Full text of DOI:10.1080/10473289.2000.10464140

This article was downloaded by: [University of Kent]
On: 22 November 2014, At: 04:45
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Journal of the Air & Waste Management Association
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uawm20
Ozone Formation in California's San Joaquin Valley:
A Critical Assessment of Modeling and Data Needs
Betty K. Pun a , Jean-François Louis a , Prasad Pai a , Christian Seigneur a , Sam
Altshuler b & Guido Franco c
a Atmospheric and Environmental Research, Inc. , San Ramon , California , USA
b Pacific Gas and Electric Company , San Ramon , California , USA
c California Energy Commission , Sacramento , California , USA
Published online: 27 Dec 2011.
To cite this article: Betty K. Pun , Jean-François Louis , Prasad Pai , Christian Seigneur , Sam Altshuler & Guido Franco
(2000) Ozone Formation in California's San Joaquin Valley: A Critical Assessment of Modeling and Data Needs, Journal of
the Air & Waste Management Association, 50:6, 961-971, DOI: 10.1080/10473289.2000.10464140
To link to this article: http://dx.doi.org/10.1080/10473289.2000.10464140
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose
of the Content. Any opinions and views expressed in this publication are the opinions and views of the
authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should
not be relied upon and should be independently verified with primary sources of information. Taylor
and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses,
damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

TECHNICAL PAPER
Pun et al.
ISSN 1047-3289 J. Air & Waste Manage. Assoc. 50:961-971
Copyright 2000 Air & Waste Management Association
Ozone Formation in California’s San Joaquin Valley: A Critical
Assessment of Modeling and Data Needs
Betty K. Pun, Jean-François Louis, Prasad Pai, and Christian Seigneur
Atmospheric and Environmental Research, Inc., San Ramon, California
Sam Altshuler
Pacific Gas and Electric Company, San Ramon, California
Guido Franco
California Energy Commission, Sacramento, California
ABSTRACT
th is con ceptual m odel, we th en con duct a critical evalua-
tion of th e m eteorological m odel an d air quality m odel.
Areas of m odel im provem en ts an d data n eeded to un -
derstan d an d properly sim ulate O3 form ation in th e SJV
are h igh ligh ted.
Data from th e 1990 San Joaquin Valley Air Quality Study/
Atm osph eric Utility Sign atures, Prediction s, an d Experi-
m en ts (SJVAQS/AUSPEX) field program in Californ ia’s San
Joaquin Valley (SJV) suggest th at both urban an d rural
areas would h ave difficulty m eetin g an 8-h r average O3
stan dard of 80 ppb. A con ceptual m odel of O3 form ation
an d accum ulation in th e SJV is form ulated based on th e
ch em ical, m eteorological, an d tracer data from SJVAQS/
AUSPEX. Two m ajor ph en om en a appear to lead to h igh
O3 con cen tration s in th e SJV: (1) tran sport of O3 an d pre-
cursors from upwin d areas (prim arily th e San Fran cisco
Bay Area, but also th e Sacram en to Valley) in to th e SJV,
affectin g th e n orth ern part of th e valley, an d (2) em is-
sion s of precursors, m ixin g, tran sport (in cludin g lon g-
ran ge tran sport), an d atm osph eric reaction s with in th e
SJV respon sible for region al an d urban -scale (e.g., down -
win d of Fresn o an d Bakersfield) distribution s of O3. Usin g
INTRODUCTION
Sum m ertim e O3 air pollution is a serious problem in
California’s San Joaquin Valley (SJV). The 1990 San Joaquin
Valley Air Quality Study (SJVAQS) and the Atm ospheric
Utility Signatures, Predictions, and Experim ents (AUSPEX)
were two com ponents of a com prehensive study to ad-
dress th is issue.1 Th e com preh en sive study in volved (1)
a large-scale air quality m easurem en t program , (2) an aly-
ses of th e causes of poor O3 air quality, an d (3) th e devel-
opm en t an d application of a region al-scale air quality
m odelin g system (SJVAQS/AUSPEX Region al Model Ad-
aptation Project, or SARMAP). Th e field program was
con ducted durin g an eigh t-week period from July 9, 1990,
to August 24, 1990. Fourteen in ten sive days were selected,
wh ich correspon ded to h igh O3 episodes, for detailed
m easurem en ts. Figure 1 depicts th e SJVAQS/AUSPEX
dom ain .1
IMPLICATIONS
The development of State Implementation Plans to ad-
dress the O3 air quality standard will require meteoro-
logical and air quality models that provide accurate rep-
resentations of the relevant atmospheric processes over
regional scales. The California SJV is an appropriate test
bed for evaluating state-of-the-science models, because
of its variety of chemical and meteorological regimes and
the availability of data to execute and test models. A
conceptual model of O3 formation in the SJV was devel-
oped and used to critically assess the suitability of me-
teorological and air quality models. Specific recommen-
dations are provided to improve existing models, thereby
increasing the level of confidence in the use of model
predictions of O3 and precursors to develop cost-effective
emission control strategies.
Tim e series of th e daily m axim um 1-h r an d 8-h r av-
erage O3 con cen tration s are sh own in Figures 2 an d 3,
respectively.2 At urban sites, th e 8-h r average con cen tra-
tion s can be 20 to 30 ppb lower th an th e 1-h r m axim um
wh en O3 con cen tration s are above 100 ppb (e.g., in
Fresn o). On th e oth er h an d, at rural sites such as Sequoia
Nation al Park, th e daily m axim um 8-h r con cen tration s
of O3 ten d to be sim ilar (< 10 ppb differen ce) to th e daily
1-h r m axim um con cen tration s. As a result, wh ile urban
areas exceed th e 1-h r average Nation al Am bien t Air Qual-
ity Stan dard (NAAQS) of 120 ppb, both urban an d rural
areas m ay exceed an 8-h r average NAAQS of 80 ppb.
Volume 50 June 2000
Journal of the Air & Waste Management Association 961

Pun et al.
Figure 1. The study domain of SJVAQS/AUSPEX (reprinted with
permission from Ranzieri and Thuillier,1 copyright CRC Press, Boca
Raton, FL, p 25).
Our objectives are to (1) construct an integrated analy-
sis of O3 form ation in the SJV and (2) use it to critically
assess wh eth er existin g m eteorological an d air quality
models can accurately describe the major relevant processes.
First, we build upon previous data analyses and develop a
conceptual m odel of sum m ertim e O3 form ation in the SJV,
which com bines our knowledge of chem ical processes with
th at of tran sport processes. Previous in vestigators h ave
analyzed the am bient chem istry and aerosol data,3 and the
m eteorological and tracer data4-6 of the SJVAQS/AUSPEX
program . However, n on e of th ose an alyses provided an
integrated analysis of the chem istry and transport processes
leading to O3 form ation in the SJV.
Figure 2. Daily maximum 1-hr average ozone concentrations at
selected sites.2
Secon d, we assess th e ability of curren t m ath em ati-
cal m odels of m eteorology an d air quality to provide an
accurate description of th e m ajor causes for O3 form ation
iden tified in th e con ceptual m odel. As discussed below,
th e SJV presen ts a wide variety of con dition s for atm o-
sph eric ch em istry an d m eteorology. Th erefore, our assess-
m en t of m odel capabilities sh ould fin d application s to
oth er areas besides th e SJV. In term s of m odel evaluation ,
our approach differs from th e m ore typical approach th at
in volves com parin g m odel prediction s with observation s.
We believe th at such com parison s provide on ly lim ited
in sigh ts in th e m odel weakn esses un less th e observation al
database is com preh en sive en ough to allow a detailed di-
agn ostic evaluation of th e m odels. Th us, our evaluation
of th e m odel form ulation with respect to th e con ceptual
m odel is com plem en tary to th e stan dard m odel perfor-
m an ce evaluation approach .
Figure 3. Daily maximum 8-hr average ozone concentrations at
selected sites.2
962 Journal of the Air & Waste Management Association
Volume 50 June 2000

Pun et al.
Following these analyses, we highlight areas for addi-
tional data collection needed for rigorous m odel evalua-
tion, including diagnostic m odel evaluation, and areas of
improvements to existing mathematical models. Additional
m odel evaluation and im provem ent activities allow us to
gain confidence that the m odels predict O3 and its response
to reductions of precursors correctly and without com pen-
sating errors, so that they can be used to develop cost-ef-
fective em ission con trol strategies for dem on stratin g
attainm ent of the O3 am bient air quality standards.
other parts of the SJV overnight (see discussion under long-
ran ge tran sport).
Except for terrain -in duced vertical m otion s of air
th rough m oun tain passes, vertical processes are of little
relevan ce in th is tran sport ph en om en on . August 6, 1990
was an exam ple wh en on sh ore flows becam e m ore dom i-
n an t after th e syn optic h igh pressure system m oved away
from th e SJV.7 High O3 con cen tration s were observed in
th e n orth western SJV (120–150 ppb n ear Tracy/Stockton
and Crows Landing) in the late afternoon. Since the m orn-
in g aloft (carryover) O3 was on ly 40–50 ppb, th e after-
n oon peak was con sisten t with th e in fluen ce of pollutan t
tran sport from th e Bay Area. In addition , pollutan t fluxes
from th e Sacram en to Valley exacerbated th e problem . Th e
com bin ed NOx fluxes in to th e n orth ern part of th e SJV
exceeded th e n orth ern valley em ission s on th at day.8
CONCEP TUAL MODEL
The conceptual m odel is a qualitative com pilation of the
physical and chem ical processes that govern the form a-
tion of O3, which, to the extent possible, is supported by
quantitative inform ation. Two scenarios help conceptual-
ize the dynam ics and chem istry of O3 pollution in the SJV:
(1) transport of O3 and precursors from upwind areas into
the SJV, and (2) precursor em issions, O3-form ing reactions,
and pollutant transport processes within the SJV. The in-
fluence of O3 and precursors from upwind areas is enhanced
by organized flow from the north/northwest into the val-
ley. Durin g m ulti-day stagn ation periods, with in -valley
processes of particular im portance are em issions and trans-
port processes including, but not lim ited to, eddy patterns,
nocturnal jets, recirculation of polluted surface layers due
to terrain-restricted outflows, large-scale subsidence events,
and vertical m ixing. These two scenarios are neither m u-
tually exclusive nor collectively exhaustive.
P r ecu r sor Em ission s, O3-For m in g Rea ct ion s, a n d
P ollu t a n t Tr a n sp or t P r ocesses w it h in t h e SJV
Alth ough th e tran sport of O3 an d precursors from upwin d
areas con tributes to O3 accum ulation in th e SJV, th e
with in -valley production of O3 was postulated to be ca-
pable of gen eratin g O3 levels th at exceed th e NAAQS.9 O3
form ation in th e valley results from an in terplay of ch em -
istry with ph ysical processes in cludin g em ission s of pre-
cursors (NOx an d VOC), aloft lon g-ran ge tran sport with in
th e valley, vertical m ixin g un der th e large-scale subsid-
en ce, an d deposition .
Emissions and Concentrations of NOx and VOC in the SJV.
According to SJV emissions inventories, the major NOx con-
tributors are m obile sources,10 consistent with observations
of m axim um m orning concentrations of NOx in urban ar-
eas. Elevated point sources m ost likely influence O3 pro-
duction farther downwind than surface em issions, except
under m eteorological conditions conducive to downwash
near the stacks.11 The chem ical m ass balance (CMB) results
show VOC contributions from vehicle exhaust, liquid gaso-
line, gasoline vapor, oil production, and acetone and uni-
dentified/unexplained VOC. During the 1990 field study,
VOC concentrations were highest in the m orning sam ples
from th e south ern SJV, wh ere sign ifican t h ydrocarbon
em ission s occurred from oil production . Motor veh icle
exhaust was the m ain contributor at all tim e periods in the
urban areas, ranging from 35 to 70% of total VOCs.12
Com pared to the CMB results, inventory estim ates of
m otor vehicle VOC em issions (exhaust and evaporation)
were significantly lower. The em issions inventory would
need to be increased by factors of 3 to 4 to be consistent
with the observed am bient concentrations at urban sites.12
The Yosem ite and Sequoia sites showed 10–15% VOC con-
tributions from biogenic em issions according to the CMB
results. However, the inventory estimates were substantially
Tr a n sp or t of O3 a n d P r ecu r sor s fr om
Up w in d Ar ea s in t o t h e SJV
O3 an d precursors from poin t, area, an d m obile sources
accum ulate in th e m ixed layer durin g th e day in th e San
Fran cisco Bay area, wh ere m obile sources accoun t for
> 50% of th e VOC an d NOx em ission s. Th e pollutan ts are
distributed th rough out th e m ixed layer by turbulen t dif-
fusion. Som e elevated point sources m ay be released above
th e m ixed layer, an d th ese ten d to rem ain as in tact plum es
for lon ger periods of tim e. Wh en syn optic con dition s are
favorable, daytim e sea breezes are fun n eled th rough th e
Carquin ez Strait an d n earby m oun tain passes, brin gin g
O3 an d precursors in to th e n orth ern part of th e SJV by
low-level advection . Som e in flow is also observed th rough
th e Pach eco Pass to th e west side of th e SJV. Fluxes of O3
an d NOx accom pan yin g sea breezes can be sign ifican t, as
m easured durin g th e SJVAQS/AUSPEX program . Wh ile
VOC fluxes were n ot directly m easured durin g th e 1990
study, both fresh an d aged VOC (e.g., from traffic in an d
out of th e Bay Area) are expected to reach th e Cen tral
Valley alon g with O3 an d NOx. Th ese tran sported pollut-
an ts con tribute prim arily to h igh O3 in th e n orth ern part
of th e SJV in th e late aftern oon an d m ay be distributed to
Volume 50 June 2000
Journal of the Air & Waste Management Association 963

Pun et al.
higher (> 50%). This discrepancy m ay be due to the trans-
port of VOC (anthropogenic em issions and secondary or-
ganic com pounds) from urban areas, the high reactivity of
isoprene, or uncertainties in the biogenic em ission esti-
m ates. Larger unidentified VOC and acetone fractions were
m easured in the afternoon sam ples than in the m orning
sam ples, reflecting the O3-producing atm ospheric reactions
that took place during the day. Sam e-day urban-scale pro-
duction resulted in high O3 downwind from urban centers
(e.g., Fresno, Bakersfield) in the late afternoon.
to an d above th e in version layer as a coh eren t plum e
before coolin g to th e am bien t n igh ttim e tem perature. Pre-
served O3 an d oth er pollutan ts are tran sported down win d
by th e n octurn al jet.
In th e m orn in g, un der th e stron g in solation , th e tem -
perature rises rapidly, form in g a m ixed layer th at brin gs
th e tran sported pollutan ts to th e surface. Sin ce th e aloft
air is ch aracterized by h igh O3 con cen tration s in m an y
parts of th e SJV, th e im m ediate effect of vertical m ixin g
in th e m orn in g m ay be th e titration of surface NO by O3
to form NO2. As th e m ixed layer grows, O3 production
m ay take place both at th e surface an d aloft with in th e
m ixed layer. At presen t, th e exten t of O3 production aloft
is un certain . Th e relative efficien cy of ph otoch em ical re-
action s at th e surface an d aloft m ay be related to th e ligh t
extin ction ch aracteristics of PM.13,14 For exam ple, UV re-
duction s associated with h igh aerosol loadin gs can result
in a decrease in O3 of several percen t due to th e absorp-
tive properties of n itrate an d arom atic aerosols.13
Aloft Long-Range Transport within the SJV. Groun d-level
winds are generally weak. Long-range transport takes place
aloft, particularly at n igh t, wh en a n octurn al jet develops
approxim ately alon g th e axis of th e SJV at an altitude of
about 400 m . Th e n octurn al jet carries pollutan ts toward
th e south ern tip of th e valley (up-valley) at 10–30 m sec-1.
By early m orn in g, th e Fresn o eddy form s due to blockage
by th e elevated terrain . Eddy flows recirculate tran sported
pollutan ts in th e south ern SJV, an d as th ey collapse, early
m orn in g em ission s in Fresn o an d Bakersfield are also re-
circulated in th e source area. Aloft con cen tration data
sh ow th at th e carry-over burden of O3 in creases from th e
n orth ern part of th e SJV to th e south ern part an d is as
h igh as 100 ppb in th e down win d area, wh ere th e h igh est
O3 con cen tration s are observed.8
Upslope circulation along the Sierras and the Coastal
Range generates divergence in the valley, strengthening the
subsidence and keeping the m ixed layer shallow. The pol-
lutants are therefore trapped in the valley, without m uch
possibility of escape except over the Chalom e Pass (west
side of the SJV, northern boundary of Kern County) and
the Tehachapi Pass. Dry deposition, especially over agri-
cultural land, appears to be an im portant physical rem oval
process for O3 within the SJV. However, dry deposition of
precursors and products has not been thoroughly studied.
Sim ilar to the characteristics of precursors in the after-
noon, aloft parcels contain little NOx, and the VOC m ix-
ture tends to be chem ically aged, with concentrations of
about a quarter of those m easured at ground level in the
m orning. Long-range transport appears to be especially im -
portant in influencing the O3 concentrations at rural sites
with no significant sources nearby. (Note that air exchange
between the southwest SJV and the San Luis Obispo area–
for exam ple, over th e Ch olam e Pass–m ay result in h igh O3
in either area under different synoptic-scale m eteorology.4)
With efficient transport by the nocturnal jet, sources hun-
dreds of m iles upwind m ay contribute to high O3 concen-
trations in the southern part of the valley when the aloft
air is m ixed into the surface layer during the next day.
Sen sit ivit y of O3 t o P r ecu r sor s
Com puter m odelin g an d data an alyses usin g observation -
based m odels sh ow th at sum m er O3 in m ost areas of th e
SJV appears to be m ore sen sitive to NOx con trols th an to
VOC reduction s alon e.15,16 However, th ere is sign ifican t
eviden ce th at sum m er O3 in th e San Fran cisco Bay area is
VOC-sen sitive. In fact, reduction s in NOx m ay result in
an in crease in O3 in th e Bay Area.15,17-19 O3 sen sitivity is a
fun ction of location an d m eteorology. Major urban cen -
ters in th e n orth ern SJV m ay be VOC-sen sitive, alth ough
oth er parts m ay be NOx-sen sitive.15 Sin ce sligh t O3 reduc-
tion s in th e n orth ern SJV result from NOx reduction s in
th e Bay Area un der certain m eteorological scen arios,15 Bay
Area em ission s con tribute to NOx budgets in th e SJV, an d
th e n orth ern SJV m ay be th e tran sition location wh ere
th e Bay Area air m ass ch an ges over from VOC to NOx
sen sitivity as it travels down win d.
Vertical Mixing. Th e large-scale subsiden ce from th e East
Pacific Ridge an d th e advection of warm , rath er m oist air
from th e south east keep th e m ixed layer sh allow (400–
1200 m ), despite h igh tem peratures (40 °C). Reduced m ix-
in g is a m ajor factor con tributin g to pollutan t buildup.
Pollutants within the daytim e m ixed layer m ay be trapped
above th e surface layer at n igh t (< ~100 m ). In sulated from
groun d-level em ission s, aloft O3 can be preserved over-
n igh t. O3 at elevated sites such as Sequoia m ay stay above
70 ppb at n igh t (e.g., August 2, 19902), an in dication of
th e exten t of O3 preservation aloft. Som e poin t sources
h ave h igh en ough buoyan cy at n igh t to carry pollutan ts
Key Kn ow led ge Ga p s Id en t ified b y
t h e Con cep t u a l Mod el
In th e form ulation of a con ceptual m odel for O3 form a-
tion in th e SJV, several kn owledge gaps were iden tified.
Of th e m ost sign ifican t con sequen ce is th e sen sitivity of
964 Journal of the Air & Waste Management Association
Volume 50 June 2000

Pun et al.
O3 ch em istry at differen t receptor location s for differen t
m eteorological scen arios. Lu an d Ch an g16 studied th e re-
spon se of SJV O3 to dom ain -wide reduction s in NOx an d
VOC em ission s an d foun d th at th e SJV is NOx-sen sitive
for th e August 4–5, 1990 scen ario. However, th e n orth -
western SJV m ay be in fluen ced by Bay Area em ission s
wh en sea breezes pen etrate in to th e valley (e.g., August
6), an d by with in -valley em ission s un der m ore stagn an t
con dition s (e.g., August 4, 5). Th e sen sitivity of O3 to its
precursors un der th e en h an ced tran sport m eteorological
scen ario h as n ot been verified. Th e sen sitivity of O3 in
th e n orth ern SJV depen ds on th e origin of O3. If O3 is
form ed in th e Bay Area an d tran sported down win d in to
th e SJV, it will probably exh ibit VOC sen sitivity.
th e exten t to wh ich th e sim ulation s of th e SJV agree with
th e available data an d exam in e wh eth er th ere are weak-
n esses or deficien cies in MM5 th at preclude it from repre-
sen tin g th e con ceptual m odel presen ted above. Th is
evaluation is based prim arily on th e work of Seam an et
al.7 an d Seam an an d Stauffer.20 We also draw on our expe-
rien ce with MM5 sim ulation s perform ed for oth er pur-
poses, m ain ly over th e Los An geles basin ,21 th e Ten n essee
Valley,22 an d th e New En glan d states.23
MM5 is a non-hydrostatic m esoscale m odel developed
at th e Pen n sylvan ia State Un iversity an d th e Nation al
Cen ter for Atm osph eric Research (NCAR).24,25 Th e m odel
can run in n ested m ode, with on e or m ore fin e m esh in -
n er grids com m un icatin g with a coarser m esh outer grid.
Th is approach m akes it possible to perform very fin e reso-
lution sim ulation s on a sm all area with out un due in flu-
en ces from th e lateral boun dary con dition s, wh ich n eed
to be im posed on ly on th e outerm ost grid. MM5 in cludes
a com plete set of ph ysical param eterization s, an d in cludes
several types of param eterization s for each ph ysical pro-
cess. A four-dim en sion al data assim ilation (FDDA) system
is also available for MM5. It is a Newton ian relaxation , or
“n udgin g,” system , in wh ich an extra term is added to
th e ten den cy equation s. Th is extra term is proportion al
to th e differen ce between th e m odel solution an d th e
observation s or a large-scale an alysis.
Seam an et al.7 perform ed sim ulation s with a triply
n ested grid, with th e in n er grid coverin g th e en tire SJV
with a h orizon tal m esh size of 4 km an d a vertical resolu-
tion of 30 layers (ran gin g from 20 m to 1 km ) from groun d
level to th e tropopause. Overall, th e MM5 sim ulation s of
th e SARMAP episodes are quite good, especially th e run s
th at assim ilate th e special observation data. All th e im -
portan t m eteorological aspects of tran sport in th e SJV are
qualitatively well sim ulated by th e m odel. Th ese in clude
th e sea breezes, th e fun n elin g of th e flow th rough th e
Carquin ez Strait an d th e oth er open in gs of th e Coastal
Ran ge, th e upslope flow durin g th e day alon g th e Coastal
Ran ge an d th e Sierra Nevada, th e n octurn al jet alon g th e
valley axis, an d th e Fresn o eddy. Alth ough th ese sim ula-
tion s are gen erally successful, we discuss below several
aspects th at could be im proved.
As a polluted air m ass travels down win d, NOx is de-
pleted faster th an VOCs, an d O3 production with in th e
air parcel becom es in creasin gly NOx-lim ited with out fresh
em ission s of precursors. If O3 is form ed down win d from
th e source area, it m ay be NOx-sen sitive. Th e precursor
NOx m ay be locally em itted or origin ate from an upwin d
location . Th e m ore likely scen ario is th at th e observed O3
in th e n orth ern SJV is a com bin ation of O3 produced up-
win d an d in th e SJV. In th is case, wh ile VOC con trols in
th e San Fran cisco Bay area m ay be ben eficial for reducin g
O3 production in th e Bay Area an d tran sported O3 at th e
receptor sites in th e n orth ern part of th e SJV, NOx con -
trols m ay be n eeded with in th e valley to reduce th e O3
production poten tial in areas affected by both fresh em is-
sion s an d aged air from th e Bay Area.
Eight-hour average concentrations of O3 exceed 80 ppb
at m any urban and rural sites (see Figure 3). Rural sites
m ay be affected by aloft transport of O3 and precursors
from upwind areas, vertical exchange with the surface layer
and free troposphere (especially for elevated sites such as
Yosem ite), local production of O3 from biogenic em issions,
and O3 scavenging by reactive VOCs. The processes/fluxes
contributing to high 8-hr concentrations at rural sites have
not been fully elucidated. The sensitivity of O3 to precur-
sors m ay also be different for the 8-hr average concentra-
tions, so that the current understanding of 1-hr sensitivities
cannot be translated to 8-hr average O3 with certainty.
Oth er kn owledge gaps in clude th e exten t of aloft pro-
duction of O3 an d th e im portan ce of deposition of pre-
cursors an d product species oth er th an O3. Sin ce th e
con ceptual m odel was form ulated prim arily based on in -
form ation from th e 1990 SJVAQS/AUSPEX episodes, th e
represen tativen ess of th e m eteorology of th ose episodes
n eeds to be con firm ed.
Dom a in a n d Resolu t ion
On e of th e m ain deficien cies of th e m eteorological sim u-
lation s is th e in ability of th e m odel to split th e flow east
of th e Carquin ez Strait in th e h igh est resolution dom ain .
Norm ally, th e low-level flow sh ou ld split in to two
bran ch es, on e goin g south east up th e SJV and the other
one turning north up the Sacramento Valley. The northern
part of the flow is generally absent in the high-resolution
dom ain , alth ough th ere are in dication s of th e flow in
th e lower resolution dom ain s. Th ere is eviden ce th at th is
ASSESSMENT OF TH E METEOROLOGICAL
MODEL MM5
In SARMAP, th e m esoscale m eteorological m odel (MM5)
was used to sim ulate m eteorology in th e SJV. We evaluate
Volume 50 June 2000
Journal of the Air & Waste Management Association 965

Pun et al.
problem is related to th e position of th e n orth ern
boun dary of th e dom ain . In on e sim ulation wh ere th e
in n er dom ain was en larged to in clude part of n orth ern
Sacram en to Valley, th e split of th e flow was m uch bet-
ter sim ulated.
Tr ea t m en t of P r ocesses
The two m ost im portant param eterization schem es used
for these sim ulations are the radiation schem e and the PBL
schem e. There was no large-scale precipitation and little
m oist convection during the period; therefore, the cloud
and precipitation param eterizations could not be tested.
The radiation schem e used by Seam an et al.7 was a surface
en ergy balan ce com puted with a un iform coolin g rate
throughout the atm osphere. The transfer of radiation be-
tween atm ospheric layers was not represented. This schem e
is sufficient to sim ulate the first-order effects of radiative
transfer, but does not reproduce accurately the details of
atm ospheric heating and cooling. In particular, the cool-
ing of the PBL at night tends to be underestim ated. The
radiative cooling of the m arine boundary layer as it m oves
beyond the San Francisco Bay area toward Sacram ento is
also poorly sim ulated by the m odel. Surface layer tem pera-
tures are 4–6 ºC too high in this area. A m ore com plete
radiation schem e28 is available for MM5, but it m akes the
com putation slightly m ore expensive.
The strength of the wind through the Carquinez Strait
an d Altam on t Pass, east of th e San Fran cisco Bay, is gen -
erally too weak in th e MM5 sim ulation s, because of th e
lack of resolution of th e very sm all-scale topograph y th at
fun n els an d accelerates win d th rough th e pass. Th is lim i-
tation could be overcom e by in creasin g th e resolution of
th e m odel at th e expen se of com putation al resources.
Wh ile th e in ability to sim ulate th e peak win d m ay result
in an in accurate tran sport tim e between th e San Fran -
cisco Bay area an d th e SJV, th e observed h igh velocities
are ch aracteristic of very sm all areas. Th e total flux of m a-
terial th rough th e straigh ts an d passes is probably m ore
accurately sim ulated th an th e m axim um win d. It is th e
win d-in duced stretch in g of th e air parcels an d th e result-
in g m ixin g of pollutan ts th at are n ot well sim ulated by
th e m odel.
Two different PBL form ulations were used in Seam an’s
sim ulation s. Th e first on e was Blackadar’s first-order
sch em e.29 Th e oth er was th e G-S h igh er-order sch em e,3
wh ich can also h an dle fog, in cludin g its radiative effects.
(Fog is n ot im portan t for explain in g th e sum m ertim e O3
form ation , alth ough it would be relevan t to win tertim e
sim ulation s in th e SJV.) Both sch em es rely on th e diffu-
sion equation to exch an ge h eat an d m om en tum between
m odel layers, with sm all differen ces in th e calculation of
th e diffusion coefficien ts. Th e Blackadar sch em e calcu-
lates diffusion coefficients from the therm odynam ic struc-
ture of th e atm osph ere, wh ile th e G-S sch em e relates th em
to th e predicted turbulen t kin etic en ergy. Overall, th e G-S
sch em e perform ed m argin ally better th an th e Blackadar
sch em e, except th at th e tim e of m axim um win d speed in
th e lowest m odel layer was 1–2 h r later th an observed.
Due to the lack of data, the initialization of the ground
tem perature an d m oisture is a problem com m on to all
weath er forecast m odels. Sin ce th e soil h as a large h eat
capacity an d m oisture reten tion capacity, errors in th e
in itial specification (up to 5 layers of soil n eed to be re-
presen ted) can affect th e sim ulation over several days.
Seam an an d Stauffer20 suggest th at an overestim ation of
t h e d eep so il t em p erat u re m ay co n t ribu t e t o t h e
overprediction of surface tem peratures at n igh t in som e
areas. It would be h elpful to m easure th ese quan tities
durin g in ten sive observation periods, but th e large vari-
ability of soil types m akes it difficult to develop a proper
m easurem en t strategy.
Da t a Assim ila t ion
Th e data assim ilation procedure used in th e MM5 sim u-
lation s is n udgin g toward a com bin ation of large-scale
an alysis an d special observation s. Th e Cressm an sch em e
used in th is case is of th e successive correction type, wh ich
is fairly prim itive an d can n ot easily take in to accoun t th e
specific error characteristics of different observing system s.
Seam an an d Stauffer20 argue again st usin g th e m ore so-
ph isticated optim al in terpolation m eth od because of th e
lack of appropriate error covarian ce m atrices. Th ese error
covarian ce m atrices are n ow available,26,27 an d can be used
to in corporate various types of data (such as radioson des,
profilers, or satellite retrievals) with th e proper weigh ts,
usin g optim al in terpolation .
Surface observation s of tem perature were n ot assim i-
lated usin g th e Blackadar plan etary boun dary layer (PBL)
sch em e (see below) because th e diagn osed PBL h eigh t,
wh ich is a critical in put to th e sch em e, is very sen sitive
to th e surface tem perature. Nudgin g toward a h igh er or
lower tem perature can easily result in an un realistic deep-
en in g or collapse of th e PBL. Th is problem could be re-
duced by usin g a sch em e (such as th e Gayn o-Seam an
[G-S] sch em e discussed below) in wh ich th e PBL struc-
ture is less sen sitive to th e surface tem perature. Th e as-
sim ilation of surface tem perature observation s will likely
im prove th e sim ulation s.
Errors in th e PBL h eigh t in th e San Fran cisco Bay area
m ay be due to th e lack of m eteorological data to th e n orth
an d east of th e region . Adequate data coverage exten din g
beyon d th e lim its of th e region of in terest sh ould be con -
sidered in future studies.
Th e vertical an d h orizon tal diffusion sch em es of th e
m odel are im portan t for win ds, h eat, an d water vapor.
The diffusion coefficient depends on the Richardson num -
ber an d is gen erally sm all, but h as a m in im um value to
966 Journal of the Air & Waste Management Association
Volume 50 June 2000

Pun et al.
en sure n um erical stability. Th is m in im um diffusion co-
efficien t n eeds to be carefully set (as low as 0.01 m 2/sec)
to avoid un realistic sm ooth in g of vertical profiles, such
as th e sh arp discon tin uity in th e n octurn al in version . Th e
h orizon tal diffusion sch em e h as little ph ysical basis, be-
sides being of the fourth order and therefore fairly strongly
scale-depen den t. It is used m ain ly for n um erical stability.
We are n ot aware of an y detailed studies of th e effects of
h orizon tal diffusion in MM5.
(output from MM5), an d in itial/boun dary con dition s,
an d sim ulates th e ch em ical an d ph ysical processes rel-
evan t to O3 form ation . Cloud processes are ign ored in
SAQM (th e RADM subm odule was bypassed) because of
th e gen eral lack of clouds (or fogs) durin g O3 episodes.
SAQM, as developed un der SARMAP, does n ot treat aero-
sol form ation , sin ce aerosols were n ot prevalen t in th e
SJV durin g th e sum m er O3 season . Recen t eviden ce13,14
sh ows th at particles m ay affect solar radiation an d in di-
rectly affect O3 production at th e surface. Th erefore, aero-
sol processes an d properties, especially th e feedback
between aerosols an d irradian ce, m ay very well be a n ec-
essary developm en t in SAQM. We discuss below som e
aspects of th e SAQM form ulation th at m ay warran t im -
provem en ts for future application s.
All m eteorological m odels, in cludin g MM5, h ave
m ore difficulty forecastin g calm con dition s th an situa-
tion s wh en th e dyn am ic forcin g is very stron g, durin g
wh ich th e dom in an t term s of th e equation s of m otion
are the advection term s, which can be com puted relatively
accurately. Un der calm con dition s, th e diabatic term s (ra-
diation, turbulence, convection) are dom inant. Since these
term s are sim ulated with param eterization s th at are far
from perfect, m eteorological m odels ten d to perform less
satisfactorily un der calm con dition s. In term s of sim ulat-
in g pollutan t con cen tration s, th e critical quan tity in calm
con dition s is th e evolution of th e h eigh t of th e boun dary
layer. Durin g th e daytim e, th e h eigh t of th e con vective
boun dary layer govern s th e con cen tration s of pollutan ts,
wh ich are n early well m ixed with in it. Un der appropriate
con dition s, cum ulus con vection can develop, ven tin g
boun dary layer pollutan ts in to th e free atm osph ere. Th is
phenom enon was not significant during the episodes stud-
ied. Ven tin g also h appen s at n igh t to som e exten t wh en
th e boun dary layer h eigh t collapses because of coolin g of
th e surface.
Dom a in a n d Resolu t ion
The SJV is affected by O3 and precursors from the San Fran-
cisco Bay area and the Sacram ento area under certain m e-
teorological conditions. In addition, air exchange also takes
place between the southern SJV and the San Luis Obispo
area.4 Polluted air exits th e SJV prim arily th rough th e
Tehachapi Pass. The m odeling dom ain used in the Air Re-
sources Board (ARB) sim ulations32 extends from just north
of th e Sacram en to m etropolitan area to th e Teh ach api
Mountains in the southeast corner, and includes the up-
wind coastal cities on the west. To properly sim ulate all
the advective fluxes, a m odeling dom ain that includes the
entire Sacram ento Valley (up to Redding in the north) and
extends farther east beyond the Tehachapi Pass is desir-
able. Because of the continuous influx from the Pacific,
Dabdub et al.33 found that 4 tim es as m uch NOx originates
from the upwind boundary than within the SJV. Aircraft
m easurem ents have shown that the polluted air m ass can
exten d for several km offsh ore, with an estim ated NOx
boun dary con dition of 3 ppb.33 Exten din g th e western
boun dary to properly sim ulate lan d/sea breezes an d th e
recyclin g of pollutan ts m ay reduce som e of th e in fluen ce
of pollutan ts from th e boun dary. However, th e recom -
m en dation for th e exact location of th e boun dary will
require better ch aracterization of th e cyclin g of pollut-
an ts due to lan d/sea breezes. Clean air boun dary con di-
tion s m ay n ot be valid even wh en lan d/sea breezes are
properly sim ulated, because of in tercon tin en tal tran sport
of O3 an d PAN.
In th e SJV episodes, th e rem n an t of th e polluted up-
per boun dary layer ten ded to be tran sported up-valley by
th e n octurn al jet. In th e n um erical experim en ts of Sea-
m an an d Stauffer, th e boun dary layer h eigh t was som e-
wh at overestim ated by th e m odel because th e upper level
subsiden ce was a little too weak. Th is would ten d to m ix
pollutan ts in a deeper layer th an observed an d would re-
sult in sligh t un derprediction s of con cen tration s in th e
source area, wh ich m ay h ave been coun terbalan ced by
th e larger vertical exten t of th e low level jet due to
sm ooth er m odel profiles. As a result, approxim ately cor-
rect n et tran sport was predicted.
ASSESSMENT OF TH E AIR QUALITY MODEL
Wh ile SAQM is bein g reviewed h ere, m an y of th e com -
m en ts in th is section are gen erally applicable to oth er
state-of-th e-art m odelin g platform s, such as Models-3.
SAQM is a th ree-dim en sion al region al-scale air quality
m odel.31 It is based on th e m odelin g fram ework of th e
Region al Acid Deposition Model (RADM), with th e m ost
fun dam en tal differen ce bein g th e reform ulation of th e
m odel to use n on -h ydrostatic m eteorological data. SAQM
takes as in puts em ission s data, m eteorological data
Bern tsen et al.34 calculated, usin g a global m odel, th at
m ean con tribution s of about 4 ppb O3 an d 26 ppt PAN in
th e n orth western Un ited States can be attributed to Asian
sources. Th e in tercon tin en tal in fluen ce is expected to in -
crease with th e in crease of Asian an th ropogen ic em ission s
in th e n ext decades.35 In ligh t of th e effects of th e upwin d
boundary conditions, offshore m easurem ents are required
to defin e th e upwin d backgroun d.
Volume 50 June 2000
Journal of the Air & Waste Management Association 967

Pun et al.
In previous studies, SAQM was applied with coarser
of plum e dispersion , win d sh ear, plum e overlappin g, an d
effect of turbulen ce on ch em ical reaction rates. Newer
plum e-in -grid m odels42 are available th at overcom e th ese
lim itation s. Such a m odel is bein g in corporated in to th e
Models-3 fram ework an d sh ould be in corporated in to
SAQM to better sim ulate m ajor SO2 an d NOx plum es from
refin eries n ear Bakersfield an d NOx plum es from fossil-
fuel fired power plan ts un til th ey reach a size com m en su-
rate with th at of th e grid cells.
resolution (12 km over th e en tire dom ain except for 4 km
n ested grids in urban areas) th an MM5. As a result, m e-
teorological m odel outputs h ave to be spatially averaged
wh en used as in puts to air quality m odels. Th e process-
in g of m eteorological fields m ay in troduce in con sisten -
cies in th e flow fields, wh ich n ecessitate m ass adjustm en t
in th e air quality m odel in areas with sign ifican t vertical
tran sport.36 For a region al dom ain such as th e Cen tral
Valley, decreasin g th e grid size to be com m en surate with
th e m eteorological m odel over th e en tire dom ain m ay be
in feasible for an air quality sim ulation because of th e as-
sociated com putation al cost. Th erefore, it is im portan t to
en sure th e con sisten cies of flow fields wh en usin g m eteo-
rological m odel outputs to drive air quality sim ulation s.
Despite h igh surface tem peratures in th e sum m er af-
ter n oon , th e m ixin g layer with in th e SJV ten ded to be
fairly low (400–1200 m ). At n igh t, NO surface em ission s
scaven ge O3 n ear th e groun d but are isolated from th e O3
aloft due to th e surface in version . A h igh m odel resolu-
tion (ten s of m eters) n ear th e surface is n ecessary to prop-
erly sim ulate th e processes with in th is surface layer. A
surface layer subm odel (SLS) is in cluded in SAQM to pro-
vide fin er resolution close to th e surface (to m atch th e
resolution of th e lowest level of MM5, with th ree layers
in th e first 60 m ). Un fortun ately, it was foun d in an ap-
plication of SAQM to th e Los An geles Basin th at, wh ile
prediction s of O3 at n igh t are im proved by th e use of SLS,
the perform ance of the m odel during the day and for other
pollutan ts is un satisfactory.37 For exam ple, th e con cen -
tration s of NO an d NO2 were seriously overpredicted dur-
in g th e day in th e bottom layer. A th orough evaluation of
th e SLS is n eeded to iden tify an y possible im plem en ta-
tion flaws an d/or oth er com pen satin g errors in th e m odel.
Transport by the Mean Wind. The num erical treatm ent of
transport is particularly im portant for regional-scale appli-
cations due to low concentrations of chem ical species in
rural areas. At surface wind speeds of < 2 m /sec observed
during the SJVAQS/AUSPEX, it is essential that num erical
errors be m in im ized in th e advection sch em e. Bott’s
schem e, used in SAQM, has undergone rigorous testing and
was shown to treat low concentrations with great accu-
racy.43 It is expected to be less susceptible to num erical dif-
fusion than the Smolarkiewicz scheme it replaced. However,
significant upstream numerical diffusion has been observed
in som e applications of Bott’s schem e.22,44 New advection
schemes may help reduce numerical diffusion problems.45,46
In addition, further testing and developm ent work m ay be
needed for applications involving low wind speeds. Reduc-
ing the grid size m ay be one way to reduce num erical dif-
fusion during stagnation.
Vertical Mixing. Vertical m ixing involves turbulent diffu-
sion , con vective m ixin g, an d relief-driven upslope an d
downslope flows. Accurate representations of m eteorology
(e.g., wind speeds, tem perature profiles, m ixing heights,
and surface heat fluxes) are necessary for the accurate pre-
diction of vertical transport, particularly in locations with
com plex terrain. However, the processing of MM5 m eteo-
rological fields results in m ass consistency problem s in ar-
eas of high vertical transport, as discussed previously.
It seem s in tuitive th at MM5 an d SAQM sh ould use
th e sam e m eth od to calculate th e vertical eddy diffusivity
(Kz) fields used to sim ulate sub-grid-scale turbulen t dis-
persion , sin ce th e turbulen t eddies respon sible for th e dis-
persion of h eat, m om en tum , an d water vapor in MM5
are also respon sible for th e dispersion of pollutan ts in
SAQM. However, as currently form ulated, SAQM estim ates
th e Kz fields in tern ally, an d th ese fields could be differen t
from th ose estim ated by MM5. Th e com putation al sig-
n ifican ce of differen t Kz prediction s in MM5 an d SAQM
m ay be tested by m odifyin g SAQM to receive extern al Kz
values from MM5 sim ulation s.
Tr ea t m en t of P r ocesses
Emissions. Em ission sources h an dled by SAQM in clude
area an d poin t sources. Th e vertical location of poin t
source em ission inputs is determ ined by the effective stack
h eigh t of th e source, estim ated by th e plum e rise m odule
of th e em ission pre-processor. It m ay be preferable to cal-
culate th e plum e rise in tern ally in SAQM to en sure th e
con sisten cy of th e m eteorological fields (e.g., win d speed
at stack h eigh t) used in th e plum e rise calculation an d
th ose used in sim ulatin g oth er processes (e.g., advective
tran sport in th e layer correspon din g to th e plum e h eigh t).
Th e ch em istry of SO2 an d NOx differs significantly in point
source plumes and in the background for area emissions.38-40
To th at effect, SAQM con tain s a sub-grid- scale treatm en t
of plum es from large poin t sources.41 Plum e-in -grid treat-
m en ts h ave been applied to sources in th e Bay Area, an d
th eir effects on O3 prediction s were foun d to be sm all.
However, th e algorith m is un satisfactory in its treatm en t
Gas-Phase Chemistry. SAQM was design ed to be used with
either the Statewide Air Pollution Research Center (SAPRC-
90) m ech an ism 47 or th e Carbon Bon d Mech an ism ,
968 Journal of the Air & Waste Management Association
Volume 50 June 2000

Pun et al.
version 4 (CBM-IV).48 Both SAPRC49 an d th e curren t ver-
sion of CBM-IV h ave been updated with revised isopren e
ch em istry. In ligh t of th e abun dan ce of biogen ic em is-
sion s in th e in ven tory9 an d th e h igh reactivity of th ese
com pounds, the m ore current versions of the m echanism s
will provide m ore accurate description s of th e ch em ical
processes with in th e SJV.
approach 52 un der differen t m eteorological scen arios is
desirable. Oxidant chem istry m ay change from being NOx-
sen sitive in th e sum m er to VOC-sen sitive in th e win ter.53
Pun an d Seign eur54 h ypoth esized th at th e win tertim e for-
m ation of PM NO3– an d HNO3 m ay be lim ited by th e avail-
ability of oxidants, which m ay be VOC-sensitive. A holistic
approach to air pollution con trol (O3 in th e sum m er an d
PM in th e win ter) requires a com preh en sive un derstan d-
in g of th e ch em istry in both season s.
In th e con ceptual m odel, we discussed th e NOx sen -
sitivity of th e O3 system in th e SJV an d th e VOC sen sitiv-
ity in th e San Fran cisco Bay area. Th erefore, it is im portan t
th at th e ch em ical m ech an ism s properly sim ulate both
NOx- an d VOC-sen sitive regim es in term s of O3; precur-
sors VOC an d NOx; oth er products, such as HNO3, PAN,
an d H2O2; an d th e radicals OH, HO2/RO2, an d NO3. Th e
results of th e SAQM/SAPRC an d SAQM/CBM-IV m odels
sh ould be com pared for con sisten cy for both NOx- an d
VOC-sen sitive ch em istry.
Solar radiation m easurem en ts are n eeded at th e sur-
face an d aloft, to evaluate ph otoch em istry as a fun ction
of altitude, an d to in vestigate th e effects of particles, if
an y, on th e production of O3. Dry deposition fluxes of
precursors an d products are n eeded to provide a better
un derstan din g of th e pollutan t rem oval processes with in
th e valley. Measurem en ts of deposition fluxes of NO, NO2,
HNO3, PAN, an d VOC are n eeded over a ran ge of surfaces
an d atm osph eric con dition s.
Deposition. Dry deposition is sim ulated in SAQM usin g
th e resistan ce-in -series approach . Th e dry deposition
m odule in SAQM was reportedly updated32 based on fin d-
in gs from th e Californ ia Ozon e Deposition Experim en t.50
Massm an et al.50 foun d th at RADM, th e precursor of
SAQM, overpredicted th e deposition velocity of O3 for two
types of plan t cover an d in correctly partition ed O3 fluxes
between tran spirin g an d n on -tran spirin g com pon en ts of
the third site. A recent process analysis com parison36 found
th at SAQM predicted h igh er deposition fluxes th an oth er
m odels, wh ich seem s in con sisten t with th e m odification s
reported in DeMassa et al.32 Sin ce dry deposition is po-
ten tially an im portan t m ass flux in th e SJV, especially in
rural areas, an y in accuracy of th e m odel represen tation
n eeds to be un derstood an d rectified.
In term s of data for m odel application an d evalua-
tion , m eteorological data (surface an d aloft win ds, tem -
perature, relative h um idity, m ixin g h eigh t) in an area
exten din g beyon d th e area of in terest, especially in th e
region s wh ere stron g advection m ay be expected, would
be used with FDDA in MM5 to better ch aracterize m eso-
scale win d flow pattern s th at exten d to th e edges of th e
air quality m odeling dom ain (e.g., flow up the Sacram ento
Valley from th e flow divergen ce th at occurs east of th e
Carquin ez Strait). It would also be useful to obtain m ea-
surem en ts of groun d tem perature an d m oisture or sur-
face fluxes at various poin ts th rough out th e dom ain . Note
th at, wh ile th e surface albedo is fairly well kn own from
satellite observation s, th e soil h eat capacity an d m ois-
ture diffusivity, wh ich are im portan t in com putin g sur-
face fluxes, are quite variable from poin t to poin t an d
are poorly kn own . This information could be used in evalu-
ating the causes of surface tem perature overprediction by
the model. Th e radiative effects of aerosols h ave also been
com pletely ign ored in th e m odel sim ulation s, an d could
also affect th e surface an d boun dary layer tem perature
prediction s.
DATA AND MODEL DEVELOP MENT NEEDS
Field Mea su r em en t s
Field data n eeded to support m odel application an d evalu-
ation h ave previously been com piled by Seign eur et al.51
Addition al data n eeded to im prove th e con ceptual m odel
or to evaluate key aspects of th e MM5/SAQM m odelin g
system are presen ted below.
Durin g SJVAQS/AUSPEX, th e NOx an d NO2 m easure-
m en ts suffered from in terferen ce from oxidized n itrogen
species such as PAN an d HNO3. Accurate m easurem en ts
of n itrogen species, especially NOx, NO2, an d PAN, are
im portan t both for th e un derstan din g of th e ch aracteris-
tics of air m asses of differen t ch em ical “age” an d for diag-
n ostic m odel evaluation purposes.
A key kn owledge gap is th e sen sitivities of O3 to VOC
or NOx as a fun ction of location , m eteorology, an d sea-
son . Eviden ce in dicates th at th e San Fran cisco Bay area is
in a VOC-sensitive regim e during sum m ertim e, while m ost
of th e SJV appears to be m ore sen sitive to NOx. Th e sen si-
tivities of O3 to its precursors in th e n orth ern part of th e
SJV m ay alter as a result of in creased tran sport from th e
San Fran cisco Bay area. An alyses are also n eeded for areas
that may experience 8-hr averaged O3 concentrations above
80 ppb. In addition to m easuring the relevant indicator
species (e.g., HNO3 + nitrates, H2O2 + organ ic peroxides,
NOy), a m ore com preh en sive evaluation of th e in dicator
NO3 measurements were taken during SJVAQS/AUSPEX.
The concentrations of NO3 in the SJV were sufficient to oxi-
dize organic compounds at the same rate in the evening as
they were oxidized by OH during the day.55 Typically, VOC
oxidation at night by NO3 is expected to be two orders of
m agn itude slower th an th e daytim e process in itiated
Volume 50 June 2000
Journal of the Air & Waste Management Association 969

Pun et al.
by OH.56 Sm ith et al.’s unexpected conclusion shows that
it is im portant to m easure other radicals in addition to NO3,
especially OH, HO2, and RO2, which have never been m ea-
sured in large-scale field program s in the SJV, and to use
this inform ation in diagnostic m odel evaluation.
In addition to th ese n ew m easurem en ts, we recom -
m en d furth er im provem en ts in th e followin g in term s of
spatial, tem poral, or species resolution :
Resources Board) for providin g valuable in puts to th e
study. Th e m aterial presen ted in th is paper does n ot n ec-
essarily represen t th e views of CEC an d PG&E.
REFERENCES
1. Ran zieri, A.; Th uillier, R. In Planning and Managing Regional Air Qual-
ity: Modeling and Measurement Studies; Solom on , P.A., Ed.; Lewis Pub-
lishers: Chelsea, MI and Pacific Gas and Electric Com pany: San Ram on,
CA, 1994; pp 1-28.
2. Californ ia Air Resources Board. ftp://themis.arb.ca.gov/pub/data/
sarmapo3/datafiles, accessed 1998.
3. Ch ow, J.C.; Watson , J.G.; Lowen th al, D.H.; Egam i, R.T.; Solom on ,
P.A.; Th uillier, R.H.; Maglian o, K.; Ran zieri, A. Atmos. Environ. 1998,
32, 2835-2844.
4. Niccum , E.M.; Leh rm an , D.E.; Kn uth , W.R. J. Appl. Meteorol. 1995,
34, 1834-1847.
5. Rappolt, T.J.; Teusch er, L.H. Iden tification of Region al Air Mass Tran s-
port Usin g Tracer Studies. In Proceedin gs of Regional Photochemical
Measurement and Modeling Studies, Vol. 1, Results an d In terpretation
of Field Measurem en ts; Ran zieri, A.J.; Solom on , P.A., Eds.; A&WMA:
Pittsburgh , PA, 1995; pp 166-180.
6. Sm ith , T.B.; Leh rm an , D.E. Lon g-Ran ge Tracer Studies in th e San
Joaquin Valley. In Proceedings of Regional Photochemical Measurement
and Modeling Studies, Vol. 1, Results an d In terpretation of Field Mea-
surem en ts; Ran zieri, A.J.; Solom on , P.A., Eds.; A&WMA: Pittsburgh ,
PA, 1995; pp 151-165.
•
Th e em ission in ven tories (NOx, VOC, etc) sh ould
be refin ed for biogen ic, m obile (e.g., weekday vs.
weeken d in ven tories, CMB con tribution s vs. in -
ven tories), oil production , an d un in ven toried
sources.
•
•
More aloft con cen tration data (O3, speciated
VOC, NO, NO2, NOy) sh ould be collected to al-
low for m odel evaluation of aloft processes.
Con cen tration data (O3, NOx) are n eeded for
specification of th e boun dary con cen tration s at
th e upwin d boun dary of th e m odelin g dom ain .
(Clean air con cen tration s sh ould n ot be assum ed
at th e western edge of th e dom ain due to th e
recyclin g of th e polluted air m ass an d th e in flu-
en ce of in tercon tin en tal pollutan t tran sport.)
7. Seam an , N.L.; Stauffer, D.R.; Lario-Gibbs, A.M. J. Appl. Meteorol. 1995,
34, 1739-1761.
8. Blu m en th al, D.L.; Lu rm an n , F.W.; Roberts, P.T.; Main , H.H.;
MacDon old, C.P.; Kn uth , W.R.; Niccum , E.M. Th ree-Dim en sion al
Distribution an d Tran sport An alyses for SJVAQS/AUSPEX; Fin al Re-
port STI-91060-1705-FR; San Joaquin Valleywide Air Pollution Study
Agen cy, c/o Californ ia Air Resources Board: Sacram en to, CA, 1997.
9. San Joaquin Valley Air Quality Study Policy Com m ittee. San Joaquin
Valley Air Quality Study Policy—Relevan t Fin din gs; Californ ia Air
Resources Board: Sacram en to, CA, 1996.
10. Maglian o, K. Descriptive An alysis an d Recon ciliation of Em ission s
an d Am bien t Hydrocarbon Data; SJVAQS/AUSPEX Tech n ical Topic
Team #5 Reportin g; Californ ia Air Resources Board: Sacram en to, CA,
1996.
11. Scire, J. A Modelin g Assessm en t of Cum ulative Air Quality Im pacts
of th e Pittsburgh District En ergy Facility an d Oth er In crem en tal
Sources; Report 700-98-006; Californ ia En ergy Com m ission : Sacra-
m en to, CA, 1998.
12. Fujita, E.M.; Watson , J.G.; Ch ow, J.C.; Maglian o, K.L. Atmos. Environ.
1995, 29, 3019-3035.
13. Jacobson , M.Z. J. Geophys Res. 1998, 103, 10593-10604.
14. Dreh er, D.B.; Harley, R.A. J. Air & Waste Manage. Assoc. 1998, 48, 352-
358.
Mod el Develop m en t
Specific areas n eeded for MM5 m odel developm en t in -
clude (1) usin g optim al in terpolation to assim ilate vari-
ous types of data with differen t error ch aracteristics, (2)
reducin g as m uch as possible th e n um erical diffusion by
usin g better algorith m s, an d (3) usin g a full radiation
sch em e an d a PBL sch em e th at is n ot overly sen sitive to
surface tem perature.
15. Blan ch ard, C.L. Application of th e Sm og Production (SP) Algorith m
to Data Collected in th e 1990 San Joaquin Valley Air Quality Study;
Fin al Report; San Joaquin Valleywide Air Pollution Study Agen cy, c/o
Californ ia Air Resources Board: Sacram en to, CA, 1996.
16. Lu, C.-H.; Ch an g, J.S. J. Geophys. Res. 1998, 103, 3453-3462.
17. Martien , P.; Um eda, T. Ph otoch em ical Model Sen sitivity Tests on th e
Effects of Utility Boiler NOx Con trols on Am bien t Ozon e Con cen tra-
tion s; Tech n ical Mem oran dum 93001; Bay Area Air Quality Man age-
m en t District: San Fran cisco, CA, 1993.
18. Altsh uler, S.; Arcado, T.D.; Lawson , D.R. J. Air & Waste Manage. Assoc.
1995, 45, 967-971.
19. Zim an , S. Presen tation at Worksh op on Draft Em ission In ven tory an d
Draft Attain m en t Assessm en t; Bay Area Air Quality Man agem en t Dis-
trict: San Fran cisco, CA, 1998.
20. Seam an , N.L.; Stauffer, D.R. SARMAP Meteorological Model Fin al Re-
port; San Joaquin Valleywide Air Pollution Study Agen cy, c/o Califor-
n ia Air Resources Board: Sacram en to, CA, 1996; p 173.
21. Hegarty, J.; Leidn er, M.; Iacon o, M. In Proceedings of 10th Joint Confer-
ence on the Applications of Air Pollution Meteorology with the A&WMA,
Am erican Meteorological Society, Boston , MA, pp 11-16.
22. Karam ch an dan i, P.; San tos, L.; Sykes, I. SCICHEM: A New Gen era-
tion Plum e-in-Grid Model; Report TR-113097; Electric Power Research
In stitute: Palo Alto, CA, 1999.
Specific areas of SAQM m odel developm en t in clude
(1) in corporatin g th e effect of aerosols on solar radiation
an d ph otolysis reaction s, (2) im plem en tin g an advan ced
sub-grid-scale treatm en t of plum es, (3) evaluatin g th e
perform an ce of th e dry deposition m odule, (4) resolvin g
m ass adjustm en t by m odifyin g th e processin g procedures
of MM5 outputs, (5) im provin g th e surface layer repre-
sen tation , an d (6) updatin g isopren e ch em istry. More re-
cen t m odelin g platform s, such as Models-3, con tain
up-to-date isopren e ch em istry an d m odules to sim ulate
th e effects of aerosols on radiative tran sfer (th ough typi-
cally n ot th e feedback of radiation on aerosol form ation ).
An advan ced sub-grid plum e m odel is bein g developed
with in th e Models-3 platform .
23. Atm osph eric & En viron m en tal Research , In c. Site for daily North east
weath er forecast. http://www.aer.com/forecast.
ACKNOW LEDGMENTS
24. Dudh ia, J. Mon. Wea. Rev. 1993, 121, 1493-1513.
25. Grell, G.A.; Dudh ia, J.; Stauffer, D.R. A Description of th e Fifth -Gen -
eration Pen n State/NCAR Mesoscale Model (MM5); NCAR Tech n ical
Note No. 398+1A, p 107; Nation al Cen ter for Atm osph eric Research :
Boulder, CO, 1993.
26. Neh rkorn , T.; Hoffm an , R.N.; Sparrow, J.; Yin , M. Developm en t an d
Testin g of Th eater An alysis Procedures: Results from Year 2; PL-TR
96-2246; Ph illips Laboratory, Han scom Air Force Base, MA, [NTIS
ADA319121], 1996.
Th is work was supported by Pacific Gas & Electric Com -
pan y (PG&E)/Californ ia En ergy Com m ission (CEC), un -
der PG&E Con tract No. 460000 84 60. Th e auth ors th an k
Gen e Sh elar (PG&E project m an ager) for h is support
th rough out th e study an d Karen Maglian o (Californ ia Air
970 Journal of the Air & Waste Management Association
Volume 50 June 2000

Pun et al.
27. Neh rkorn , T.; Hoffm an , R.N. In Proceedings of 11th Conference on Nu-
merical Weather Prediction, Am erican Meteorological Society, Boston ,
MA, 1996, pp 91-93.
51. Seign eur, C.; Ch in kin , L.R.; Morris, R.E.; Kessler, R.C. In Planning and
Managing Regional Air Quality: Modeling and Measurement Studies;
Solom on , P.A., Ed.; Lewis: Ch elsea, MI an d Pacific Gas & Electric Co.:
San Ram on , CA, 1994; pp 37-78.
28. Dudh ia, J. J. Atmos. Sci. 1989, 46, 3077-3107.
29. Zh an g, D.L.; An th es, R.A. J. Appl. Meteorol. 1982, 21, 1594-1609.
30. Gayn o, G.A. M.S. Th esis, Pen n sylvan ia State Un iversity, 1994.
31. Ch an g, J.S.; Jin , S.; Li, Y.; Beauh arn ois, M.; Lu, C.-H.; Huan g, H.-C.;
Tan rikulu, S.; DaMassa, J. Th e SARMAP Air Quality Model; Fin al Re-
port; Californ ia Air Resources Board: Sacram en to, CA, 1997.
32. DaMassa, J.; Tan rikulu, S.; Maglian o, K.; Ran zieri, A.J.; Niccum , L.
Perform an ce Evaluation of SAQM in Cen tral Californ ia an d Attain -
m en t Dem on stration for th e August 3-6, 1990 Ozon e Episode; Cali-
forn ia Air Resources Board: Sacram en to, CA, 1996.
33. Dabdub, D.; DeHan n , L.L.; Sein feld, J.H. Atmos. Environ. 1999, 33,
2501-2514.
52. Sillm an , S. J. Geophys. Res. 1995, 100, 14175-14188.
53. Jacob, D.J.; Horowitz, L.W.; Mun ger, J.W.; Heikes, B.G.; Dikerson , R.R.;
Artz, R.S.; Keen e, W.C. J. Geophys. Res. 1995, 100, 9315-9324.
54. Pun , B.K.; Seign eur, C. Atmos. Environ. 1999, 33, 4865-4875.
55. Sm ith , N.; Plan e, J.M.C.; Nien , C.-F.; Solom on , P.A. Atmos. Environ.
1995, 29, 2887-2897.
56. Atkin son , R. J. Phys. Chem. Ref. Data 1997, 26, 215-290.
34. Bern tsen , T.K.; Karlsdóttir, S.; Jaffe, D.A. Geophys. Res. Lett. 1999, 26,
2171-2174.
35. Jacob, D.J.; Logan , J.A.; Murti, P.P. Geophys. Res. Lett. 1999, 26, 2175-
2178.
36. Wan g, Z. Person al com m un ication , Un iversity of North Carolin a,
Ch apel Hill, NC, 1998.
37. Pai, P.; Vijayaragh avan , K.; Seign eur, C. In Proceedings of 10th Joint
Conference on the Applications of Air Pollution Meteorology with the
A&WMA, Am erican Meteorological Society, Boston , MA, 1998, pp
510-513
38. Karam ch an dan i, P.; Koo, A.; Seign eur, C. Environ. Sci. Technol. 1998,
32, 1709-1720.
39. Gillan i, N.V.; Meagh er, J.F.; Valen te, R.J.; Im h off, R.E.; Tan n er, R.L.;
Luria, M. J. Geophys. Res. 1998, 103, 22593-22616.
40. Karam ch an dan i, P.; Seign eur, C. J. Air & Waste Manage. Assoc. 1999,
49, PM-175-PM-181.
41. Myers, T.C.; Guth rie, P.D.; Wu, S.-Y. Th e Im plem en tation of a Plum e-
in -Grid Module in th e SARMAP Air Quality Model (SAQM); Fin al
Report; San Joaquin Valleywide Air Pollution Study Agen cy, c/o Cali-
forn ia Air Resources Board: Sacram en to, CA, 1996.
42. Sykes, R.I.; San tos, L.P.; Seign eur, C.; Karam ch an dan i, P. In Proceed-
in gs of 90th An n ual Meetin g & Exh ibition , A&WMA, Pittsburgh , PA,
1997.
43. Bott, A. Mon. Weather Rev. 1989, 117, 1006-1115.
44. Zh an g, Y.; Seign eur, C.; Sein feld, J.H.; Jacobson , M.Z.; Bin kowski, F.S.
Aerosol Sci. Technol., 1999, 31, 487-514.
45. Nguyen , K.; Dabdub, D. Atmos. Environ., subm itted for publication .
46. Ch ock, D. Ford Motor Co., Dearborn , MI. Private com m un ication ,
1998.
47. Carter, W.P.L. Atmos. Environ. 1990, 24A, 481-518.
48. Gery, M.W.; Wh itten , G.Z.; Killus, J.P.; Dodge, M.C. J. Geophys. Res.
1989, 94, 12925-12926.
49. Carter, W.P.L.; Lu, D.; Malkin a, I.L. En viron m en tal Ch am ber Studies
for Developm en t of an Updated Ph otoch em ical Mech an ism for VOC
Reactivity Assessm ent; Report prepared for the California Air Resources
Board, Sacram en to, CA, Con tract 92-345, Coordin atin g Research
Coun cil Atlan ta, GA, Project M-9, an d th e Nation al Ren ewable En -
ergy Laboratory, Golden , CO, Con tract ZF-2-12252-07.
50. Massm an , W.J.; Pederson , J.; Delan y, A.; Gran tz, D.; den Hartog, G.;
Neum an n , H.H.; On cley, S.P.; Pearson , R., Jr.; Sh aw, R.H. J. Geophys.
Res. 1994, 99, 8281-8294.
About the Authors
Dr. Betty Pun (corresponding author) is a staff scientist in
the Air Quality Division of Atmospheric & Environmental
Research, Inc. (AERI), 2682 Bishop Drive, Suite 120, San
Ramon, CA 94583; Dr. Jean-François Louis is principal sci-
entist in the Numerical Weather Prediction Group at AERI,
840 Memorial Drive, Cambridge, MA. Dr. Prasad Pai was a
senior scientist in the Air Quality Division; he is currently a
program manager at Sun Microsystems, 901 San Antonio
Road, Palo Alto, CA. Dr. Christian Seigneur is vice presi-
dent and head of the Air Quality Division of AERI. Mr. Sam
Altshuler is a unit supervisor in Technical and Ecological
Services at Pacific Gas & Electric Company, San Ramon,
CA. Mr. Guido Franco is an associate mechanical engineer
with the California Energy Commission, Sacramento, CA.
Dr. Betty Pun can be reached at phone: (925) 244-7125;
fax: (925) 244-7129; and e-mail: pun@aer.com.
Volume 50 June 2000
Journal of the Air & Waste Management Association 971