III. OZONE PRECURSOR EMISSIONS

Unlike other photochemical pollutants, including acid pollutants and aerosols, the precursors of ozone, VOC and NOx, are peculiar in that they not only produce ozone but they also destroy it. As a consequence, ambient ozone does not depend linearly on either VOC or NOx. Ozone accumulates in concentrations that affect human health (>80 ppb) and ecosystems (>60 ppb) when the ambient concentrations of the two precursors are at an optimum ratio and is suppressed when either one of the two precursors is in large excess relative to the other. For this reason, it is critically important that VOC and NOx emissions in all non-attainment regions be characterized both with respect to their absolute rates and also with respect to the relative ambient concentration ratios they create in the atmosphere. Such characterizations are extremely difficult in the SOS region mainly because of the abundance of biogenic VOC emissions and of the complex influences on emissions of the unusually intensive solar radiation, temperature, and relative humidity conditions in that region. The SOS program aimed at collecting improved data mainly on concentrations and variability of the VOC and NOx emissions from motor vehicles and other anthropogenic sources, and from natural sources, with emphasis on VOC emissions from vegetation and NOx emissions from well fertilized crops and pastures.

For this reason, it is important that VOC and NOx emissions in non- attainment areas, and in vertically and horizontally upwind areas, be characterized both with respect to the absolute amounts of NOx and VOC emissions and also with respect to the relative ambient concentration ratios they create in non-attainment atmospheres. Such characterizations are difficult to achieve in the SOS region mainly because of the abundance of highly reactive biogenic VOC emissions (such as isoprene) and also because of the complex influences on emission rates of the intensive solar radiation, temperature, and relative humidity conditions in the SOS region. The SOS research program was designed to produce improved data and information regarding the rates, amounts, sources, and ratios of VOC and NOx emissions from natural biogenic sources, motor vehicle sources, electric utility sources, and other natural and anthropogenic sources.

During the years since initiation of SOS, the ozone and particulate matter precursors of concern and both natural (N) and anthropogenic (A) sources have become progressively more numerous:

ANOx + AVOC (Haagen-Smit, 1952)

ANOx + AVOC + NVOC (Chameides et al., 1988)

ANOx + AVOC + NVOC + NNOx (Valente & Thornton, 1993)

ANOx + AVOC + NVOC + NNOx + ACO (Daum et al., 2000b)

ANOx + AVOC + NVOC + NNOx + ACO + CH4 (Goldan et al., 2000)

ANOx + AVOC + NVOC + NNOx + ACO + CH4+ N/ANH3 + ASO2 (Chameides et al., 1999)

A. Biogenic and Other Natural Sources of Ozone Precursors (BG)

Investigation of the occurrence and role of biogenic and other natural precursors of ozone has been one of the two main research themes in the SOS research program (the other was the development and application of observational methods and observation- based models). Such emphasis was justified by: 1) The unusually large abundance of biogenic VOC in the SOS region, 2) The difficulty in obtaining reliable biogenic VOC emission inventory data, and 3) The extremely complex role that such organics play in the development of ozone control strategies - for example, the question of whether ozone attainment in an ozone non-attainment area should be pursued through VOC control or NOx control is closely linked to the role of the biogenic VOC in the SOS region.

The SOS program included studies of nearly every aspect of biogenic VOC role in the photochemical ozone pollution problem including: 1) the extremely important issue of land use, 2) vegetation species identification, 3) biogenic VOC emission rates, 4) atmospheric chemistry of biogenic VOC - especially isoprene, and 5) the effect of biogenic VOC emissions on VOC or NOx control requirements in ozone non-attainment areas. Given the fact that biogenic VOC are ubiquitous - they occur in urban areas in both the eastern and western regions of the US, Canada, Mexico, and other parts of the world - the biogenic research findings from the SOS should be of general interest to air quality managers.

The biogenic VOC compounds studied by SOS scientists included:

  1. A wide variety of hydrocarbons including both alkanes and alkenes (especially isoprene from hardwood forest trees, ethylene from many different species of healthy and diseased plants, and methane from plant, animal, and insect sources);
  2. Many aromatic VOC, especially alpha- and beta-pinenes from softwood (coniferous) trees, wind-downed and ice damaged conifers, and from harvesting, chipping, sawing, drying, and other processing of softwood timber in pulp, paper, lumber, and plywood manufacturing,
  3. A large variety of oxygenated biogenic VOC including carbon monoxide from wild fires and controlled burning operations, aldehydes ( especially formaldehyde), both saturated and unsaturated alcohols (especially methanol and ethanol), ketones, and organic acids, and
  4. Alkyl sulfides.

BG1. Vegetation is a major source of reactive volatile organic compounds (VOC) in both urban and rural areas throughout the SOS region (Chameides et al., 1988; Guenther et al., 1993, 1995, 1996a, 1996b; Guenther 1997).

BG2. Biogenic hydrocarbons (mainly isoprene and monoterpenes) play a major role in ozone formation and accumulation in both urban and rural areas in large parts of the eastern United States, especially in the summertime (Chameides et al., 1988; Williams et al., 1997; Kleinman et al., 1997; Frost et al., 1998; Helmig et al., 1998; Roberts et al., 1998; Nouaime et al., 1998; Starn et al., 1998a, 1998b).

The implication of this finding is that pursuing an ozone abatement strategy that ignores the effect of natural VOC emissions can incur substantial error.

BG3. Based on measurements made at the Rural Oxidants in the Southern Environment (ROSE) Site in 1990, isoprene was the largest OH-consumer VOC species (the fraction of OH reacting with isoprene was 0.71) (Cantrell et al., 1992).

The term 'reactivity' in Finding BG3 is meant to describe reactivity in terms of reaction with the OH radical, and not reactivity in terms of ozone production efficiency. Isoprene has a very high ozone production reactivity, which it displays in VOC-limited atmospheres, but isoprene also has a very large ozone inhibition/destruction reactivity, which it displays in ozone-rich and NOx-deficient atmospheres.

BG4. Isoprene is mainly of biogenic origin and accounts for a large part of the total VOC reactivity in the SOS region. Various species of oaks are the most important biogenic sources of isoprene (Geron et al., 1994, 1995, 1997).

BG5. The temperature dependence of isoprene emissions arises from the temperature response of the enzyme isoprene synthase. Due to the progressive effect of warm temperatures on the activation and stimulation of isoprene synthase, the highest emissions of isoprene occur in the late summer (Monson et al., 1994; Wildermuth and Fall, 1996)

BG6. Water stress, temperature, and light intensity all have substantial effects on isoprene emissions from vegetation. These effects are especially notable in leaves of kudzu plants, which are very common in many parts of the SOS region. In kudzu, isoprene emissions typically represent a significant fraction of the total carbon fixed in photosynthesis (as much as 0.67 grams of isoprene per gram of carbon fixed). Also, isoprene emission rates in kudzu are more sensitive to temperature than that in other species of isoprene-emitting plants (Loreto and Sharkey, 1993a, 1993b; Sharkey and Loreto, 1993; Fang et al., 1996; Wildermuth and Fall, 1996).

BG7. Isoprene shows large regional and seasonal variations in emission rates, especially on small temporal (e.g., diurnal) and spatial (e.g., a few km2) scales. Because of the short lifetime of isoprene in the atmosphere, a few hours or less, these variations are important to predict accurately (Guenther et al., 2000).

BG8. Concentrations of biogenic VOC decrease slowly with altitude in the mixed layer while surface layer concentrations show much more variability, based on tethered balloon measurements at ten North American sites and one Amazon site (Greenberg et al., 1999).

BG9. Isoprene concentrations in the planetary boundary layer (mixed layer) of the atmosphere remain fairly constant in the middle of the day, in contrast to isoprene concentrations at canopy level, which continue to increase until evening. Daytime emissions, which increase with temperature and solar radiation, are balanced by changes in entrainment and oxidation (Greenberg et al., 1999).

BG10. Measurements made at 40-100 meters above ground yield the most reliable measures of average boundary layer concentrations of reactive organics such as isoprene (Andronache et al., 1994).

BG11. Estimates of isoprene emission fluxes based on ambient concentrations of isoprene and monoterpene emissions measured at two rural sites in Alabama and Georgia in 1990 were within a factor of two of fluxes predicted based on enclosure measurements and landscape data (Guenther et al., 1996b).

BG12. Oxygenated organic compounds (mainly aldehydes and alcohols) also contribute significantly to air concentrations of VOC in the SOS region. The identification of oxygenates produced from isoprene photooxidation, such as methylvinylketone (MVK) and methacrolein (MACR), at rural sites is consistent with a photochemical mechanism involving their production from isoprene and their subsequent photooxidation. Photooxidation of isoprene is probably a major local source of peroxyacetyl nitrate (PAN) arising from production of MVK and methylglyoxal and their subsequent oxidation to produce peroxyacetyl radical, the immediate precursor of PAN (Lee and Zhou, 1993, 1994; Montzka et al., 1993; Kleinman et al., 1994; Lee et al., 1995, 1998; Stroud et al., 2001).

BG13. Simultaneous measurements of peroxymethylacrylyl nitrate (MPAN), peroxypropionyl nitrate (PPN), and peroxyacetyl nitrate (PAN) provide a method for apportioning photochemically produced ozone into a fraction resulting from oxidation of biogenic VOC (mainly isoprene) and a fraction resulting from oxidation of anthropogenic VOC (Nouaime et al., 1998).

BG14. The contribution of biogenic emission sources to ambient VOC concentrations can be determined quantitatively through radiocarbon (14C) measurements (Lewis et al., 1999).

Scientific finding BG14 is extremely important because it identifies a reliable method for determining by direct measurements what fraction of the ambient VOC are from recently fixed carbon (and therefore biogenic sources), compared with historically fixed (and therefore fossil-fuel-based, anthropogenic sources) of carbon. Reliable measurement of ambient biogenic VOC helps to evaluate or obtain reliable biogenic VOC emission inventory data.

BG15. Destruction of about 20% of the urban forests of Atlanta, GA from 1979-1988 caused an approximate 2° C intensification of Atlanta's urban heat island and may have resulted in a net increase rather than decrease in Atlanta's total biogenic emissions of isoprene (Cardelino and Chameides, 1990).

City planning and construction practices that modulate the intensity of urban heat islands, through placement of 'green spaces ' in the urban core of cities and use of highly light-reflective building materials, may aid in ozone pollution abatement by decreasing urban temperatures and thus decreasing emissions of biogenic (as well as anthropogenic, evaporative) VOC.

BG16. More detailed studies of biogenic emissions of alkane and aromatic VOC, including research at SOS sites in Georgia, showed that biogenic emissions are much less than those reported in an earlier national inventory of emissions of these same classes of biogenic VOC. The primary factor in previous overestimates was misinterpretation of chromatographic data in the earlier study (Guenther et al., 2000; Zimmerman, 1979).

BG17. Biogenic emissions of isoprene are more important to urban ozone production in Nashville, TN (Roberts et al., 1998) and Atlanta, GA (Chameides et al., 1992), than in Houston, TX (Wiedinmyer et al., 2001). The major differences in biogenic emissions between Houston and both Atlanta and Nashville may be explained in part by the greater abundance of isoprene-emitting trees (mainly oak forests) in the land cover of the suburban and rural areas surrounding Atlanta and Nashville than in similar rural areas near Houston.

BG18. Much of the photochemically produced ozone in the southern part of the multi- state Nashville/Middle Tennessee study area resulted from oxidation of biogenic VOC. This percentage decreased from south to north within the study region (Goldan et al., 2000).

BG19. Much of the spatial variability in air concentrations of isoprene and NOx can be explained by differences in patterns of land use (mainly crop vs. forest), forest type (especially amount of oak), and meteorological conditions (mainly temperature and wind speed) (Thornton et al., 1997).

BG20. NO emissions from well-fertilized soils used for row crops and intensively managed pasture lands are a significant regional source of NO emissions in parts of the SOS region which are dominated by agriculture. NO emission rates: 1) increase significantly after light rains, 2) increase exponentially with temperature, and 3) are highest in soils with high nitrate fertilizer application rates (Williams and Fehsenfeld, 1991; Williams et al., 1992; Meyers and Baldocchi, 1993; Valente and Thornton, 1993; Kim et al., 1994; Valente et a1., 1995; Thornton and Shurpali, 1996; Davidson and Kingerlee, 1997; Potter et al., 1997; Thornton et al., 1997).

BG21. The average summertime contribution of soil NO to the overall NO inventory for nine states within the southeastern US (Alabama, Florida, Georgia, Kentucky, Mississippi, North Carolina, South Carolina, Tennessee, and Virginia) was 4.1 percent (Thornton et al., 1997).

The NO released from agricultural activities plays an important role in the photochemical air pollution problems; namely, it contributes significantly to regional build-up of ozone and particulate nitrates by reacting photochemically with biogenic VOC. Viewed in this respect, agricultural activities may be characterized as a quasi-anthropogenic source of pollution.

BG22. Urban-metropolitan soils were not an important source of soil NOx during the Nashville/Middle Tennessee Ozone Study (Thornton et al., 1997).

BG23. For the middle Tennessee non-attainment area, soil biogenic NOx contributed from 7 to 10 percent of the daily average NOx budget during the months of June through August, 1995. But measurements made during the hottest days of July 1995 indicated that soil biogenic sources contributed more than 17 percent of the total NOx emissions for the state of Tennessee (Thornton et al., 1997).

BG24. NO emissions from lightning, although still highly uncertain in amount, cannot be ruled out as a source of ozone precursors in the SOS region. Lightning appears to be a negligible source of NOx on large space (regional) and time (weeks to months) scales but can be a significant source of NOx on smaller space and time scales in the SOS region (Biazar and McNider, 1995).

BG25. Emissions from distant forest fires can contribute to regional concentrations of carbon monoxide and to regional accumulations of high ozone concentrations. This conclusion is based on measurements made during the Nashville/Middle Tennessee Ozone Study and back-trajectory modeling results for July 1995 that showed an influence of Canadian forest fires on CO and ozone concentrations in both the central and eastern US; ambient ozone concentrations were increased by 10-30 ppb (McKeen et al., 2002; Wotawa and Trainer, 2000).

The policy implication of this finding is that, in addition to local sources and regional sources of ozone precursors, occasionally far-distant sources of precursors can contribute to ozone accumulation in a given region.

B. Motor Vehicle and Other Anthropogenic Sources of Ozone Precursors (AN)

Motor vehicles are the most important source of VOC and NOx emissions in most urban areas throughout the world. The National Academy of Science '1999 Rethinking' report expressed concerns about automobile emissions inventories and recommended both that tunnel studies of on-road motor vehicle emissions be conducted, and that remote sensing methods be used in such studies (NAS, 1999). The anthropogenic emissions part of the SOS program focused mainly on such tunnel studies, on refining methods for traffic volume measurement, on studies of driving pattern and roadway factors and their effects on motor vehicle emissions, and also on determining and characterizing other anthropogenic sources, especially power plants in the case of NOx and other industrial emissions in the case of VOC.

Noteworthy scientific findings from SOS' anthropogenic emissions studies are presented below.

AN1. Motor vehicles are a major human source of ozone precursor chemicals in urban and rural areas throughout the SOS region (LeBlanc et al., 1995).

AN2. Direct on-road measurements of speciated VOC, NOx, and CO emissions were made in the Fort McHenry Tunnel under Baltimore Harbor and the Tuscarora Tunnel on the Pennsylvania Turnpike during the summer of 1992. These on-road measurements showed that the MOBILE4.1 and MOBILE5.0 models provided reasonably good estimates (most of them within ± 50%) of actual VOC, NOx, and CO emissions for the fleet of passenger cars and light-duty trucks as well as for the fleet of heavy-duty trucks (gasoline- and diesel-powered) using these two interstate highway tunnels (Gertler et al., 1996; Pierson et al., 1996; Robinson et al., 1996).

In interpreting this scientific finding (AN2), it should be recognized that neither the relatively new and well-maintained fleets of vehicles using these tunnels, nor the manner in which they were operated during these tunnel experiments (relatively constant interstate highway speeds), are fully representative of the total fleet and more variable conditions of operation of motor vehicles, especially in many urban areas of the United States.

AN3. VOC emissions from light-duty passenger cars and trucks were very different from those emitted by heavy-duty gasoline-powered and diesel trucks. More than half the VOC from heavy-duty vehicles contained more than 10 carbon atoms per molecule, while only 10 to 20 percent of the VOC from light-duty vehicles contained more than 10 carbon atoms (Zielinska et al., 1996).

These differences in chemical composition of VOC described in AN3 above were not predicted by the MOBILE4.1 or MOBILE5.0 emissions models. It is also noteworthy that the high molecular weight VOC emitted by heavy duty vehicles have a smaller ozone production efficiency than the lower molecular weight VOC emitted by light-duty vehicles. But these high molecular weight VOC also produce larger amounts of photochemically induced particulate matter.

AN4. Roadway grade had a large effect (uphill values were 100% greater than downhill values) on both the amount of CO, VOC, and NOx emitted and on the composition of VOC emitted by passenger cars and light-duty and heavy-duty trucks (Pierson et al., 1996).

These large effects of roadway grade on the amount and chemical composition of motor vehicle emissions should not be ignored in developing the next generation of motor vehicle emissions models, or in using these models to prepare mobile source emissions inventories for use in State Implementation Plans.

AN5. During the SOS intensive measurement campaign in 1992, traffic counters were found to be a very useful tool for developing day-specific estimates of mobile source emissions of VOC and NOx. These day-specific estimates were significantly different from those obtained using the defaults contained in the Emissions Preprocessor System Version 2 (EPS2) of the Urban Airshed Model. These discrepancies included different temporal patterns of emissions as well as different magnitudes of total emissions. In general, the EPS2 defaults tended to overestimate emissions, especially during the morning and evening rush hours in Atlanta and Nashville (Cardelino, 1998).

AN6. Acceleration, roadway grades, and other heavy engine loads leading to 'power enrichment' significantly increased motor vehicle emissions and should be accounted for in mobile source emissions inventories. On-road tests on the urban vehicle fleet in Atlanta, GA indicate that power enrichment occurred about 1 percent of the fleet's total operating time and thus constituted a significant part of VOC emissions in the Atlanta metropolitan area (Fehsenfeld et al., 1994).

AN7. In the Atlanta metropolitan area, the temporal distribution of motor vehicle emissions was strongly dependent on the area (urban or rural), the type of road (interstate, principal, secondary or local), and the day of the week during which the measurements were made. Motor vehicle emissions were substantially different on week days (Monday through Thursday) compared to weekend days (Friday through Sunday). However, ozone predicted by models with customary resolution did not reflect comparable variability (Cardelino, 1998).

AN8. The vehicle classification distribution by road type obtained from traffic counters in Atlanta in 1992 was significantly different from the default distribution contained within the Mobile5a computer model. When applied to a specific-day inventory, the use of observed data, as opposed to default data, produced decreases in emissions that varied by 8 percent for VOC, 9 percent for CO, and 19 percent for NOx in Atlanta (Cardelino, 1998).

AN9. Compared to a typical summer day, the day-to-day range of mobile emission variability was 26 to 28% for urban areas and 13 to 19% for rural areas around Atlanta (Cardelino, 1998).

AN10. The daily variability in point-source NOx emissions was found to be as much as 24 percent with respect to typical summer day emissions in the Atlanta metropolitan area. The daily variability of point-source VOC emissions was as large as 28 percent, but their contribution to total VOC emissions in Atlanta was not very significant (W. Chang et al., 1996).

This substantial day-to-day variability of both the motor vehicle emissions and the point source emissions make it imperative that day-specific emission inventory data be used in modeling simulations of control strategies.

AN11. Numerical simulations suggest that changes in point-source NOx emissions can have either a positive or a negative effect on ambient ozone concentrations, depending on the geographical location of the NOx sources (rural vs urban areas) (M. Chang et al., 1996).

The extremely important implication of this scientific finding (AN11) and findings from other studies is that because of the ozone-destruction effect of NOx under VOC-sensitive conditions and the widespread occurrence of such conditions (e.g., within intensely urbanized areas), air quality managers should view NOx control with caution.

AN12. Use of day-specific roadway traffic information resulted in significantly different emissions estimates for motor vehicles in the 1992 SOS Intensive Field Study in Atlanta and the 1995 SOS Nashville/Middle Tennessee Ozone Study than those obtained using the defaults contained in the Emissions Preprocessor System Version 2 (EPS2) of the Urban Airshed Model. In general, the defaults tended to overestimate emissions especially during the morning and evening rush hours (Cardelino, 1998).

The important implication of this scientific finding (AN12) is that the emission inventory data used in model simulations of ozone and PM concentrations should be day-specific rather than default data.

AN13. Observations of the amounts, types, and variability of CO, VOC, and NOx emissions from motor vehicles in Houston, TX - including passenger cars, light-duty trucks, and both diesel-powered and gasoline-powered heavy duty trucks - were essentially identical to similar observations in other urban areas in the southern US such as Nashville, TN and Atlanta, GA (Allen and Durrenberger, 2003).

AN14. VOC and NOx emissions in industrial areas of Houston, Texas, showed substantial spatial and temporal variability (Allen and Durrenberger, 2003).

One important implication from this scientific finding (AN14) is that model simulations of ozone problems in such areas should be conducted using high spatial and temporal resolution (i.e., 1 km or less cell size, and 1-hr averaging time).

AN15. NOx, SO2, and CO2 emissions from fossil-fueled power plants in eastern and central Texas were accurately estimated in inventories, but CO emissions show significant discrepancies between direct measurements and inventory estimates at some power plants (Nicks et al., 2003).

AN16. The uniquely rapid formation and accumulation of ozone ('ozone spikes') in Houston, Texas was caused primarily by photochemical processing of industrial emissions (Daum et al., 2003; Ryerson et al., 2003; Allen and Durrenberger, 2003).

AN17. Burning of biomass both in open fields and in pulp and paper mills was identified as an important source of NOx in various rural parts of the SOS region (Buhr et al., 1995b).

An important consequence of scientific finding AN17 is the large contribution of such NOx to regional ozone accumulation.

C. Emissions Inventories and Evaluation Methods (EIE)

Reliable ozone precursor emissions inventories are indispensable in the development and implementation of ozone management strategies. Such data are obtained either through direct measurement of precursor emission rates and amounts, or through calculations based on guidelines issued by the USEPA. Different applications of inventory data require different temporal and spatial resolutions, and such requirements are not always met. This latter problem is particularly serious in development of State Implementation Plans for ozone through use of three- dimensional grid models. SOS efforts in this arena were aimed at assessing and improving the reliability of emission inventory data.

Noteworthy scientific findings from SOS' emissions inventory improvement research studies are presented below.

EIEl. Certain aspects of the guidance currently followed in developing emissions inventories of ozone precursor chemicals in rural and urban areas of the SOS region:

  1. Do not accurately reflect the land-use distributions in many southern cities;
  2. Are not consistent with direct observations of NOx, VOC, and CO in rural and urban areas; and/or
  3. Do not properly quantify biogenic and other natural emissions of VOC and NOx (Fehsenfeld et al., 1994).

E1E2. The fraction of ozone produced from photooxidation of isoprene decreased from south to north during SOS' 1994-95 and 1999 field measurement campaigns in the multi-state area centered over the Nashville, TN Metropolitan area. This decrease was predicted by the BEIS2 emission model (Ryerson et al., 1998).

EIE3. Ground-based measurements of biogenic emissions of NOx from soils and isoprene emissions from vegetation agreed well with NOx and isoprene emissions estimates provided by BEIS2, a second-generation mathematical model for estimating biogenic sources of VOC and NOx (Geron et al., 1994; 1997).

EIE4. Boundary layer concentrations of isoprene and NOx varied greatly from one geographical locale near Nashville, Tennessee to another. BEIS2 estimates of this spatial variability agreed well with aircraft measurements of boundary layer isoprene concentrations when land use, forest type, canopy temperature, and wind speed above the canopy were considered (Pierce et al., 1998).

EIE5. BEIS2 uses more temporally resolved environmental corrections (hourly versus monthly), more spatially resolved vegetative cover (county-level versus latitude/longitude grid cells), and more resolved vegetative emission factors (genus versus broad biome classes) than BEIS1. Higher isoprene emissions are obtained with BEIS2, which are about a factor of 5 higher than BEIS1 during warm, sunny conditions (Pierce et al., 1998).

EIE6. When BEIS2 was used with the RADM model, areas of elevated concentrations of ozone went from being VOC-sensitive to NOx-sensitive across much of the RADM modeling domain. The new system yielded better agreement with observations. Using BEIS2 in RADM resulted in mean near-surface isoprene predictions that were slightly lower (25%) than observed (Pierce et al., 1998).

EIE7. Isoprene emissions derived from inverse modeling were 2 to 10 times higher than any of the accepted BEIS-based emission estimates for Atlanta (M. Chang et al., 1996, 1997).

EIE8. Isoprene emissions significantly increased the concentration of ozone observed within the plumes of nitrogen oxides emitted from large point sources (M. Chang et al., 1996, 1997).

EIE9. Uncertainties in mixing height were not likely to be responsible for underpredictions of isoprene in the Urban Airshed Model seen in a modeling study in Atlanta (M. Chang et al., 1996, 1997).

EIE10. Inhomogeneities in the spatial distribution of emissions can severely limit the application of the inverse method for analysis of source-receptor relationships (M. Chang et al., 1996, 1997).

To the limitations of the 'inverse method,' described in scientific finding EIE10, one may add the uncertainties introduced by the fast disappearance of isoprene in air in reactions with OH and ozone.

EIE1l. Correlations between SO2, CO, and NOy in the Nashville urban area indicated only a minor impact from power plant emissions. Plume-like excursions of high SO2 occurred less than 5% of the time during the Nashville-Middle Tennessee Ozone Study (Kleinman et al., 1998).

EIE12. The 1990 NAPAP emission estimates for VOC and NOx emissions can be brought into reasonable agreement with the values observed during the 1995 SOS Nashville Intensive if the VOC emission rate is decreased by 30 percent and the CO emission rate is increased by about 35 percent (Kleinman et al., 1998).