Reconciling Urban VOC/NOx (Volatile Organic
Corapounds/NOx) Emission Inventories with
Ambient Concentration Data
(U.S.) Environmental Protection Agency
Research Triangle Park, NC
Jun 87

-------
EFA/60Q/D-87/199
June 1987
RECONCILING URBAN VOC/NOX EMISSION INVENTORIES WITH
AMBIENT CONCENTRATION DATA
by
Jason K.S. Ching, Joan H. Novak, and Kenneth L. Schere
Meteorology and Assessment Division
Atmospheric Sciences Research Laboratory
US Environmental Protection Agency
Research Triangle Park, NC 27711
and
Noor V. Gillani
Department of Mechanical Engineering
Washington University
St. Louis, MO 63130
EPA Project Officer
Jason K. S. Ching
ATMOSPHERIC SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711

-------
TECHNICAL REPORT DATA
(Please read Instruction! on the reverse before completingI
•1. REPORT NO.
EPA/600/D-87/199
2.
3. RECIPIENT S ACCESSION NO.
P287 2 0 2 8 91 /AS
4. TITLE AND SUBTITLE
RECONCILING URBAN V0C/N0X EMISSION INVENTORIES WITH
AMBIENT CONCENTRATION DATA
5. REPORT DATE
June 1987
6. PERFORMING ORGANIZATION CODE
7. AUTMORtS)
J.K.S..Ching, J.H. Novak, K. Schere and
N. V. Gillani
8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND ADDRESS
J.K.S. Ching, J.H. Novak, and K.L. Schere
Meteorology and Assessment Division, ASRL, RTP, NO
N.V. Gillani
Washington State University, St. Louis, MO
10. PROGRAM ELEMENT NO.
N104/C/06-6234 fFY-87)
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Atmospheric Sciences Research Laboratory-RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Rpsparnh Trianale Park. N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/09
IS. SUPPLEMENTARY NOTES
X
\

i
i _
1 . \
,-i/o
/
16. ABSTRACT %
" A review of the current state of emission inventories of/V0C, and vNox^,
data compiled for urban areas 1n the U.S. is presented. The study reveal
great differences in the gross emission magnitudes when compared with corres-
ponding ambient air concentration data. The VOC emissions data are_,in general,
shown to be underestimated by factors exceeding 3 or more for most cities." \
Also, large differences are apparent when different methods of preparation,
are applied to develop the emissions inventory. At this time there is no
acceptable method to check the accuracy of the overall emission output from
these major ares sources# The mass balance approach to determining the urban
area source strength is in principle, able to provide an absolute measure of
such emissions rate. The method is reviewed, and an approach to conducting a
feasibility study is discussed. >The aggregated emission from urban areas is
determined from measurements of excess crosswind and vertically integrated
, pollutant concentration over background in the urban plume just downwind of
the emissions area.;
-------
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse
ment or recommendation for use.
-------
INTRODUCTION
The problems of acidic deposition and photochemical smog, among others,
originate with the emission of the precursor species SOg, N0X, and a myriad of -
volatile organic compounds (VOC). These precursors are chemically transformed
in the atmosphere to acidic products (e.g., sulfuric acid, nitric acid, organic
acids), visibility reducing aerosols (e.g., sulfates), and to such noxious
and harmful products as O3 and PAN (an organic nitrate). Precursor photo-
chemistry provides the important coupling between the phenomena of acidic
deposition and photochemical smog, and is a principal source of non-linearity
in the source-receptor relationships pertaining to these two phenomena.
Regulations to mitigate these regional pollution problems are developed
after analysis of control strategies and atmospheric modeling to estimate the
impacts of the controls. Atmospheric models used in this analysis are region-
al models which simulate atmospheric dispersion, transformation, and deposi-
tion processes. EPA is currently sponsoring two major efforts aimed at devel-
opment and evaluation of two independent and sophisticated Eulerian models:
the Regional Acid Deposition Model (RADM) and the Regional Oxidant Model
(ROM). The validity and utility of these models will be determined criti-
cally, in part, by the reliability of their input emissions data. An accurate
representation of the magnitude and temporal and spatial distribution of the
SO2, N0X, and VOC emitted into the atmosphere is of central importance to
reliable assessment and control of the problems of acidic deposition and
local as well as regional oxidant burdening of the atmosphere. Thus, accurate
emissions data is crucial to the selection of control strategies and regula-
tions which will effectively mitigate pollution problems.
The National Acid Precipitation Assessment Program's (NAPAP) Task Group
on Emissions and Controls has the responsibility for development of anthropo-
genic precursor emission inventories in support of these and other related
research and assessment activities. Traditionally, these anthropogenic inven-
toring has been developed from the U.S. EPA's National Emissions Data Survey.
The accuracy of this database and emissions inventories developed from it is
a topic of much debate. Emissions are typically estimated by multiplying an
activity or fuel consumption level by an average emission factor. The com-
pleteness of this database as well as different assumptions or methodologies
for calculating emissions can have a significant impact on the resulting
emission estimates.
Because inventories are developed from emissions estimates rather than
measurements, there can be no conclusive check on how well the estimated
emissions represent actual emissions. There is, however, suggestive evidence
based on ambient measurements of nitrogen oxides (NOx) and nonmethane organic
compounds (NMOC) in urban areas that existing VOC emission inventories may be
considerably underestimated. This evidence is strengthened by the limitations
of the National Emissions Data Survey. Currently, states are not required to
report emissions from point sources that emit less than 100 tons per year.
In instances were significant quantity of emission may be emitted from smaller
sources and when surrogate emission indicators are difficult to quantify, the
likelihood of underestimation is high. The ambient measurements which suggest
2
-------
the likelihood of problems with the emission inventories are themselves point
measurements and hence not the best surrogates for urban emissions. Thus an
independent means for determining actual emissions is needed to better assess
the accuracy of emission inventories.
As a candidate, the mass balance approach in principle, has the potential
to provide a meaningful and reliable measure of the actual emissions of any
precursor species from an urban area over any given period of time. To date,
utilization and studies to demonstrate feasibility of this approach towards
determination of emissions rate has not been extensive, however the purpose
of this paper is to 1) present emissions and-ambient measurement data which
support the need for an independent measure of emissions and 2) to discuss
pertinent properties and requirements of the mass balance approach as an
independent means of determining aggregated source strengths of N0X and VOC
for an urban area. Part of this overview will explore observational require-
ments for reliable estimates of emissions fluxes with temporal, spatial, and
species resolutions consistent with regional modeling emissions requirements.
In a companion paper1, Rillani et al., 1987, describes the mass balance
method in greater detail and also discusses the existing data hases from which
the mass balance approach will be tested.
EMISSIONS
Recognizing the importance of providing some estimate of the uncertainty
associated with emission estimate the Han-Made Sources Task Group of the Inter-
agency Task Force on Acid Precipitation sponsored a workshop on "Accuracy in
National Scale Air Pollutant Emissions Data Bases"2 in January 1984 which
addressed many of the difficulties surrounding the question of accuracy in
emission estimates. The Workshop viewed accuracy of emissions as a measure
of how close the emissions estimate is to the actual emissions value However,
since actual emissions are not well known, the task of determining accuracy
of the emissions inventory becomes one of estimating the uncertainty surround-
ing inventory emission estimates. This, in turn, involves quantitative
measures of 1) the certainty that the actual emissions value lies between
some defined limits, 2) the internal consistency of the emissions estimates,
and 3) the bias in emission estimates. An assessment of the internal consist-
ency of the emission estimates through statistical methods and the use of
expert teams is discussed in Benkovitz, 19R53. That report presents prelimi-
nary uncertainty estimates for one component or category of factors that must
be evaluated to determine inventory uncertainty for the NAPAP 1980 base year
inventory. These preliminary results indicate' that area source emission
factors and speciation factors for VOC are the components with highest uncer-
tainty, in the range of 100%. Even with improvements in precision of these
components the major problem still remaining is to determine how close the
emissions estimates are to actual emissions. This is the absolute measure of
accuracy of emissions inventories.
The basic information for estimating annual emissions are .taken from
EPA's National Emissions Data System (NEDS). Primarily the data consist of
statistically averaged emission factors and fuel consumption estimates which
are used to calculate annual emissions of the five criteria pollutants (TSP,
S0X, N0X, VOC and CO). Point source emission estimates are based on individual
sources or processes whereas area source emissions are based on county level
3
-------
fuel consumption or activity estimates. Such statistically averaged para-
meters may not adequately represent the full population of sources. Such
inaccuracies in the basic data probably constitute an important source of
error in the emission inventories even though the precision of these estimates
have been significantly improved in consecutive NAPAP updates. Further uncer- -
tainties of the NAPAP inventories used in regional scale models result from the
application of spatial, temporal, and species allocation algorithms to annual
county and point emissions. For example, countywide area source emissions
are allocated to a subcounty grid scale (1/4° longitude by 1/6° latitude);
annual emissions estimates are allocated to seasonal, daily and hourly resolu-
tion; and total VOC emissions are broken into about a dozen VOC classes
specifically corresponding to model chemistry mechanism requirements. Also,
several known VOC source categories which were not included in the 1980 emis-
sions are now scheduled for inclusion in the NAPAP 1985 inventory. These
include: publically owned treatment works (POTW's), hazardous waste treatment,
storage and disposal facilities (TSDF's), fugitive emissions from synthetic
organic chemical manufacturing and petroleum refinery operations, bulk ter-
minal and bulk plants, pharmaceutical and synthetic fiher manufacturing* etc.
The combined uncertainty from all of these factors is probably quite large, hut
not quantifiable.
A review of emission inventory development efforts during the past few
years provides a startling awareness of the impact of emission inventory
updates on the gridded emission estimates used in Eulerian regional scale
model development and evaluation activities. To date these emission inven-
tory efforts have primarily focused on providing a 1980 base year inventory.
Table I summarizes some of the methodology updates over a four year period.
This summary and related emission comparisons shown in Figures 1 thru 4
provide some indication of the impact of methodology changes on successive
estimates of the 1980 base year emissions.
The Northeast Corridor Regional Modeling Project (NECRMP) was one of the
earliest attempts to provide detailed ozone precursor (VOC, N0X) emissions
required by regional models using nonlinear chemistry mechanisms. Figures 1
and 2 compare 200 grid cells with highest emissions (point and area sources
combined) of N0X and VOC respectively in the Northeastern U.S. The regression
lines in both bases show that the NAPAP 4.2 inventory updates have effectively
reduced the estimated source strengths of grids that were strongest in the
NECRMP data set and raised the estimated strengths of weaker grids. These
differences may be attributed to spatial shifts due to use of different
census data for spatial allocation or to emission factor differences introduc-
ed by the AP-42 emission factor supplement updates. For the most part, the
NAPAP values are within a factor of 3 of the corresponding NECRMP estimates.
The uncertainties are largest, however, in the N0X data where discrepancies
as large as a factor of 100 occur. Figure 3 compares two successive versions
of the NAPAP NO inventory. The bulk of the data shows high correlation; no
bias appears, but the magnitude of the outliers raises concern about the
accuracy of current emission methodologies. Since motor vehicles typically
contribute 50% of the N0X emissions the update from Mobile 2 emission factors
to Mobile 3 emission factors might possibly explain the observed differences.
Finally Figure 4 compares the NAPAP 4.2 and 5.3 grid emissions of sulfur diox-
ide (SOj). SO2 emissions are thought to be most accurately estimated because
4
-------
TABLE I Proposed uncertainties for the NAPAP 1980 base year inventory
from: Benkovitz (1985)
yncerta-jnty Range (f)
Across all
Parameter	S0X N0X VOC Pollutants-
Fuel S content	10
Point source emission factors
Method 1 (stack tests)
25
• 25
25
2 (material balance)
. 10
25
50
3 (AP-42 emission factors)



confidence rating A
10
10
10
R .
25
25
25
C
50
50
50
0
75
75
75
E
100
100
100
5 (state emission factor)
25
50
50
4,6,7,0 and blank
100
100
100
Point source production throughput
15
15
15
Control equipment efficiency
25
25
25
Area source emission factors



Mobile sources
*
50
100
"Other" sources
25
25
100
Area Source activity level



Mobi1e



Other



Point source temporal apportionment



Seasonal profiles
10
10
10
Daily profiles
25
25
25
Hourly profiles
50
' 50
75
Area sources temporal apportionment



Seasonal profiles
10
10
10
Daily profiles
25
25
25
Hourly profiles
50
50
75
Area sources spatial apportionment factors
25
25
25
Chemical speciation factors



N0/N02



VOC



~Apply same criteria as for point source emission factors.
5
-------
X
o
o
O
N
NAPAP 4.2 vs mem*
LOG NOX
5-
3X .X
' a
- X
¦ ¦ ¦ .,**
~ ¦ *J? A# ¦
. /
¦ ¦ * ¦ /
¦ nA-rr-i- ¦ ,/
/
X/3
REGRESSION
"I1 11 	 1	 1 	1 1 I " "T	"""I	 " 'I	
NECRMP (LAG NOX)
Figure 1. Comparison of NAPAP 4.2 and NECRMP gridded NOX emissions for the
200 strongest source celia In the Northeastern U. S. ROM domain.
NAPAP 4.2 va msm
LOG VOC
3X.
w
i"y ««»
a
>
o
7-
3
8-
X/3
NECRMP (LOG VOC)
Figure 2. Comparison of NAPAP 4.2 and NECRMP gridded VOC emissions for the
200 strongest source cells Sn the Northeastern U. S. ROM domain.
6
-------
o
z
o
Q
CM
Q_
t
7-r-
NAPAP 6.3 vs NAPAP 4.2
LOG NO
/

|K	*
"•J*' V
" \
KEBRESSJOk
—t--—I	[	1	1	,	j	p-
5	6	7
NAPAP 5.3 (LOG NO)
Figure 3. Comparison of NAPAP 5.3 and NAPAP 4.2 gridded NO emissions for the
200 strongest source cells In the Northeastern U.S. ROM domain.
NAPAP 6.3 vs NAPAP 4.2
LOG S02
CM
o
tn
o
Q
EM
•4
NAPAP 5.3 (LOG S02)
Figure 4. Comparison of NAPAP 5.3 and NAPAP 4.2 gridded S02 emissions for the
200 strongest source cells In the Northeastern U.S. ROM domain.
7
-------
the bulk of emissions are generated hy utilities and industrial combustion
sources whose emissions factors have been more accurately determined. As
illustrated, however, even gridded SOg emissions exhibit significant dif-
ferences in the latest NAPAP update. The Eulerian models primarily use emis-
sions in gridded form. While the bias is small, the "noise" level between
successive estimates of gridded 1980 base year emissions is a factor of 10.
This brings into question the significance of the model calculations them-
selves because the response of O3 concentrations to changes in VOC and NO x
emissions is a function of the base emission levels. Therefore, if the
estimated base emissions are in error, the simulated response of O3 concen-
trations to emissions controls could also be in error.
AMBIENT MEASUREMENTS
There is suggestive evidence based on ambient measurements in urban
areas that current emissions inventories, especially of VOC, may be signifi-
cantly underestimated by as much as a factor.of 5 for some urban areas. For
example, in a study in Atlanta in July 1981 the mean value of the NM0C/N0x
ratio based on ambient measurements in the city was 8.fi compared to the value
of 1.5 for the same ratio based on the emissions inventory values of NMOC and
N0X (ratios are in molar units of carhon and N0X). Application of a simple
box model using the emissions data yielded good agreement between the predicted
and observed N0X concentrations but the predicted anthropogenic VOC concen-
tration (99 ppbC) was about five tirps less than the observed concentration
(491 ppbC). In another recent study , similar N!10C/N0X ratios were computed
for 22 cities in the eastern and central U.S. from 1375 samples of ambient NMOC
and N0X measurements taken specifically in central business districts. The
NMOC measurements were based on analyses of the ambient air samples by two
independent state-of-the-art techniques. The N0X data are also considered to
be of state-of-the-art accuracy. The results of the ratio NM0C/N0x for the 22
cities are summarized in Table II. The median values for the 22 cities range
between 8.3 and 50.0 (ppmC/ppm), with a median of medians values for all
cities of 13,9. For comparison Table III shows the values of the NM0C/N0x
ratio for the five eastern U.S. urban counties with the highest emission flux
of NMOC, and five counties with the highest emission flux of N0X. These
ratios are based on gridded precursor emission values extracted from the NAPAP
4.2 inventory and range between 1.49 and 8.25 with the means and medians of
the two sets ranging between 3 and 5. These ratios are about 3 to 5 times
smaller than 13.9, indicating once again that the VOC emissions in the inven-
tories for urban areas may be underestimated.
The Mass Balance Method for Estimating Urban Emissions
The mass balance method has been used in past studies^ to estimate
changes in the mass flow rate of given pollutants between two crosswind
sections of a pollutant plume as a result of plume kinetics (chemistry and dry
deposition). In these applications the contribution of pollutant mass by
primary emissions was either assumed known or was not a factor (e.g., contrast-
ing two cross-sections, both downwind of the source region). In principle,
by comparing the mass inflow rate of a given pollutant into an urban source
area with the corresponding mass outflow rate just downwind of the source, an
estimate can be made of the source contribution (i.e., the primary emission)
of the pollutant. The direct difference between these two mass flow rates
8
-------
Table II. Median ambient urban NMOC/NOx concentration ratios during 1984.
EPA Region	Site	Ratio (No. of data)
III
Philadelphia
19.6
(45)

Wilkes Barre
14.3
(53)

Richmond
10,5
(62)

Washington
9.4
(54)
IV
Memphis
13.9
(35)

Chattanooga
16.8
(37)

Charlotte
10.4
(55)

Bi rmingham
11.7
(51)

Atlanta
. 10.5
(52)

Miami
13.3
(15)

W. Palm Beach
14.3
(60)
V
Akron
12.8
(49)

Cincinnati
9.1
(51)

Indianapolis
8.3
(50)
VI
Beaumont
25.3
(45)

Clute
23.7
(52)

Dallas
16.0
(69)

El Paso
15.3
(60)

Fort Worth
11.6
(58)

W. Orange
50.0
(41)

Texas City
37.7
(52)
VII
Kansas City
-

Median of Medians	13.9 (21)
* In the Central Business District
Table III. NM0C/N0x emissions ratios of urban counties (based on NAPAP
4,2 emissions inventory).
Urban Counties with highest	Urban Counties with highest
VOC Emissions Flux:	N0X Emissions Flux:
County
Ratio
County
Ratio
New York, N.Y.
3.15
New York, N.Y.
3.15
Oueens, N.Y.
3.93
Hudson, N.J.
1.49
Baltimore City MD
8.25
Oueens, N.Y.
3.93
Philadelphia, PA.
3.70
Union, N.J.
.2.91
Richmond, N.Y.
4.97'
Philadelphia, PA.
3.70
Mean
4.80
Mean
3.04
Median
3.93
Medi an
3.15
9
-------
is, in fact, equal to the species emission rate if the meteorology is in
steady state (e.g., constant wind field, mixing height, etc.), and if the
pollutant is a conservative species (chemically inert), and nondepositing.
In a dynamic and non-uniform meteorological environment, the change in the
pollutant mass flow rate between urban upwind and downwind sections will not -
only be due to urban primary emissions but also to mass exchange at the top
and sides of the urban plume, deposition losses at the surface, and chemical
transformations within the plume. In such cases, the primary emission contri-
bution may be isolated by separately estimating the contribution of these
secondary effects through the use of an appropriate transport/transformation/
removal model. It is important, however, to recognize that the estimation of
these secondary effects must necessarily be based on some assumed primary
emission input to the model. Such primary emission, input may be based either
on the emission inventory (catch 22!), or may be inferred iteratively in the
model application to match the sum of the observed change in the upwind/
downwind mass flow rates and the net sink due to all secondary effects. It
is advisable to select situations for which the magnitude of the secondary
effects is negligible (ideally) or, secondary effects is negligible (ideally)
or, at least, small so that the difference between the inflow/outflow mass
accounts directly for most of the emission mass.
Irrespective of the magnitude of the secondary effects and the need for
the model application, there is a fundamental need in the implementation of
the mass balance method to estimate the mass flow rates across one or more
crosswind vertical planes. The method for doing this is illustrated below
for the case of a conservative species in a uniform, steady meteorological
environment. Subsequently, the significance of non-simple meteorology and
non-conservative species kinetics will be considered. This illustration is
focused on the urban source scenario. The discussion below is based on White
et al., 1983. Consider the following scenario for a flow leaving an urban
source complex (Figure 5). The local mean flow field is assumed to be spati-
ally uniform, both horizontally and vertically, and locally steady. The
urban plume spreads vertically within a mixed layer of" uniform and steady
"background" chemical composition. (This will not be the case when the back-
ground is chemically changing, or being diluted by entrainment of chemically
different air from aloft as when the mixing layer is growing in the mid-morn-
ing period, or when the background is being modified by removal processes.)
In Figure 5 the primary mass emission rate (Qi) of any arbitrary species
i and the wind speed (U) are shown as functions of time merely to reflect that
their values may vary diurnally and seasonally. They are, however, assumed
to be temporally constant for the duration of a specific mass balance experi-
ment (for which 0 can be considered quasi-stationary).
The following expression describes the primary emission rate in this
idealized scenario:
*	CO CO
Oi « TT J J (C7 - C1)bg)dydz	(1)
	o
where and C-j,hg are ^he total and background concentrations respec-
tively. This approximation applies to any source group in which the
wind speed U over the time of sampled emission is constant. White et
10
-------
URBAN PLUME
URBAN SOURCE
Qi (t)
U(t)
Figure 5. Surface projection of uniform horizontal flow
uniform
horizontal
of
influenced by an urban source
al., 1983, refers to the quantify on the right hand side of (1) as the
virtual emission rate, Qi* of species i. It represents the fraction of
emission of species, i, which would be determined at downwind distance, x,
after the emissions have undergone deposition, chemical transformation, and
dilution. Therefore, for SO2 and other primary emissions, Qi* will be less
than the actual emission rate Qi, while for secondary product species such as
O3, Qi* will be greater than their primary emission rate (zero for O3).
The difference between Q^ and Q^*(x) must represent the net effect of all
intervening sink processes affecting the concentration from the source region
to x. Thus,
Qsi(x) = Qi - Q*(x)	(2)
and for a conservative species, Qi* is not a function of x and thus Qi = Q^*.
Note that Q^ is the primary emission rate aggregated for the urban area.
Together, Qi and Qsi can provide an estimate of Qj which must then be used
for comparison with Qeii, the emission rate according to the emission inven-
tory.
While this approach is promising, it must be demonstrated that Qi* and
QSi are measurable or semi empirically determined quantities. Figure 0 is a
schematic which defines the parameters of the mass balance approach for a
primary pollutant specie. The current emission inventory estimate is Q£i and
is qualitatively indicated as small compared to Qi. The goal of the approach
is the determination of Qi, the aggregate anthropogenic emission of arbitrary
species i from the urban source region. (Hereafter, the subscript i will be
dropped for convenience.)
11
-------
There are obviously differences in the magnitude of 0 for different
pollutant compounds, depending on their reactivity, surface uptake and magni-
tude of the background values. Confidence in the use of the mass balance
approach will he greatest for the situation in which 0S is small relative to
0. However, 0S is difficult to determine directly and will typically require
the use of theoretical or modeling approaches. Therefore modeling "noise",
DQS, will necessarily he introduced and it should be clear that due to these
model uncertainties, the smaller the relative magnitude of 0S relative to 0,
the more "exact" will be the estimated emissions from the mass budget approach.
The modeling uncertainties derive from the obvious contributions of-measure-
ment "noise" as well as a uncertainties in the chemical modeling due to the
requirement to use highly condensed chemical mechanisms and approximate rate
constants for the complex ambient air mixtures. Clearly, the mass balance
approach achieves greatest potential accuracy when the relative magnitude of
DQS is small compared to 0S, or what is better, when the magnitude of 0 is
closest in value to Q. Field measurements to determine Q* will contain the
usual measurement errors, DQ*, due to sampling errors arising from subgrid
scale variations, instrument response, etc. The mass balance technique will
require such errors to be much less than 0*.
Figure 7 illustrates another property of the mass budget approach to be
considered. The relative magnitude of Q to 0S will decrease with increasing
distance from the downwind edge of the urban source region. This is simply
because the role of removal, transformation, and dispersion will increase
with x, and since 0 is fixed, 0 must decrease. Additionally, D0S will in-
crease commensurately and the resulting confidence in the mass balance tech-
nique (Eq. 2) will diminish. There is, of course, the need to measure Q * far
enough downwind of the emissions area so that the pollutants can disperse
vertically to heights for which sampling, if performed hy aircraft, is pos-
sible. The selection of optimum sampling distance is therefore an experiment
design problem. It is quite clear that not all meteorological conditions
will be suitable for mass budget studies. Sampling platforms and sampling
methods have limitations which must be recognized and, moreover, the sampling
strategy does require a finite time period to complete. Remote sensing
systems such as lidars, sodars, cospecs, etc., provide current technical
capabilities in obtaining requisite measurements of key parameters. For
example, airborne lidar systems are available to provide real time information
on the width and composition of the urban plume, and with range resolution
can be used to provide strategic operational information to the other sampl-
ing aircraft platforms making in situ measurements. Doppler sodar systems
provide detailed wind measurements for determining the air flow fields. As a
practical measure, it would be necessary to be selective of meteorological
conditions for which relatively steady conditions prevail, for which the mixed
layer is and remains fairly shallow over the duration of the experiment, and
for which fairly sizeable emissions occur. Further, the wetness of the sur-
face between the sources and x will complicate the parameter relationship of
models which governs the quantification of the losses of some pollutants at
the surface by deposition processes. Overcast conditions are desireable to
reduce the amount of solar radiation and therefore reduce the amount of photo-
chemical conversion of the primary emissions. It may be possible to use the
mass balance approach to determine the emissions for the afternoon commuter
traffic period. In this situation, the outflow from the urban area may enter
12
-------
0
EI
existing
Inventory
Q (actual)
Q.
cr
DO.
DQ*
f
Figure 6. Parameters of the mass balance approach
1
Figure 7. Downwind variation of mass flow parameters
for a primary pollutant.
13
-------
into a mixed layer whose height is now considerably shallower than during the
afternoon (winter/fall condition), and moreover, sunlight intensity is strong-
ly reduced dropping to zero at sunset. Operational difficulties at dusk will
be more difficult of course. Accurate and detailed measurements to character-
ize the meteorology and pollutant fields over the temporal and spatial domain -
of the study will be required, and obviously, suitable sampling instrumenta-
tions and mobile (surface and airborne) platforms must be properly deployed.
Instrumented helicopters have been utilized in previous studies over urban
areas that have permitted nearsurface pollutant characterization. Obviously,
the list of observational degrees of freedom are large and need to be carefully
limited and optimized.
The mass balance approach can be complimented by Lagrangian studies of
tagged air parcels into which a specific source type of emissions has been
injected. The detailed characterization of and changes to the pollutants in
this air parcel during their transit to the sampling plane from specific
sub-urban scale sources provide useful diagnostic information on the magnitude
of the 0S term and on the accuracy of the model determination of the Qs and
00s terms. These special emissions areas might include major roadway inter-
changes located far upwind of the downwind sampling plane, or for major
industrial non-mobile source complexes, etc. These studies will provide
information on changes in chemical composition and the magnitude of the
pollutant mass gains/losses. It will provide the means to separate those
emissions of mobile origin from industrial and other sources.
Tracers and parcel markers are currently available which provide the
means to conduct these Lagrangian experiments. Methods which utilize pollu-
tant -to-tracer ratio techniques and fingerprinting of specific source types
could be incorporated into the experimental design where possible to provide
information on emission rates from sub-urban source regions for which inven-
tory uncertainties might be expected to be large. The ratio of various species
of VOCs from mobile sources to the acetylene content appears to be a promising
method of mobile source fingerprinting which provides interpretive aid to
these diagnostic studies8.
The magnitude of the emissions will be variable in time; point and area
sources of N0X and VOC will vary diurnally. Figure 8 describes the temporal
allocation factors for N0X and VOC; clearly, the magnitude of the diurnal
variation is large. Ohviously, the consideration of the diurnal variation
is a design requirement of the mass balance approach. The straightforward
solution is, of course, the mass balance determination of 0 for different
times of the daily cycle. Clearly, characterizing the diurnal' variation is a
major aspect of the feasibility study. An initial step in judging feasibility
is the use of existing data to determine emissions using the mass flow rate
technique. In consideration of the major expense of and the operational
difficulties to obtain reliable measurements required to make accurate esti-
mates of emission rates, it is prudent to conduct pilot studies using both
existing data bases, and current models to examine in sufficient detail, the
requirements for mass budget studies. This is discussed in greater detail by
Gillani et al1. It should be noted though that none of the existing ambient
data bases were collected with the objective of reconciling source emissions
with ambient data and therefore may be incomplete for this purpose.
14
-------
NOX EMISSIONS FOR U. S.
50
/—N
«
¦o
c
o
«
3
O
JC
&
40-
POINT SOURCES
I
™ 10 "
D>
AREA SOURCES
12
20
24
HOUR (GMT)
Figure 8. Diurnal variation of NOX from NAPAP 4,2 emissions inventory.
SUMMARY AND CONCLUSIONS
Improvements are continuously being made to emission inventories, both
raw data corrections/updates and methodology improvements. Since these
improvements have significant impact on emissions inputs to Eulerian regional
scale models there is a real need to determine the accuracy of the emission
inventories as well as the uncertainty of various components used in estimat-
ing emissions. In addition, ambient measurement data indicate a potential
underestimation of VOC emissions, however, these point measurements are not
necessarily representative of larger scale emissions.
This paper has proposed a mass balance approach as a potential means of
obtaining an independent estimate of area wide VOC and N0X emissions which
can be used to determine some level of accuracy of existing emission estimates
in urban areas. This ambient measurement technique uses measured air concen-
tration data downwind and downwind of an urban area to estimate emissions for
the area over a specified unit of time. The net effect of all processes influ-
encing the concentration from the emitting sources to the point of downwind
15
-------
measurement must be accounted for and underlying uncertainties quantified.
The feasibility of this approach is currently being explored by analyzing
existing ambient measurement data bases.
We conclude that confidence in the accuracy of current emissions inven-
tory estimates are low. Certainly, the bias and the "noise" arising from
both inadequacies in the emission input data as well as in the differences in
methodologies adopted to develop the emission inventories is very large.
Such inaccuracies, when used in urban or regional scale models, will lead to
erroneous and inconclusive results. In fact, due to the great uncertainty in
the emission inventory accuracy, the modeling results might well lead to air
pollution control strategies that are either ineffective, or, even worse,
wrong. The mass balance technique offers one possible approach to determining
the accuracy of emission estimation methodologies for urban areas. The
problem is significant enough to emphasize the need for more research to
determine an effective approach for quantifying accuracy of emission inven-
tories.
References
1.	N. V. Gillani, W. H. White and J. K. S". Ching: A semi-empirical mass
balance approach for estimating primary emissions of reactive species for
an urban-industrial complex. Preprint Volume, 80th Annual APCA Conference,
June 1987, New York City, NY.
2.	Record of A Workshop on Accuracy in National-Scale Air Pollutant Emissions
Data Basis! Prepared by the MITRE Corporation. January lO-lI^ 1984.
3.	C. M. Renkowitz, 1985: Uncertainty analysis of NAPAP emissions inven-
tory. Preprint volume, Second Annual Acid Deposition Emissions Inventory
Symposium. Charleston, SC. Nov. 12-14.
4.	H. Restberg and B. Lamb, 1985: Ozone production and transport in the
Atlanta Georgia region. EPA Project Report, Grant CR 809221, ASRL, RTP,
NC.
5.	H. G. Richter, F. F. McElroy and V. L. Thomson, 1985: Measurement of
ambient NMOC concentrations in 22 cities during 1984. Paper 85-22.7,
Preprint volume, 78th Annual Meeting Air Pollution Control Assoc., Detroit
Mich, June 16-21.
6.	R. R. Husar, J. D. Husar, N. V. Gillani, S. B. Fuller, W. H. White, J-. A.
Anderson, W. M. Vaughan and W. E. Wilson Jr., 1976: Pollutant flow rate
measurements in large plumes: Sulfur budget in power plant and area
source plumes in the St. Louis region. Paper presented at Div. of Environ
mental Chemistry. American Chemical Society Annual Meeting, New York.
7.	W. H. White, 0. E. Patterson and W. E. Wilson, Jr., 1983: Urban export
to the nonurban troposphere: Results from Project MISTT. .Journal of
Geophys. Research, 88(C15):10745-10752.
8.	W. A. Lonneman, R.L. Seila and S.A. Meeks: Monmethane organic composi-
tion in the Lincoln Tunnel. Unpublished manuscript.
16
-------