Preview of 1994 Ozone Precursor Concentrations in the Norteastern U.S.
5.1 Total (Targeted) VOCs
While 6 to 9 AM measurements of total non-methane hydrocarbon (TNMHC) concentrations are readily available from PAMS sites, a major strength of the PAMS program is that speciated hydrocarbon measurements are made on a continuous (hourly or 3-hour) basis for up to 55 targeted hydrocarbon (and 3 carbonyl) compounds. The abundance of targeted VOCs (i.e., compounds for which the automated GCs are calibrated) from the five sites for which hourly data is available is shown in Table 5.1. Dashed lines show the proposed limit of quantitation (LOQ) above which, target precision of 25% is expected. This sensitivity is well within the capabilities of the systems in use and although rigorous detection level tests are not fullyin place, it is felt that the indicated LOQ is appropriate. The last column in Table 5.1 shows the regional average VOC levels arrived at by averaging the hourly averages of each compound from each site.
Figure 5.1 Speciated and Unspeciated VOCs at NESCAUM PAMS Sites During 7/94 Episodes
Figure 5.1 displays the fraction of average TNMHC concentrations which are accounted for by targeted compounds (above LOQ) at NESCAUM PAMS sites during the 1994 episode periods. Two of the Type II urban sites, Lynn and East Hartford, show very similar total VOC values and are about four times higher than the totals detected at the remote Cape Elizabeth site. Chicopee, the Springfield Type II site, shows a uniquely high total (see Section 6.3), while total VOC levels at the Stafford, type III site are significantly below urban site levels but well above remote levels. At the urban Type II sites, 65 to 75% of the average TNMHC concentration is accounted for by targeted species present above the LOQ. This fraction drops to about 45% at the rural Stafford, CT site and 25% at the remote Cape Elizabeth site - where a high fraction of compounds are below the (assumed) 1 ppbc LOQ. It is anticipated that the quantified fractions of targeted species will increase in the future.
Figure 5.2 is a graphical representation of the data in Table 5.1 and includes all targeted hydrocarbons (and carbonyls) regardless of relation to LOQ. The 10 most abundant VOCs at Chicopee account for 66% of the total, at East Hartford the top 10 are 71% of the total and at Lynn the top 10 comprise 74% of the total. A different pattern emerges at the more remote Stafford and Cape Elizabeth sites where the 10 most abundant VOCs account for 93% and 100% (9 compounds only), respectively, of the total targeted VOCs.
Table 5.1 VOC Abundance in ppbc (Hourly Averages for 7/6-8/94 and 7/20-22/94)
Figure 5.2 Average Hourly Concentration by Compound (ppbc)
Regardless of the proportion of "top 10 to total", the most prevalent species are remarkably consistent across the region for the time period of interest. Seven compounds: isopentane, toluene, propane, ethane, n-butane, m&p-xylene and n-pentane are on the top 10 most-occurring lists at all sites but Cape Elizabeth (five of the seven, however are on that sites' top ten list). Isoprene (a biogenic), also ranks high on the abundance list at four of the five sites but was eleventh in abundance at East Hartford where a misidentification is thought to have biased the average on the low side. This pattern of similar abundances throughout the urbanized portion of the region (and to some extent, the remote portions i.e. Cape Elizabeth) is perhaps to be expected given the ubiquity of mobile sources.
Table 5.2 Most Abundant Anthropogenic VOCs in Selected Measurement Campaigns
An earlier study, limited to the 6-9 a.m. time period in five Northeastern cities3, found six of the same seven compounds, as noted above, leading the VOC abundance list. Another, more extensive study of speciated VOCs4 in the Los Angeles area, found the same seven anthropogenic compounds to be most abundant and, with the exception of propane, in virtually the same rank of occurrence (see Table 5.2). Note also the similar distribution of HCs at the New York City site in Section 8.
It is worth noting that all carbonyl compounds required under the PAMS program (formaldehyde, acetaldehyde and acetone) displayed a similar pattern across the region. On a ppbc basis (Figure 5.3), acetone is most abundant followed by formaldehyde and then acetaldehyde. This is a departure from the pattern found in the Los Angeles area in that formaldehyde occurs as the second most abundant carbonyl in the Northeast while it is fourth in abundance in L.A
Figure 5.3 Average Carbonyls (ppbc) at NESCAUM PAMS Sites during July, 1994 Episodes ]
Figure 5.4 Average Carbonyls (ppbv) at NESCAUM PAMS Sites during July, 1994 Episodes
The units of ppbc are used (above) because dispersion models utilizing hydrocarbon input require those units. Converting to ppbv (units to which human lungs can relate), shows (Figure 5.4) that in the northeast, formaldehyde is the most abundant followed by acetone and acetaldehyde. While all sites exhibit a similar rank of occurrence, the average levels of formaldehyde are highest at Lynn, acetaldehyde highest at East Providence and acetone highest at Chicopee.
Figure 5.5 Diurnal Carbonyl Cycles for July, 1994 (aggregated for all NESCAUM Type II Sites)
Figure 5.5 displays the average diurnal patterns for the carbonyl concentrations aggregated for the East Hartford, Lynn, Chicopee and East Providence sites for the entire month of July, 1994. On a regional average basis, formaldehyde exhibits the most extreme diurnal variation - suggesting a predominant photochemical production mechanism. Conceivably, the secondary morning and afternoon peaks might be related to primary emissions from automotive sources. Alternatively, the slight, "mid-day dip" might be indicative of a photochemical destruction mechanism.
VOCs vary in their ability to form ozone because they differ in both composition (what they are made of) and in structure (how they are put together). Control strategists face the dilemma of A) treating all VOCs the same (an oversimplification) or B) applying different reactivities to each VOC (a difficult and imperfect process). Traditional reactivity scales have been referenced to a single "base" compound (e.g., carbon monoxide, propene or ethylene) for simplicity but such systems have been criticized as only representing initial reaction rates when, in fact, polluted atmospheres vary considerably during the course of the day as compounds are depleted and replenished constantly. Carter5 has developed a more sophisticated scheme that uses a mixture of all VOCs (except biogenics) as the base.
Regardless of the base, systematic similarities appear in reactivity schemes. For example, members of the alkene family generally posses the greatest ozone forming potential (OFP). Aldehydes too, are great ozone promoters, although acetone (measured along with formaldehyde and acetaldehyde) is not an aldehyde and has a very low reactivity.
Figure 5.6 attempts to show the relative ozone forming potential of the various VOC and carbonyl species. The hourly average abundances have been scaled by Carters maximum incremental reactivities (MIRs) and displayed in units of (potential) ppbv of ozone. Under the appropriate VOC/NOx conditions, the various anthropogenic compounds could produce the amount of ozone indicated. Note that while isopentane and acetone are near the top of the abundance list, their low reactivities result in rather low ozone forming potential. Note also the extreme OFP of formaldehyde which averaged 5-6 ppbc at regional sites but could potentially produce 20-30 ppbv of ozone. Although this data is preliminary and covers a short time span it is believed to be a valid "snapshot" of upper limit conditions that typically exist in the northeast. Future control strategies should be preceded by more rigorous and more encompassing analyses that take into account both the abundance and ozone forming potential of regional VOCs.
Figure 5.6 Average Reactivity-Weighted (MIR) VOC Concentrations (Y-scale units are potential ozone in ppbv)
Figure 5.7 Diurnal Isoprene and m/p Xylene (averaged for 6 NESCAUM PAMS Sites for July, 1994 Episodes)
Figure 5.7 displays the average diurnal concentrations for m/p Xylene and Isoprene, averaged over the 6 NESCAUM PAMS sites for the two July, 1994 episodes. These highly reactive VOCs have similar MIR reactivities (7.4 and 9.1, respectively).
The diurnal pattern for m/p xylene is similar to a number of other reactive, anthropogenic VOCs (toluene, o-xylene, isopentane, etc.), with emissions generally dominated by automotive-related sources. Isoprene is emitted predominantly by deciduous vegetation, as a function of solar radiation and temperature. It is not clear whether its diurnal emission pattern is similar to other reactive biogenic VOCs which are not (yet) quantified by PAMS methods.
While the morning (6 to 9 AM) levels (and MIR values) of m/p xylene and isoprene are quite similar, the reactive anthropogenic pollutants are generally depleted rapidly during the day. Although isoprene also reacts rapidly, its rate of production exceeds its rate of destruction during mid-day. Formaldehyde also has a very high MIR value (Figure 5.6) and exhibits a strong mid-day peak (Figure 5.5). Consequently, the contributions of isoprene and formaldehyde to ozone production may be grossly underestimated by methods which fail to consider differential reactivity, and/or which focus on morning values only. Note also the distinct morning and afternoon dips in the otherwise smooth regional isoprene pattern in figure 5.7. Conceivably, these might be related to temporary rush hour increases in NOx emissions, and subsequent photochemical reactions. These isoprene dips also correspond approximately to the slight increases in morning and afternoon formaldehyde levels.
Isoprene is a naturally occurring VOC produced mainly by deciduous trees and shrubs. Trace quantities of this substance have also been reported in the exhaust of spark ignited engines (SEIs), although it is anticipated that anthropogenic isoprene emissions are relatively insignificant in comparison to biogenic emissions. The most complete research on the role of biogenic VOC emissions on ozone photochemistry has been done by the Southern Oxidants Study (SOS). Recent conclusions drawn by SOS research scientists are:
a) biogenic emissions in the South play an overwhelming role in oxidant production because they are the dominant sources of photochemically active carbonyls and radicals in both rural and urban areas,
b) isoprene, mono-terpenes and alcohols are important biogenic compounds, however, isoprene is the dominant biogenic VOC,
c) averaged over the summer, biogenic VOC emissions exceed anthropogenic VOC emissions, and
d) biogenic emissions from southern forests exceed those of any other region in the United States6,7.
Average diurnal Isoprene levels for the NESCAUM PAMS sites during the July 1994 episode periods are displayed in Figures 5.8 and 5.9 (note factor of 2 difference in Y scales).
Figure 5.8 Average Diurnal Isoprene at Selected NESCAUM PAMS Sites During July, 1994 Episodes
Figure 5.9 Average Diurnal Isoprene at Selected NESCAUM PAMS Sites During July, 1994 Episodes
Concentrations at the Stafford, Lynn and Chicopee sites are similar to each other, roughly twice as high as levels at the E. Providence, E. Hartford and Cape Elizabeth sites, and exhibit a relatively distinct mid-day peak. Each of these sites in Figure 5.8 also exhibits distinct mid-morning and mid-afternoon dips. The sites displayed in Figure 5.9 show lower concentrations and a different diurnal pattern - with peak concentrations in early morning and late afternoon, and a distinct dip at mid-day.
Figure 5.10 Isoprene vs. Temperature at Stafford, CT PAMS Site During July, 1994 Episodes (for temperatures >= 70 Degrees F)
Biogenic isoprene emissions from a given mix of vegetation types are theoretically a function of leaf surface temperature - which relates in turn to both ambient temperature and solar radiation. In the absence of cloud cover, we would expect isoprene emissions to exhibit a relative smooth diurnal cycle - which approximately tracks the diurnal cycles of solar radiation (and temperature). Isoprene concentrations should also track this smooth diurnal cycle, but depart from it as a function of meteorological influences and photochemical isoprene destruction processes.
The relationship between isoprene concentration and temperature for the rural, forested Stafford, CT site during the July, 1994 episodes is displayed in Figure 5.10. At temperatures < 70 degrees F, there was little if any correlation, but above 70, the correlation is relatively strong (R2 = 0.61). This relationship will likely improve if solar radiation data (not available for this site/time) were used in place of temperature, and/or if only daytime data were employed (isoprene levels typically fall to zero at night when most leaf stomata are closed).
Figure 5.11 Hypothetical (temperature-based) Diurnal Isoprene Emissions & Concentrations at Stafford CT during 7/94 Episodes
Figure 5.11 compares a hypothetical diurnal isoprene emission curve (based on the equation in Figure 5.10) with average ambient levels at the rural Stafford, CT site. While the emission estimate is intended as only a crude approximation, it may be instructive to consider the difference between these 2 curves as representing the extent to which isoprene is diluted by meteorological influences and/or consumed by photochemical reactions - or more precisely, the extent to which isoprene destruction processes exceed isoprene production processes. For the sites in Figure 5.9, we might also envision a relatively smooth diurnal isoprene production curve, but at these sites, mid-day isoprene dilution or destruction rates may substantially exceed production rates.
Alternative explanations for the unexpected diurnal patterns in Figure 5.9 include: measurement errors (probable at E. Hartford, and currently under investigation); mid-day wind shifts (for example to off-shore winds at coastal/ near coastal Cape Elizabeth and E. Providence); vegetation characterized by isoprene/temperature relationships which differ from those at Stafford. It is also quite possible that isoprene emissions may be reduced at mid-day under high temperature conditions, which may cause leaf stomata to close to conserve moisture. Conceivably, vegetation influencing isoprene levels at E. Hartford, E. Providence and Cape Elizabeth may have been subject to a greater degree of local moisture stress than was experienced near the other 3 NESCAUM PAMS sites during these episodes.
For comparison, average diurnal isoprene values from (urban and rural sites in) the Atlanta area during the summer of 1990 are displayed in Figures 5.12 and 5.13. Values in Figure 5.12 are averaged over the months of July and August, 1990, while Figure 5.13 is based on a 10-day period (August 15-20) of extremely high temperature (considered more directly comparable to the NESCAUM July, 1994 episode periods).Two features of the Atlanta isoprene data are noteworthy by comparison to the July, 1994 NESCAUM episode data in Figures 5.8 and 5.9. Peak concentrations during high temperature episodes in the Southeast and (3 sites in) the Northeast and are approximately the same (it had previously been assumed that isoprene levels were substantially higher in the Southeast).
Figure 5.12 Average Diurnal Isoprene at Atlanta Sites during July and August, 1990
Figure 5.13 Average Diurnal Isoprene at Atlanta Sites during August 15 to 25, 1990
Second, the diurnal patterns in Atlanta - peaks in the morning and late afternoon with a mid-day dip - are similar to the patterns at E. Providence, E. Hartford and Cape Elizabeth (suggesting a common mid-day isoprene dilution, destruction and/or emissions suppression mechanism in both regions).
Again, it should be cautioned that the July 1994 NESCAUM episodes represent a very small sample (6 days at 6 sites), and that these episodes were generally characterized by exceptionally high temperatures - such that the associated isoprene measurements may be atypical. On the other hand, such multi-day, multi-site episodes occur with some regularity in the Northeast each summer, and account for a high fraction of the region's ozone exceedances - such that these high isoprene concentrations may be quite typical of levels on days when federal ozone standards are exceeded. Given its high concentrations and high incremental reactivity, isoprene may be one of the most important VOCs influencing ozone production in the Northeast. Isoprene is also thought to be an important precursor in the formation of other reactive (and/or toxic) species, including formaldehyde, methyl vinyl ketone, peroxyl radicals and peroxyacetyl nitrate (PAN).
Recent research results from the Southern Oxidant Study (SOS, 1995) indicate that EPA's currently recommended Biogenic Emission Inventory System (BEIS) underestimates isoprene emissions in the Southeast by a factor of 5 to 10, and that the Urban Airshed Model (UAM) consistently under predicts isoprene concentrations. Recently proposed revisions to the biogenic emissions algorithms (BIES2) include substantial increases in emission factors for oaks and spruce - which may have significant implications in the Northeast. It may also be noteworthy that current models disaggregate isoprene emission estimates on an hourly basis by algorithms which are primarily a direct function of ambient temperature and solar radiation. If diurnal isoprene emissions can be suppressed by leaf stomatal closure under high temperature/ moisture stress conditions, these model algorithms may need to be reconsidered. This possibility may also have implications for ozone effects research, particularly if stomata are frequently closed during mid-day periods of maximum ozone exposure.
A number of the targeted PAMS compounds are inherently toxic. Several, like benzene and formaldehyde are known or suspected carcinogens, and may pose greater health risks from direct human exposures than through their (indirect) contributions to ozone formation. Benzene, formaldehyde and a number (?) of other
targeted PAMS compounds have also been measured in several National toxics monitoring programs, most notably the EPA Urban Air Toxics (UAT) program, initiated in 1988. During 1993 and 1994, the UAT program was temporarily folded in to the EPA CASTNet program (Clean Air Status and Trends Network). The UAT and CASTNet programs employ analytical methods which are generally comparable to PAMS, although the sample duration (24 hour), frequency (usually every 12th day) and seasonality (year-round) are different. Like PAMS, the toxics sampling programs also have a wide range of site locations - from dense emission areas to small rural villages. Therefore we might expect the PAMS and UAT/CASTNet data to be relatively comparable. The toxics networks will enhance the National spatial and seasonal coverage for PAMS; while the high time resolution of PAMS data can provide additional information about human diurnal exposure patterns, source origins and photochemical production and destruction mechanisms.
Some CASTNet and PAMS benzene data are available for comparison for the Summer of 1993. For CASTNet sites, the data include 12th-day, 24-hour samples during the months of July through September, 1993. For PAMS sites, the approximately comparable data includes hourly or 3-hour samples during the months of June through August, 1993. The average values for these 1993 benzene data are plotted below in Figure 5.14. For comparison, the average values for the NESCAUM PAMS sites during the July, 1994 episode periods are also displayed. The estimated cancer risks are based on an assumption of long-term (70 yr.) exposure, and an assumed cancer risk of 10-6 from a long-term exposure of 0.12 ug/m3. These data are presented in this way for illustrative purposes only, as there are substantial uncertainties in the quality of the data, the representativeness of data for long-term exposures, and the appropriate reference dose for benzene. With these qualifications, several features are noteworthy:
- average benzene concentrations during Summer, 1993 are very similar at PAMS and CASTNet sites,
- estimated cancer risk from long-term exposures to these kind of benzene levels is between 10-4 and 10-5 at urban sites throughout the country, and somewhat lower (10-5 to 10-6) at more rural sites,
- average benzene levels at the NESCAUM sites during the studied July, 1994 episode periods were generally lower than from the Summer, 1993 PAMS and CASTNet sites.
Figure 5.14 Estimated Cancer Risk from Benzene (Assuming Long-Term Exposures) from Summer, 1993 CASTNet and PAMS Data, and during July, 1994 Ozone Episodes
This latter observation (based on a very small sample size) does not necessarily imply that benzene levels are lower in the Northeast than in other regions, or that benzene levels have declined from 1993 to 1994. An alternative explanation is that Northeastern benzene levels may have been atypically low during the studied episode periods - which were specifically selected because of high regional ozone levels. Benzene is relatively non-reactive (MIR = 0.42), and is not anticipated to make a substantial contribution to ozone formation under most conditions. Diurnal (automotive-related) and seasonal emissions cycles of benzene, combined with effects of meteorological dispersion, are likely to be out of sync with ozone formation cycles. Photochemical conditions that contribute to maximum ozone formation are also likely to lead to maximum benzene destruction. So it is perhaps not surprising that the studied July 1994 episodes - with high regional ozone levels - are associated with relatively low benzene concentrations.
Figure 5.15 Ozone vs. Benzene at 1993 PAMS Sites
Figure 5.16 Ozone vs. Formaldehyde at '93 PAMS Sites
Figure 5.15 scatters hourly ozone and benzene from all available US PAMS sites for the Summer (JJA) of 1993, and shows a weak, negative correlation (O3 = 36 - 1.3 benzene, R=0.21). For comparison, Figure 5.16 scatters hourly ozone with 3-hour formaldehyde from the available 1993 PAMS sites. In this case, each plotted data point compares a 3-hour formaldehyde level with the hourly ozone at the end of that 3-hour period. Ozone and formaldehyde show a weak, positive correlation (O3 = 18 + 2 HCHO, R = 0.29).
Figure 5.17 Estimated Cancer Risks from Formaldehyde (Assuming Long-Term Exposures) from Summer, '93 PAMS and CASTNet Data, and from July, '94 - (Monthly Means and Ozone Episodes)
Figure 5.17 shows the estimated cancer risks from Formaldehyde based on limited measurements from Summer 1993 CASTNet and PAMS sites for assumed long-term (70 yr.) exposures and an assumed 10-6 risk associated with long-term concentrations of 0.08 ug/m3. NESCAUM site values for the month of July, 1994 and for the studied July, 1994 episode periods are also plotted here for comparison. As with the estimated risks displayed in Figure 5.14, these data are plotted this way for illustrative purposes only, are characterized by high uncertainties, and should be interpreted with caution.
With the above qualifications, several features of Figure 5.17 are notworthy:
- Average formaldehyde levels during Summer, 1993 at (the small number of available) PAMS sites are somewhat higher than Summer, 1993 levels at (the few available) CASTNet sites.
- Estimated cancer risks from formaldehyde are somewhat higher than from benzene, approaching 10-4.
- Formaldehyde levels during the studied July, 1994 ozone episodes were higher than average values for that month at these sites, and were as high or higher than at Summer 1993 PAMS and CASTNet sites.
The relatively high regional formaldehyde and low regional benzene during the studied ozone episode periods are consistent with the (Figures 5.16 and 5.15) positive and negative correlations for all 1993 PAMS sites. These (weak) statistical relationships do not necessarily indicate positive or negative causality.
However, it appears that periods of high ozone exposure frequently tend to be associated with relatively high formaldehyde (and low benzene). High ozone levels in the Northeast also frequenty occur concurrently with episodes of PM-10, fine particles, toxic metals, acidic sulfates and organic aerosols (see NESCAUM Report: "1992 Regional Ozone Concentrations in the Northeastern U.S.")8. For example, July 7, 1994 - the only 6th day PM-10 sampling day during the studied episode periods - was characterized by high regional PM-10 concentrations, exhibiting a similar spatial pattern to the high, afternoon ozone values on that day (note: PM-10 data from PA and more southerly states were not available at time of extraction).
Figure 5.18 Comparison of Daily PM-10 and 3 PM Ozone Concentrations on July 7, 1994
From From a toxicological perspective, the nature (and effects) of combined (short and long-term) exposures to multiple pollutants is an important consideration and area for needed future investigations. The PAMS, Urban Air Toxics, and pending future fine particle data promise to provide much useful information to support these investigations, and to evaluate potential effects of alternative control strategies.
