III. CHEMISTRY OF OZONE FORMATION

Unlike the traditional laboratory (smog chamber) approach to studying atmospheric chemistry, the approach taken in SOS' research program was use of direct field observational methods. Key atmospheric chemistry issues studied by SOS included:

  1. The relative ozone-production efficiencies of VOC and NOx in various environments - an issue that is linked to the relative ozone-management benefits associated with VOC and NOx controls;
  2. The existence of predictive relationships between ambient ozone concentrations and concentrations of VOC and NOx photo-degradation products; and
  3. The chemical mechanism of the atmospheric photooxidation of biogenic VOC, especially isoprene.

The SOS program was especially effective in developing innovative observational methods for:

  1. Defining ambient condition-regimes for which decreases in ozone exposures should be pursued through VOC controls or though NOx controls;
  2. Determining the ozone-production efficiencies of NOx from different types and sizes of sources, and
  3. Evaluating the accuracy of atmospheric VOC-photooxidation mechanisms by determining both the identity and yield of photooxidation products.

A. General Features of Ozone Chemistry (OC)

General features of ozone chemistry studied by SOS included:

These four aspects of the ozone chemistry were investigated by SOS scientists in the belief that scientific findings from these studies will provide valuable evidence regarding the causes of ozone non-attainment problems in the SOS region and other parts of the US, Canada, and Mexico.

Noteworthy scientific findings from SOS' studies of ozone chemistry are summarized below.

OC1. In the SOS region, rates of ozone accumulation, and rates of NOx removal from air near the ground, were more rapid than in other parts of the United States. These differences in rates were caused in part by the higher air concentrations of biogenic VOC in the SOS region than in other regions within the US (Fehsenfeld et al., 2003)

This scientific finding (OC1) suggests two conflicting effects of biogenic VOC on ozone: 1) a positive, direct effect that favors ozone production, and 2) a negative and indirect effect - removing NOx(the precursor that catalyzes the ozone formation) and thus inhibits ozone production. One important implication of these conflicting effects is that in atmospheres with abundant NOx, biogenic VOC enhance ozone formation, whereas in NOx-deficient or NOx-depleted atmospheres, biogenic VOC inhibit ozone formation.

OC2. Various oxygenated species of VOC, including methylvinyl ketone (MVK), methacrolein (MACR), and peroxyacetyl nitrate (PAN), are produced as a result of photochemical oxidation of isoprene. Photochemical oxidation of isoprene is a major source of peroxyacetyl nitrate (PAN) in the SOS region (Lee and Zhou, 1993, 1994; Montzka et al., 1993; Kleinman et al., 1994; Lee et al., 1995, 1998).

OC3. Maximum ozone concentrations in the Atlanta metropolitan area occur when plumes from power plants or other point sources are embedded in a broader urban plume which in turn is embedded in a regional 'tide' of ozone resulting in part from the interaction of NOx with the 'sea' of isoprene in rural areas surrounding isolated metropolitan areas (Imhoff et al., 1995; St. John et al., 1998; St. John and Chameides, 2000).

OC4. Isoprene chemistry dominated the formation of ozone in forested rural areas near Nashville, Tennessee during the 1995 study (Helmig et al., 1998; Starn et al., 1998a, 1998b; Roberts et al., 1998; Frost et al., 1998).

OC5. The removal rate of VOC (rate of decrease in concentrations as the Nashville urban plume aged) was proportional to the reactivity of the VOC under consideration. Thus, highly reactive VOC (such as isoprene) showed more rapid rates of decrease in concentration as an urban plume aged than less reactive VOC (Nunnermacker et al., 1998).

OC6. Anthropogenic VOC (including CO) accounted for about two-thirds of OH - VOC reactivity during the 1995 Nashville/Middle Tennessee Ozone Study (Daum et al., 2000b).

The OH- VOC reactivity is the reactivity of VOC with respect to their reaction with OH radicals. It is different from and unrelated to the ozone-production reactivity of VOC.

OC7. Chlorine was shown to enhance ozone production in chamber experiments with captured Houston air, although it appears not to be the dominant mechanism of ozone formation in Houston (Tanaka et al., 2003a, 2003b).

OC8. In the southeastern US, the highest ozone concentrations occur under stagnation conditions. Model calculations and observations show that stagnation conditions promote VOC sensitivity. In Nashville, the sensitivity of peak ozone concentrations is somewhere between the strongly VOC-sensitive condition typical of Los Angeles and the strongly NOx-sensitive condition typical of rural areas in most US states. In all likelihood, a dual (NOx and VOC) control strategy may be required for efficient and cost-effective management of ozone concentrations in the SOS region. Such a strategy will have to take into account the role of biogenic VOC emissions since their emissions cannot be controlled (Valente et al., 1998; Banta et al., 1998).

OC9. Ozone photochemistry was rapid in the Nashville urban plume during the Nashville/Middle Tennessee Ozone Study. The urban plume examined on July 3 and on July 18, 1995 consumed about half of its NOx and half its supply of anthropogenic hydrocarbons within two hours (Nunnermacker et al., 1998).

OC10. Biogenic VOC became more important as the Nashville urban plume was advected into the rural surroundings during the 1995 Nashville/Middle Tennessee Ozone Study. As the urban plume moved into surrounding rural areas the biogenic VOC contribution to ozone production increased from about 25 percent to about 36 percent while the anthropogenic contribution to ozone production decreased from about 55 percent to about 44 percent (Nunnermacker et al., 1998; Luria et al., 2000).

OC11. During a stagnation episode on July 18, 1995, lidar cross-sections and profiles and wind profiler data showed that the layer of ozone aloft over Nashville mixed out during the day and became part of the surrounding suburban and rural mixed layer during the next day (Banta et al, 1998).

OC12. Measurements made at ground level at a suburban Tennessee site on July 1, 1994 showed that of the 120 ppbv of ozone recorded on that date, 80 ppb was due to vertical transport of ozone-rich air from aloft. This ozone-rich layer of air was a remnant of the previous day's photochemistry. An additional 40 ppb was added due to the current day's urban plume. These observations demonstrated that urban ozone can impact next day's ozone at downwind suburban locations (Baumann et al., 2000).

OC13. Based on modeling studies of ozone formation within an urban plume as it advects and mixes with the background atmosphere, Duncan and Chameides (1998) made the following conclusions.

1) A given day's photochemistry often builds on a regional background of ozone that was created in previous days.

2) Regional background ozone almost always is NOx-sensitive, and often cannot be attributed to a single NOx source or even a single NOx source region.

3) Urban plume ozone, on the other hand, can be either NOx-sensitive or VOC- sensitive.

4) In stagnation episodes, the ozone formed in an urban plume during the first day of an ozone episode can be very VOC-sensitive, and even increase further in ozone concentration if NOx emissions are decreased.

5) Model calculations show that decreases of NOx emissions in an urban area also decrease export of ozone from that urban area. This was true even in cases when the peak concentration of ozone accumulated during the ozone pollution episode is more effectively decreased by VOC emission controls than NOx controls.

6) Under advection conditions, decreases in NOx emissions generally provided more effective control of ozone pollution in both rural and urban areas of the SOS region.

7) Model calculations under advection conditions, also indicated a tendency for NOx sensitivity to increase and VOC sensitivity to decrease as the meteorological situation shifts from the stagnation conditions typical of extreme ozone episodes to more typical advection conditions.

8) Ozone production over the Nashville and/or Atlanta urban center can still be VOC-sensitive; but in contrast to stagnation episodes, however, the peak ozone concentration is more likely to occur downwind than in the urban core of the city.

9) Thus, optimally efficient and effective management of urban ozone and regional background ozone often can require different ozone precursor emissions control measures (Duncan and Chameides, 1998).

An importantpolicy implication from this set of scientific findings (OC13 above), is that control strategies aimed at decreasing emissions of NOx from sources that impact rural areas, especially well-vegetated areas, always are beneficial in terms of decreasing ozone exposures in rural areas. By contrast, however, decreasing NOx emissions in urban areas may be either beneficial or counter-beneficial, depending on whether the urban area of concern is VOC-sensitive or NOx-sensitive.

B. Ozone Production Efficiencies of VOC and NOx (OPE)

Resolution of the issue of relative ozone production efficiencies of VOC and NOx is perhaps the most important achievement of the SOS research and assessment program. Its importance lies, first, in the fact that the issue itself is critically important as it pertains to the relative merits of VOC and NOx controls. Thus, air quality managers need reliable evidence regarding such efficiencies for the purpose of determining whether to focus control efforts on VOC emissions or on NOx emissions (or on both).

Also, ozone production efficiencies of VOC and NOx serve as bases for emission trade-off strategies. The SOS achievement is remarkable also because the scientific issue of ozone production efficiencies of VOC and NOx is an extremely complex one, as the absolute and relative efficiencies of VOC and NOx are subject to influences from numerous factors and that these influences are often conflicting. Please note in the findings described below, that ozone formation conditions are often referred to as 'VOC-limited/sensitive' or 'NOx- limited/sensitive, meaning that, under such conditions, the VOC or the NOx precursor, respectively, is the more efficient producer of ozone.

OPE1. In the Houston-Galveston area of Texas during the month-long Texas 2000 Air Quality Study, ozone was produced very rapidly and very efficiently in downwind areas dominated by industrial sources. The rate of ozone formation in and around the industrial-source dominated areas in Houston was very high; ozone formation rates ranging between 50 ppbv/hr and 150 ppb/hour were measured on multiple days during the Texas 2000 Air Quality Study. These rates of ozone production are much greater than those observed in other urban areas in the US and Canada, which almost always are less than 40 ppb/hour (Daum et al., 2002).

OPE2. In Houston, Texas, high rates and high efficiencies of ozone formation can be explained by co-located emissions of VOC and NOx from industrial sources. High rates and high efficiencies of ozone production in the industrial plumes are driven by high concentrations of reactive hydrocarbons in the presence of NOx. The industrial plumes exhibiting rapid and efficient ozone formation also tend to exhibit a complex spatial structure (Daum et al., 2002; Kleinman et al., 2002).

OPE3. Ozone formation in the Nashville, Tennessee urban plume was VOC-limited during the extreme stagnation conditions observed in July 1995. Observed indicator ratios showed VOC-sensitive conditions within the Nashville urban core and NOx-sensitive conditions by the edges of the urban plume and in the background air (Valente et al., 1998).

OPE4. In Nashville under stagnation conditions, the urban plume started out being VOC-limited in the morning and remained that way for the remainder of high ozone days (Daum et al., 2000b).

OPE5. In Nashville, the amount of ozone formed under stagnation and advective conditions was similar. The total amount of ozone formed in the Nashville urban plume could be approximated by multiplying the daily average NOx emissions rate times the ozone production efficiency (OPEx equals the number of molecules of ozone formed per molecule of NOx that reacts). OPEx under stagnation and advective conditions was observed to be about the same. The plume from small power plants such as the Gallatin plant, which had an average NOx emission rate of 11,000 tons of NOx per year, had an OPEx similar to the Nashville urban plume (OPEX of about 3.0). By contrast, large power plants such as the Paradise plant, which hadan average NOx emission rate of 120,000 tons of NOx per year, had a lower OPEx of about 2.0 (Nunnermacker et al., 1998; Daum et al., 2000a).

OPE6. NOy concentrations in rural areas of the SOS region are generally very low, i.e., less than 4 ppbv. Under such conditions, ozone accumulation is largely limited by emissions of NOx (Kleinman, 1994).

OPE7. The yield of ozone from NOx is sometimes relatively large - as many as about 10 molecules of ozone being produced for each molecule of NOx emitted (Trainer et al., 1993). In other situations, however, the yield of ozone per molecule of NOx emitted can be relatively small, with as few as 1.0-7.0 molecules of ozone formed per molecule of NOx emitted. Observations in SOS's Nashville/Middle Tennessee Ozone Study indicated that the ozone production efficiency is lower (OPEx of 0.8 to 1.7 moles of ozone per mole of NOx) for isolated large power plants than for isolated small power plants (OPEx of 3.0 to 7.0). This difference in ozone production per unit of NOx emissions appears to be related to rapid NOy removal and may be associated with a more rapid conversion of NOx to HNO3 in power plant plumes than in urban plumes. Within 50 km from a small rural power plant, NOx oxidation proceeded most rapidly when the NOz:NOy ratio was greater than 0.7 (Ryerson et al., 1998).

OPE8. Measurements of total NO, and different chemical species of NOy (NO, NO2, nitric acid, PAN, organic and inorganic nitrates, and occasionally nitrous acid) provide significant insight into the chemical reactions that control the rate of ozone accumulation in many rural and some urban areas in the SOS region. It appears that a significant part of the variability in ozone concentrations from time to time and from place to place in the SOS region can be explained on the basis of variation in NOx concentration, as inferred from variation in observed NOy concentration (Trainer et al., 1993; Sillman and Samson, 1993; Trainer et al., 1995; Valente et al., 1998; Sillman et al., 1998; Nunnermacker et al., 1998, 2000; Daum et al., 2000a, 2000b).

OPE9. SOS evaluations of a variety of 'photochemical indicators' for use in identifying VOC-sensitive or NOx-sensitive areas and conditions have indicated that the VOC:NOx ratio is not a reliable diagnostic tool but that ozone sensitivity to VOC and NOx correlates well with four other photochemical indicators: 1) NOy concentration, 2) ozone:NOz ratio, 3) (HCHO concentration - minus 5ppbv):NOz ratio, and 4) H2O2:HNO3 ratio, with the last being most robust (Milford et al., 1994; Kleinman et al., 1994, 1995, 1997; Sillman, 1995a, 1995b, 1999; Sillman et al., 1997; Tonnesen and Dennis, 2000a, 2000b).

OPE10. Data from the 1992 SOS Intensive Field Study in Atlanta indicate that Atlanta is close to the transition between VOC-sensitivity and NOx-sensitivity, but that the city was in fact in the NOx-sensitive regime at the time of this 1992 Intensive Field Study (Imhoff et al., 1995; St. John et al., 1998; St. John and Chameides, 2000).

OPE11. Diagnostic analyses of photochemical smog reactions have uncovered a fundamental property of the photochemical system that explains why the atmosphere switches from a VOC-sensitive and NOx-sensitive regime as NOx concentrations decrease. This shift reflects the two different ways in which the atmosphere processes emissions inputs and the effect of these processes on the pool of reactive free radicals {Kleinman, 1994; Weinstein-Lloyd et al., 1998):

1) VOC-sensitive regimes within the atmosphere contain an excess of NOx sources over free radical sources. Thus, such regimes are characterized by: a) an abundance of NOx, b) a deficiency of free radicals, and c) HNO3 concentrations greater than H2O2 concentrations.

2) By contrast, NOx-sensitive regimes within the atmosphere contain an excess of free radicals over NOx sources. Thus such regimes are characterized by: a) a deficiency of NOx, b) an abundance of free radicals, and c) H2O2 concentrations greater than HNO3 concentrations.

OPE12. During the summers of 1994 and 1995, the urban plume of Nashville, Tennessee showed a stronger tendency to VOC limitation than the urban plume of Atlanta, Georgia during the summers of 1990, 1991, and 1992, or that of Los Angeles, California (Sillman et al., 1997).

OPE13. During the 1994-1995 Nashville/Middle Tennessee Ozone Study, power plant plumes typically decreased ozone concentration in the near field (0 to 20 km downwind) and increased ozone concentrations above the regional background further downwind (> 20 km). Ozone concentration increases as large as 50 ppb above the regional background were observed in these plumes. When the plume from a nearby power plant mixed with the Nashville urban plume on July 12, 1995, the power-plant contribution to the total increase on ozone concentration was estimated to be about 17 ppb or about 28 percent of the total increase in ozone concentration observed (Ryerson et al., 1998; Gillani et al., 1998a, 1998b).

OPE14. Most of the largest utility sources of NOx are found in rural areas of the SOS region, which have high emissions of biogenic VOC (USEPA, 1997).

OPE15. Ozone production efficiency was observed to have an inverse relationship with NOx emission rate at several coal-fired power plants in the SOS region during the summer of 1995. (Ryerson et al., 1998; Sillman, 2000; Nunnermacker et al., 2000).

The policy implication of scientific finding OPE15 is that, for large power plant sources, the benefits of decreasing emissions of NOx are partially offset by an increase in the efficiency with which ozone is formed.

OPE16. NOy concentrations decreased very rapidly in summertime urban plumes and power plant plumes near Nashville, Tennessee. The rate of depletion of NOy in these plumes was more rapid than can be accounted for by any known NOy-removal process (Gillani et al., 1998a, 1998b).

The unexpectedly rapid rates of depletion described in scientific finding OPE16 suggest that either:

  1. These plumes produce an unknown reaction product that was not detected at any of the ground-based or aircraft-based NOy measurement systems used in the Nashville/Middle Tennessee Ozone Study; or
  2. Rates of deposition of NOy to vegetation or other natural or man-made surfaces near the ground are much greater than have been reported before.

The rapid rate of NOy depletion described in scientific finding OPE16 also implies that:

  1. Ozone production rates based on ozone:NOy or ozone:NOz relationships may significantly overestimate the ozone production efficiency of both urban plumes and power plant plumes, or
  2. NOx emissions may not be transported as far as is commonly assumed in current air quality models such as UAM-IV, UAM-V, MM5, ROM; RADM, etc.

OPE 17. Peroxide concentrations were lower in both urban plumes and power plant plumes than in background air near Nashville, Tennessee. The lower rates of formation of peroxides within these plumes are believed to be the result of higher NOx concentrations within than outside the plumes; this maximizes radical consumption by NO2 to form nitric acid as opposed to radical combination reactions that form peroxides (Jobson et al., 1998; Weinstein-Lloyd et al., 1998).

OPE18. Carbon monoxide and methane made significant contributions (~19 and ~3 percent, respectively) to ozone formation in regions where isoprene emissions were relatively small, i.e., in the Nashville urban plume (Nunnermacker et al., 1998).

The scientific finding described in OPE18 indicates that carbon monoxide and methane will need to be considered as more important precursors of ozone than has generally has been recognized in past ozone management strategies. The role of methane and carbon monoxide in ozone formation is expected to increase still further as air quality managers continue to decrease emissions of the more reactive anthropogenic VOC.