Ozone patterns for three metropolitan statistical ... - NCSU

*Corresponding author. Tel.: #1-919-515-3711; fax: #1-919-515-7802.

E-mail address: viney}[email protected] (V.P. Aneja)1Eastern Research Group, Inc., P.O. Box 2010, 1600 Per-

imeter Park, Morrisville, NC 27560-2010, USA.2U.S. Environmental Protection Agency, Integrated Policy

and Strategy, Division, MD-15, RTP, NC 27711, USA.

Atmospheric Environment 33 (1999) 5081}5093

Ozone patterns for three metropolitan statistical areasin North Carolina, USA

Viney P. Aneja!,*, R.G. Oommen!,1, A.J. Riordan!, S.P. Arya!, R.J. Wayland!,2,George C. Murray"

!Department of Marine, Earth and Atmospheric Sciences, North Carolina State University, Raleigh, NC 27695-8208, USA"Division of Air Quality, North Carolina Department of Environmental and Natural Resources, Raleigh, NC 27626-0580, USA

Received 15 December 1998; accepted 14 April 1999

Abstract

As part of an e!ort by the state of North Carolina to develop a State Implementation Plan (SIP) for 1-h peak ozonecontrol, a network of ozone stations was established to monitor surface ozone concentrations across the state. Between19 and 23 ozone stations made continuous surface measurements between 1993 and 1995 surrounding three majormetropolitan statistical areas (MSAs): Raleigh/Durham (RDU), Charlotte/Mecklenburg (CLT), and Greensboro/HighPoint/Winston-Salem (GSO). Statistical averages of the meteorological and ozone data were performed at eachMetropolitan Statistical Area (MSA) to study trends and/or relationships on high ozone days (days in which one of theMSA sites measured an hourly ozone concentration*90.0 ppbv). County emission maps of precursor gases, wind roses,total area averages of ozone, total downwind averages of ozone deviations, upwind averages of ozone, and a modi"eddelta ozone analysis were all obtained and analyzed. The results of this study show a reduction in the delta ozone relativeto an earlier study at RDU, but no average signi"cant change at CLT (no comparison can be made for GSO). Thestatistical data analyses in this study are used to quantify the importance of local contributions and regional transport, toozone air pollution in the MSAs. ( 1999 Elsevier Science Ltd. All rights reserved.

Keywords: Urban ozone; Pollutant transport

1. Introduction

Ambient ozone concentrations found in the lowertroposphere continue to be a major air pollution prob-lem in the United States (Logan et al., 1985; NRC, 1991;Aneja et al., 1992; Sillman et al., 1993; Vukovich, 1994).Areas that violated the previous hourly National Ambi-ent Air Quality Standard (NAAQS) of 0.12 parts permillion by volume (ppmv) for ozone, as set by the US

Environmental Protection Agency (EPA), were desig-nated as non-attainment for ozone control (EPA, 1993).A day in which any 1-h average ozone concentration wasgreater than 0.12 ppmv was considered an exceedanceday. An area which produces four or more exceedancesover a 3-year period is considered in violation of NAAQS(EPA, 1993). The ozone standard has been revised morerecently to 0.08 ppmv for the 8-h averaged concentration.

Ozone, a secondary pollutant, is designated as a cri-teria pollutant under the Clean Air Act Amendments(CAAA) of 1990 (EPA, 1993). In 1989, the EPA estimatedthat 67 million people lived in areas of NAAQSexceedances for ozone (NRC, 1991). Health e!ects of highozone concentration exposure include eye irritation(NRC, 1991) and breathing trouble for asthmatics (EPA,1994). Crop losses from high ozone exposure are esti-mated to be $1 to $2 billion annually in the US (EPA,

1352-2310/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved.PII: S 1 3 5 2 - 2 3 1 0 ( 9 9 ) 0 0 2 4 1 - 1

Table 1Information on the three metropolitan statistical areas (MSAs) in North Carolina

MSA name Population(as of 1995)

Number of NAAQS Closest national weatherservice station

Latitude and Longitudeozone exceedances of national weather(1980}1992) service station

Raleigh/Durham (RDU) 1,000,000 23 Raleigh-DurhamInternational Airport

35352'N, 78347'W

Charlotte/Mecklenburg (CLT) 1,200,000 72 Charlotte-DouglasInternational Airport

35313'N, 80356'W

Greensboro/High Point/Winston-Salem (GSO)

1,000,000 47 Piedmont-TriadInternational Airport

36305'N, 79357'W

Source: NC DEHNR.

Fig. 1. Locations of the three North Carolina MSAs used in thestudy. Asterisks (*) indicate MSAs greater than 900,000 people.(Source: US Census).

1996). Additionally, non-crop losses have been documen-ted by Moy et al. (1994), who report an alarming decreaseof white pine trees at Shenandoah National Park in ruralVirginia attributed to elevated ozone concentrations.

Tropospheric ozone is produced through reactions ofhydrocarbons (HCs), (commonly referred to as volatileorganic compounds or VOCs), nitrogen oxides (NO

x"

NO and NO2) and ultraviolet solar radiation (hl). In

regions of high VOC concentrations (primarily throughproduction of mobile, biogenic and industrial sourceemissions and transport of these species by the prevailingwind), a disruption in the photostationary state of ozoneproduction occurs.

The ozone production cycle is driven by sunlight (hl);however, other meterological parameters, such as cloudcover, air temperature, relative humidity, atmosphericpressure, and wind speeds in#uence the kinetics for ozoneproduction and distribution (Aneja et al., 1994). Further-more, this reaction is maximized during the summertime,when the incoming solar radiation is greatest togetherwith high temperatures.

High ozone concentrations in the southeastern UnitedStates continue to be a pervasive problem for both the pre-vious 1-h NAAQS and the newly revised 8-h NAAQS. (Inaddition to the change in averaging time, the new stan-dard was lowered to 0.080 ppmv over this 8-h period.)Wind direction associated with synoptic-scale weatherpatterns is important everywhere; for the Southeast US,ozone and its precursors can be transported primarilyfrom the Northeast, Midwest, as well as other regions(Vukovich, 1994).

2. Background

North Carolina has three Metropolitan StatisticalAreas (MSAs) that periodically experience high concen-trations of ozone: Raleigh/Durham (RDU); Charlotte/Mecklenburg (CLT); and, Greensboro/High Point/Winston-Salem (GSO) (Fig. 1). Table 1 gives the number

of 1-h NAAQS ozone exceedances over a 12-yr period foreach of these MSAs. In 1990, the CLT MSA was desig-nated as a non-attainment area for ozone. Two yearslater (June 1992), GSO and RDU were similarly desig-nated. However, all three MSAs are currently consideredto be attainment/maintenance areas with respect to therecently revised NAAQS (EPA, 1998).

According to the NAAQS, ozone control strategieswere to be formulated at the state level. For our study

5082 V.P. Aneja et al. / Atmospheric Environment 33 (1999) 5081}5093

Fig. 2. County Mobile Emission Maps of CO, VOC, and NOx. EPA estimates for mobile emissions of CO, VOCs, and NO

xby county.

Asterisks (*) indicate National Weather Service stations. The ozone sites are in italics, interstates are denoted by thick lines, while USroutes are denoted by thinner lines. Regions with high CO, VOCs, and NO

xmatch well with the three MSAs. (Source: NC DEHNR).

period (1993}1995), there were 16 exceedances over 15 din the North Carolina MSAs; RDU and GSO had threeeach, while CLT had 10 (there was one day in whichRDU and GSO each had an exceedance).

The estimates of county mobile emissions of carbonmonoxide, VOCs and NO

xfor 1990 by US EPA (EPA,

1995) show that counties with high carbon monoxideemissions match favorably with the locations of the threeMSAs and the highway links between Raleigh and Char-lotte. The EPA considers a county to be high in emissionsfor CO, if the emissions are greater than 50 Mg d~1

(1 Mg"1]106 g). For a county to be high in VOC andNO

x, emissions must be greater than 10 Mg d~1. The

reason that CO is of interest is that carbon monoxide canbe used as a tracer for mobile emissions and also acts asa precursor for ozone formation (Warneck, 1988). Alongwith the counties of high CO emissions, the mobile NO

xemissions and VOC emissions matched well with thehighways and MSA locations, indicating that a majorprecursor gas for ozone formation can be attributed tothese mobile emissions (Fig. 2).

Additionally, the 1990 EPA estimations of above aver-age to high biogenic VOCs are strongly related to coun-

ties of heavy forest canopies in western and southernNorth Carolina. The EPA considers a county to haveabove average emissions of biogenic VOCs if emissionsare greater than 50 Mg VOC d~1, while high VOC countyemissions are greater than 60 Mg VOC d~1 (EPA, 1995).Farming in south central North Carolina accounts forthe above average to high biogenic NO

x. The EPA con-

siders a county to be above average if biogenic NOx

emissions are greater than 1.1 Mg NOxd~1, while high

indicates greater than 3 Mg NOxd~1 (Fig. 3). In rural

North Carolina, contributions of naturally-occurringVOCs, such as isoprene (Lawrimore et al., 1995) andNO

xfrom agricultural activity (Sullivan et al., 1996) also

play an important role in ozone formation.Due to the high emissions of naturally occurring NO

xand VOCs in the southeastern US, rural backgroundconcentrations of ozone are typically higher than those inother regions. Altshuller and Lefohn (1996) recentlyreported background ozone concentrations from acrossthe US and Canada for inland sites of between 25 and45 ppbv, for a speci"c 7-h time period from 0900 to 1559Local Standard Time (LST). From their measurements ata rural site in central North Carolina, Aneja et al. (1996)

V.P. Aneja et al. / Atmospheric Environment 33 (1999) 5081}5093 5083

Fig. 3. County Mobile Emission Maps of VOC, and NOx. EPA estimates for biogenic emissions of VOCs, and NO

xby county.

Asterisks (*) indicate National Weather Service cations. The ozone sites are in italics, interstates are denoted by thick lines, while USroutes are denoted by thinner lines. Most of south central NC has above average to high emissions of biogenic VOCs, while biogenicemissions are low. (Source: NC DEHNR)

obtained an average background ozone concentration of27 ppbv for the daytime period in summer.

The North Carolina Department of Environmentaland Natural Resources (NC DENR), formerly known asNorth Carolina Division of Environment, Health, andNatural Resources (NC DEHNR), in response to ozoneexceedances at RDU, GSO, and CLT, expanded itsstatewide monitoring network from exclusively withinthe MSA to include more sites downwind of the threeMSAs. During the years 1993, 1994, and 1995, there were18, 21, and 22 active ozone stations in and around thethree MSAs, respectively (Table 2). The purpose forthe additional sites was to measure ozone concentrationsat optimal downwind locations in order to developa North Carolina State Implementation Plan (SIP) forozone control. A site in York County, SC was added forthis study. This location is southwest of the CLT MSA.

This study examines the ozone episodes experiencedby the three MSAs and attempts to determine the role oflocal contributions versus regional transport of ozoneand its precursors in producing high ozone levels withinand downwind of the MSAs in NC. The methods used insupport of this objective are: (1) to examine surface wind"elds at the three MSAs; (2) to compare these wind "elds,within an appropriate transport range of each MSA, inrelation to precursor source locations; and (3) to examinethe changing ozone concentrations as air masses moveover the MSA.

3. Data retrieval and reduction

The three MSAs provided ideal locations for an ozonestudy because in the center of each MSA is a National


Tab

le2

Ozo

nem

onitoring

site

sin

the

Ral

eigh

/Dur

ham

(RD

U),

Gre

ensb

oro

/Hig

hP

oint/

Win

ston-

Sal

em(G

SO

),an

dC

har

lott

e/M

eckle

nbur

g(C

LT

)M

etro

pol

itan

Stat

istica

lA

reas

(MSA

s)

MSA

Dow

nwin

dEPA

Dow

nwin

dst

atio

nnam

eC

ounty

Lan

duse

1993

1994

1995

Ele

vation

(m)

Lat

itude

Longi

tude

stat

ion

AIR

SN

ort

hW

est

iden

ti"er

ID(D

DM

MSS)

(DD

MM

SS)

RD

UBT

O37

-077

-000

1Butn

erG

ranv

ille

Res

iden

tial

XX

X91

3608

2778

4609

UFO

37-0

69-0

001

Fra

nklin

ton

Fra

nklin

Res

iden

tial

XX

X13

536

0551

7827

49M

IO37

-101

-009

9M

icro

John

ston

Agr

icul

tura

lX

X52

3534

1178

1109

JOO

37-1

01-0

002

John

ston

Cou

nty

John

ston

Agr

icul

tura

lX

127

3530

0078

2615

UQ

O37

-183

-001

6Fuqu

ay-V

arin

aW

ake

Res

iden

tial

XX

117

3535

0678

4741

UPO

37-0

37-0

004

Pitts

boro

Cha

tham

Agr

icul

tura

lX

X40

035

4532

7909

55D

KO

37-0

63-0

013

Duke

St.

Ora

nge

Com

mer

cial

XX

X39

036

0208

7854

17SN

C37

-183

-001

5St

.A

ugus

tine's

Wak

eR

esid

ential

XX

127

3547

1878

3720

MLO

37-1

83-0

014

Milb

rook

Wak

eR

esid

ential

XX

X10

035

5122

7834

32U

TO

37-1

83-0

017

Aubu

rnTow

erW

ake

Agr

icultura

lX

XX

320

3541

0078

3300

WFO

37-1

83-2

001

Wak

eFor

est

Wak

eA

gric

ultu

ral

X87

3558

1578

3927

GSO

MC

O37

-081

-001

1M

cLea

nsv

ille

Guilf

ord

Res

iden

tial

XX

X22

936

0647

7942

16U

CO

37-0

33-0

001

Che

rry

Gro

veC

asw

ell

Agr

icul

tura

lX

X24

136

1810

7928

00U

BO

37-1

57-0

099

Bet

hany

Scho

olR

ocki

ngha

mA

gric

ultu

ral

XX

X77

3618

3279

5131

FC

A37

-067

-002

7H

olly

ber

ryL

ane

Fors

yth

Agr

icul

tura

lX

271

3614

1180

2438

RH

O37

-067

-000

7R

ura

lH

all

Fors

yth

Res

iden

tial

XX

X29

536

0956

8014

13H

AO

37-0

67-0

022

Hat

tie

Ave

nue

Fors

yth

Res

iden

tial

XX

X28

736

0638

8013

36U

KO

37-0

67-1

008

Pie

dm

ont

Mem

orial

Fors

yth

Res

iden

tial

XX

X28

536

0303

8008

38FO

O37

-059

-009

9Fork

Dav

ieA

gric

ultu

ral

XX

244

3552

0180

2545

GC

O37

-067

-000

6G

oodw

illC

hurc

hFors

yth

Res

iden

tial

X24

436

1259

8005

00C

LT

UY

O45

-091

-000

6Y

ork

York

(SC

)A

gric

ultura

lX

XX

222

3456

0881

1342

AW

O37

-119

-100

5A

rrow

wood

Mec

kle

nbu

rgIn

dus

tria

lX

XX

195

3506

4780

5511

ULO

37-1

09-0

004

Cro

use

Lin

coln

Res

iden

tial

XX

270

3526

1881

1631

CLO

37-1

19-0

034

Pla

zaR

dM

eckle

nbu

rgR

esid

ential

XX

X23

935

1455

8045

59C

OO

37-1

19-1

009

Coun

tyLin

eM

eckle

nbu

rgA

gric

ultura

lX

XX

216

3520

5580

4137

UR

O37

-159

-002

1R

ockw

ell

Row

anC

omm

erci

alX

X24

035

3454

8023

33EN

O37

-159

-002

2Enoc

hville

Row

anR

esid

ential

X27

035

3205

8041

55LO

O37

-109

-009

9Lin

coln

Coun

tyLin

coln

Agr

icul

tura

lX

351

3526

1681

1038


Fig. 4. Comparison of the number of high ozone days for RDU,CLT, and GSO for 1993}1995.

Weather Service (NWS) reporting station, which recordshourly meteorological surface observations. NWS datawere retrieved from the meteorological archive at NorthCarolina State University (NCSU) and the participatingNWS sites. Here, we assume that the measured windspeed and direction at the NWS reporting station arerepresentative of the mean transport velocity within thespeci"ed distance from the NWS site. Hourly ozone datawere obtained from the US EPA Aerometric InformationRetrieval System (AIRS) for the study period. Additionalozone sites not in the AIRS database were obtained fromNC DENR.

Ozone was measured by DENR site operators usingthe reference or equivalent ultraviolet photometric detec-tion technique with Dasibi model 1003 AH ozoneanalyzers. These instruments and measurement tech-nique are designated by the EPA as `equivalent referencemethodsa. Precision and accuracy checks were done ona routine basis at each monitoring site.

A `mean %a is the average percent di!erence for allthe precision or accuracy checks. For the entire statewidenetwork, ozone data has a `mean %a accuracy (at the5% signi"cance level of 0.5% [$3.0%] in 1993, 0.2%[$4.5%] in 1994, and !0.1% [$3.1%] in 1995. Thisinformation was provided by NC DENR (Oommen,1996).

3.1. High ozone day criteria

Because of the elevated background ozone concentra-tion in the Southeast, it was necessary to identify anozone concentration threshold which was distinguish-able from the background value. Consequently, ozonedata for this study were selected speci"cally for HighOzone Days (HODs), which met the criteria that at leastone of the MSA ozone monitoring sites reported at leastone hourly concentration greater than or equal to90 ppbv and that this concentration occurred between0900}1459 LST.

Previous studies have used lower threshold values suchas 60.0 ppbv (Adams, 1992) and 80 ppbv (Lindsay et al.,1989). Even though the minimum threshold was 90 ppbvfor our study, enough data were available to performa statistically robust analysis. There were no HODs inOctober through March, and thus, data from thesemonths were not used in this study.

While previous investigators have used di!erent timeperiods such as 0700}1359 LST (Altshuller, 1988),1100}1459 LST (Altshuller, 1986), and 1100}1559 LST(Lindsay et al., 1989; Adams, 1992), in this study we chosethe 6-h period between 0900 and 1459 LST because this istypically the optimum time of day for ozone productiondue to high solar insolation. The ozone concentrationsaround the MSAs during this time period are generallyhigher than the background values reported by Altshul-ler and Lefohn (1996).

After the HOD criteria were met, vector-averagedwind speeds from NWS hourly observations for thosedays were computed for each 6-h period. Two 3-h timeperiods (0900}1159 LST and 1200}1459 LST) wereestablished for determination of downwind e!ects andvector-averaged wind speeds for these periods for eachHOD were also computed.

4. Results

4.1. Number of high ozone days

Even though the o$cial ozone season in NorthCarolina is from April 1 to October 31, our study periodfocused from April 1 to September 30, for a total of 549 dbetween 1993 and 1995. For the three MSAs of interest,1993 had the highest frequency of HODs (72/183, 39.3%)followed by 1995 (59/183, 32.2%) and 1994 (38/183,20.7%). Thus there were a total of 169 HODs (169/549,30.8%) during the study period when at least one of thethree MSAs experienced high ozone concentrations.Additionally, there were 100 HODs (100/549, 18.2%)when at least two of the three MSAs experienced highozone concentrations, and 46 HODs (46/549, 8.4%)when all three MSAs experienced HOD conditions.

Fig. 4 shows the inter-relationship of the 169 HODsfor the three MSAs. During this period, there were 69(69/549, 12.6%) days on which only one MSA had highozone, and only one of these days had an MSA (CLT)exceeding the pre-existing 1-h NAAQS. There were 54 d(54/549, 9.8%) on which only two MSAs had high ozone,and on these days, there were only three NAAQS ozoneexceedances (all in CLT). However, there were 46 d(46/549, 8.4%) on which all three MSAs experienced highozone. On these days, there were 11 ozone NAAQSexceedances, suggesting that the greatest proportion ofozone exceedances occurred when all three MSAs experi-


enced high ozone. This also suggests that the high ozoneexperienced by all three MSAs may be part of theregional pollution associated with synoptic scale weatherpatterns, rather than due solely to local production.

Over the 3 yr period, CLT had the highest frequency ofhigh ozone days, followed by RDU and GSO. In theRDU MSA, of the 549 total sampling days during the 3-yperiod, 102 d met the high (90 ppbv) ozone day criterion.This translates to about 18.6% of the days, or about onehigh ozone day for every 5 d. Similar, frequencies ofHODs occurred in the other two MSAs; CLT and GSOexperienced high ozone on 115 and 98 d (20.9% and17.9%, respectively).

4.2. Determination of downwind sites using three-hourtransport criteria

Eight wind direction regimes were established for thepurpose of designating an ozone station as a downwindsite. For each of these regimes, 3-h forward trajectoriesfrom near the center of the MSA (i.e., the airports) wereconstructed by vector-averaging the hourly winds inorder to determine the range of air motion (Oommen,1996). This procedure provided a rough check that theozone sites used for statistical computations were allwithin the calculated range of downwind transport dis-tances for parcels travelling from the MSA to the down-wind ozone-monitoring site over a 3-h period. Table 3summarizes these results, and identi"es the downwindozone-monitoring sites. Not all wind regimes could beincluded since not all had a suitable downwind ozone-monitoring site. Note that the distance of the downwindmonitoring site from the airport was rarely below theminimum transport distance that an air parcel wouldtravel in three hours.

The 3-h forward trajectories and ozone concentrationsused in our study were averaged (vectorally for the winds;arithmetically for the ozone concentrations) and statist-ically analyzed using a student's t-test at the 5% signi"-cance level. Three-hour averaged ozone concentrationswere calculated at each appropriate downwind site foreach high ozone day. The time averaging reduced thevariability of the ozone data and allowed enough time forthe air to move from the MSA to the downwind station.All potential sites that met the transport criterion wereused.

RDU had a potential maximum of seven sites outsideof the MSA that could be designated downwind sitesfrom the analysis, while CLT and GSO had four and "ve,respectively (Figs. 1 and 2). The corresponding winddirections are provided in Table 3. The average resultantwind speed for high ozone days was also calculated foreach MSA. It should be noted that not all sites were atoptimal spacing, but each site used for upwind anddownwind analysis met the 3-h transport criterion.

5. Trends and/or relationships

Local and regional e!ects of winds, associated withhigh ozone days were studied utilizing a number ofspatial averages and deviations. Wind roses wereconstructed on high ozone days, and were compared withclimatological wind roses. Several types of ozone aver-ages including area means and upwind}downwind di!er-ences were then computed at each MSA for high ozonedays. Results were compared with the background ozoneconcentration determined by Altshuller and Lefohn(1996).

5.1. Meteorological 3-h wind roses

Three-hour transport wind roses were constructed forthe three MSAs for high ozone days during the studyperiod; these were compared with climatological windroses (e.g., 1963}1992) provided by the National WeatherService (Oommen, 1996). Climatology for RDU, CLT,and GSO show peaks for south and southwest winds.Secondary peaks were associated with north and north-easterly winds at all three MSAs.

When examining all wind directions on high ozonedays, there was a predominance of SW winds (Oommen,1996). Some of the di!erences among the other winddirections may be explained by the upwind emissions. AtRDU, the wind directions from the southwest and westlikely result in the transport of polluted air from CLTand GSO. Combining these directional sectors at RDUaccounts for over 41% of the high ozone days at RDU.From climatology, we should expect winds from thosedirections only 27% of the time. Thus, the air arrivingfrom the directions of GSO and CLT is often enrichedwith ozone and its precursors.

The di!erences in wind roses are not as large for CLT.Days when the air is coming from the northeast and eastinto CLT account for almost 24% of all the HODs. Thiswas very close to the climatological frequency (21.6%) forthese winds. However, the number of days on which theair was coming from the southwest into CLT accountedfor almost 26% of all HODs, which is signi"cantlygreater than what is expected from climatology (18.2%).Thus, it appears that a signi"cant source of ozone precur-sors could be to the southwest of the CLT MSA. At GSO,the number of HODs when the air was coming fromRDU (east) and CLT (southwest) accounted for nearly35% of the total HODs, which is slightly higher thanwhat is expected from climatology (27.3%), suggestingthat some of the ozone or ozone precursors may comefrom CLT and RDU.

5.2. Total area averages

The total area average (TAA) is the spatial average ofozone for all sites in and surrounding the MSAs for all


Tab

le3

Res

ultsfrom

the

tota

lare

aav

erag

e(T

AA

),th

eto

tald

ownw

ind

aver

age

(TD

A),

the

ozone

dev

iation

(OD

),th

eup

win

doz

one

conce

ntr

atio

n,an

dth

em

odi"ed

delta

ozo

nean

alys

is(M

DO

A).

The

stat

istica

lave

rage

sfo

rth

eab

ove

par

amet

ersar

epre

sente

dher

e.V

alues

with

stat

istica

llysign

i"ca

ntdi!er

ence

are

indi

cate

dby

aste

risk

s( *

).The

stan

dar

der

rorfo

rea

chpa

ram

eter

isin

brac

kets

and

was

eval

uate

dat

the

95%

con"den

cein

terv

al.T

he

bol

din

dica

tes

the

MD

OA

for

each

MSA

TA

A(p

pbv)

TD

A(p

pbv)

Win

dre

gim

eA

vera

gew

ind

spee

d(m

/s)

Dow

nw

ind

Tra

nsport

3-h

range

(km

)

Dow

nw

ind

stat

ion

iden

ti"er

Dow

nw

ind

stat

ion

nam

eO

D(p

pbv)

Upw

ind

(ppbv

)M

DO

A(p

pbv)

RD

U76

.89

[0.1

9]82

.24

[0.5

8]SW

3.0}

3.7

32.5}40

.0U

FO

Fra

nklin

ton

4.50

[2.3

0]H

67.0

0[4

.00]

7.24

[2.9

4]H

NE

2.0}

3.3

21.5}35

.8U

PO

Pitts

boro

5.90

[4.7

0]H

69.0

0[3

.00]

14.0

6[5

.36]H

S2.

2}3.

024

.0}31

.9BTO

Butn

er6.

50[3

.10]H

N/A

9.06

[10.

49]

N2.

4}4.

825

.9}52

.0U

QO

Fuqua

y-V

arin

a8.

12[5

.14]H

N/A

!1.

72[1

4.65

]N

W2.

3}3.

925

.1}41

.7U

TO

Aub

urn

Tow

er4.

20[9

.20]

N/A

N/A

aver

age"

8.32

[3.9

7]H

CLT

77.2

9[1

.43]

83.8

7[1

.18]

SW

3.2}

3.8

24.6}32

.8C

OO

Cou

nty

Lin

e10

.48

[2.1

8]H

62.0

4[4

.44]

21.3

4[3

.94]H

NE

2.3}

3.0

34.0}41

.4U

YO

Yor

kC

ount

y2.

12[2

.67]

81.3

9[4

.69]

1.94

[4.6

9]av

erag

e"13

.58

[4.1

0]H

GSO

75.1

7[1

.87]

81.5

0[3

.50]

W1.

7}2.

518

.6}26

.5M

CO

McL

eansv

ille

12.2

1[2

.90]H

65.8

0[9

.25]

23.6

7[8

.10]H

E1.

8}2.

918

.9}31

.7FC

AFors

yth

Cou

nty

A!

0.20

[5.9

0]N

/AN

/AHH

SW

3.1}

3.8

33.8}49

.2U

CO

Che

rry

Gro

ve5.

67[4

.09]H

61.7

8[2

.81]

18.7

2[1

0.67

]HN

E1.

3}2.

214

.4}23

.9FO

OFork

Rec

reat

ion

0.52

[5.1

4]N

/AN

/AHH

S1.

9}3.

220

.5}34

.1U

BO

Bet

hany

Scho

ol2.

86[5

.07]

N/A

N/A

aver

age"

16.7

1[8

.08]H

**Forea

choft

heE

asta

nd

Nor

thea

stre

gim

esin

theG

SO

regi

on,t

here

was

only

oneda

tapoi

ntw

hic

h"tt

hecr

iter

iaof

this

stud

y,an

dth

usth

est

uden

t'st-te

stco

uld

notb

eca

lcula

ted

forth

ese

regi

mes

.H

owev

er,t

hes

edat

apo

ints

wer

ein

clud

edin

the

over

allG

SOM

DO

Aan

alys

is.


HODs during the study period (Oommen, 1996). TheTAA was calculated for RDU, CLT, and GSO MSAs(Table 3). These values are compared to the rural back-ground values given by Altshuller and Lefohn (1996).

The TAA for RDU was calculated to be 76.89 ppbv[$0.19 ppbv]. The small con"dence interval suggeststhat all the ozone sites surrounding the MSA were high,and not in#uenced by an anomalous measurement. Thisvalue is more than twice the rural background value of35 ppbv [$10 ppbv] (Altshuller and Lefohn, 1996).Thus, there is a large di!erence between the RDU TAAand the background ozone value. Also of importance isthat each TAA is a 3-h average, which indicates highexposure of ozone.

Similarly, the TAAs for CLT and GSO were not signif-icantly di!erent from RDU, but were nearly twice therural background concentrations. Since the averagevalues at all three MSAs were not signi"cantly di!erentfrom each other, this again suggests that high ozone daysare a regional rather than a local problem, but each MSAadds substantially to the regional background ozone onthese days.

5.3. Total downwind averages

The total downwind average (TDA) is computed ateach MSA (Table 3) to determine if there was a statist-ically signi"cant di!erence between the TAA and theTDA (Oommen, 1996). If the ozone concentration at thedownwind site was equal to the TAA, then there is no netozone being added by the city, suggesting that the rate oflocal production of ozone over the city is equal to the rateof destruction between the city and the downwind sites.

The TDA for RDU was computed to be 82.24 ppbv[$0.58 ppbv], which was statistically signi"cantly higherthan its TAA. Again, a small con"dence interval suggeststhat all sites surrounding the MSAs were high. Similarly,the TDAs for CLT and GSO (see Table 3) were statist-ically signi"cantly higher than their respective TAAs, butnot statistically signi"cantly di!erent from RDU, againsuggesting that the three MSAs are a!ected by similarsynoptic-scale weather patterns. The consistently higherdownwind averages suggest that each MSA does add toits downwind ozone level.

5.4. Ozone deviations

The idea of separating the downwind ozone averagesby wind direction was then investigated to see if anyparticular wind direction produced a signi"cantly higherdownwind ozone average. Previous studies (Altshuller,1988) have separated downwind ozone concentrations bywind direction. The di!erence between the downwindconcentration for a particular wind regime and the TAA(Table 3) is called the ozone deviation (OD) (Oommen,1996).

For RDU, the greatest ozone deviation occurs whenthe wind is from the north. That is, ozone concentrationin an air parcel moving from the RDU MSA southwardwill increase by 8.12 ppbv [$5.14 ppbv] when measuredat the downwind station. Other signi"cant ozone devi-ations occur for S, SW, and NE wind directions. Fromthe county precursor maps (EPA, 1995), there is strongevidence that the ozone precursors are being transportedfrom the RDU MSA to regions downwind. Under theabove wind regimes, biogenic VOCs and NO

xemissions

are above average, but not high. However, mobile emis-sions of the precursor gases are high. A possible scenariois that as an air parcel leaves the RDU MSA, it is loadedwith VOCs and NO

xdue to mobile emissions. As this air

parcel is transported by the prevailing wind, ozone isbeing produced giving large concentration at the down-wind site.

For CLT, the SW direction was the only wind regimefor signi"cant ozone deviation. Ozone concentrationsincreased by an average of 10.48 ppbv [$2.15 ppbv]when the air parcel travels from the MSA to the down-wind site. Anthropogenic emissions of VOCs and NO

xlikely contribute to this high ozone concentration devi-ation. A major source for these precursors is likely theexhaust from mobile sources traveling along a highway.As the plume leaves the CLT MSA and moves northeast-ward, it is likely that the ozone precursors from thehighway as a continuous line source (e.g., Interstate-85,which is oriented southwest-northeast) add to the plume(EPA, 1994). The ozone accumulation cycle is furtherdisrupted, causing an increase in the downwind ozoneconcentration.

The highest ozone deviation for GSO occurs when thewind is from the west. That is, as an air parcel leavesthe GSO MSA and travels eastward, 12.21 ppbv [$2.90ppbv] of ozone is added, when measured at the down-wind site. The only other wind regime for signi"cantozone deviation was SW. Like CLT, these increases arebest explained by mobile emissions of VOC and NO

x(EPA, 1994). Interstate-40/85 is aligned east-west fromthe GSO MSA to the RDU MSA, while I-85 is orientedsouthwest-northeast into GSO from CLT. Emissions ofthe precursor gases from mobile sources on Interstate-40/85, and the subsequent conversion to ozone, couldexplain this increase on the eastern side of the MSA.Under a SW wind regime, this suggests that VOC andNO

xemissions from these highways may be transported

into GSO, form ozone, and then are transported to thedownwind site. This situation is also likely consideringthat the counties to the northeast of the GSO MSA donot emit high amounts of VOC and NO

x(EPA, 1995).

Transport of the VOC and NOx

gases in the MSAsappear to play an important part in the higher downwindozone concentrations. Yet, an MSA is not the sole sourcefor the high ozone downwind. If that were true, thenall ozone concentrations at sites downwind should be


signi"cantly higher than the corresponding TAAs. How-ever, this is not the case, indicating the precursor gases arealso being transported into the MSA from other regions.

5.5. Upwind averages

Upwind averages were also computed for the deltaozone analysis (Table 3). The upwind averages were allsigni"cantly higher than the background ozone valuescomputed by Altshuller and Lefohn (1996), indicatingthat transport of ozone and its precursors provides theelevated upwind values for high ozone days. The upwindaverage values were not signi"cantly di!erent from eachother (between 61 and 69 ppbv), except for the muchlarger value of 81.4 ppbv for the northeasterly wind intoCLT. This high value further indicates that ozone mightbe transported from a region further upwind of CLT.

5.6. Modixed delta ozone analysis

To further understand transport of ozone and its pre-cursors, a modi"ed delta ozone analysis (MDOA) wasperformed for the three MSAs. Such an analysis wasperformed by Lindsay et al. (1989) to determine if emis-sion controls were working for the Atlanta, Georgia (GA)MSA, where ozone concentrations have been among thehighest in the country. Delta ozone, *(O

3), is de"ned as

the di!erence between the ozone concentration measuredat a downwind site and that measured at an upwind site.This should re#ect the net increment of photochemicalozone added to an air parcel as it moves over the city.Lindsay et al. (1989) found that the city of Atlanta addedfrom 30 to 40 ppbv for southwest winds as an air parcelpasses through the city. Adams (1992) performed a sim-ilar analysis on ozone data for Raleigh from 1981 to 1991,which showed that about 19 ppbv of ozone was added tothe air parcel as it moved over the city under southwestwinds.

In 1990, the Clean Air Act Amendments (CAAA) settighter standards for ozone control. Primarily throughmore stringent inspection/maintenance (I/M) programs,emissions of the precursor gases (from mobile sources)were to be reduced signi"cantly. A modi"ed delta ozoneanalysis was performed, using the 1993}1995 data fromthe three MSAs in the current study (Oommen, 1996).Several modi"cations have been made to the originaldelta ozone analysis of Lindsay et al. (1989). First, weused 90 ppbv as a minimum (threshold) hourly ozoneconcentration for de"ning a high ozone day whereasLindsay et al. (1989) and Adams (1992) used 80 and60 ppbv, respectively. Second, our MDOA uses a di!er-ent averaging period (0900}1459 EST), as compared to1000}1559 LST used by Lindsay et al. (1989) and Adams(1992). This modi"ed time period is centered aroundnoon and probably better represents the solar insolationaround its maximum value, but still avoids the morning

and evening heavy tra$c periods. The time also startslate enough in summer for the ozone in the residual layerto be mixing downward into the unstable boundary layer(Oommen, 1996), while the boundary layer growth ratemay not reach its maximum (Stull, 1988). Third, morewind direction regimes were used in this study, whencompared to the work of Lindsay et al. (1989), whichrestricted their study to only two wind regimes. It isimportant to recall that not all wind directions had ozonemonitoring sites which "t the 3-h transport criterion;further, not all sites which "t this criteria were suitable forthe modi"ed delta ozone analysis in this study. Addi-tional sites provided four wind regimes for the 1993}1995RDU and GSO data sets, which were found suitable forthe modi"ed delta ozone analysis, whereas CLT still hadtwo suitable wind directions. It should be noted that thedownwind ozone sites may not have optimum locationsfor the speci"ed wind directions, but the sites used arewithin the estimated range of the downwind transportdistances.

The MDOA does show that the Raleigh/DurhamMSA on average added 8.32 ppbv [$3.97 ppbv] as anair parcel passes through the MSA for all suitable winddirections. The above average value of delta ozone isabout half of that found in the earlier analysis performedby Adams (1992), who calculated an average value ofabout 15.00 ppbv [$10.00 ppbv]. Di!erences in thethreshold ozone concentrations, averaging periods, andmonitoring sites used in the two studies could explain theobserved di!erences.

There are four wind regimes (N, NE, S, and SW) whichhave been used in our delta ozone analysis. When thewind is from the NE, the RDU MSA adds 14.06 ppbv[$5.36 ppbv] of ozone (Table 3), whereas it addsonly half that amount for a southwesterly wind. Adams(1992) also separated the total delta ozone analysisinto southwesterly and northeasterly wind regimes, butfound quite di!erent delta ozone values for southwesterly(19.00 ppbv [$7.00 ppbv]) and northeasterly (4.00 ppbv[$1.00 ppbv]) winds. However, it is important to notethat the downwind site for the Adams study utilized theWake Forest (WFO) station. This station stopped takingmeasurements after the 1993 ozone season. Further, thissite does not "t the 3-h transport criterion for a south-west wind. This site is approximately 23 km from RDU'sNWS station. For our study period, the Franklinton(UFO) station, a site further downwind, was utilized(Figs. 2 and 3).

Only emissions of biogenic VOCs are high to thenortheast of RDU, yet the MSA adds approximately14 ppbv of ozone when the wind is northeasterly.Medium to long-range transport of NO

xand additional

VOCs from regions northeast of RDU are plausibleexplanations. Similarly, biogenic VOC emissions to thesouthwest of RDU are high, which makes it harder todistinguish between urban/anthropogenic transport from


CLT or transport from the biogenic-rich regions ofVOCs.

There is no signi"cant contribution to delta ozonewhen winds are from the north, but southerly winds givean average delta ozone of 9.06 ppbv [$10.49 ppbv].The negative delta ozone for a north wind was a littlesurprising, but this value was based on only three datapoints and is not statistically signi"cant. Adams (1992)did not consider north and south winds, so no compari-son can be made with that study.

The MDOA showed that the CLT MSA adds onaverage 13.58 ppbv [$4.10 ppbv] of ozone as an airparcel passes through the MSA. This value shows a slightreduction from the mean delta ozone value of 17 ppbv[$5 ppbv] reported by Adams (1992).

We separated our delta ozone analysis into two windregimes (SW and NE). When the wind is from the south-west, the CLT MSA adds 21.34 ppbv [$3.94 ppbv] ofzone (see Table 3). This is much higher compared to thedelta ozone value for the RDU MSA for a southwesterlywind. Adams computed a delta ozone of 17.50 ppbv[$9.50 ppbv] for a southwesterly wind; however, thissmall di!erence is not statistically signi"cant. The highamount of ozone being added by CLT suggests thatVOCs and NO

xare transported medium to long-range

via the prevailing winds from regions southwest of CLT.The average delta ozone for a northeasterly wind is muchsmaller and not statistically signi"cant, but comparableto that reported by Adams (1992). Thus, there is nosigni"cant change in delta ozone at CLT between the twostudies.

It is important to note that in this study, a site furtherto the southwest of the CLT MSA was chosen. Adams(1992) chose the Arrowood (AWO) site, which we conside-red to be within the CLT MSA (Figs. 3 and 4). Instead,we utilized the York County (UYO), South Carolina, sitewhich is outside of the CLT MSA, but within the 3-htransport range. The di!erences in locations would prob-ably explain the small di!erences in the average deltaozone values found in the two studies for the NE regimein the CLT MSA.

The MDOA shows that the GSO MSA adds, on aver-age, 16.71 ppbv [$8.08 ppbv] of ozone as an air parcelpasses through the MSA. This mean is the highest amongthe three MSAs and twice the mean delta ozone for theRDU MSA. When separating by wind regimes (Table 3),with the west wind the GSO MSA added 23.67 ppbv[$8.10 ppbv] of ozone, whereas for the southwest wind,it added 18.72 ppbv [$10.67 ppbv]. For the east andnortheast winds, delta ozone could not be calculated forthe lack of signi"cant data. Because emissions of VOCsare above average and NO

xis low to the west of GSO

(EPA, 1995), transport of VOCs and NOx

into GSO areconsidered the likely explanation for the high amount ofozone being added by the city. Under a southwesterlywind, there appears to be a mixture of medium to long-

range transport from CLT, coupled with mobile emis-sions along Interstate-85.

6. Summary and conclusions

Statistical analyses were performed on hourly ozoneconcentration data from 1993 to 1995 for three metro-politan statistical areas (MSAs) in North Carolina:Raleigh/Durham (RDU), Charlotte/Mecklenburg (CLT),and Greensboro/High Point/Winston-Salem (GSO). Theresults of this study show that even though stricter regu-lations on emission controls for the precursors of ozonehave been implemented, as a result of the 1990 CAAA,these MSAs still experience high ozone days (e.g., anyhourly ozone concentration*90 ppbv) and exceedancesof the previous 1-h NAAQS (e.g., hourly ozone concen-trations*120 ppbv). The three MSAs had a total of 16exceedances of the ozone standard over 15 d. Of the549 d (April}September, 1993}1995) in the study period,there were 169 d which had high ozone (ozone concentra-tion*90 ppbv) in at least one of the three MSAs. Wind"eld patterns and averages of ozone were analyzed in ane!ort to determine whether these high ozone days(HODs) occur due to local contributions versus regionaltransport.

Local processes of ozone formation were found to beimportant in determining HODs at the three MSAs, asthere were signi"cant emissions of VOCs and NO

x.

County mobile precursor emission maps showed that thethree MSAs have higher than average mobile emissionsof VOCs and NO

x. Additionally, the three MSAs were

above average in biogenic VOCs, but low in biogenicemissions of NO

x. Thus, the large amounts of precursors

in these MSAs in conjunction with strong solar insola-tion and high daytime temperatures during late springand summer periods accounted for the large number ofHODs.

A comparison of the wind roses for high ozone daysand those based on the climatology showed some inter-esting features of the three MSAs. On high ozone days,winds at RDU were frequently from the directions ofCLT and GSO. Similarly, GSO measured a high percent-age of its high ozone days when winds were from RDUand CLT. The frequencies of above wind directions onHODs were signi"cantly greater than those obtainedfrom climatology at both RDU and GSO. Yet, CLT'swinds on high ozone days were similar to the clima-tological values. However, southwesterly winds at CLTon high ozone days were more frequent than the clima-tological value, indicating a signi"cant source for CLTshigh ozone comes from regions to the southwest of thatMSA.

The total area average (TAA) for the three MSAswere signi"cantly higher than the regional backgroundvalue of ozone reported by Altshuller and Lefohn (1996),


indicating the importance of the local production ofVOCs and NO

xin these MSAs, especially on those

selected HODs. Both the regional transport and localproduction of ozone within the MSA contributed to thehigh ozone values downwind of the three MSAs, as thetotal downwind averages (TDAs) were signi"cantly high-er than their TAAs. However, there are no signi"cantdi!erences among either the TDAs or TAAs for the threeMSAs. This suggests that the three MSAs are similar and,in general, high ozone concentrations occur under thesame regional/synoptic conditions.

An analysis of ozone deviation showed the importanceof photochemical production and regional transport ofozone from the MSA to the downwind site. Highestozone deviations occurred at RDU under a north wind,CLT under a southwest wind, and GSO under a westwind. Signi"cant deviations also occurred at RDU underS, SW, and NE winds and at GSO under southwestwinds. Local emissions of anthropogenic VOC and NO

xfrom mobile sources contributed to these high ozonedeviations. If the MSAs were solely responsible for thedeviation, then all wind directions should have producedsigni"cant ozone deviations, whose magnitudes woulddepend on the spatial distribution of emission sources aswell as on the transport distance over the MSA. Sincethis was not strictly true, regional, and potentially long-range transport of VOCs and NO

xmight also be

in#uencing the ozone deviations.For all three MSAs, upwind ozone concentrations

were found signi"cantly higher than the regional back-ground ozone value. Lower background concentration isa result of averaging in both space and time (this includesall days, not just the HODs). Much elevated upwindconcentrations on HODs could be attributed to regionaltransport of ozone and its precursors (VOCs and NO

x)

by the prevailing winds and conversion to ozone beforereaching the upwind site. The upwind concentrationswere signi"cantly lower than the total area averagedconcentrations, except for site northeast of CLT. Themodi"ed delta ozone was the contribution of the MSA tothe local ozone production. The modi"ed delta ozoneanalysis (MDOA) values at RDU and CLT were com-pared with those computed by Adams (1992) for anearlier (1981}1991) data set. The comparison showeda 50% reduction of delta ozone at RDU, but no signi"-cant change at CLT. The decrease in the RDU MDOmay partly be attributed to the tougher emission stan-dards as required by the Clean Air Act Amendments of1990; however, 3 years of data is a short record fromwhich to qualitatively make this decision. It is not clearwhy the same regulations did not reduce MDO at CLT.No comparison between the two data sets was made forGSO.

The MDOA was also separated by wind regimes ateach MSA. At RDU, the highest MDO occurred undera northeast wind. Since only biogenic VOCs are high

northeast of RDU, this suggests medium to long-rangetransport of ozone precursors from regions furthernortheast by the prevailing winds. For CLT and GSO,the highest MDOs occurred under southwest winds, al-though a west wind at GSO also gave a large MDO.Lower, county-averaged emissions of the precursor gasesto the southwest of CLT and west of GSO may explainrelatively large values of MDO for these MSAs undersouthwest winds.

We "nd that on the high ozone days experienced bythe RDU, CLT, and GSO MSAs ozone pollution wasin#uenced by the local production of VOCs and NO

xdue to mobile sources, the interaction of the wind "eldsbetween the MSAs, and regional transport of the precur-sor gases. Additionally, upwind ozone averages wereelevated even in regions where VOC and NO

xemissions

were low, suggesting that long-range transport of precur-sor gases could occur from other regions, such as themid-Atlantic, Ohio Valley, and to the southwest of theCLT MSA.

Acknowledgements

This research has been partly funded through a con-tract from North Carolina Department of Environmentand Natural Resources (Contract No. EA 8001). Thanksto Mr. B. Timin, NC Division of Air Quality, for com-ments and suggestions on the manuscript. Thanks to Ms.M. DeFeo for the word processing of the manuscript.The mention of trade names or commercial or noncom-mercial products does not necessarily constitute endorse-ment or recommendation for use.

References

Adams, A.A., 1992. An observational-based analysis of ozonetrends and production in North Carolina. M.S. Thesis NorthCarolina State University, Raleigh, NC, 80 p.

Aneja, V.P., Yoder, G.T., Arya, S.P., 1992. Ozone in the UrbanSoutheastern United States. Environmental Pollution 75,39}44.

Aneja, V.P., Li, Z., Das, M., 1994. Ozone case studies at highelevation in the Eastern United States. Chemosphere29(8), 1711}1733.

Aneja, V.P., Kim, D., Das, M., Hartsell, B.E., 1996. Measure-ments and analysis of reactive nitrogen species in the ruraltroposphere of Southeast United States: Southern oxidantstudy site SONIA. Atmospheric Environment 30(4),649}659.

Altschuller, A.P., 1986. Relationships between direction of wind#ow and ozone in#ow concentrations at rural locationsoutside of St. Louis, MO. Atmospheric Environment20(11), 2175}2184.

Altschuller, A.P., 1988. Some characteristics of ozone formationin the urban plume of St. Louis, MO. Atmospheric Environ-ment 22(3), 499}510.


Altshuller, A.P., Lefohn, A.S., 1996. Background ozone in theplanetary boundary layer over the United States. J. Air& Waste Management Association 46, 134}141.

Environmental Protection Agency (EPA), 1993. The Plain Eng-lish Guide to the Clean Air Act, EPA, 400-K-93-001.

Environmental Protection Agency (EPA), 1994. Measuring AirQuality: the Pollutants Standards Index, EPA, 451-K-94-001.

Environmental Protection Agency (EPA), 1995. National AirPollutant Emission Trends, 1900}1994, EPA, 454/R-95-001.

Environmental Protection Agency (EPA), 1996. National AirQuality & Emissions Trends Report, 1995, 454/R-96-005,p. 21.

Environmental Protection Agency (EPA), 1998. Ozone, CarbonMonoxide, Particulate Matter, Sulfur Dioxide and LeadNonattainment Areas, 1998, August 4.

Lawrimore, J.H., Das, M., Aneja, V.P., 1995. Vertical distribu-tion of non-methane hydrocarbons. Journal of GeophysicalResearch 100(D11), 22785}22793, Nov 20.

Lindsay, R.W., Richardon, J.L., Chameides, W.L., 1989. Ozonetrends in Atlanta, GA: have emission controls been e!ective?.Journal of Air Pollution Control Association 39, 40}43.

Logan, J.A., 1985. Tropospheric ozone: seasonal behavior,trends, and anthropogenic in#uence. Journal of GeophysicalResearch 90 (D6), 10463}10482.

Moy, L.A., Dickerson, R.R., Ryan, W.F., 1994. Relationshipbetween back trajectories and tropospheric trace gas concen-tration in rural Virginia. Atmospheric Environment 28 (17),2789}2800.

National Research Council (NRC), 1991. Rethinking the OzoneProblem in Urban and Regional Air Pollution. NationalAcademy Press, Washington, DC.

North Carolina Division of Environment, Health, andNatural Resources (NC DEHNR): Air Quality Section,2728 Capitol Blvd, P.O. Box 29580, Raleigh, NC 27626-0580.

Oommen, R.G., 1996. Boundary layer ozone variations overcentral North Carolina. M.S. Thesis North Carolina StateUniversity, 149 p.

Sillman, S., Samson, P.J., Masters, J.M., 1993. Ozone productionin urban plumes transported over water: photochemicalmodel and case studies in Northeastern and MidwesternUnited States. Journal of Geophysical Research 98 (D7),12687}12699.

Stull, R.B., 1988. In: An Introduction to Boundary LayerMeteorology. Kluwer Academic Publishers, Dordrecht.

Sullivan, L.J., Moore, T.C., Aneja, V.P., Robarge, W.P., Pierce,T.E., Geron, C., Gay, P., 1996. Environmental variablescontrolling nitric oxide emissions from agricultural soils inthe Southeast United States. Atmospheric Environment 30(21), 3573}3582.

Vukovich, F.M., 1994. Boundary layer Ozone variations in theEastern United States and their associations withmeteorological variations: long-term variations. Journal ofGeophysical Research 99 (D8), 16839}16850.

Warneck, P., 1988. Chemistry of the Natural Atmosphere. Inter-national Geophysics Series, Vol. 41. Academic Press, Inc,San Diego.


Ozone patterns for three metropolitan statistical ... - NCSU

Documents

Transcript of Ozone patterns for three metropolitan statistical ... - NCSU