Document Sample
v3 Powered By Docstoc
					Final Report, Volume III: Summaries of Individual
               AQAWG Analyses
     OTAG Air Quality Analysis Workgroup
          Dave Guinnup and Bob Collom, Co-chair

                      May 26, 1997
                                     Table of Contents

SUMMARIES OF INDIVIDUAL AQAWG ANALYSES                                                                  1
  Spatial Pattern of Daily Maximum Ozone Over the OTAG Region                                            2
  Pattern of 8-Hour Daily Maximum Ozone Over the OTAG Domain                                             4
  Weekly Pattern of Ozone over the OTAG Region                                                           6
  Spectral Decomposition of Ozone Time Series                                                            8
  Ozone Exceedances Data Analysis: Representativeness of 1995                                           12
  Ozone Measurement Trend Studies in the Northeast                                                      14
  SOS Nashville/Middle Tennessee Ozone Study                                                            17
  Air Trajectory Analysis of Long-Term Ozone Climatology                                                19
  Source Regions of Influence for High and Low Ozone Conditions in the Eastern US                       22
  Ozone/Tracer/NOY Relationships at Three SOS-SCION Sites                                               24
  Analysis of Ozone, NOY and Tracer Data from a Site in South-Central Pennsylvania                      26
  A Comparison of Modeled and Measured Ozone, NOY and CO at Nine Regional Monitoring Stations during the
  1995 OTAG Episode                                                                                     28
  Ambient Monitoring Sites for OTAG (time series) Model Evaluation                                      30
  Comparison of OTAG UAM-V/BEIS2 Modeling Results with Ambient Isoprene Observations                    31
  Comparison of SOS Nashville Data to OTAG 1995 Base Model                                              36

Cover Image: a) Frequency of back trajectories for 22 receptor sites for low ozone days. These
back trajectories point to outside the OTAG domain as the source of low ozone. b) Frequency of
back trajectories for high ozone days (upper 50th percentiles). These back trajectories point to the
OTAG domain as the source of high ozone.

Summaries of Individual AQAWG Analyses                                                                  ii

A brief summary and critical review was          discussions of each study‟s limitations and the
prepared for most of the AQAWG analyses          scientific and policy implications for the
identified in Table 1 in Volume II: Summary      issues being addressed by the OTAG. The
and Integration of Results. These summaries      summaries of findings, limitations, and
provide a description of the purpose,            implications were used as the basis for the
methodology, and major findings of each          integrated assessment presented above.
study and are intended as a resource for quick   Summaries are presented on the following
reference in lieu of turning to the full study   pages in the order in which projects are listed
report(s). These summaries also include brief    in Table 1.

Summaries of Individual AQAWG Analyses                                                        1
Spatial Pattern of Daily Maximum Ozone Over the OTAG Region

Participants: Rudy Husar, CAPITA, Washington University, St. Louis, MO
Reference: Spatial Pattern of Daily Maximum Ozone Over the OTAG Region, Rudy Husar 16
September 1996 (

Purpose: Identify spatial patterns in daily       in the corners of the domain.
maximum 1-hour ozone concentrations over          2. Mean daily maximum concentrations are
the OTAG domain.                                  highest in the Washington-New York urban
Method: A five year (1991-1995) data set of       corridor, with elevated concentration “hot
June-August daily maximum 1-hour ozone            spots” also observed in the vicinity of several
concentrations for the eastern U.S. was           other (but, surprisingly, not all) major urban
obtained by combining data from AIRS,             centers.
CASTNET, and several other monitoring             3. A large region of elevated (65 - 70 ppb)
networks. All sites with at least 25 percent      mean daily maximum concentrations is
complete data were included in the analysis       evident along a band stretching across
(thus, some of the sites used may only have       southern Indiana and Ohio, just north of the
data for less than a single season) . The final   Ohio River.
data consisted of a total of 715 monitoring
sites. Six of these sites were discarded as a     4. East-west sections across the center of the
result of quality control procedures. Average     domain show concentrations increasing
concentrations and various percentiles of the     consistently from west to east.
concentration distribution were calculated for    5. In the industrial Midwest, from Missouri
each monitoring site and contour plots            through Illinois, Indiana, Ohio, Kentucky, and
prepared for each summary statistic. Specific     the Virginias, low (10 percentile)
emphasis was placed on the spatial                concentrations range from 42-47 ppb, about
distribution of the mean, 10, 50, and 90th        15 ppb above values observed along the edges
percentiles.                                      of the OTAG domain. However, in contrast to
Findings: Principal findings of this analysis     seasonal mean values, tenth percentile
were:                                             concentrations in urban areas are
                                                  indistinguishable from their surroundings.
1. Mean daily maximum values range from a
value close to 40 ppb (which is within the        6. 90th percentile concentrations are highest
range of northern hemisphere tropospheric         in urban areas, including those urban areas
background values) in the “corners” of the        which did not show mean concentrations
OTAG domain (south Florida, south Texas           elevated above regional background values.
Gulf Coast, upper Midwest/northern Plains,        Elevated 90th percentile values are also found
and northern Maine), up to about 80 ppb in the    over a large region covering southern Indiana
vicinity of some major urban centers. Mean        and southern Ohio.
values along the southern portion of the          7. Differences between 90th and 10th
western boundary in Texas, Oklahoma, and          percentiles show that daily maximum ozone
Nebraska are about 10 ppb higher than those       concentrations are most variable in urban

Summaries of Individual AQAWG Analyses                                                         2
areas and least variable in the corners of the     ozone concentrations due to variations in
OTAG domain. This is consistent with notion        meteorological conditions. Thus, blocks of
that monitored values in the corners of the        missing values at key groups of monitoring
domain represent broad-scale background            sites could be biasing the results in unknown
ozone levels.                                      ways. Finally, it should also be noted that
                                                   many or all of the days contributing to the low
Limitations: These results are dependent on
                                                   10th percentile concentrations observed in
the spatial distributions of monitors, the
                                                   many of the urban areas may have been
spatial interpolation methods used, and the
                                                   impacted by NOx titration under conditions
accuracy, precision, and completeness of the
                                                   leading to high NOx concentrations and
data.     Quantitative estimates of spatial
                                                   limited photochemical ozone production.
interpolation errors applicable to the
                                                   This is in contrast to the authors conclusion
AIRS/CASTNet data subset of this study are
                                                   that the low urban 10th percentile values
discussed by Falke (1996) but the impact of
                                                   reflect well ventilated conditions under which
these errors are not discussed in connection
                                                   urban and surrounding rural concentrations are
with the principal findings of this study. A
                                                   similar in magnitude. Additional analysis of
complete spatial analysis of interpolation
                                                   the conditions contributing to the 10th
errors for the integrated data set used in the
                                                   percentile values will be needed to resolve this
spatial analysis would be needed to confirm
some the findings noted above. In particular,
since urban, near-urban, and rural sites were      Scientific Implications: Any realistic model
weighted equally in the interpolation              of ozone/precursor relationships in the OTAG
procedure, it is possible that the interpolation   domain should be able to reproduce (or at
from urban or near urban sites (which make         least be consistent with) the principal features
up most of the network) produced a bias in the     of the monitoring data described in this
results in some locations. Also, since extreme     analysis. Of particular importance are the
concentrations (e.g., 90th percentiles) exhibit    tropospheric background levels found in the
larger spatial gradients than mean or median       corners of the domain and the apparent
values, the interpolation procedure may give a     narrowing and shift to the right (i.e., towards
less accurate picture of the spatial pattern of    higher concentrations) in the distribution of
extreme events as compared to average events.      daily maximum values in southern Ohio and
 It should also be noted that the author uses a    Indiana.
25 percent data completeness criteria for
                                                   Policy Implications: Human activities must
including monitors in the analysis rather than
                                                   be responsible for some or most of the ozone
the more commonly used 75 percent criteria.
                                                   excess in the central portions of the eastern
No analysis of the effect of the less stringent
                                                   U.S. over hemispheric background levels.
completeness criteria on uncertainties in
                                                   Furthermore, these results suggest that
computed summary statistics is provided.
                                                   precursor sources (presumably major NOx
This is particularly of concern for the more
                                                   point sources) along the Ohio River Valley are
extreme (i.e., 10 and 90) percentiles and the
                                                   associated with a broad region of elevated
exceedance statistics. Application of the
                                                   ozone levels not associated with any specific
completeness criteria to the integrated five
                                                   major urban centers. These results by
year data set rather than year-by-year is also a
                                                   themselves, however, do not indicate whether
potential concern since missing values are
                                                   or not this broad region of elevated ozone
unlikely to be randomly distributed over the
                                                   concentrations is of concern from the stand
five-year period and individual years are
                                                   point of limiting exceedances of the ambient
known to differ markedly in mean and peak
Summaries of Individual AQAWG Analyses                                                           3
ozone standard. Nor do these results by           associated with this region will have a
themselves indicate whether or not control of     beneficial effect on reducing exceedances in
the major NOx sources geographically              the OTAG domain.

Pattern of 8-Hour Daily Maximum Ozone Over the OTAG Domain

Participants: Rudy Husar, Washington University, St. Louis, MO
Reference: Pattern of 8-Hour Daily Maximum Ozone and Comparison with the 1-Hour Standard.
Rudy Husar, 16 September 1996 (

Purpose: Compare and contrast the spatial         example, the average ratio at Greenwich, CT
patterns of daily maximum 8-hour and 1-hour       (adjacent to New York City) is 0.82 with
ozone concentrations over the OTAG region.        ratios corresponding to daily maximum 1-hour
                                                  values above 100 ppb averaging about 0.75.
Method: A five year (1991-1995) data set of       Other locations exhibit ratios in between these
June-August       hourly     average     ozone    extremes. For the OTAG region as a whole,
concentrations for the eastern U.S. was           the average ratio is 0.86 with slightly higher
obtained by combining data from AIRS,             values in the northwestern portion of the
CASTNET, and several other monitoring             region and slightly lower values in the
networks. All sites with at least 25 percent      southwestern portion. Exceedances of a 1-
complete data were included in the analysis       hour, 120 ppb level are most common in the
(thus, some of the sites used may only have       northeastern urban corridor and near other
data for less than a single season). The final    major urban centers. The spatial pattern of 8-
data consisted of a total of 715 monitoring       hour exceedances of 102 ppb is similar.
sites. Six of these sites were discarded as a     Moving to a 1-hour, 94 ppb or an 8-hour, 80
result of quality control procedures. Daily       ppb level broadens the geographic extent of
maximum 1-hour and 8-hour averages were           the exceedances to include a band of eight or
extracted for each day with sufficient data and   more exceedance days per year just north of
means, 10, 50, 90th percentiles, and              the Ohio river stretching from Illinois to Ohio.
exceedance counts for various 1-hour and 8-
hour thresholds were computed.                    and on into western Pennsylvania. Although
                                                  the overall spatial patterns of 1-hour and 8-
Results: Comparisons of daily maximum 1-          hour exceedances are similar, examination of
hour values with 8-hour values for the same       a difference plot (8-hour exceedances of 80
day indicated that the 8-hour to 1-hour ratios    ppb minus 1-hour exceedances of 94 ppb)
were lowest in urban areas where high peak 1-
hour ozone concentrations are observed and
approaches 1:1 at rural remote sites.1 For        ratio exceeds one. This is due to EPA‟s
                                                  proposed method of assigning a given 8-hour
                                                  average to the day in which the first hour of
        In some cases at these sites in which     the 8-hour window falls.
ozone concentrations increasing overnight, the
Summaries of Individual AQAWG Analyses                                                          4
indicates that the 8-hour exceedances are          to exhibit relatively sharp 1-hour peaks. As a
relatively more frequent within those broad        result, on a same day basis daily maximum 8-
areas north of the line separating Kentucky        hour averages at such sites are relatively small
and Virginia from Tennessee and North              compared to 1-hour maxima. In contrast, rural
Carolina which are away from the direct            remote sites are characterized by low
influence of a major urban area. With the          amplitude diurnal cycles in 1-hour values and
exception of North Carolina, differences in        therefore maximum 8-hour values tend to be
exceedances are much less pronounced south         only slightly less than corresponding 1-hour
of this line. Interestingly, Dallas and Houston,   maxima.
TX show quite different responses, with 8-         Maps presented in this study show that
hour exceedances relatively more common in         multiple exceedances per year of an 8-hour, 80
Dallas and less common in Houston.                 ppb level covered a much broader area
Maps of 90th and 10th percentile 8-hour daily      between 1991-1995 as compared to multiple
maximum ozone concentrations are nearly            exceedances per year of a 1-hour, 120 ppb
identical to corresponding maps for the 1-hour     level: the latter are confined to the immediate
daily maxima. Similarly, 8-hour and 1-hour         vicinity of a few large urban areas while the
maps of the difference between 90th and 10th       former cover much of the OTAG domain with
percentiles are also very similar.                 the exception of the extreme northwest,
                                                   northern Main, and the deep South indicating
Limitations: Limitations noted for Husar‟s
                                                   the broad spatial extent of such “mid-level”
analysis of 1-hour daily maxima also apply to
                                                   ozone concentrations. The threshold levels
this study with the exception that an
                                                   for which the frequency of 8-hour
appropriately     more       stringent    data
                                                   exceedances is comparable to the frequency of
completeness criterion was applied to the
                                                   1-hour exceedances of 120 ppb for the OTAG
analysis of 8-hour exceedances, thus reducing
                                                   domain as a whole is approximately 102 ppb.
the potential for distortions resulting from
missing data in these results. This study was      Policy Implications: Results presented in this
done prior to EPA‟s announcement of plans to       study indicate that daily maximum 1-hour and
replace the current 1-hour ozone NAAQS of          daily maximum 8-hour concentrations are
0.12 ppm with an 8-hour, 0.08 ppm standard         very closely related both spatially and
and therefore did not directly compare areas of    temporally. Differences in the areas over
violation of the current and proposed              which 1-hour vs. 8-hour standard violations
standards, although exceedance maps for these      occur depend on the level and form of the 8-
two concentration levels are provided. Such        hour standard and were not explicitly
area of violation maps have, however, been         addressed here. However, the above results
produced by EPA and others.                        do indicate that at locations experiencing the
                                                   highest 1-hour concentrations in the OTAG
Scientific Implications: One-hour and eight-
                                                   domain (e.g., downwind of major urban areas
hour daily maximum concentrations are highly
                                                   such as in southern Connecticut and along the
correlated both temporally and spatially. The
                                                   eastern shore of Lake Michigan), the 1-hour
spatial patterns of threshold exceedances and
                                                   peak exceeds the 8-hour peak by a greater
percentiles for 1-hour and 8-hour daily
                                                   amount than in other areas. Thus, all else
maxima are very similar to one another.
                                                   being equal, an 8-hour standard may be
Comparisons of the magnitude of same-day 1-
                                                   relatively less stringent in these locations.
hour and 8-hour peaks discussed above
indicate that, under ozone episode conditions,
concentrations at high ozone urban sites tend

Summaries of Individual AQAWG Analyses                                                           5
Weekly Pattern of Ozone over the OTAG Region

Participants: Rudy Husar, CAPITA, Washington University, St. Louis, MO
Reference: Weekly Pattern of Ozone Over the OTAG Region, Rudy Husar 16 September 1996

Purpose: Analyze the influence of weekly          notably absent (actually reversed with slightly
cycles in anthropogenic emissions on daily        higher values on Sundays than on Fridays)
maximum ozone concentrations.                     along a broad band stretching northeastward
                                                  from western Tennessee through Kentucky,
Method: A five year (1991-1995) data set of       West Virginia near the Ohio border,
June-August daily maximum 1-hour ozone            Pittsburgh, and then northwestward to
concentrations for the eastern U.S. was           Cleveland. With the exception of Pittsburgh,
obtained by combining data from AIRS,             this band overlaps the region of relatively high
CASTNET, and several other monitoring             10th percentile ozone concentrations noted by
networks. All sites with at least 25 percent      Husar (1996a). This “negative” Sunday effect
complete data were included in the analysis       is also evident at a few other isolated
(thus, some of the sites used may only have       locations. No significant weekly cycle is
data for less than a single season) . The final   evident in average or 50th percentile
data consisted of a total of 715 monitoring       concentrations.
sites. Six of these sites were discarded as a
result of quality control procedures. After       Weekly patterns in the frequency of
disagregation by day of week, average             exceedance of 120 ppb are similar to those for
concentrations and various percentiles of the     the 95th percentile of daily maximum
concentration distribution were calculated for    concentrations with one third as many
each monitoring site and contour plots            exceedances on Sundays as on Fridays. The
prepared for each summary statistic. Specific     number of exceedances increases nearly
emphasis was placed on the spatial                linearly from Monday through Friday. In most
distribution of the mean, 10, 50, and 90th        cases, the largest positive weekday - weekend
percentiles and on exceedances of 120 ppb.        differences in exceedances (i.e., more
                                                  exceedances on Fridays than on Sundays) are
Results: For the OTAG region as a whole,          over the urban areas with smaller differences
seasonal peaks (90 and 95th percentiles) of       elsewhere. In some urban areas, however,
daily maximum ozone concentrations were           exceedances are actually somewhat more
found to be about seven percent lower on          common on Sundays than on Fridays. Most of
Sundays than during the week.           After     these urban areas fall within the geographic
correcting for a constant 40 ppb background,      region noted above in which 90th percentile
the weekday to Sunday difference is seen to       concentrations on Sundays are slightly higher
represent about 33 percent of the OTAG            than on Fridays. Several other isolated areas
region ozone excess above background. This        also exhibit a weak “negative” Sunday effect.
“Sunday effect” is most pronounced in the         The weekly cycle noted at the 120 ppb
90th percentiles around the major urban areas     exceedance threshold is progressively damped
where peak concentrations are highest but is      at lower thresholds (100, 80, 60 ppb). Almost

Summaries of Individual AQAWG Analyses                                                          6
no weekend effect is evident at 60 ppb.            has not been examined and the extent of any
                                                   weekday/weekend differences are unknown.
Scientific Implications: Results of this study
indicate that the changes in average emissions     Limitations:       These results are highly
between weekdays and weekends are                  dependent on the spatial distributions of
sufficient to significantly reduce peak ozone      monitors, the spatial interpolation methods
concentrations on weekends while mean              used, and the accuracy, precision, and
concentrations are largely unaffected.             completeness of the data. Quantitative
Unfortunately, since the nature (magnitude of      estimates of spatial interpolation errors
emission reductions, relative source mix,          applicable to the AIRS/CASTNet data subset
diurnal pattern) of the differences in emissions   of this study are discussed by Falke (1996) but
on weekends as compared to weekdays have           the impact of these errors are not discussed in
not been quantified, the mechanism resulting       connection with the principal findings of this
in lower peak ozone levels on weekends in          study. A complete spatial analysis of
urban areas is not yet fully understood. The       interpolation errors for the integrated data set
gradual increase in upper percentiles of the       used in the spatial analysis would be needed to
concentration distribution and in the numbers      confirm some the findings noted above. In
of exceedances of high concentration               particular, since urban, near-urban, and rural
thresholds (e.g., 120 ppb) during the course of    sites were weighted equally in the
the work week is indicative of the influence of    interpolation procedure, it is possible that the
carry over of ozone and precursors from one        interpolation from urban or near urban sites
day to the next. This carry over keeps             (which make up most of the network)
Saturday concentrations higher than they           produced a bias in the results in some
would otherwise be but by Sunday the full          locations. Also, since extreme concentrations
influence of the weekend emission changes          (e.g., 90th percentiles) exhibit larger spatial
are evident. The weekend effect is most            gradients than mean or median values, the
pronounced at the upper end of the                 interpolation procedure may give a less
concentration distribution and therefore           accurate picture of the spatial pattern of
changes in urban areas, where high                 extreme events as compared to average events.
concentrations occur most frequently, tend to       It should also be noted that the author uses a
be larger than in rural areas. In some urban       25 percent data completeness criteria for
areas, however, peak concentrations on             including monitors in the analysis rather than
Sundays are actually slightly higher than on       the more commonly used 75 percent criteria.
Fridays (e.g., Pittsburgh, PA). The reasons for    No analysis of the effect of the less stringent
this anomalous behavior are not known but          completeness criteria on uncertainties in
may be related to geographic variations in         computed summary statistics is provided.
combinations of several factors (mix of source     This is particularly of concern for the more
types, activity patterns, VOC/NOx ratios).         extreme (i.e., 10 and 90) percentiles and the
Geographically, the overlap between areas of       exceedance statistics. Application of the
relatively high 10th percentile concentrations     completeness criteria to the integrated five
and areas of negative weekend effect in the        year data set rather than year-by-year is also a
central portion of the OTAG domain may be          potential concern since missing values are
related to the impacts of elevated NOx sources     unlikely to be randomly distributed over the
associated with base-load utility boilers          five-year period and individual years are
characterized by relatively small weekly           known to differ markedly in mean and peak
variations in emissions.         However, the      ozone concentrations due to variations in
temporal pattern of emissions in this region       meteorological conditions. Thus, blocks of
Summaries of Individual AQAWG Analyses                                                           7
missing values at key groups of monitoring           by roughly a factor of three. Unfortunately,
sites could be biasing the results. Finally,         the exact nature of weekday/weekend
reasons for the “negative” weekend effect            emission differences is not currently known.
found in some areas are not well understood          Not only the magnitude of the emission
and require further study. In particular, no         reductions on weekends, but also changes in
analysis     is    presented     here     of         the relative mix of VOC vs. NOx emissions
weekday/weekend differences in precursor             and of low-level vs. elevated NOx sources are
emissions.                                           likely to be important. Weekday/weekend
                                                     differences in the diurnal pattern of emissions
Policy Implications: Results of this analysis
                                                     are also likely to be important: differences in
imply that a control scenario mimicking
                                                     the timing of mobile source emissions may be
weekday to weekend emission reductions
                                                     as important in reducing ozone concentrations
would result in decreases in peak ozone
                                                     as differences in the total amount of
concentrations in most urban areas. For the
OTAG region as a whole, exceedances of the
current NAAQS (120 ppb) would be reduced

Spectral Decomposition of Ozone Time Series

Participants: P.S. Porter, University of Idaho, Idaho Falls, ID 83405; S.T. Rao, I. Zurbenko, and E.
Zalewsky, State University of New York, Albany, NY 12222; R.F. Henry and J.Y. Ku, NY State
Department of Environmental Conservation, Albany, NY 12222.
References: Statistical Characteristics of Spectrally-Decomposed Ambient Ozone Time Series Data,
P.S. Porter, S.T. Rao, I. Zurbenko, E. Zalewsky, R.F. Henry, and J.Y. Ku, Final Report, August,

Purpose:        Estimate       the       spatial     logarithms of daily maximum 1-hour and 8-
representativeness of data collected at routine      hour average ozone concentrations into long-
ozone monitoring sites and the typical size of       term (annual), seasonal, and short-term
airsheds (zones of influence).         Identify      (synoptic scale) components. This is done at
principal features of the spatial and temporal       all routine ozone monitoring sites in the U.S.
characteristics of daily maximum 1-hour and          reporting to EPA‟s AIRS database for which
8-hour ozone concentrations within three             few values are missing for the period 1983 -
distinct frequency domains: long-term (i.e.,         1994. Monitors are segregated into those that
annual), seasonal, and short-term (i.e.,             operate year-round and those that operate only
synoptic variations on time scales of less than      during the summer ozone season.
about 30 days).                                      Characteristics of each frequency domain are
                                                     investigated, including temporal variance and
Methodology: Using multiple pass moving              autocorrelation, spatial patterns in means and
average filters (Kolmorgorov-Zurbenko, or            variances, spatial correlations, and the
KZ filters) described by Rao and Zurbenko            influence of meteorological conditions.
(1994), the authors develop a spectral
decomposition of the time series of natural          Findings: Using two KZ filters, one a 15-day

Summaries of Individual AQAWG Analyses                                                            8
average, 5-pass filter and the other a 365-day,    zero mean, the authors found that it is possible
3-pass filter, the authors find that the time      to reproduce the observed median
series of log-transformed daily maximum            concentrations at nearly all sites (to within the
concentrations can be cleanly separated into       95% confidence interval of the predicted
(1) a long-term component representing the         value); 95th percentiles of the daily maximum
influence of trends in emissions and (possibly)    values were less accurately reproduced.
short-term climatic fluctuations, (2) a seasonal   Exceedance counts were similarly evaluated,
component representing the influence of the        but it is not clear how to interpret these results
earth‟s rotation about the sun, and (3) a short-   since most monitors will have had very few or
term component representing the influence of       no exceedances over the 12-year period
fluctuating synoptic meteorological conditions     examined. Rao et al. (1996) provide
and random processes (noise). An analysis of       comparisons of observed and modeled
variance showed that covariances between the       exceedance counts and 95th percentiles at five
different components were generally small          locations which experienced multiple
(less than 2% of the total variance). Long-        exceedances per year (at least prior to 1989)
term component variances were less than 10%        which show generally close agreement with
of the total at more than ninety percent of all    some exceptions.
monitoring sites, with median values of 3.6%       A trends analysis for the period 1985 - 1995
for summer season only monitors. For               was also presented by the authors. Trends
monitors with year-round data, the seasonal        were computed for the ozone baseline (long-
component accounts for up to 73% of the total      term plus seasonal component) both before
variance (median value of 51%) while the           and after adjusting for variations in
median contribution of the short-term              temperature. Raw (unadjusted) baseline
component is 51%. For summer season only           trends are mixed, with generally negative
monitors, the median seasonal component            trends in the southern tier states, Illinois, and
variance contribution was 12%, while the           the Philadelphia - New York urban corridor
median short-term component was 77%.               but increases at most sites in Indiana, Ohio,
Seasonal components are higher in the              West Virginia, Western Pennsylvania, and the
Northeast than in the South. Contributions of      Washington D.C. area. After adjusting for
short-term variance to the total are highest in    temperature fluctuations, negative trends were
coastal areas and lowest in the Midwest. In        found at a much larger number of sites.
absolute terms, the highest short-term
variances are in the Northeast and along the       Spatial correlations in baseline and short-term
Gulf Coast.                                        components of both ozone and surface
                                                   temperature time series were examined. For
An evaluation of the statistical properties of     the Washington D.C. area, temperature
the short-term component, W(t) revealed it to      baseline correlations remained constant and
be significantly negatively skewed and
                                                   near unity over the maximum distance
therefore not normally distributed at nearly all   examined (1,280 km or 800 miles), while
monitoring sites, while exp(W) was more            correlation in the short-term component
nearly normally distributed in most locations.     decreases linearly with distance over this
Autocorrelations in W(t) at one day lag had a      radius, dropping to 0.3 at 1,120 km (700
median value of 0.35, indicating some day-to-      miles). Ozone baseline spatial correlations
day dependence resulting from persistence in       drop off more rapidly but also respond
meteorological conditions. Nevertheless,           approximately linearly with distance, at least
modeling the short-term component as a             out to the maximum distance considered (800
random variable normally distributed with
Summaries of Individual AQAWG Analyses                                                             9
km or 500 miles), at which point the                monitors in high ozone locations in the
correlation coefficient is approximately 0.8.       Northeast. Generally lower reductions are
Short-term ozone component spatial                  required during the latter period due to the
correlations decrease exponentially with            progress towards attainment already achieved.
distance over this range, dropping to a value
                                                    Limitations: From a regulatory perspective,
of 0.3 at about 720 km (450 miles) along the
                                                    much of the value in these analyses is based
“direction of transport”2. A map of spatial
                                                    on the viewpoint that the baseline component
correlation isopleths for the short-term
                                                    of the ozone time series (i.e., the component
component is presented by the authors for five
                                                    remaining after removal of short-term
cities in the OTAG domain (Atlanta, Chicago,
                                                    fluctuations using the KZ15,5 filter) is
Cincinnati, Greenbelt MD, Pittsburgh). In all
                                                    “deterministic” in the sense that it is
cases, correlations drop to 0.4 within 560 to
                                                    controlled entirely by seasonal meteorological
640 km (350 to 400 miles) of the selected
                                                    fluctuations and year-to-year trends in
monitors. It is suggested that these are the
                                                    emissions.       The remaining, short-term
spatial domains which should be considered in
                                                    component, on the other hand, resemble
designing control strategies.
                                                    random fluctuations. Thus, management of
Comparisons of time series of 1-hour average        the ozone problem should focus on the effects
daily maximum ozone to 8-hour average daily         of control strategies on the baseline rather than
maximums indicated that variances are higher        being concerned with a highly fluctuating
for the 8-hour averages, both in an absolute        peak statistic (e.g., 4th highest concentration
sense, relative to the mean, and for the short-     in three years) in which much of the
term component only. Looking at daily               variability is associated with “noisy”
maximum n-hour averages at one site                 processes that cannot be accurately modeled.
(Cliffside Park, NJ), the authors found that        However, in order to make use of this
variances of the short-term components peak         approach, one must be able to accurately
at n=10 hours ( = 0.38 as compared to  =          estimate the impact of future control strategies
0.27 at n=1 and  = 0.35 at n=24). The              on the baseline, and such methods have yet to
statistical significance of these differences are   be successfully demonstrated (i.e., it is not
not discussed.                                      clear how to adapt episodic photochemical
                                                    models to models of baseline ozone). In
Finally, using the technique described above
                                                    addition, one must be able to estimate the
of modeling the short-term component, W(t)
                                                    impact (or convincingly demonstrate that there
as white noise, the authors compute the
                                                    is no impact) of control scenarios on the
percent reduction needed in the ozone baseline
                                                    statistical characteristics of the short-term
component to achieve attainment of the ozone
                                                    component.       To date, neither of these
NAAQS for the 1987-1989 and 1991-1993
                                                    conditions have been met. Of course, it
attainment periods at each monitoring site.
                                                    should also be noted that the ability of
These required reductions range from 0% (for
                                                    episodic photochemical models to accurately
monitors already in attainment), to 50% at
                                                    predict the impact of future-year control
                                                    strategies on ozone levels has also yet to be
        2                                           successfully demonstrated.
         This direction (also referred to by the
authors as the “major axis”) is oriented in the     Although the trends analysis results presented
direction of the prevailing wind at a particular    by the authors are in general agreement with
location (e.g., SW-NE along the Washington -        results of other studies (see review prepared
New York urban corridor).                           by Morris, 1996), the fact that nearby sites

Summaries of Individual AQAWG Analyses                                                            10
often exhibit opposing trends is indicative of a    Furthermore, correlations at non-zero lag
complex spatial structure in trend statistics       times (e.g., one or two day lags) also need to
which may not be fully resolvable with the          be considered. Spatial scales of correlations
current monitoring network. Due to the large        in the short-term temperature component
number of sites examined, the trend maps            appear to be reasonable from a synoptic scale
presented by the authors are difficult to read in   meteorological perspective. The fact that
urban areas with many closely spaced                these scales are greater than those for the
monitoring sites due to the overlapping of          ozone component may indicate that much of
green disks (denoting negative trends) on top       the remaining variance in the ozone
of red ones (denoting positive trends); green       component (i.e., that portion not related to
disks are always on top, thus obscuring an          temperature) is associated with other
unknown number of sites with positive trends.       meteorological factors that operate on smaller
Tables of trend values for each site are            spatial scales.
available from the authors for review and Rao
                                                    Implications: Perhaps the most important
et al. (1995) present a table of trends for sites
                                                    implication of the above findings is that, when
in the eastern U.S. for 1983 - 1992. Similar
                                                    only the summer high ozone season is
trend maps for the northeastern U.S., but
                                                    considered, short-term fluctuations are the
showing spatially smoothed trends (where the
                                                    dominant source of variability in daily
smoothing was done with spatial KZ filters)
                                                    maximum concentrations (either 1-hour or 8-
are presented by some of the same authors
                                                    hour average). Thus, identifying the impact of
(Zurbenko, et al., 1995) and these maps do not
                                                    control strategies on ozone levels is extremely
suffer from this difficulty. Unfortunately, the
                                                    difficult due to the low signal to noise ratio.
spatial smoothing may obscure some
                                                    Furthermore, areas close to the attainment
important small scale trend differences. Also,
                                                    /nonattainment threshold are likely to jump in
interpretation of trend statistics would be
                                                    and      out      of     attainment      (where
enhanced by further characterization of trends
                                                    attainment/nonattainment is determined on the
by site type (i.e., rural, remote, suburban,
                                                    basis of one to three years of monitoring data),
downwind urban peak, urban center, etc.).
                                                    thus complicating management efforts. This
Published demonstrations of the independence        study shows that the problem appears to be
of short-term components of daily maximum           particularly severe in the Northeast - precisely
ozone and temperature (Rao and Zurbenko,            where accurate estimates of control strategy
1994; Flaum, Rao, and Zurbenko, 1996) are           impacts are most needed.
not conclusive because seasonal influences on
                                                    When seasonal and long-term trends are
the correlation were not resolved. Therefore,
                                                    removed from the ozone time series, the above
it has not been demonstrated that the short-
                                                    results show the spatial scale of the remaining
term ozone fluctuations during the summer
                                                    short-term fluctuations to be 560 - 640 km
season are completely independent of
                                                    (350 - 400 miles), suggesting a scale for the
meteorological effects.          Thus, spatial
                                                    coherence of same-day fluctuations in peak
correlations in this component may be
                                                    ozone which may be useful for air quality
partially attributable to spatial correlations in
                                                    management efforts. However, the degree to
meteorological conditions in addition to
                                                    which this spatial scale is determined by
emissions forcing and defining a spatial
                                                    spatial correlations in meteorological
domain for control strategy development on
                                                    conditions as compared to same-day transport
the basis of zero lag time spatial correlations
                                                    of ozone and precursors is unknown.
in the short-term components may not
properly address the area of source influence.      Additional References:

Summaries of Individual AQAWG Analyses                                                           11
Morris, 1996. Review of Recent Ozone               Eastern United States.      Environmental
Measurement and Modeling Studies in the            Manager, January, 1996, pp. 17-31.
Eastern United States.            ENVIRON          Flaum, J.B., Rao, S.T., and I.G. Zurbenko,
International Corp., 21 February 1996.             1996.      Moderating the influence of
Rao and Zurbenko, 1994. Detecting and              meteorology      on      ambient     ozone
Tracking Changes in Ozone Air Quality. J. Air      concentrations. J. Air & Waste Manage.
and Waste Management Assoc., Vol. 44:              Assoc., 46: 35-46, 1996.
1089 - 1092, 1994.                                 Zurbenko I.G., Rao, S.T. and R.F. Henry,
Rao, S.T., Zurbenko, I.G., Porter, P.S., Ku,       1995. Mapping Ozone in the Eastern United
J.Y. and R.F. Henry, 1996. Dealing with the        States. Environmental Manager, Vol. 1 pp.24-
Ozone Non-Attainment Problem in the                30, January, 1996.

Ozone Exceedances Data Analysis: Representativeness of 1995

Participants: Lyle Chinkin, Richard Reiss, Douglas Eisenger, Timothy Dye, and Christopher Jones,
Sonoma Technology, Inc., Santa Rosa, CA
Reference: Ozone Exceedances Data Analysis: Representativeness of 1995. L. Chinkin et al.,
Sonoma Technology, Inc., 1996.

Purpose: Assess the representativeness of the      concentrations were evaluated for the 1980s
1995 ozone season in the Northeast in the          and 1990s to determine to what extent 1995
context of the past 15 ozone seasons. The          was consistent with these trends. Year-to-year
motivation for this study was the realization      fluctuations in the frequency and severity of
that the summer of 1995 was one of the             meteorological conditions conducive to ozone
hottest on record for the Northeastern United      formation were accounted for using several
States, and that the 1995 summer season was        different methods including: 1) An
therefore a strong candidate to experience         examination of year-to-year differences in the
high ozone concentrations.                         relationship between the frequency or severity
                                                   of high temperature days (e.g., the number of
Methodology: The authors undertook an              days above 90F) and the frequency or
investigation of whether the above normal          severity of ozone episodes (e.g., number of
temperatures during the summer of 1995 were        exceedance days), 2) the Poison regression
associated with above normal ozone                 model developed by Cox and Chu (1995), and
concentrations. The authors utilized air           3) The synoptic typing scheme of Comrie and
quality and meteorological data from seven         Yarnal (1992).
metropolitan areas within the Northeast‟s
Ozone Transport Region (OTR), as well as the       Findings: It was found that despite near
OTR overall. The metropolitan areas studied        record breaking temperatures, the 1995 ozone
include Baltimore, Boston, Hartford, New           season was no worse than the average ozone
York, Philadelphia, Providence, and                season experienced during the 1990s. In fact,
Washington, D.C. Numerous regional and             1995 is consistent with overall trends showing
subregional trends in 1-hour average ozone         declining ozone concentrations in the

Summaries of Individual AQAWG Analyses                                                        12
Northeast. The major findings cited by the        particular, the Cox and Chu Poison regression
authors are as follows:                           method assumes underlying (i.e., emission-
   The number of days per year above the         related) trends are linear in time - this is a
    national ozone air quality standard has       rough approximation at best. In addition,
    declined by about 70 percent from 1981-       year-to-year changes in the ratio of the number
    1995. The summer of 1995, despite being       of days in a given season above a temperature
    one of the hottest on record, is consistent   threshold to the number of ozone exceedance
    with that trend.                              days are very sensitive to the temperature and
   The severity of individual ozone events is    concentration thresholds used and are subject
    declining and 1995 is consistent with that    to numerous factors not related to emissions.
    trend. For example, on days above the         These ratios thus provide only a very rough
    national      standard,     the    average    indication of actual precursor emission trends.
    concentration by which the standard was       Scientific Implications: Results of this study
    exceeded has declined by about 70 percent     are generally consistent with those of other
    from 1981-1995.                               recent trends analyses which show concurrent
   The extent of the geographic area             declines in peak ozone levels and VOC
    experiencing days above the ozone             emissions.     Data on ambient precursor
    standard is declining.                        concentration trends are needed to verify
   During the 1990s ozone concentrations         estimated trends in emissions.
    above the national standard occurred at
                                                  Policy Implications: Although not entirely
    higher temperatures than in previous years:
                                                  definitive for the reasons noted above, this
    the most plausible explanation is that
                                                  study supports other independent analyses of
    ozone precursor emissions have decreased.
                                                  northeastern ozone and emission trends,
   Between 1985 and 1994, VOC emissions          strongly suggesting that recent VOC emission
    in the Northeast (EPA Regions I, II, and      reductions resulted in decreases in peak 1-
    III) decreased 13 percent, while NOx          hour ozone levels over the routine monitoring
    emissions decreased only 5 percent.           network. Concurrent but relatively small NOx
Limitations: As with most other trends            emission reductions may have contributed to
studies, the emphasis here is on peak             the ozone reductions. Of course, these studies
concentrations in urban areas in the Northeast.   do not provide any direct information on the
 However, the study did include a comparison      relative effectiveness of future VOC controls
of trends at sites designated as urban or         in the region.
suburban in the AIRS database with sites          Additional References:
determined to be in rural locations based on a    Comrie, A.C. and B.            Yarnal, 1992.
review of their map coordinates. This analysis    Relationships       between     synoptic-scale
indicated no discernable difference between       atmospheric       circulation    and    ozone
urban and rural trends. It should also be noted   concentrations in metropolitan Pittsburgh,
that uncertainties in the emission inventory      Pennsylvania. Atmos. Environ., 26B, 301-
trends (especially in the NOx trends) are not     312.
evaluated. This is particularly important given   Cox, W.M. and S-H Chu, 1996. Assessment
the lack of attention to ambient NOx trends       of interannual ozone variations in urban
which could be used to corroborate the            areas from a climatological perspective.
inventory trend (assuming sufficient high         Atmospheric Environment, 30(14): 2615-.
quality NOx data are available). Uncertainties
in the correction for meteorological conditions
must also be taken into consideration. In
Summaries of Individual AQAWG Analyses                                                        13
Ozone Measurement Trend Studies in the Northeast

Participants: Ralph Morris, ENVIRON, 101 Rowland Way, Novato, CA (reviewing work
performed by USEPA (1994 Air Quality Trends Report), S. T. Rao et al., New York State
Department of Environmental Conservation, George Wolff and Patricia Korsog, General Motors
Corporation (GMC), William Cox and Shao-Hang Chu, OAQPS, USEPA, Kay Jones, Zephyr
Reference: Review of Recent Ozone Measurement and Modeling Studies in the Eastern United
States. R.E. Morris, ENVIRON International Corp., March, 1996 and references therein.

Purpose: Review and prepare an integrated         meteorological-adjusted ozone concentrations
summary of recent major northeastern U.S. air     illustrated a definite downward trend in ozone
quality and emission trend analyses.              over the period studied.         EPA‟s 1995
                                                  emissions trends report (EPA, 1995) provides
Methodology: Several researchers have             trends in national ozone precursor emissions
performed trends analysis of ozone in the         over the period of 1900-1994, with particular
eastern U.S. in recent years. These analyses      focus on the last ten-year period of 1985-
differ mainly in the techniques used to account   1994. During this ten-year period, there was a
for the impact of meteorological variations in    distinct downward trend in VOC emissions
the ozone trends. A brief summary of              (14 percent reduction) and a slight increase in
methodologies and results were prepared for       NOX emissions (6 percent increase)
each analysis. These summaries were then          suggesting that the downward trend in ozone
used as the basis for an integrated assessment    is likely due to the reductions in VOC
of recent ozone and emission trends in the        emissions.
                                                  Accounting for Meteorological Variations in
Findings: Summaries of methods and results        Ozone Trends by NYSDEC/SUNY:
for each individual study:                        Researchers from the State University of New
EPA National Trends Reports: EPA‟s 1994           York (SUNY) at Albany and the New York
annual trends report (EPA, 1994) examined         Department of Environmental Conservation
both longer-term (1984-1993) and shorter-         (NYSDEC) have developed a statistical
term (1991-1993) trends in ozone using            technique to eliminate the high frequency
national composite averages (EPA‟s 1995           variations in ozone trends due to
trends reports was not available). Between        meteorological fluctuations (Rao and
1984 and 1993, the composite average              Zurbenko, 1994). Rao and co-workers applied
exceedances over the 532 monitoring sites         this technique to analyze the trends in
examined decreased 60 percent, a statistically    meteorological-adjusted ozone concentrations
significant amount. The EPA national trends       for the period of 1980-1992. They found a
report also examined the trends in ozone          definite downward trend in ozone for the
concentrations after the effects of               Northeast     Corridor    stretching     from
meteorological variations have been removed       Wilmington, Delaware up to Boston,
using the parametric regression technique         Massachusetts, and essentially no change in
developed by Cox and Chu (1992). The              the meteorological-adjusted ozone in the

Summaries of Individual AQAWG Analyses                                                        14
Baltimore-Washington D.C. and southern              researchers‟ analysis of the trends in VOC-to-
Maine regions. Zalewsky and co-workers              NOX ratios also suggest a downward trend in
(1994) examined the trends in VOC and NOX           the cities studied from a value of
emissions in the Northeast Corridor during          approximately 10 in 1986 to around 5-6 in
this period and found a downward trend in           1991. These results suggest that urban
VOC emissions and very little change in NOX         photochemistry is becoming more VOC-
emissions. This suggests that the decrease in       limited in these cities (i.e., further VOC
ozone concentrations are attributable to the        emission reductions will be effective in
VOC emissions reductions.                           lowering ozone concentrations, while NOx
                                                    emission reductions may result in local
General Motors Corporation Studies on Ozone
                                                    increases in ozone concentrations).
Trends: A series of studies performed by
researchers at General Motors Corporation           Trends in Meteorological-Adjusted Ozone
(GMC) analyzed ozone trends at several cities       Concentrations using the Cox and Chu
in the eastern U.S. for the period of 1980          Method: Cox and Chu (1993, 1995) used a
through 1993, as well as trends in morning          probability model to examine meteorological-
VOC-to-NOX ratios and the Reid Vapor                adjusted ozone trends with statistical
Pressure (RVP) of gasoline as a surrogate for       confidence estimates. For the period of 1981
VOC emissions (Wolff, 1993; Wolff and               to 1990, statistically significant declines in
Korsog, 1994).          Although the GMC            ozone concentrations were calculated in New
researchers did not perform adjustments to the      York, Philadelphia, Pittsburgh, and Baltimore,
ozone trends to account for meteorological          and small, but not statistically significant,
variations, trends were calculated for several      declines in Washington D.C. and Boston were
different ozone parameters, including 8-hour        calculated. Cox and Chu also used their
averages and number of days the maximum 1-          probability model to estimate the relative
hour and 8-hour ozone exceeds threshold             potential for the occurrence of meteorological
levels. They also calculated confidence             conditions conducive for ozone formation in
intervals to determine whether trends were          each urban area for the 40 year record of 1953
statistically significant.        They noted        to 1993. For northeastern urban regions, the
statistically significant downward trends in        rank order statistics suggest that 1988 had the
many of the ozone summary statistics in New         most conducive meteorological conditions for
York, Newark, Milwaukee, and Chicago, and           ozone formation in the 40 year record
downward trends in maximum and second               followed by 1953, 1955, 1983, 1991, and
highest 1-hour ozone, although not always           1993. The return time of the extremely
statistically significant, in all cities studied    adverse conditions seen in 1988 was estimated
except Boston and Grand Rapids. It is               to be a 1-in-40 year event in most northeastern
interesting to note that in Atlanta and             cities.
Philadelphia (and to a lesser extent in
                                                    Meteorological-Adjusted Trends in Ozone
Baltimore), there were downward trends in           Concentrations Calculated by Kay Jones: Kay
peak 1-hour ozone concentrations, but upward        Jones of Zephyr Consulting performed several
trends in the number of days in which the 8-        analyses of trends in yearly maximum, second
hour ozone concentration exceeded 60 ppb.           highest, and three-year design value (fourth
These results suggest that the distribution in --   highest ozone value in three years) ozone
hour ozone concentrations in these cities is        concentrations. Year-to-year variability in
becoming narrower and urban peaks are               meteorology were accounted for by the
declining, but mid-level (baseline) ozone           number of days per year the temperature in
concentrations are increasing. The GMC
Summaries of Individual AQAWG Analyses                                                          15
Philadelphia exceeds 90 F. A steep decline in     nonattainment areas, thus the trends are biased
meteorological-adjusted ozone concentrations      towards urban areas; trends in rural areas are
in all northeastern cities was estimated. Using   not specifically analyzed. It should also be
more recent ozone data to calculate design        noted that uncertainties in the emission
values suggests there would be fewer ozone        inventory trends (especially in the NOx trends)
nonattainment areas than the three-year period    are not evaluated. This is particularly
that was used to classify them in the 1990        important given the lack of attention to
CAAA (which included 1988).                       ambient NOx trends which could be used to
                                                  corroborate the inventory trend (assuming
Integrated Results: According to the author‟s
                                                  sufficient high quality NOx data are available).
review, the measurement studies demonstrate
that: (1) meteorological conditions play an       Scientific Implications: Recent declines in
important role in the occurrence of ozone         (predominantly urban) peak 1-hour ozone
exceedances; and (2) after accounting for         levels in the northeastern U.S. (after at least
meteorological variations, there is a definite    partially accounting for variations in
downward trend in ozone concentrations over       meteorological conditions) are coincident with
the last ten years in the eastern U.S. Because    decreases in VOC emissions. In contrast,
VOC emissions have been reduced during this       inventory NOx emissions are relatively stable
period while NOX emissions have remained          during this period. In at least one case,
fairly stable, this air quality improvement can   opposing trends in 8-hour vs. 1-hour
be attributed to reductions in VOC emissions.     concentrations have been noted, implying
Although most of the studies focused on 1-        either that VOC controls may be decreasing
hour ozone concentrations, one study also         peak concentrations or that diurnal variations
considered        8-hour       average   ozone    in ozone are getting narrower. Data on
concentrations. This study found that while       ambient precursor concentration trends are
the peak 1-hour concentrations decreased, the     needed to verify estimated trends in emissions.
8-hour concentrations increased. This may         Additional information on rural area trends is
indicate either that the diurnal variation in     also needed.
ozone is getting narrower or that the VOC
                                                  Policy Implications: Although not entirely
reductions are effectively reducing the urban
                                                  definitive for the reasons noted above, these
peak 1-hour ozone concentrations, but they
                                                  independent analyses of northeastern ozone
have less effect on regional ozone (which is
                                                  and emission trends strongly suggest that
more closely represented by the 8-hour ozone
                                                  recent VOC emission reductions resulted in
                                                  statistically significant decreases in peak 1-
Limitations: The principal focus of this          hour ozone levels over the routine monitoring
review was on modeling studies; thus the          network. Of course, these studies do not
review of measurement studies was not             rovide any direct information on the relative
comprehensive.      Furthermore,   most           effectiveness of future VOC controls in the
measurement programs are focussed on              region.

Summaries of Individual AQAWG Analyses                                                         16
SOS Nashville/Middle Tennessee Ozone Study

Participants: Jim Meagher, Tennessee Valley Authority
References: Slides from presentation given to the OTAG AQAWG meeting, Washington D.C., July

Purpose: Analyze ambient data from the 1995       occurred over July 11-15, 1995 appears to be
Nashville SOS study to investigate ozone          “home grown,” i.e., the area of high ozone
formation in the Nashville area and in major      was quite localized around Nashville and
point source plumes released into a southern      therefore appears to have been related to
environment.                                      Nashville emissions. Ozone concentrations in
                                                  the mixed layer around Nashville appear to
Methodology: The SOS Nashville data base          have been elevated by up to about 80 ppb
for 6 weeks in June and July, 1995, includes      above the regional ozone concentration of 60-
enhanced measurements of air quality and          70 ppb. The mixed layer depth extended up to
meteorological parameters, e.g., surface/aloft    about 2 km and ozone concentrations above
VOC, CO, SO2, NOX and NOY concentration           the mixed layer were about 50 ppb (aircraft
data, airborne LIDAR ozone data, ozone            data).
sondes, and RADAR/RASS atmospheric
profilers. There were over 200 scientists         The author also compared O3 to NOZ
involved, and Nashville 1995 was one of the       relationships in urban and power plant plumes
largest air quality studies ever conducted in     using two approaches. In one method,
the U.S. This period overlaps with the July 9-    O3/NOZ relationships measured at three
18, 1995 OTAG episode. The study also             ground stations were compared for periods of
covered a four week period in 1994, but           high and low SO2 levels, where the SO2 levels
aircraft measurements were much more              were used to indicate whether the airmass was
limited in 1994 than 1995. For 1995, surface      associated with power plant or urban
and aircraft ozone and NOY data were              emissions. The ozone production efficiency
analyzed for ozone productivity, defined as       was 70% higher for air masses with low SO2
the ratio of O3 to NOZ (where NOZ = NOY -         levels, possibly indicating that ozone
NOX) which may be considered as the number        production is more efficient (in terms of NOX
of ozone molecules produced per NOX               utilization) in urban plumes than in power
molecule lost (ozone productivity analyses are    plant plumes. O3/NOZ relationships were also
also discussed in the review of presentations     evaluated from aircraft data obtained in urban
by Bob Imhoff of TVA). Surface and aloft          and power plant plumes. On one day, July 16,
ozone data were analyzed to characterize the      the TVA helicopter made measurements in
spatial extent (horizontal and vertical) of the   both a power plant plume and an urban plume
Nashville ozone episode that occurred over        and observed different ozone production
July 11-15, 1995. Aircraft data collected in      efficiencies, with the urban plume
several large point source NOX plumes were        approximately 65% more efficient than the
analyzed to characterize the spatial/temporal     power plant plume.
scales for conversion of NOX to NOZ and the       Plumes from the Cumberland (~350 tons per
concomitant ozone production.                     day of NOX) and Johnsonville (~80 tons per
Findings: The Nashville ozone episode that        day of NOX) power plants were followed by

Summaries of Individual AQAWG Analyses                                                       17
helicopter flights from close to the source out   thus travel relatively long distances before
to about 150 km downwind. At this distance,       taking part in photochemical reactions the
NOX was observed to be essentially totally        following day.
converted to NOZ, and NOX levels had
                                                  Scientific Implications: For the July 11-15,
dropped to a point where ozone production
                                                  1995 Nashville ozone episode, enhancement
was essentially over (this situation is
                                                  of ozone above background due to local
characterized as one where the NOX reactivity
                                                  emissions appeared to dominate over regional
is exhausted). For the Johnsonville plume on
                                                  ozone enhancement.           Of course, the
June 23, ozone levels were enhanced by up to
                                                  characteristics of this episode may differ from
~25 ppb over a background of ~75 ppb at the
                                                  other episodes in the region. The reactivity of
point where NOX reactivity was exhausted.
                                                  NOX in power plant plumes released into the
This plume filled the mixed layer and plume
                                                  southern, daytime summer boundary layer
conditions were sampled at an altitude of
                                                  appears to be rather short-lived (30 to 100 km
about 1200 m M.S.L. Based on the plumes
                                                  downwind). Also, these observations suggest
followed in this field study, the authors
                                                  that ozone production is more efficient (in
concluded that NOX reactivity is exhausted
                                                  terms of NOX utilization) in urban plumes
(and ozone formation is over) within 30 to
                                                  than in power plant plumes
100 km downwind.
                                                  Policy Implications: If Nashville is
Limitations: The SOS Nashville data base
                                                  representative of other urban areas in the
appears to be rich and robust, so that
                                                  south, and if the July 11-15, 1995 episode
limitations of these findings are likely to be
                                                  which was the focus of the results described
related the number of events captured during
                                                  here is typical of ozone episodes in these
the study (4 weeks in 1994 and 6 weeks in
                                                  areas, then the results presented here suggest
1995). The July 11-15, 1995 Nashville ozone
                                                  that effective ozone control measures should
episode is only one event and may not be
                                                  focus on reducing local urban emissions. The
representative of all Nashville ozone episodes,
                                                  short-lived reactivity of NOX in power plant
thus not all Nashville episodes may be “home
                                                  plumes found in this study suggests that
grown.” The conclusions regarding the
                                                  during the day, the ozone formation generated
lifetime of NOX reactivity in power plant
                                                  by the plume is completed within about 100
plumes are for daytime meteorological
                                                  km of the source. On the other hand, plumes
conditions typical of the Nashville area in the
                                                  released at night may travel relatively longer
summer, and may not be applicable to other
                                                  distances before taking part in ozone
regions and/or times of year. Also, power
                                                  generating reactions the following day.
plant plumes released at night are generally
subject to less turbulent mixing and may
remain intact longer. These NOx plumes may

Summaries of Individual AQAWG Analyses                                                        18
Air Trajectory Analysis of Long-Term Ozone Climatology

Participants: Rich Poirot and Paul Wishinski, Vermont Department of Environmental
References: VT DEC Air Trajectory Analysis of Long-Term Ozone Climatology: Status Report
to OTAG Air Quality Analysis Workgroup, 12/3/96. R. Poirot and P. Wishinski, 3 December
1996 (
VT DEC Air Trajectory Analysis of Long-Term Ozone Climatology: Status Report to OTAG Air
Quality Analysis Workgroup, 11/7/96. R. Poirot and P. Wishinski, 7 November 1996
VT DEC Air Trajectory Analysis of Long-Term Ozone Climatology. Status Report to OTAG
AQA Workgroup: 8/15/96. R. Poirot and P. Wishinski, 15 August, 1996
Air Trajectory Residence Time Analysis Investigation of Ozone Transport Pathways: 1989-95.
P. Wishinski and R. Poirot, 20 February, 1996 Summary Draft

Purpose: Perform a back-trajectory based          resulting grid of residence times are displayed
analysis of relationships between high            as a gridded density plot called a residence
regional ozone levels and transport pathways.     time plot.
 Develop trajectory climatologies for             Trajectories ending at a given receptor can be
regionally representative receptor locations      sorted and grouped by various criteria. For
throughout the OTAG domain and use                concentration based sorting, trajectories are
trajectory sorting techniques to identify         sorted on the basis of the ozone concentration
predominant      synoptic-scale      transport    occurring at the time of their arrival at the
pathways when ozone is high at the receptors.     receptor (e.g., above average, above 60 ppb).
Methodology: The basis for the approach is        The resulting residence time plot addresses the
calculation of trajectories running backwards     question “if the concentration at this site was
in time from a selected receptor/time.            high (or low or very high), where did the air
Multiple trajectories for the same receptor but   come from?” Residence time probabilities are
different end-times (e.g., 9 a.m. on all summer   computed for each grid square as the ratio of
days) are assemble into families that define a    the total residence time in the square from a
trajectory climatology.         The trajectory    group of trajectories (e.g., those corresponding
climatology suggests where air arriving at the    to above average ozone at their end points) to
receptor within the selected family of end        the total residence time over all grid squares
times had previously resided. A technique         for the trajectory group. Such probabilities
known as “residence time analysis” is used to     tend to be highest in the grid squares closest to
display integrated results for large numbers of   the receptor since all (or nearly all) trajectories
back trajectories. The approach is to grid the    must pass through these grid squares
trajectory domain, keep track of the time each    regardless of the end point ozone
trajectory spends over each grid square along     concentration. However, by subtracting the
its path, and integrate the residence times for   residence time probability for a selected group
all trajectories over all grid squares. The       of trajectories from the residence time
                                                  probability for all trajectories, a gridded
Summaries of Individual AQAWG Analyses                                                            19
density plot of the incremental residence time     mean of 0 and standard deviation of 1. This
probability is obtained. Incremental residence     standardization technique was used to allow
time probability plots indicate the areas which    data from different hours of the day and from
are relatively more likely to be upwind under      different sites to be merged and sorted into
the selected set of conditions and thus            consistent subsets of the statistical distribution
eliminate the “near field” bias.                   (i.e. upper 50th, 10th or 5th percentiles). Z-
                                                   scores were used to define various subsets of
A different type of sorting is location based
                                                   trajectories and were also employed in a
sorting. Here, an average (or median) ozone
                                                   variety of inter-site correlations, time-lagged
level is calculated for each grid square as the
                                                   auto-correlations, and time-lagged inter-
weighted average concentration at the end
                                                   regional correlations - for comparison to the
points of all trajectories that passed through
                                                   trajectory results. For the inter-regional
the grid square. Trajectories are weighted
                                                   analyses, six subregions within the OTAG
according to their residence times in the grid
                                                   domain (New England, Mid-Atlantic,
square. Resulting plots display average
                                                   Northwest, Midwest, Southeast, and
concentrations at the receptor for prior airmass
                                                   Southwest) were defined on the basis of
locations on the grid and address the question
                                                   groups of monitoring sites in close geographic
“if the air had previously been here (or there),
                                                   proximity with high inter-site correlations.
what‟s the average concentration at the
receptor?”        Typically, these average         Findings: For the complete trajectory
concentrations are expressed in terms of           ensemble (i.e. unsorted), eastern receptors
normalized deviations from the receptor mean       such as Whiteface Mountain (NY) and Look
concentration. Location based sorting was          Rock (TN) have 2 to 4 day back trajectories
used to draw maps showing the locations of         covering almost the entire OTAG domain,
grid squares where the residence time              suggesting the potential for transport of air
weighted average end point concentrations          parcels to these high elevation sites from
were above or below the average receptor           across the whole OTAG region. Residence
concentration by varying amounts.                  time analyses of trajectories ending at both the
Back-trajectories were calculated using the        high and low elevation sites sorted for high
NOAA HY-SPLIT model with archived wind             end point ozone levels (based on normalized
fields from the NGM weather forecasting            deviations from the receptor site means)
model (Draxler, 1992; Phillips, 1975).             suggest high ozone is generally associated
Trajectory end times examined were 3 am, 9         with transport from the heart of the OTAG
am, 3 pm, and 9 pm for June-August 1989-95.        domain rather than from around the edges.
 Results are presented for six high elevation      Similarly, location based sorting of
sites in the Appalachian Mountain chain from       trajectories indicates that trajectories which
Look Rock, TN to Whiteface Mountain, NY.           traversed areas within the central portion of
Results are also presented for 19 low elevation    the OTAG domain produced higher end point
sites in an area ranging (approximately) from      ozone concentrations on average, regardless of
the Gulf Coast to the Canadian border and the      the location of the receptor. The incremental
Mississippi River to the Atlantic Coast.           probability plots for many receptors or groups
                                                   of receptors show that, as the analysis is
In most of the analyses, raw ozone data were       increasingly restricted to the highest ozone
transformed to “Z-scores”, based on geometric      concentrations, the area of highest incremental
mean and standard deviations - computed            probability becomes increasingly focussed on
separately for each site, for each hour of day.    an area around the Ohio River Valley. Within
All resultant Z-score distributions have a         the southern portion of the OTAG domain,
Summaries of Individual AQAWG Analyses                                                            20
trajectories associated with above average         end point ozone concentrations do not
ozone tend to extend a smaller distance from       necessarily indicate a cause and effect
the receptor than in the northern portion of the   relationship between emissions in a given area
domain, indicating a relatively greater            and end point ozone concentrations - these
influence of stagnation in the South.              relationships may simply be an expression of
                                                   the prevailing meteorological conditions under
The finding of relatively more stagnation in
                                                   which high ozone concentrations at the end
the south is supported by an analysis of the
                                                   point receptor are favored.
autocorrelation in ozone concentrations (mean
3:00 pm Z-scores) within each of the six           Because all the analyses are limited to
subregions within the OTAG domain:                 regionally representative receptors, the high
autocorrelations at one-day lag times were         ozone levels observed are often below the
above 0.4 in all subregions, while                 ozone NAAQS. This fulfills the objective of
autocorrelations remained well above 0.2 even      identifying transport conditions for high
at a three-day lagtime in the Southwest and        regional ozone levels, but this study does not
Southeast subregions.          In contrast,        specifically address how such conditions
autocorrelations dropped below 0.1 at two-day      relate to transport conditions during ozone
lagtimes in the New England, Mid-Atlantic,         NAAQS exceedances. It must also be
and Northwest subregions.                          recognized that these analyses are limited by
                                                   uncertainties and/or biases in the back
Limitations: Residence time analysis of
                                                   trajectory calculations, and these uncertainties
trajectory climatologies can suggest regional
                                                   can be expected to grow as the trajectories are
airflow patterns associated with high ozone
                                                   computed further back in time. Trajectories
regimes at a receptor, but without information
                                                   computed from NGM wind fields with
on the age of the elevated ozone reaching the
                                                   relatively coarse vertical resolution are not
receptor, inferring the location (i.e., distance
                                                   likely to reflect local or mesoscale flow
upwind) of the emissions that caused the
                                                   features such as low-level nocturnal jets, sea
elevated ozone is problematic; the trajectory
                                                   breezes, and other small scale flow features
results do not account for chemical
                                                   likely to be important to ozone and precursor
transformation or deposition processes. These
                                                   transport under ozone episode conditions.
problems apply to all trajectories but become
                                                   Some trajectory uncertainties have been
particularly acute for trajectories extending
                                                   investigated by Schichtel and Wishinski
back more than about 12 hours. Location
                                                   (1996) by comparing alternate trajectory
based sorting seems as if it may address this
                                                   models, and these uncertainties do not appear
issue, but it is difficult to know how to
                                                   to exert a major influence on the results
interpret the plots of location sorted
                                                   presented here. Although some trajectories
trajectories. One difficulty is that location
                                                   are calculated for as long as four days at
plots sorted for high and low ozone frequently
                                                   which time the uncertainties are relatively
show no contribution from grid squares
                                                   large, most trajectories are truncated at 2 to 3
immediately upwind of the receptor. These
                                                   days for various reasons. Furthermore, the
squares tend to be classified as average
                                                   authors indicate that they are unaware of any
because many trajectories of both high and
                                                   sources of bias in the analyses (nor were any
low ozone pass through them The effects of
                                                   biases found by Schichtel and Wishinski) and
this „far-sighted‟ bias in the location based
                                                   therefore that results averaged over a large
sorting have not been estimated. These
                                                   number of trajectories calculated under a
limitations imply that relationships between
                                                   particular set of conditions should provide a
the patterns of trajectory residence times and
                                                   reliable indication of the mean trajectory
Summaries of Individual AQAWG Analyses                                                          21
pattern under those conditions. Ultimately,        confirmed solely on the basis of these studies
statements about potential uncertainty/bias        given the limitations of the methodology
should be related to the spatial scales of ozone   discussed above.
transport patterns inferred from the results.
                                                   Additional References:
Scientific Implications: Results of both           Draxler, R.R. 1992: Hybrid Single-particle
location based and concentration based sorting     Lagrangian Integrated Trajectories (HY-
of trajectories reveals the locations favored by   SPLIT): Version 3.0 - User’s Guide and
trajectories associated with above average         Model Description.       NOAA Technical
ozone concentrations at the trajectory end         Memorandum ERL ARL-195, Air Resources
points (receptors). The central portion of the     Laboratory, Silver Spring Maryland.
OTAG domain is such a favored location for
receptors in many different locations              Phillips, 1975: NOAA Technical Report
throughout the eastern U.S. As noted above,        NWS-22
however, these results do not indicate the         Rolph, 1992: NGM Archive. NCDC Report
relative contributions of different source         TD-6140, National Climatic Data Center,
regions to ozone levels at the receptor site.      July, 1992.
Policy Implications: Emission controls in the      Schichtel, B. and P. Wishinski, 1996. HY-
areas suggested by the trajectory calculations     SPLIT and CAPITA Monte Carlo Model Back
to be consistently upwind of receptor sites        Trajectory Comparison. Summary Draft, 1
experiencing high regional ozone levels may        July                                    1996
be effective in reducing regional ozone levels.    (
 For many of the receptor sites examined           p/trajcomp.html)
throughout the eastern U.S., the area
implicated is centered on the Ohio River
Valley. However, this conclusion cannot be

Source Regions of Influence for High and Low Ozone Conditions in the
Eastern US

Participants: Bret Schichtel and Rudolf Husar, Washington U., St. Louis, MO.
References: Update on the Characterization of Transport over the Eastern US. Schichtel and
Husar, February 1, 1997.
Source Regions of Influence for High and Low Ozone conditions in the Eastern
U.S.,Schichtel and Husar, 8/08/96 Summary Draft.

Purpose: Develop a plume based method for          transport patterns during the summers (June _
investigating the spatial impact of ozone and      August) of 1991_95 for both average
ozone precursors from selected source areas        conditions and periods of high and low ozone
(the source region of influence). Apply the        concentrations, as well as during the 1991,
technique to the Eastern US to characterize        1993, and 1995 OTAG episodes.
potential source regions of influence and          Method:     Forward particle plumes were
Summaries of Individual AQAWG Analyses                                                        22
calculated using the CAPITA Monte Carlo             manner. The results can be used to
Model (Schichtel, 1995) with archived wind          qualitatively interpret relative changes in the
fields from the NGM weather forecasting             size of regions of influence under different
model. The forward plumes represent                 conditions. Results aggregated over all days
simulations of air parcel transport away from a     (June_August in 1991-95) suggest two
source, showing the direction and speed of the      characteristic transport regimes over the
transport. Ranges of influence downwind             Eastern US. One is from Texas north, through
(“source regions of influence”) were estimated      the Midwest and then east through New
by imposing one or two day lifetimes on the         England to the Atlantic. The other regime is in
released particles. By aggregating trajectories     the Southeast with shorter transport distances.
for multiple days the authors obtained              These regimes are consistent with summer
climatological transport directions, air parcel     mean wind fields over the eastern U.S. which
residence times, and regions of influence.          are dominated by the Bermuda High pressure
Naturally, the region of influence was larger       system. Mean residence times increase
for an assumed two day lifetime than a one          approximately 50% from the northern half of
day lifetime. Two methods were used for             the OTAG domain (~0.2 s/m) to the southern
visualizing the results. In the first method, six   half (~0.3 s/m).
sites were selected throughout the OTAG             Aggregating results for days with the highest
domain and the source region of influence           and lowest 10 percent average daily maximum
around each site plotted. The region of             ozone at each grid point showed that high
influence is elongated along the predominant        ozone in the Southeast is associated with
transport direction(s) and, viewed together, the    longer residence times indicating relatively
six regions of influence convey a sense of the      stagnant air masses with net transport from the
main transport directions and the one/two day       north. Low ozone is associated with shorter
regions of influence. In the second method,         residence times and more persistent transport
504 sources were evenly distributed on a grid       from the Gulf of Mexico and the Atlantic
covering most of North America and one day          Ocean. In the Midwest, high ozone conditions
regions of influence were calculated around         are also associated with relatively long
each source. A vector was plotted at each           residence times indicating stagnant conditions
source showing the characteristic (i.e.,            under which a high pressure system is located
predominant) transport direction where the          over the Midwest. Low ozone in the Midwest
length of the vector was a measure of the
                                                    is associated with shorter residence times and
persistence of transport in the direction of the    more persistent, rapid flow from the West. In
vector. The resulting plot has the appearance       contrast, high ozone episodes in New England
of a gridded wind field plot. In addition,          are associated with persistent, relatively rapid
residence times were computed as the inverse        transport (relatively short residence times)
of the characteristic transport speed. The          from the west-southwest whereas low ozone is
characteristic transport speed was in turn          associated with transport primarily from the
estimated by dividing the radius of circle with     northwest. Overall, residence times are lower
area equal to the source region of influence by     than average during the low ozone events with
the assumed pollutant lifetime.                     transport from outside of the OTAG domain
Findings: The source regions of influence are       inward. While the high ozone events have
highly dependent upon the pollutant lifetime,       longer than average residence times with
and, as the lifetime of ozone is difficult to       transport from the center of the OTAG
characterize, it is not possible to interpret the   domain outward.
source regions of influence in a quantitative
Summaries of Individual AQAWG Analyses                                                           23
Calculations of residence times and transport      instance, based on the study findings, a
vectors during each of the 1991, 1993, and         plausible scenario for significant transport
1995 OTAG episodes shows different spatial         from the Midwest to New England might be a
patterns with some similarities between the        stagnation event in the Midwest followed by
1991 and 1995 episodes. The 1993 episode is        the onset of organized transport from the
characterized by long residence times in the       Midwest to the New England, but it is unclear
south as was the case for the other two            whether the analyses performed would be able
episodes but also by shorter than average          to detect such an interaction. The potential
residence times in much of the rest of the         effects of trajectory uncertainty and bias
domain.                                            should be discussed in the context of the study
Limitations: The techniques developed are
quite effective at illustrating climatological     Implications: This study suggests the
transport patterns, but less effective at          potential for transport from the Gulf Coast to
showing regions of influence because it is         the Midwest and the Midwest to the
unclear what lifetime to assume. The analyses      Northeast, but does not provide information
do not shed light on the extent to which           on whether such transport actually occurs.
transport actually was important under             High ozone events in the southeast and
different conditions, they can only address the    Midwest appear to be associated with
potential for transport. The vector plots focus    relatively stagnant conditions, which suggests
on the predominant transport direction, but the    that ozone levels in theses areas are less
region of influence plots show significant         strongly influenced by inter-regional transport.
potential for transport in different, even          However, the net transport under these
opposing directions. Therefore, as noted by        conditions is from the Midwest suggesting
the authors, the vector plots may obscure          potential contributions to high ozone in the
important details of the transport patterns. For   South.

Ozone/Tracer/NOY Relationships at Three SOS-SCION Sites

Participants: Eric S. Edgerton and Benjamin E. Hartsell, ESE Environmental Inc. (Currently
              affiliated with Atmospheric Research and Analysis Inc.)
References: Slides from presentation given to the OTAG-AQA meeting, Washington D.C., July
              1996 and written summary posted to the OTAG-AQA web site on November 4,

Objective: Identify the sources contributing       sites.
to elevated ozone at SCION sites in the South      Methods: Data for the 1995 ozone season
by using CO and SO2 as tracers to distinguish      (may to October) from three sites in the
between contributions from urban areas and         SCION network in the rural south were
major point sources. Investigate the spatial       analyzed. The sites were: Oak Grove (about
extent of elevated ozone events by looking at      120 km NW of Mobile, AL), Centreville
the coincidence of elevated ozone at different     (about 85 km SW of Birmingham, AL) and
Summaries of Individual AQAWG Analyses                                                          24
Yorkville (about 65 km NW of Atlanta, GA).        mixed > major point.
The measurements were for O3, NOY, CO and
                                                  Urban and point source plumes could be
SO2 and were for 1 minute averaging times.        distinguished at these sites because key tracer
Elevated ozone days were defined as those         species had high degrees of covariance. This
having 8-hour average ozone greater than 80       suggests that the plumes were sufficiently
ppb. These days were classified as having         recent in origin to retain their physical
high ozone due to urban emissions, major          identity, i.e., they had not yet been dispersed
point source emissions, or mixed (i.e., periods   into the regional airmass. The age of the
of both urban and major point source impacts      plumes can‟t be estimated directly from these
within the same day) by examining CO/NOY          data, however, the locations of the monitoring
and SO2/NOY ratios. Periods with relatively       sites relative to nearby major urban areas
high SO2 and low CO were attributed to major      suggests that the urban plumes must have
point sources, whereas periods with relatively    been transported at least 65 to 120 km to
low SO2 and high CO were attributed to urban      influence these sites.
emissions. An 80 ppb background for CO
was subtracted before calculating the CO/NOY      Limitations: The implications of the results
ratios.                                           are limited in the number of sites (3), the
                                                  geographic distribution of sites (rural
Findings: All three sites observed elevated       southeast) and the number of years
ozone days in the 1995 season, but there were     investigated (1995). The unfortunate paucity
fewer at Oak Grove (5) than Centreville (19)      of data required for this type of analysis
or Yorkville (29). Elevated ozone was             prevents generalization to other parts of the
observed at one or more site(s) on 39 days, of    OTAG domain. On the other hand, the data
which 13 involved 2 sites and one involved all    considered appear very rich mainly by virtue
three sites. Therefore, about two thirds          of the high temporal resolution (1-minute)
(25/39) of the elevated ozone days involved
                                                  which permit robust inter-species correlation
only a single site, suggesting that on these      analyses. The high degrees of covariance
days the spatial scale of elevated ozone was      between species during elevated ozone periods
less than the distance between any two sites      may suggest that the plumes were of recent
(250-300 km).                                     origin, but there is no way to estimate how
A strong relationship between O3 and NOY          recent or therefore what range of influence is
was observed at all sites on clear sunny days.    implied.
There was some scatter in the relationships       The selection of elevated ozone days was
and there were instances of departure from the    based on an 8-hour period which may tend to
predominant relationship. There was evidence      focus away from days that had short term
of non-linearity in the relationship with a       impacts. Point source plumes might be
lower slope of O3/NOY at higher NOY (5-10         expected to result in shorter term impacts than
ppb) than lower NOY (1-5 ppb).                    urban plumes by virtue of there smaller
It was possible to distinguish periods of urban   (narrower) spatial scale, and this is born out
and major point source plume impacts at all       by some of the data presented by the authors.
three sites. At Centreville and Yorkville,        Therefore, the basis for selecting elevated
there were elevated ozone days classified         ozone days may influence the relative
urban, major point and mixed; but at Oak          numbers of days classified as being urban,
Grove all five elevated ozone days were           point or mixed influence.
classified as mixed. Across the network the       Since the Oak Grove site is much closer to the
rank order of classifications was urban >         Gulf Coast than the other two sites, it can be
Summaries of Individual AQAWG Analyses                                                        25
expected to have a quite different ozone          strongly suggest that both urban emissions and
climatology due to the influence of marine air.   point sources can cause elevated ozone levels
 Therefore, looking at the simultaneous           at these three sites. Furthermore, urban
occurrence of elevated ozone at all three sites   emissions played a more dominant role than
may not be a good way of examining the            major point source emissions for the kind of
regionality of high ozone episodes in South.      days selected (8 hour O3 > 80 ppb) at these
Looking for concurrent events at Centreville      three sites.
and Yorkville may provide a somewhat more
                                                  Policy Implications: For the three southern
reasonable estimate of frequency of periods of
                                                  sites considered here, the majority of days
widespread high ozone in the south. Better
                                                  with 8-hour ozone exceeding 80 ppb were
still would be to use the more extensive multi-
                                                  associated with urban signatures, sometimes
year database of routine ozone measurements.
                                                  in combination with point source signatures.
Scientific Implications:     The analyses         There were also some days when 8-hour
performed in this study provide a good            ozone exceeded 80 ppb that had
example of how tracer SO2/CO/NOY analyses         predominantly point source signatures. These
can be used to distinguish point and urban        findings suggest that controls on urban area
source signatures in relation to ozone            emissions may be more effective at reducing
formation. Much of the power of the tracer        the frequency of such exceedances than
analyses stems from the high temporal             controls on point source emissions for this set
resolution of the SCION data. The data            of sites and this type of ozone episode.

Analysis of Ozone, NOY and Tracer Data from a Site in South-Central

Participants: Eric S. Edgerton and Benjamin E. Hartsell, ESE Environmental Inc. (Currently
              affiliated with Atmospheric Research and Analysis Inc.)
References: Written summary posted to the OTAG-AQA web site on January 6, 1997

Objective:      Identify types of sources         200 m of the site. The data include
contributing to elevated ozone at a rural site    continuous measurements of O3, NO, NOY,
toward the West of the Ozone Transport            CO and SO2 reported at 1, 15 or 60 minute
Region (OTR) by using CO and SO2 as tracers       averaging times recorded between June 17 and
to distinguish between contributions from         September 30, 1995. Elevated ozone days
urban areas and major point sources.              were defined as those having 8-hour average
                                                  ozone greater than 80 ppb. These days were
Methods: The Arendtsville site is located         classified as having high ozone due to urban
about 90 km Northwest of Baltimore in a rural     emissions, major point source emissions, or
area toward the West of the OTR. The site is      mixed (i.e., periods of both urban and major
located on top of a small hill and there is no    point source impacts within the same day) by
major point source of SO2 or NOx is within 40     examining CO/NOY and SO2/NOY ratios.
km of the site, and no secondary road within      Periods with relatively high SO2 and low CO

Summaries of Individual AQAWG Analyses                                                        26
were attributed to major point sources,          be used in extrapolating the results for
whereas periods with relatively low SO2 and      Arendtsville to the Western OTR in general.
high CO were attributed to urban emissions.      The selection of “high ozone days” was based
An 80 ppb background for CO was subtracted       on an 8-hour period and a threshold of 80 ppb.
before calculating the CO/NOY ratios. The         An analysis for ozone events that are of
cut-points for assigning episode types were:     shorter duration but higher concentration (e.g.,
                                                 1 hour, 120 ppb) might reflect a different mix
                      SO2/NOY         CO/NOY     of sources.
       Urban          >10             <0.5       Six of the 21 high ozone days were classified
       Mixed          >5              >0.5       as likely urban dominated using a criterion of
       Major Pt. Source<5             >1.5       SO2/NOY < 0.5, but 5 of these 6 days had
These cut-points were used to determine if       SO2/NOY ratios that were between 0.4 and
one type of source was probably the dominant     0.5. Thus, a relatively small change in this
contributor, meaning that the source type        classification criterion to SO2/NOY < 0.4
could have provided at least 75 percent of the   would result in only 1 of 21 days being
NOY. This analysis is similar to one             classified as urban with the remainder all
performed by the same authors for three          classified as mixed. Similarly, several of the
southern sites and described in the summary      “mixed” days exhibited SO2/NOy ratios of
“Ozone/Tracer/NOY Relationships at Three         about 0.6 and CO/NOy ratios greater than 10
SOS-SCION Sites.”                                such that, if the criterion for SO2/NOy ratios
                                                 were a little higher, there would be a few more
Findings: Of the 101 days in the 1995 field
                                                 days classified as urban.
season, 21 had 8-hour average ozone greater
than 80 ppb. As in the analysis of data from     It appears that high ozone days at Arendtsville
the SOS-SCION sites, the authors found that      usually are influenced by a mix of urban and
ozone was correlated with NOY levels on high     point sources. The authors investigated
ozone days, and that there were correlations     several approaches to estimating the relative
between NOY and CO and or SO2. The high          contributions of these source types by
temporal resolution in the concentration data    analyzing ratios of pollutant concentrations,
greatly enhance the ability to discern such      but the analyses are complicated by the need
correlations. However, compared to the SOS-      to account for the effects of atmospheric
SCION sites, there were fewer periods that       removal on the CO, SO2 and NOY
were clearly dominated by one type of source     concentrations measured at Arendtsville.
in the Arendtsville data. Most of the days       These      analyses     contain     significant
were diagnosed as having “mixed” influence       uncertainties because the age of emissions
(i.e., both urban and major point source         impacting Arendtsville, and the atmospheric
emissions contributed to the high ozone), a      lifetimes of NOY, SO2 and, to a lesser extent,
few (six) were diagnosed as urban dominated      CO, in the surrounding environment are hard
and none were diagnosed as major point           to quantify.
source dominated. Thus, even for a site at the   Scientific Implications:        The analyses
western edge of the OTR, urban emissions         performed in this study provide a good
were associated with all 8-hour averages > 80    example of how tracer SO2/CO/NOY analyses
ppb.                                             can be used to investigate the contributions of
Limitations: The implications of the results     point and urban sources to ozone formation at
are limited because only one site in the         a site where high temporal resolution data are
Western OTR was evaluated. Caution should        available. Data for Arendtsville, in the

Summaries of Individual AQAWG Analyses                                                        27
western OTR, indicate that a mix of urban and   influenced the Arendtsville site.
point source emissions causes elevated ozone
                                                Policy Implications: For Arendtsville, almost
levels. Urban emissions appear to be involved
                                                all days with 8-hour ozone exceeding 80 ppb
in all episodes, while point source emissions
                                                were associated with mixed urban and point
appear to be involved in the majority of
                                                source signatures, manifested by characteristic
episodes. The O3 to NOY ratio measured at
                                                ratios of CO and SO2 to NOy. This suggests
Arendtsville was about 4, which suggests that
                                                that controls on urban or point source
ozone was formed under conditions that were
                                                emissions have the potential to reduce
VOC limited or intermediate between VOC
                                                elevated ozone levels at Arendtsville. Given
and NOX limited. Ratios at Arendtsville
                                                the rural nature of this site, as well as its
should be compared to other sites in the
                                                location in the western OTR, it is reasonable
western OTR to investigate whether the
                                                to expect that the urban character of episodes
relatively low values measured are
                                                would be more dominant in most other parts
representative of air masses transported into
                                                of the OTR.
the OTR or whether local emissions may have

A Comparison of Modeled and Measured Ozone, NOY and CO at Nine
Regional Monitoring Stations during the 1995 OTAG Episode

Participants: Benjamin E. Hartsell and Eric S. Edgerton, ESE Environmental Inc. (Currently
              affiliated with Atmospheric Research and Analysis Inc.)
References: Written summary posted to the OTAG-AQA web-site on November 24, 1996.

Objective: Use data from 9 rural monitoring     “large clearings with excellent fetch and were
stations to evaluate the OTAG UAM-V model       well removed from the influence of local
performance for the 1995 episode. The           sources.” Model predicted concentrations for
questions addressed are how does the model      the site locations were from the OTAG D2
chemistry perform (in terms of O3/NOY) and      base case. 1-hour averages of observed and
how well does the model predict observed O3,    modeled data were compared.
NOY and CO levels. Additional analyses          Findings: Scatter plots of O3 against NOY
currently in progress will include additional   across all sites and all hours between 1100 and
measurements and look at some issues in         1900 LST were prepared for the observations
more detail.                                    and model predictions. The plots showed
Methods: Air quality data were assembled        significant variation in the relationship for
for the 1995 episode (July 7-18) from four      both the observations and model predictions,
southern sites in the SOS/SCION network         however the general appearance of the plots
(sites in MS, AL, GA and TN) and five           was similar. When the data were sorted by
northern sites from the NARSTO-NE study         ozone concentration and put into ten bins
(sites in PA, NY and MA). All stations          containing equal numbers of points, linear
measured O3 and NOY and four measured CO.       relationships between O3 and NOY emerged
 All sites are described as being located in    for both the observations and the model.
                                                Within each bin, the distribution of NOY
Summaries of Individual AQAWG Analyses                                                       28
values tails toward high values and so this      generally predicted better for the southern site
analysis tends to emphasize the lower NOY        in MS (close to the Gulf) and for the one
values that likely represent cases where ozone   northern site in PA. The time series for NOY
is most strongly NOX limited. Therefore, this    showed generally good agreement with some
analysis effectively compares the modeled and    tendency for the model to predict higher
observed O3/NOY relationships (i.e., the ozone   maximum NOY values than were observed for
yield per NOX) under NOX limited conditions,     the northern sites.
finding good generally agreement between the
                                                 Limitations: The paper acknowledges several
model (slope 8.7) and observations (slope
                                                 limitations that apply generally to model
9.2). There was more scatter in the modeled
                                                 performance evaluations of this type: (1)
O3/NOY relationship than in the observed
                                                 observations at single points are compared to
relationship, which is consistent with the
                                                 model predictions that are volume averages
model over predicting the frequency of fresh
                                                 over large grid cells. Grid cells were about 12
NOX emission impacts at these sites.
                                                 km by 12 km by 50 m at most sites, but 36 km
Plots were prepared of bias (modeled -           by 36 km by 100 m for the MS site. (2) The
measured) against modeled values with points     number of sites and days available for
averaged into 10 bins of equal numbers of        comparison is limited. Relative to other
points. These plots are difficult to interpret   analyses of model performance for OTAG
without scatter plots of the underlying data.    episodes, this limitation is more severe for
For instance, the plots of ozone residuals       ozone than the other species because there are
against modeled ozone and ozone residuals        a large number (hundreds) of other ozone sites
against measured ozone are quite different:      available but only a small number of NOY and
one plot suggests model over predictions of      CO sites are available.
about 20 ppb for the highest bin while the
                                                 The modeled NOY does not include organic
other plot suggests model under predictions of
                                                 nitrates (Carbon Bond species NTR) because
about 5 ppb for the highest bin. A clearer
                                                 NTR concentrations were not saved in the
picture is obtained from analyses of ozone
                                                 OTAG model runs. Thus, modeled NOY
values exceeding thresholds of 80, 100 and
                                                 concentrations are always biased low.
120 ppb. For these rural sites, the modeled
                                                 However, the magnitude of this bias is not
and observed ozone exceeded 120 ppb in only
                                                 well known and likely varies in time and
a few instances and did not exceed 100 ppb at
                                                 space depending upon the age of NOY (the
all sites. This shows that ozone levels above
                                                 bias might be as high as 20 percent for aged
100 ppb represent the upper tail of the
                                                 air masses). Accounting for the missing
distribution of ozone values for these sites.
                                                 contribution of NTR to modeled NOY would
The ratio of modeled to observed hours above
                                                 increase the model under estimation of the
80 and 100 ppb was 342/259  1.3 and 102/59      O3/NOY relationship (perhaps by up to 20
 1.7, respectively. The ratio of modeled to     percent).
observed days with values exceeding 80 and
100 ppb was 47/39  1.2 and 19/11  1.7,         The statistical analyses averaged data across
respectively. This shows a tendency for the      all sites so that there is no indication whether
model to over-estimate the frequency and         significant differences existed for individual
duration of high ozone values at these sites.    sites or between groups of sites (e.g., North
                                                 vs. South).
The time series for CO showed a tendency for
over prediction of CO concentrations for two     Scientific Implications: The comparison of
southern sites in GA and TN. CO levels were      modeled and observed O3/NOY relationships
                                                 revealed good agreement between the model
Summaries of Individual AQAWG Analyses                                                  29
(slope 8.7) and observations (slope 9.2).            model performance problem. The apparent
These values can be interpreted as estimates of      tendency for the model to over predict CO at
the ozone yield per NOX under NOX limited            two sites in GA and TN may be indicative of
conditions. However, the model value is              emissions or atmospheric mixing problems in
biased high by an uncertain amount because           this region, and should be investigated.
organic nitrates were not included in the
                                                     Policy Implications: Analyses for nine rural
model NOY values: If possible, the magnitude
                                                     sites in the South and the Northeast indicate
of this bias should be investigated to
                                                     that the model has a tendency to over predict
determine whether it significantly influences
                                                     the frequency and duration of ozone levels in
this finding. Also, the comparison of modeled
                                                     80-100 ppb range. If confirmed by more
to measured O3/NOY relationships suggests
                                                     extensive analyses, this finding would indicate
that the model may be over predicting the
                                                     a potential problem with using OTAG results
frequency with which fresh NOX emissions
                                                     to provide boundary conditions for urban scale
impact these sites.
                                                     modeling. The good performance of model
For these nine rural sites, there was a tendency     chemistry for O3/NOY relationships under
for the model to over-estimate the frequency         NOX limited conditions suggests that overall
and duration of high ozone values: this should       there is no gross problem with the model
be investigated for other rural sites to find        chemistry.
whether it is indicative of a more general

Ambient Monitoring Sites for OTAG (time series) Model Evaluation

Participants: Rich Poirot, Vermont Department of Environmental Conservation. Vermont Air
Program, Building 3 South, 103 S Main Street, Waterbury, VT 05676
Reference: Ambient Monitoring Sites for OTAG (time series) Model Evaluation, 11 January 1996
memo from Rich Poirot, Vermont Department of Environmental Conservation to S.T. Rao and Rudy
Husar Available via WWW at

Purpose: Recommend ozone monitoring sites            carbonyls; and (4) Good spatial coverage of the
in the Northeast to be used for evaluation of the    modeling domain. The author noted that these
OTAG base case regional photochemical model          four criteria for site selection are to some degree
simulations for the July 1988, 1991, 1993, and       in conflict with each other. For example, the
1995 episodes.                                       PAMS data sites have more complete chemical
                                                     species measured but are more urban oriented
Methodology: Four ideal objectives were              (i.e., aligned along the Northeast Corridor) and
identified for site selection: (1) Sites should be   primarily have data available only for the 1995
spatially representative of approximately 12 x       episode although some data are available for the
12 km2 model grid squares; (2) Data should be        1993 episode as well.
available for all OTAG model episodes; (3)
Data should include hourly measurements of           Findings: Given the conflicting objectives
ozone, NOx, and speciated hydrocarbons and           noted above, the author developed a list of 21

Summaries of Individual AQAWG Analyses                                                             30
ozone sites in the OTR which meet objectives          these areas would throw into question the
1, 2, and 4 (with some NOx measurements) and          accuracy of model transport estimates.
an additional set of 13 PAMS/NARSTO sites
                                                      Scientific Implications: Statistics summarizing
where more complete chemical species data are
                                                      model performance are likely to be highly
available but which primarily operated only
                                                      dependent on the choice of sites to include in
during the 1995 episode (with some data
                                                      the analysis. The author is able to identify a
possibly available at Whiteface Mtn, NY and
                                                      number of predominantly rural sites that meet
Wye River and Shenandoah NP, VA for 1993).
                                                      his criteria. However, as noted above, OTAG
 The author suggests that if this list is too long,
                                                      must also be concerned with model
priority should be given to: a)sites above 500
                                                      performance in and near nonattainment areas.
meters; b)sites with NOx data; and c)rural
PAMS sites.                                           Policy Implications: Meeting OTAG‟s goals
                                                      requires development of photochemical model
Limitations: Although the site selection
                                                      simulations that meet acceptable performance
approach used in this study provides a means of
                                                      criteria. Appropriate selection of data sets to
selecting optimal sites for evaluating the
                                                      be compared with model predictions is critical
rural/regional model predictions, the emphasis
                                                      to the calculation of model performance
on rural sites with broad spatial
                                                      statistics relevant to OTAG‟s goals. This study
representativeness (e.g., sites above 500 m or
                                                      has identified monitoring sites in the
rural PAMS sites) tends to preclude
                                                      northeastern portion of the OTAG domain that
recommendation of sites in nonattainment
                                                      can be used to summarize model performance
areas. As the fundamental purpose of OTAG is
                                                      in rural and remote areas. Selection of sites in
to identify options for reducing the transport of
                                                      nonattainment areas and in other portions of the
ozone and precursors into nonattainment areas,
                                                      OTAG domain are not addressed here.
OTAG must also be concerned with model
performance on the edges of and within
nonattainment areas since poor performance in

Comparison of OTAG UAM-V/BEIS2 Modeling Results with Ambient
Isoprene Observations

Participants: Eric Edgerton and Ben Hartsell, Atmospheric Research & Analysis, Inc.; Ralph
Morris, Keane Lee, and Greg Yarwood, ENVIRON International Corporation.
References: Comparison of Modeled versus Observed Isoprene Concentrations at Rural and
Suburban        Sites     Across  the      Eastern   U.S.         E.    Edgerton,      1997
Phase I Comparison of OTAG UAM-V/BEIS2 Modeling Results with Ambient Isoprene and Other
Related Species Concentrations. R. Morris, K. Lee, and G. Yarwood, 31 January 1997

Purpose: Estimates of biogenic emissions used         emissions are significantly higher than those
in the OTAG UAM-V modeling are based on               generated by its predecessor model, BEIS1,
the BEIS2 emissions model. BEIS2 isoprene             resulting in a biogenic emissions estimate (of

Summaries of Individual AQAWG Analyses                                                           31
which isoprene is the largest single component)       Pennsylvania (one). Other portions of the
that constitutes 70 - 80 percent of the total VOC     OTAG domain (notably the Midwest, Mid-
emissions in the OTAG modeling domain.                Atlantic region, Gulf Coast outside of LA,
Thus, control strategy impact estimates obtained      southern Florida, and Texas) are not well
from the OTAG modeling depend critically on           represented by these databases. Edgerton‟s
the accuracy of the BEIS2 biogenic emission           analysis concentrated primarily on 16 rural and
estimates. Two independent analyses of the            suburban sites while Morris et al. focussed on
available ambient data were conducted to              15 urban/suburban (or urban influenced) sites.
evaluate the UAM-V/BEIS2 isoprene                     The two studies have six sites in common. At
concentration predictions.                            some sites, (primarily PAMS and NARSTO-
                                                      NE) observations were made every hour by
Method: UAM-V/BEIS2 predictions of
                                                      automated gas chromatographs (auto-GCs),
ambient isoprene concentrations in surface layer
                                                      while at others, integrated canister samples
grid cells (roughly 12 km by 12 km horizontally
                                                      were collected at set times. For the auto-GC
by 50 m in the vertical) corresponding to
                                                      data, comparisons with model predictions were
ambient VOC monitor locations were compared
                                                      based on 11:00 - 17:00 LST averages (11:00 -
with       observed      afternoon       isoprene
                                                      19:00 LST in the Morris et al. study); at the
concentrations for the July, 1995 OTAG
                                                      canister sites, afternoon integrated canister
episode. Isoprene observations were obtained
                                                      samples were compared with model predictions
from      ambient     speciated      hydrocarbon
                                                      for the corresponding time intervals. In this
measurements made at PAMS network
                                                      way, comparisons were restricted to the period
monitoring sites as well as sites operated as part
                                                      of maximum isoprene emissions and turbulent
of the NARSTO-Northeast and SOS-Nashville
                                                      diffusion when comparisons of surface-based
Intensive field studies. These sites are primarily
                                                      observations (5 - 15 m agl) with grid cell
concentrated in the Northeast and in the
                                                      (volume average) predictions are most
Tennessee-Kentucky areas where the two field
                                                      meaningful. All comparisons were done on a
studies were conducted in 1995. Additional
                                                      site-by-site (spatially matched) basis.
sites are located in Virginia (one), Georgia
(one), Baton Rouge, LA (two), and western
Table 1. Summary of bias and gross error results.

                                  Max. Bias3                         Gross Error4

                           Under-         Over-             Mean                    Range
                          prediction    prediction
 Morris et al.            -60 %         243 %        7.7 ppbC               2 - 21 ppbC

 Edgerton                 -71 %         245 %        10.7 ppbC              1.1 - 34.9 ppbC

     In the Morris et al. study, bias is defined as the average of the differences (predicted minus
observed) expressed as a fraction of the mean observed value; in the Edgerton study, bias is defined
as the mean percent difference.
     Gross error is defined as the average absolute difference (predicted minus observed) in both
Summaries of Individual AQAWG Analyses                                                            32
Distributions of the biases across sites are summarized in Table 2.

Table 2. Distribution of bias across sites.5

                  Within ±25%        Within ±50%       Under prediction Over      prediction
                                                       greater than 50% greater than 50%

                  No. of    % of No. of % of No. of % of No. of                       % of
                  Sites     Sites Sites Sites Sites Sites Sites                       Sites

 Morris et al.    2         13%      8         53%     1          7%       6          40%

 Edgerton         2         12%      6         38%     1          6%       9          56%

   In this comparison, bias is defined as the average of the differences (predicted minus observed)
expressed as a fraction of the mean observed value in both studies.
Summaries of Individual AQAWG Analyses                                                           33
Results: Afternoon observations included in the       ppbC, all but one exhibited over predictions on
analysis showed a wide range of concentrations        average. While this relationship is partially a
from near zero to over 60 ppbC; mean values           reflection of the lack of correlation between
across the network ranged from 2 ppbC to              observed and predicted isoprene concentrations,
almost 30 ppbC. Comparison with predictions           the strength of the correlation between the
matched in time showed a wide range of                differences and predicted values in this case
prediction errors. Bias and gross error results       suggests that some additional factor is
are summarized in Table 1. Results are quite          contributing to the relationship. Thus, in an
similar for both studies with maximum positive        absolute sense, model performance is worse at
biases (over predictions) being larger in an          sites with high mean isoprene concentrations
absolute sense than the maximum negative              (e.g., in the southeast) than at sites with low
biases (under predictions).                           mean concentrations. Morris et al. did not
                                                      conduct a similar analysis of residuals.
Bias was within ±50% at just 6 of the 16 sites
examined by Edgerton, with all but one                Edgerton examined two potential sources of
(Brookhaven, NY) of the remaining sites               bias between observations made at a fixed point
exhibiting over predictions. While bias was           roughly 10 m above the surface and model
within ±25% at just two sites in each study,          predictions made for a 50 m deep grid cell: 1)
slightly better agreement overall was found by        the decrease in isoprene concentrations with
Morris et al.: just over half of the 15 sites         height (which would cause predicted values to
examined showed biases within ±50 percent.            be less than observed), and 2) the use of
One site (Ware, MA) under predicted by more           temperature measurements made at roughly 1.5
than 50 percent, with over predictions of greater     m agl to estimate isoprene emissions from
than 50% at the remaining six. Thus, more than        forest canopies located further above the ground
three-fourths of the sites examined in these          where cooler temperatures can be expected as a
studies exhibited biases greater than ±25% and        result of near adiabatic or super adiabatic lapse
nearly half the sites exhibited mean biases of        rates typical of sunny summer afternoons
greater than ±50% of mean observed values.            (which would cause predicted isoprene values
Nearly all the biases over ±50% were over             to be greater than observed). In some cases,
predictions.6 Edgerton found no discernable           forest canopy temperatures may also be lower
geographic pattern of bias. Edgerton‟s analysis       than the near-surface observations due to
of residuals (mean differences as a function of       enhanced evapotranspiration.          Edgerton‟s
mean predicted) indicates a fairly strong linear      review of the isoprene profile estimates of
relationship (r2 of 0.78) with a positive slope: of   Andronache and Chameides (1994) indicates
the eight sites with mean predicted                   that observations taken at 10 m should be only
concentrations below 10 ppbC, the model under         about 2 - 6 percent higher than the 10 - 50 m
predicted on average at three sites while for the     layer average. As for the temperature effect,
eight sites with mean predictions above 10            Edgerton states that daytime vertical
                                                      temperature gradients in rural eastern U.S.
                                                      environments as measured at CASTNet
    The large Under prediction at the                 meteorological monitoring sites are typically -
Brookhaven National Laboratory site in Long           0.5 to -1.0 deg. C between 2 m and 9 m. If this
Island noted by Edgerton is likely due to the         is assumed to be the same order of magnitude as
relatively large fraction of the corresponding        the temperature difference between 1.5 m and
modeling grid cell that is over water where           the vegetation canopy, it could cause a possible
there are no isoprene emissions.
Summaries of Individual AQAWG Analyses                                                            34
overestimation of emissions on the order of 7 to    do for some species that are not unique to
15 percent. Thus, the influence of vertical         biogenic sources or are difficult to measure
gradients in isoprene and temperature appear to     accurately).
be negligible in comparison to the relatively
                                                    Scientific Implications: Because isoprene is so
large biases between observed and predicted
                                                    abundant in the OTAG domain and as a result
isoprene concentrations found in these studies.
                                                    of its high reactivity, significant errors in
Of course, this does not rule out the possibility
                                                    isoprene predictions have the potential to
of other temperature-related problems such as
                                                    introduce significant errors into predictions of
biases in the 1.5 m measurements used to
                                                    the sensitivity of ozone to anthropogenic VOC
generate gridded wind fields for the model.
                                                    and NOx controls. Isoprene is a highly reactive
Limitations: Comparison of observed and             primary species with emissions that vary
predicted isoprene concentrations are primarily     strongly as a function of solar insolation,
limited by the spatial representativeness of the    temperature, and land cover. As a result, one
monitoring data as well as the ability of the       might expect relatively large differences
model to characterize vegetation within a grid      between volume average predictions made for a
cell. Isoprene emissions exhibit strong spatial     12 by 12 km by 50 m deep grid cell and
and temporal variations due to differences in the   measurements made at a single fixed location
amount of plant material and species mix and        within that cell. Measurement errors would
the response of plants to changes in                also contribute to these differences. Results of
environmental conditions. In addition, isoprene     these studies bear out this expectation, showing
is highly reactive and undergoes rapid              much larger differences between UAM-
transformation once released.              Thus,    V/BEIS2        isoprene      predictions     and
measurements made at fixed locations with           corresponding observations than is the case for
varying degrees of exposure to isoprene sources     ozone, even when the comparison is limited to
can provide only an approximate indication of       the afternoon hours when high emission rates
the volume average concentration represented        and strong turbulent diffusion should act to
by the model predictions. Additional analysis       minimize the expected differences. This
of both the ambient measurements and the land       expectation not withstanding, the magnitude of
use and temperature data input to the model are     the prediction errors and the tendency towards
needed to better diagnose the factors               over predictions indicates a potentially serious
contributing towards the prediction errors noted    problem with the UAM-V/BEIS2 modeling
in these two studies and to rectify the apparent    system. The fact that over predictions occur at
inconsistencies between these results and the       many of the rural sites examined by Edgerton is
emission flux measurements used to support          especially troublesome - over predictions at
development of BEIS2 (see Guenther et al.,          some urban sites could be explained by the fact
1996 and references therein). If the underlying     that the corresponding modeling grid cells
formulation of BEIS2 is sound as suggested by       include vegetation cover beyond the urban
the earlier studies, then the OTAG over             boundaries. Isoprene emission rate biases for
predictions point to a potential problem with the   tall forest canopies introduced by the use of
procedures used to calculate the OTAG               inappropriately high near surface temperatures
biogenic inventory. It should also be noted that    in BEIS2 do not appear to be large enough to
isoprene represents only one component of the       explain the observed over predictions, although
total biogenic VOC inventory, all though it is      other potential sources of error in the
the dominant species. Similar comparisons for       temperatures used to drive the model have yet
other biogenic species are not included in these    to be explored.
studies (and may be difficult or impossible to
Summaries of Individual AQAWG Analyses                                                         35
Policy Implications: Incorrect representation of     Additional References:
biogenic VOC emissions in the OTAG UAM-
V/BEIS2 modeling system would result in              Andronache, C. and W.L. Chameides, 1994.
errors in the model‟s predictions of ozone           Vertical distribution of isoprene in the lower
concentrations and the model‟s responses to          boundary layer of the rural and urban southern
anthropogenic emission control strategies. If        United States. J. Geophys. Res. 99:16989-
isoprene emissions in the OTAG inventory are         16999.
significantly overestimated (for whatever            Guenther, A., P. Zimmerman, L. Klinger, J.
reason), the over representation of biogenic         Greenberg, C. Ennis, K. Davis, W. Pollack,
VOCs would cause the model to underestimate          H. Westberg, G. Allwine, and C. Geron, 1996.
the expected response to anthropogenic VOC            Estimates of regional natural organic
controls. In addition, the model would tend to       compound fluxes from enclosure and ambient
be overly sensitive to NOx controls in all but the   measurements. J. Geophys. Res. Vol. 101, p.
largest urban areas.                                 1345.

Comparison of SOS Nashville Data to OTAG 1995 Base Model

Participants: Robert E. Imhoff, Tennessee Valley Authority
References: Summary of, and slides from, presentation given to the OTAG AQAWG meeting at
              Cherry Hill, NJ, August, 1996.

Purpose: Use ambient data from the 1995              NOX molecule lost) was assessed at these two
Nashville SOS study to evaluate how well the         sites by comparing measurements and model
OTAG air quality modeling for July 9-18, 1995        results during the 11:00 to 18:00 daytime period
is simulating atmospheric chemistry and              when the atmosphere was well mixed. The
physics in a southern environment.                   third data set was for vertical profiles of ozone
                                                     over a suburban site near Nashville. The
Methodology: UAM-V results for the July 9-           accuracy of vertical mixing processes in the
18, 1995 OTAG D2 base case were obtained             model was assessed by comparing measured
from the New York State Department of                and modeled ozone profiles. The ozone sonde
Environmental Quality. Model results were            data were collected by Georgia Institute of
compared to three ambient data sets. The first       Technology, the remainder of the data were
data set was morning surface CO, NOY, and            collected by TVA.
VOC concentration data for an urban site in
Nashville for June 19 to July 28, 1995. The          Findings: For morning surface concentration
measured species ratios were compared to their       data at the Nashville urban site, measured
relative proportions in the emissions inventory.     CO/NOY slopes were 7.7 for June 19 to July 28
 The second data set was for ozone, NOX and          and 10.6 for July 9-18. The modeled slope for
NOY at a suburban and a rural site. Ozone            July 9-18 was 12.6, and the emission inventory
productivity (i.e., the relationship of O3 to NOZ    slopes were 8.2 for mobile sources and 7.5 all
where NOZ = NOY - NOX and where O3/NOZ is            sources. The apparent background of CO in the
the number of ozone molecules produced per           ambient data was 150 - 200 ppb, compared to

Summaries of Individual AQAWG Analyses                                                           36
about 100 ppb in the model. The measured            formation at the Youth site to be VOC limited
VOC/NOY slope for July 9-18 was 2.5                 on July 18, whereas the observations show the
compared to a modeled slope of 2.9. The             site to be more NOX limited. Thus the model
apparent background of VOC in the ambient           sensitivity to VOC versus NOX emission control
data was quite close to zero, compared to about     around the Youth site would likely be opposite
90 ppb in the model. Agreement between the          to that predicted by the observations for this
modeled and ambient CO/NOY and VOC/NOY              day.
slopes was reasonably good. Modeled CO/NOY          The comparison of observed and modeled
slopes would be expected to be higher than          vertical profiles for ozone showed that the
inventory slopes because VOC oxidation is a         model consistently overestimated the surface
secondary source of CO, and because NOY is          and layer 4 ozone concentrations (measured
deposited more rapidly than CO.                     layers 2 and 3 were not available for
Comparing observed and modeled ozone                comparison because the resolution of the sonde
concentrations at the suburban and rural sites      was too coarse in these layers). The model
revealed a fair amount of variability between       consistently estimated a slight increase in ozone
the model and observed ozone on a day to day        from layer 1 through layer 4, with a sharp drop
basis, but no indication of bias overall. At the    above layer 5. The top of the model's mixed
rural site the model estimated the ozone            layer was in layer 5 during this period. The
productivity very well, with least squares          model's estimated ozone concentration in layers
regressions for both model and observations         6 and 7 were consistently too low. One
showing O3/NOZ slopes of about 6, consistent        possible cause is that the previous day's mixed
with NOX limited ozone formation. At the            layer in the model was too low and thus
suburban site, there was more variability in both   material was not mixed to layers 6 and 7.
observed and, in particular, modeled O3/NOZ         Another possibility is that the model does not
ratios. The observed ozone productivity (least      handle localized mixing due to convective cells
squares regression slope about 3.5) was lower       which were common during the meteorological
than at the rural site consistent with ozone        conditions prevalent during this period. These
formation being less NOX limited at the             cells can inject significant amounts of air from
suburban site than at the rural site. The model     near the surface above the mixed layer. Once
ozone productivity at the suburban site was         above the mixed layer the material can have
about 4.3 on July 9-17 which is similar to the      long residence times. Another possibility is that
observed ozone productivity (about 3.5).            the "Topcon" concentration used at the upper
However, on July 18 the model O3                    boundary in the model (35 ppb) is too low.
concentrations were much lower than on July 9-
                                                    Limitations: The modeled NOY used in these
17 resulting in lower O3/NOZ ratios.
                                                    comparisons does not include organic nitrates
Further investigation of the model performance      (Carbon Bond species NTR) because NTR
for the suburban site (Youth) on July 18 showed     concentrations were not saved in the OTAG
that the model predicted high concentrations of     model runs.         Thus, UAM-V NOY
NOX (10-15 ppb) between 11:00 and 18:00,            concentrations are always biased low.
contrary to the observations. In other words,       However, the magnitude of bias is not well
the model predicted that Youth was impacted         known and likely varies in time and space
during the middle of the day by a substantial       depending upon age of NOY (the bias might be
source of fresh NOX that was not observed on        as high as 20 percent for aged airmasses).
July 18, or on July 9-17 for that matter. As a      Accounting for the missing contribution of
result of this error, the model predicts ozone      NTR to modeled NOY might: a) further improve

Summaries of Individual AQAWG Analyses                                                          37
agreement between the observed and modeled           ozone profile data may suggest that the model is
CO/NOY and VOC/NOY ratios for the urban              not properly distributing mass vertically or that
site; b) further improve agreement between           the top boundary concentration for ozone is too
observed and modeled ozone productivity at the       low. Underestimating vertical mixing might
suburban site, but; c) degrade agreement             tend to overestimate concentrations at the
between observed and modeled ozone                   surface, even though the chemistry and
productivity at the rural site.                      emissions      inventory       are        accurate.
                                                     Underestimating the top boundary concentration
Scientific Implications: The relatively good
                                                     for ozone might have a small impact of biasing
agreement between modeled and ambient
                                                     surface ozone concentrations low. While no
CO/NOY and VOC/NOY ratios supports the
                                                     prediction bias for surface ozone was noted at
validity of these parameters in the emission
                                                     the sites examined in this study, it is interesting
inventory, in overall terms. However, the
                                                     to note that, overall, the model was found to
discrepancy between the apparent background
                                                     over predict surface ozone in the southeastern
for VOC of 90 ppb in the model compared to
                                                     U.S. (EPA, 1996).
near zero in the ambient data possibly indicates
a source of VOC in the model that is not also a      Policy Implications: Although not conclusive,
source of CO. Further investigation is needed        this study raises the possibility that the UAM-V
to identify the source of this discrepancy. The      model runs are not treating concentrations in the
modeled and observed ozone productivities            upper layers of the modeling domain properly,
were consistent at the rural site, and were          thereby introducing errors in the estimated
consistent at the suburban site on all but one       impact of transport on downwind areas. The
day. The generally good agreement in ozone           inability of the model to correctly predict the
productivity suggests that the model chemistry       influence of NOX emissions on ozone formation
is performing well for rural and suburban sites      at the suburban site on one day sends a
in the Nashville area. However, the inability of     cautionary message for the use of OTAG model
the model to correctly predict the ozone             results at suburban, and presumably urban sites.
productivity at the suburban site on July 18          Further technical analyses will be needed to
cautions that OTAG model performance is              sort through these issues.
subject to important day-to-day variabilities that
have the potential to influence emission control
decisions. The discrepancy in the vertical

Summaries of Individual AQAWG Analyses                                                             38

Shared By: