pro/et rion.\ jar rhc, Sourh C‘oasr .4ir Busin. Santa Barbara, t‘ahfornia. General Research Corporation document RM1905-A. Ow,ers M. J. and Powell A. W. ( 1974) Deposition velocity of sulphur dioxide on land and water surfaces using a “S tracer method. Armospherit Encironmenr 8, 63--68. Pierson W. R. and Brachaczek W. W. (1976) Parficulafe Maffc~ Assocraft~d with Vehicles on fhe Road. Society of Automotive Engineers Paper 760039. Pierson W. R. (1977) Ford Motor Company. personal communication, 3 August 1977. Roberts P. T. (1975) Gas to particle conversion: sulfur dioxide in a photochemically reactive system. Ph.D. Thesis, California Institute of Technology. Sander S. P. and Semfeld J. H. (19761 Chemical kmetics of homogeneous oxidation of sulfur dioxide. Enrrr. Sci. 7Lchnol. IO, 1 I14 II 23. Shair F. H. (1977) Cahfornia Institute of Technology, personal communication, 30 Sept. 1977. Forwarded preliminary data from SF, tracer studies performed under Cahfornta Air Resources Board Contract A6202-30. Sheih C. M. (IY77) Application of a statistical trajectory model IO the simulation of sulfur pollution over northeastern United States. .Armospheric, Enrironmc*nr I I, 173 178. SJovold A. R. (lY73) Elerrricul Encryy Projecfionsjor rhe Los .4nyele~ Region, IYRO-2000. Santa Barbara, California. General Research Corporation, document RM- 1859 (as cited by Nordsieck. 1974). South Coast Air Quality Management District (1978) Sulfur dioxide:sulfate control study. El Monte, Califomta, a staff report hy the South Coast Air Quality Management District, May 1978 edition. Thomas M. F. (1976) Federal Power Commission, personal communication, 20 February 1976. Forwarded copies of FPC Form 67 showing electric utility boiler stack characteristics for each generating station of interest. TriJonrs J.. Richard G., Crawford K.. Tan R. and Wada R. ( IY 75) An implementation plan for suspended particulate matter m the Los Angeles region. TRW Transportation and Environmental Engineering Operations, Redondo Beach, California, March 1975 edition. Cost cited is for the sulfur dioxide control features of Strategy 1. Trijonis J. and Yuan K. (1978a) Visibilify in rhe Sourhwesf: An Erolorurion offhe _ Hisrorical Dora Base. Research Trianalc .

Sulfate air quality control Park, North Carolina, U. S. Environmental Protection Agency, document EPA-600/3-78-039.

Trijonis J. and Yuan K. (1978b) Yisibility in the Notcheast: Long-term Visibility Trends and Visibility/Pollutant Relationships. Research Triangle Park, North Carolina, U.S. Environmental Protection Agency, document EPA600/3-78-075. Waggoner A. P., Vanderpol A. J., Charlson R. J., Larsen S., Granat L. and TriigdrdhC. (1976)Sulphate-light scattering ratio as an index of the role of sulfur in tropospheric optics. Nature 261, 126122.

1249

Weiss R. E., Waggoner A. P., Charlson R. J. and Ahlquist C. (1977) Sulfate aerosol: its geographical extent in the Midwestern and Southern United States. Science 195, 97%981. White W. H., Heisler S. L., Henry R. C., Hidy G. M. and Straughan I. (1978) The same-day impact of power plant emissions on sulfate levels in the Los Angeles air basin. Atmospheric Environment 12, 779-784. White W. H. and Roberts P. T. (1977) On the nature and origins ofvisibility-reducing aerosols in the Los Angeles air basin. Atmospheric Enoironment 11, 803-812.