United States
                   Environmental Protection
                   Agency
 Atmospheric Research and
 Exposure Assessment Laboratory
 Research Triangle Park, NC 27711
                    Research and Development
^ EPA      Project  Summary
 EPA/600/SR-92/067  September 1992
                   Sensitivity  of  Modeled
                   Ozone  Concentrations  to
                   Uncertainties  in Biogenic
                   Emissions
                   Shawn J. Roselle
                     This study examines the sensitivity
                   of regional ozone (O3) modeling to un-
                   certainties in biogenic emissions esti-
                   mates. The United States Environmen-
                   tal Protection Agency's (U.S. EPA) Re-
                   gional Oxidant Model (ROM) was used
                   to simulate the photochemistry of the
                   northeastern United States for the pe-
                   riod July 2-17, 1988. An operational
                   model evaluation showed that ROM had
                   a tendency to underpredict O3 when
                   observed  concentrations were above
                   70-80 ppb and  to overpredict  O3 when
                   observed values were below this level.
                   On average, the model underpredicted
                   daily maximum O3 by 14 ppb. Spatial
                   patterns of O3, however, were repro-
                   duced favorably by the model.
                     Several  simulations were performed
                   to analyze the  effects of uncertainties
                   in brogejiic^rnjssipns ^>n predicted O3
                   arid to study* the effectiveness of 'two
                   strategies of controlling anthropogenic
                   emissions for  reducing high  O3 con-
                   centrations. Biogenic hydrocarbon
                   emissions were adjusted by a factor of
                   3 to account for the existing range of
                   uncertainty in  these  emissions. The
                   impact of biogenic emission uncertain-
                   ties on O3 predictions depended upon
                   the  availability of NO,. In some  ex-
                   tremely NO,-limited areas,  increasing
                   the amount of biogenic emissions de-
                   creased O3 concentrations.
                     Two control strategies were com-
                   pared in the simulations: (1)  reduced
                   anthropogenic hydrocarbon emissions,
                   and (2) reduced anthropogenic hydro-
                   carbon and NO, emissions. The simu-
                   lations showed that hydrocarbon emis-
sion controls were more beneficial to
the New York City area, but that com-
bined NO, and hydrocarbon controls
were more beneficial to other areas of
the Northeast. Hydrocarbon controls
were more effective as biogenic hydro-
carbon  emissions were  reduced,
whereas combined NO, and hydrocar-
bon controls were more effective as
biogenic hydrocarbon emissions were
increased.
  This Project Summary was developed
by EPA's Atmospheric Research and
Exposure Assessment Laboratory, Re-
search Triangle Park, NC, to announce
key findings of the research project
that is fully documented in a separate
report of the same title  (see Project
Report ordering information at back).

Introduction
 --Ozone (O3)~in "the'troposphere  is pro-"-
duced by photochemical reactions of ni-
trogen oxides (NO,) in the presence of
hydrocarbons. NO, and hydrocarbons are
emitted into  the atmosphere from both
natural (biogenic) and man-made (anthro-
pogenic)  sources. In the United States,
anthropogenic sources contribute most of
the NO,, with fossil fuel combustion ac-
counting for  88% of the total. Biogenic
NO, sources, including soils and lightning
flashes, contribute much  less to the total
NO, budget.  Biogenic sources of  hydro-
carbons include plants, shrubs, and trees
and are on the same order of magnitude
as anthropogenic hydrocarbon emissions.
Because of the increase in anthropogenic
emissions  due to industrialization, urban
O3 concentrations have reached serious
                                                                   Printed on Recycled Paper

-------
levels. Additionally, an increase in rural O3
concentrations was noted from observa-
tional data taken over the past 40 years.
  High concentrations of  O3 present sev-
eral problems to human  health and wel-
fare.  Ozone in high concentrations can
cause respiratory illnesses in humans. It
also deteriorates rubber materials such as
automotive tires, wiper blades, etc., short-
ening the lifetime of these materials. Addi-
tionally, O3 affects the growth  of vegeta-
tion and can damage plants and trees.
With the many problems  of high O3 con-
centrations, a national standard for O3 was
developed to assure that people in the
United States  would  be  protected  from
the dangers of  this pollutant. A 1-hour
average National Ambient Air Quality Stan-
dard  (NAAQS) for O3 was set such that
the concentration would  not exceed  120
parts  per billion  by volume (ppb)  on one
or fewer days per year on average.
  Because O3 is a secondary pollutant,
not directly emitted into  the atmosphere
but rather formed by chemical reactions, it
has been difficult to develop schemes for
reducing its concentration. The  precursors
to Oj formation are NO, and hydrocar-
bons. Therefore, precursor emissions must
be reduced if O3 concentrations are to be
lowered. Because of the presence of bio-
genie emissions of hydrocarbons and NO,,
however, development of a procedure for
reducing O3 becomes complicated  and the
outcome of any emission controls is not
intuitively obvious. Also, the chemistry of
Oa formation is nonlinear. A further com-
plicating factor is that O3 has  a 1-2 day
residence time in the atmosphere, permit-
ting  its transport for distances of up to
1000 km. Thus, an O3  problem  in  one
location may not be caused by  local emis-
sions, but from sources far upwind.
  To help understand this complex prob-
lem,  computer models have been devel-
oped to simulate the physical and chemi-
cal processes of the atmosphere.  A num-
ber of  models  now exist, ranging from
simple box models  to  more complex
Eulerian and Langrangian models. These
computer models  provide valuable tools
for studying the  behavior and structure of
the atmosphere under various conditions
and also provide test-beds for  developing
emission  control strategies  for reducing
O, concentrations.
   A commonly heard phrase in the "mod-
eling world" is that computer models are
only  as good as their input  data. This
statement is true, provided the model is
capable of perfectly  simulating physical
processes. Models are  not perfect and
inherently contain various degrees of un-
certainty In their formulation (e.g. in the
chemical mechanism, treatment of physi-
cal processes, and numerical algorithms).
In most cases, however, uncertainties in
the input data are  more important, espe-
cially when the model is used in control
strategy evaluation. Emissions data,  the
most fundamental  data required for air
quality modeling, are laden with major un-
certainties. It is extremely important that
emissions uncertainties  be minimized in
order to provide reasonable simulations of
the real atmosphere. Development of more
accurate emission  inventories will require
many years of research of improving emis-
sion factor estimates,  source  activity  lev-
els, etc. In the interim, a model's sensitiv-
ity to uncertainties in emissions can be
studied to provide  insight into the reliabil-
ity of model results. This research project
focuses on  uncertainties in the biogenic
emissions database. Uncertainties in the
anthropogenic data are also important, but
are beyond the scope of this study.
  The  purpose of  this study is to further
examine the hypothesis that biogenic emis-
sions have a significant impact on O3 pro-
duction, by analyzing the differences in O3
predictions caused by the range of uncer-
tainty in estimates  of biogenic emissions.
Additionally, the results from this research
should give guidance to air quality plan-
ners on the effects of biogenic emissions
uncertainties and the  role they may play
in control strategy selection. An  opera-
tional model  evaluation  performed in  this
study also provides more  insight into the
performance of EPA's ROM  beyond the
evaluations  that have already been per-
formed.

Procedure
   The ROM, which is capable of simulat-
ing most of the  physical and  chemical
processes responsible for the formation
and transport of O3 on regional scales
(10ekm2), was used to perform simulations
of the  period July 2-17, 1988. Most ofthe
northeastern United States was included
in the  modeling domain. The region was
defined by a curvilinear coordinate system
with a horizontal grid resolution of (1/4)°
longitude by  (1/6)°latitude, or about
18.5km x 18.5km,  with  a  total  of  3,328
(64 columns x 52 rows) model grid cells in
each layer. The planetary boundary layer
and capping inversion (or cloud layer) were
simulated using  three  dynamic  layers,
which  were  free to locally  expand  and
contract in  response to changes in the
physical processes occurring therein.
   Anthropogenic  emissions  used in the
simulations were derived  from the 1985
National  Acid  Precipitation  Assessment
Program  (NAPAP) inventory. Emissions
from anthropogenic sources of volatile or-
ganic compounds  (VOC),  nitrogen oxides
 (NOJ, and carbon  monoxide (CO) were
 provided by the NAPAP inventory. Major
 point, area, and mobile source emissions
 were included in the inventory and varied
 daily (Saturday, Sunday, and weekday)
 and hourly.  Additionally, mobile source
 evaporative emissions were adjusted for
 daily temperature variations using the em-
 pirical model MOBILE3.9. Some station-
 ary  VOC sources were  adjusted for the
 average daily temperature and for activity
 levels typical of high O3 episodes in the
 northeastern United States. The inventory
 also accounts for existing emission con-
 trols.
   Biogenic emissions were obtained from
 the  Biogenic Emissions Inventory System
 (BEIS). BEIS produced estimates of iso-
 prene, a-pinene, monoterpenes,  and uni-
 dentified  hydrocarbons  from  trees  and
 crops, and also estimated NOX emissions
 from grassland  soils.  For  these simula-
 tions,  no NOX emissions from  lightning
 flashes were accounted for by BEIS.
   Nine model simulations were performed
 with the  ROM  using  different levels  of
 emissions. The different levels of anthro-
 pogenic emissions included (1) base case
 emissions, (2)  emissions reduced  by hy-
 drocarbon controls, and (3) emissions re-
 duced by hydrocarbon and NOX  controls.
 Three different  levels of biogenic emis-
 sions were used with  each of these an-
 thropogenic emissions data sets: (1) the
 reference amount of biogenic emissions,
 (2)  biogenic emissions reduced by a fac-
 tor  of 3, and (3)  biogenic emissions in-
 creased by a factor of 3. With the different
 combinations of  emissions  data sets, a
 matrix  of  nine model simulations  was
 formed.

 Results and Discussion
   The  ROM was operationally evaluated
 using  observed hourly O3 concentrations
. taken during the period of simulation^The
 evaluation showed that the model tended
 to underpredict O3 if observed concentra-
 tions  were  above 70-80  ppb  and  to
 overpredict O3 if concentrations were be-
 low this  level.  On  average, the model
 underpredicted daily maximum O3 by 14
 ppb (15%). Spatial patterns of maximum
 O3 were predicted well by the model, even
 though magnitudes  were underpredicted.
 Analysis of several time series, taken from
 different locations within the modeling do-
 main, showed that the model predicted
 diurnal O3 variations better in urban areas
 than in rural areas. Also, the time-series
 analysis  showed  that  the   model
 overpredicted concentrations during  the
 nighttime hours.
   The effect of uncertainties in  biogenic
 emissions on O3 production was studied.

-------
 Decreasing biogenic hydrocarbon  emis-
 sions by a factor of 3 substantially low-
 ered the predicted concentration of reac-
 tive organic gases (ROG) by more than
 30%. in  response to the change in bio-
 genic emissions,  predicted NO, concen-
 trations  increased slightly. Ozone predic-
 tions were lower in most locations with the
 decrease in biogenic hydrocarbon emis-
 sions; episodic maximum O3 decreased
 by 5% to 15% over much of the modeling
 domain.  However, analysis  of  the  rural
 interior  portion of New England, where
 the predicted  chemical system was ex-
 tremely  NO,-limited, showed ah increase
 in O3 as the level of biogenic hydrocarbon
 emissions was decreased. Analysis of daily
 and  hourly data showed that in most ar-
. .e.as_of ,tne,Nortbe,as.t,jie.creasing .bjogenic,..__
 hydrocarbons  caused the smallest pre-
 dicted O3 values to increase  but the larg-
 est predicted values decreased.
   Simulations using biogenic hydrocarbon
 emissions increased by a factor of 3 were
 also examined. In response to the increase
 in  these emissions, modeled ROG con-
 centrations increased by well over 100%
 in  most  of the model domain but NO,
 decreased slightly.  Predicted  episodic
 maximum O3 increased by 15% over many
 areas of the  Northeast.  Increases were
 equally  apparent  in  both the daily and
 hourly data. Some rural locations,  which
 were NO,-limited in the model, were pre-
 dicted to  have lower O3  concentrations
 with  increased emissions of  biogenic hy-
 drocarbons. With  increased  amounts of
 biogenic hydrocarbons,  the  smallest  O3
 predictions were reduced but the largest
 predictions were increased.
   The effectiveness of two simulated con-
 trol strategies  for  reducing O3  was also
 examined; one strategy reduced only an-
 thropogenic hydrocarbon emissions and
 the other reduced both anthropogenic hy-
•'drocarbon and NO~emissionsr Both -strat	-«
 egies resulted in  a reduction in modeled
 O3 concentrations. An  exception  to this
 was  in extremely NO,-limited areas of the
 Northeast where application  of  hydrocar-
 bon  controls  alone caused  a  slight  in-
 crease in O3.  ROG concentrations were
 similar for the two control strategy simula-
 tions, but NO., concentrations were lower
 for the strategy that reduced both hydro-
 carbons and NO,  emissions. Modeled  O3
 was  reduced in NOx-limited areas with the
 application of the  combination of both hy-
 drocarbon and NO, controls. Most of the
 Northeast was predicted to be NO,-limited
 except in the vicinity of major NO, sources,
 such as are found in the New York City
 area. For these NO,-rich  (hydrocarbon-
 limited) areas, the hydrocarbon-only con-
 trol strategy performed better in the simu-
 lations.  The usefulness or effectiveness of
 the combined strategy (hydrocarbon and
 NO,) for reducing O3 predictions increased
 over  NO,-limited areas of the  Northeast
 as the   level  of biogenic emissions in-
 creased. The strategy  of only  reducing
 hydrocarbon emissions, however, was pre-
 dicted to become more effective when bio-
 genic emissions were decreased  in the
 simulations. For the most part, however,
 the combined  strategy of hydrocarbon and
..NO, .controls,  was pxedjctecJLto J3e_mpre_
 effective than  the control of hydrocarbons
 alone, except in the vicinity of New York
 City. This may result in part from the fact
 that the controls are not spatially or tem-
 porally homogeneous.

 Conclusions and
 Recommendations
   The sensitivity of  regional ozone (O3)
 modeling to uncertainties in biogenic emis-
 sions was studied using EPA's ROM. The
 ROM was  used to  simulate the  photo-
 chemistry of the northeastern United States
 for a 15-day period in July 1988. A matrix
 of nine simulations was performed using
 different levels of biogenic and anthropo-
 genic emissions in each simulation. Bio-
 genic emissions were  increased and de-
 creased by a factor of 3 to account  for the
 existing range of uncertainty in these emis-
 sions. Anthropogenic emission  estimates
 were reduced  by two different control strat-
 egies—one using hydrocarbon controls
 and another using both hydrocarbon and
 NO, controls.
   The model results suggest that biogenic
-hydrocarbon-emissions-contribute  signifi-
 cantly to the formation of O3 in the north-
 eastern United States. Uncertainties in bio-
 genic emissions used in the model may
 lead to  significant errors in O3 predictions.
 Because of horizontal transport, the influ-
ence of biogenic  emissions (even  when
 biogenic emissions were reduced by a
 factor of 3) on the New York City area can
 be very  important.
   There appears to be a broad spectrum
 of predicted chemical regimes within the
 Northeast, ranging from the extremes  of
 NO,-limited to  hydrocarbon-limited envi-
 ronments. Based upon modeled  concen-
 trations, most of the Northeast appears to
 be  NO,-limited  except in  the  vicinity of
 major NO, sources, such as are found in
 the New York City area. Reduction of NO,
 emissions appears more beneficial to most
 of the Northeast,  whereas  hydrocarbon
 emission reductions appear  better  in the
 vicinity of New York City (and other hydro-
 carbon-limited areas). Application of one
 strategy type for the whole Northeast may
 not be the most effective means for re-
 ducing high O3 concentrations.  Instead,
 the design of  control strategies for  the
 Northeast should consider the geographic
 distribution of emissions and the chemical
 regime of the  area, whether  it  is NO,-
      ^                      _______
   Because of the model's  tendency "to
 underpredict  higher O3  concentrations,
 analysis  of control strategies  could  not
 fully assess the regional  effectiveness of
 the strategies for reducing high  O3 con-
 centrations.  Also,  the  simulations have
 shown that uncertainties in biogenic emis-
 sions may lead to  errors  in predicting O3
 concentrations and thus could  possibly
 lead one to choose a less effective control
 strategy.
   The following recommendations are sug-
 gested from results of this study: (1)  Un-
 certainties in  biogenic emission estimates
 need to  be reduced to  improve  predic-
 tions of  ozone concentrations.  (2) Field
 validation of rural hydrocarbons and NO,
 is  needed  because of the sensitivity of
 predicted O3  to biogenic emissions and
 background NO, concentrations. Monitor-
 ing programs planned for the  Southeast
 Oxidant Study and for the Lake Michigan
 Ozone Study should help immensely  be-
 cause  of the  existing dearth of NO, and
 rural hydrocarbon measurements. (3) New
 York City and other high  concentration
.-areas should  be- treated as special -cases-
 and examined in further detail with urban
 scale models. (4) Additional modeling stud-
 ies should be conducted to determine the
 relative effectiveness of hydrocarbon ver-
 sus NO, controls. A project recently initi-
 ated by  the  EPA, which uses ROM to
 simulate  the  effects of across-the-board
 hydrocarbon  and  NO,  emissions reduc-
 tions, will provide a systematic approach
 to  determining the  effectiveness of emis-
 sion control strategies.
                                                                                       •U.S. Government Printing Office: 1992 — 648-080/60059

-------
   The EPA author, Shawn J. Roselle, (also the EPA Project Officer, see below) is with
     the Atmospheric Research and Exposure Assessment Laboratory, Research
     Triangle Park, NO 27711.
   The complete report, entitled "Sensitivity of Modeled Ozone Concentrations to
     Uncertainties In Biogenic Emissions," (Order No.  PB92-192202/AS; Cost:
     $35.00; subject to change) will be available only from:
          National Technical Information Service
          5285 Port Royal Road
          Springfield, VA 22161
          Telephone: 703-487-4650
   The EPA Project Officer can be contacted at:
          Atmoshparic Research and Exposure Assessment Laboratory
          U.S. Environmental Protection Agency
          Research Triangel Park, NC 27711
 United States
 Environmental Protection
 Agency
Center for Environmental
Research Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
EPA/600/SR-92/067

-------