Ozone Excerpts 3

Excerpts Taken From

Integrated Science Assessment for Ozone and
Related Photochemical Oxidants

U.S. Environmental Protection Agency
EPA/600/R-10/076B
September 2011

Excerpted by Michael H. Klein

Page 3-5 line 19 (ARP 148)

Anthropogenic NO_X emissions are associated with combustion processes. Most emissions are in the form of NO, which is formed at high combustion temperatures from atmospheric nitrogen (N₂) and oxygen (O₂) and from fuel nitrogen (N). According to the 2005 National Emissions Inventory, the largest sources of NO_X are on- and off-road mobile sources and electric power generation plants. Emissions of NO_X therefore are highest in areas having a high density of power plants and in urban regions having high traffic density. Dallman and Harley compared NO_X emissions estimates from the National Emissions Inventory, mobile sector data with an alternative method based on fuel consumption and found reasonable agreement in total U.S. anthropogenic emissions between the two techniques (to within about 5%). However, emissions from on-road diesel engines in the fuel based inventory constituted 46% of total mobile source NO_X compared to 35% in the EPA inventory. As a result, emissions from on-road diesel engines in the fuel based approach are even larger than electric power generation as estimated in the 2005 NEI, and on-road diesel engines might represent the largest single NO_X source category. Differences between the two techniques are largely accounted for by differences in emissions from on-road gasoline engines. Uncertainties in the fuel consumption inventory ranged from 3% for on-road gasoline engines to 20% for marine sources, and in the EPA inventory uncertainties ranged from 16% for locomotives to 30% for off-road diesel engines. It should be noted that the on-road diesel engine emissions estimate by Dallman and Harley is still within the uncertainty of the EPA estimate (22%).

Page 3-6 line 4 (ARP 149)

Major natural sources of NO_X in the U.S. include lightning, soils, and wildfires. Uncertainties in natural NO_X emissions are much larger than for anthropogenic NO_X emissions. Fang et al. estimated lightning generated NO_X of ~0.6 MT for July 1 2004. This value is ~40% of the anthropogenic emissions for the same period, but Fang et al. estimated that ~98% is formed in the free troposphere and so contributions to the surface NO_X burden are low because most of this NO_X is oxidized to nitrate containing species during downward transport into the planetary boundary layer. The remaining 2% is formed within the planetary boundary layer. Both nitrifying and denitrifying organisms in the soil can produce NO_X, mainly in the form of NO. Emission rates depend mainly on fertilization amount and soil temperature and moisture. Nationwide, about 60% of the total NO_X emitted by soils is estimated to occur in the central corn belt of the U.S. Spatial and temporal variability in soil NO_X emissions leads to considerable uncertainty in emissions estimates. However, these emissions are relatively low, only ~0.97 MT/year, or about 6% of anthropogenic NO_X emissions. However, these emissions occur mainly during summer when O₃ is of most concern.

Hundreds of VOCs, containing mainly 2 to ~12 carbon (C) atoms, are emitted by evaporation and combustion processes from a large number of anthropogenic sources. The two largest anthropogenic source categories in the U.S. EPA’s emissions inventories are industrial processes and transportation. Emissions of VOCs from highway vehicles account for roughly two-thirds of the transportation-related emissions. The accuracy of VOC emission estimates is difficult to determine, both for stationary and mobile sources. Evaporative emissions, which depend on temperature and other environmental factors, compound the difficulties of assigning accurate emission factors. In assigning VOC emission estimates to the mobile source category, models are used that incorporate numerous input parameters (e.g., type of fuel used, type of emission controls, and age of vehicle), each of which has some degree of uncertainty.

On the U.S. and global scales, emissions of VOCs from vegetation are much larger than those from anthropogenic sources. Emissions of VOCs from anthropogenic sources in the 2005 NEI were ~17 MT/year (wildfires constitute ~1/6 of that total and were included in the 2005 NEI under the anthropogenic category, but see Section 3.4 for how wildfires are treated for background.), but were 29 MT/year from biogenic sources. Uncertainties in both biogenic and anthropogenic VOC emission inventories prevent determination of the relative contributions of these two categories, at least in many areas. Vegetation emits significant quantities of VOCs, such as terpenoid compounds (isoprene, 2-methyl-3- 32 buten-2-ol, monoterpenes), compounds in the hexanal family, alkenes, aldehydes, organic 33 acids, alcohols, ketones, and alkanes. The major chemicals emitted by plants are isoprene (40%), other terpenoid and sesqui-terpenoid compounds (25%) and the remainder consists of assorted oxygenated compounds and hydrocarbons according to the 2005 NEI. Coniferous forests represent the largest source on a nationwide basis because of their extensive land coverage. Most biogenic emissions occur during the summer because of their dependence on temperature and incident sunlight. Biogenic emissions are also higher in southern states than in northern states for these reasons and because of species variations. The uncertainty in natural emissions is about 50% for isoprene under midday summer conditions and could be as much as a factor of ten higher for some compounds. In EPA’s regional modeling efforts, biogenic emissions of VOCs are estimated using the BEIS model with data from the Biogenic Emissions Landcover Database (BELD) and annual meteorological data. However, other emissions models are used such as MEGAN (Model of Emissions of Gases and Aerosols from Nature), especially in global modeling efforts.

Anthropogenic CO is emitted primarily by incomplete combustion of carbon-containing fuels. In general, any increase in fuel O₂ content, burn temperature, or mixing time in the combustion zone will tend to decrease production of CO relative to CO₂. However, it should be noted that controls mute the response of CO formation to fuel-oxygen. CO emissions from large fossil-fueled power plants are typically very low since the boilers at these plants are tuned for highly efficient combustion with the lowest possible fuel consumption. Additionally, the CO-to-CO₂ ratio in these emissions is shifted toward CO₂ by allowing time for the furnace flue gases to mix with air and be oxidized by OH to CO₂ in the hot gas stream before the OH concentrations drop as the flue gases cool,. Nationally, on-road mobile sources constituted about half of total CO emissions in the 2005 NEI. When emissions from non-road vehicles are included, it can be seen from Figure 3-2 that all mobile sources accounted for about three-quarters of total anthropogenic CO emissions in the U.S.

Page 3-15 line 15 (ARP 158)

Except when activities such as photocopying or welding are occurring, the major source of O₃ to indoor air is through infiltration of outdoor air. Reactions involving ambient O₃ with NO either from exhaled breath or from gas-fired appliances, surfaces of furnishings and terpenoid compounds from cleaning products, air fresheners and wood products also occur in indoor air as was discussed in the previous O₃ AQCD. The previous O₃ AQCD also noted that the ozonolysis of terpenoid compounds could be a significant source of secondary organic aerosol in the ultrafine size fraction. Chen et al. examined the formation of secondary organic aerosol from the reaction of O₃ that has infiltrated indoors with terpenoid components of commonly used air fresheners. They focused on the formation and decay of particle bound reactive oxygen species (ROS) and on their chemical properties. They found that the ROS content of samples can be decomposed into fractions that differ in terms of reactivity and volatility, however the overall ROS content of samples decays and over 90% is lost within a day at room temperature. This result also suggests loss of ROS during sampling periods longer than a couple of hours.

Page 3-15 line 29 (ARP 158)

As might be expected based on the temperature dependence of many reactions involved in the production and destruction of O₃ and the temperature dependence of emissions processes such as evaporation of hydrocarbon precursors and the emissions of biogenically important precursors such as isoprene, ambient concentrations of O₃ also show temperature dependence. Bloomer et al. determined the sensitivity of O₃ to temperature at rural sites in the eastern U.S. They found that O₃ increased on average at rural (CASTNET) sites by ~3.2 ppbv/ºC before 2002, and after 2002 by ~2.2 ppbv/ºC. This change in sensitivity was largely the result of reductions in NO_X emissions from power plants. These results are in accord with model predictions by Wu et al. showing that the sensitivity of O₃ to temperature decreases with decreases in precursor emissions. However, this study was basically confined to the eastern U.S., but results from sites downwind of Phoenix, AZ showed basically no sensitivity of O₃ to temperature (R²=0.02). However, Wise and Comrie did find that meteorological parameters (mixing height and temperature) typically accounts for 40 to 70% of the variability in O₃ in the five southwestern cities (including Phoenix) they examined. It is likely that differences in the nature of sites chosen (urban vs. rural) accounted for this difference and is at least partially responsible for the difference in results. Jaffe et al. regressed O₃ on temperature at Yellowstone and Rocky Mountain NP and found weak associations (R² = 0.09 and 0.16). They found that associations with area burned by wildfires are much stronger. These results demonstrate that the associations of O₃ with temperature are not as clear in the West as they might be in the East. Other sources as discussed in Section 3.4 might also be more important in the West than in the East.

Page 3-29 line 1 (ARP 172)

Background concentrations of O₃ have been given various definitions in the literature over time. In the context of a review of the NAAQS, it is useful to define background O₃ concentrations in a way that distinguishes between concentrations that result from precursor emissions that are relatively less directly controllable from those that are relatively more directly controllable through U.S. policies. North American (NA) background O₃ can include contributions that result from emissions from natural sources (e.g., stratospheric intrusion, biogenic methane and more short-lived VOC emissions), emissions of pollutants that contribute to global concentrations of O₃ (e.g., anthropogenic methane) from countries outside North America. In previous NAAQS reviews, a specific definition of background concentrations was used and referred to as policy relevant background (PRB). In those previous reviews, PRB concentrations were defined by EPA as those concentrations that would occur in the U.S. in the absence of anthropogenic emissions in continental North America (CNA), defined here as the U.S., Canada, and Mexico. For this document, we have focused on the sum of those background concentrations from natural sources everywhere in the world and from anthropogenic sources outside CNA. North American background concentrations so defined facilitate separation of pollution that can be controlled directly by U.S. regulations or through international agreements with neighboring countries from that which would require more comprehensive international agreements, such as are being discussed as part of the United Nations sponsored Convention on Long Range Transboundary Air Pollution Task Force on Hemispheric Air Pollution. There is no chemical difference between background O₃ and O₃ attributable to CNA anthropogenic sources, and background concentrations can contribute to the risk of health effects. However, to inform policy considerations regarding the current or potential alternative standards, it is useful to understand how total O₃ concentrations can be attributed to different sources.

Contributions to NA background O₃ include photochemical reactions involving natural emissions of VOCs, NO_X, and CO as well as the long-range transport of O₃ and its precursors from outside CNA, and the stratospheric-tropospheric exchange (STE) of O₃. These sources have the greatest potential for producing the highest background concentrations, and therefore are discussed in greater detail below.

Natural sources of O₃ precursors include biogenic emissions, wildfires, and lightning. Biogenic emissions from agricultural activities in CNA are not considered in the formation of NA background O₃. Sources included in the definition of NA background O₃ are shown schematically in Figure 3-7. Definitions of background and approaches to derive background concentrations were reviewed in the 2006 O₃ AQCD and in Reid et al. In addition to emissions from North America, emissions from Eurasia have contributed to the global burden of O₃ in the atmosphere and to the U.S Because the mean tropospheric lifetime of O₃ is 30-35 days, O₃ can be transported from continent to continent and around the globe in the Northern Hemisphere and O₃ produced by U.S. emissions can be recirculated around northern mid-latitudes back to the U.S. High elevation sites are most susceptible to the intercontinental transport of pollution especially during spring.

Page 3-33 line 4 (ARP 176)

The basic atmospheric dynamics and thermodynamics of STE [stratospheric-tropospheric exchange – editor] were outlined in the 2006 O₃ AQCD; as noted there, stratospheric air rich in O₃ is transported into the troposphere. Ozone is produced naturally by photochemical reactions in the stratosphere as shown in Figure 3-1 in Section 3.2. Some of this O₃ is transported downward into the troposphere throughout the year, with maximum contributions at mid-latitudes during late winter and early spring mainly coming from a process known as tropopause folding. These folds occur behind most cold fronts, bringing stratospheric air with them. The tropopause should not be interpreted as a material surface through which there is no exchange. Rather these folds should be thought of as regions in which mixing of tropospheric and stratospheric air is occurring. This imported stratospheric air contributes to the natural background of O₃ in the troposphere, especially in the free troposphere during winter and spring. STE also occurs during other seasons including summer.

Page 3-35 line 24 (ARP 178)

Biomass burning consists of wildfires and the intentional burning of vegetation to clear new land for agriculture and for population resettlement; to control the growth of unwanted plants on pasture land; to manage forest resources with prescribed burning; to dispose of agricultural and domestic waste; and as fuel for cooking, heating, and water sterilization. Globally, most wildfires may be ignited directly as the result of human activities, leaving only 10-30% initiated by lightning. However, because fire management practices suppress natural wildfires, the buildup of fire fuels increases the susceptibility of forests to more severe but less frequent fires in the future. Thus there is considerable uncertainty in attributing the fraction of wildfire emissions to human activities because the emissions from naturally occurring fires that would have been present in the absence of fire suppression practices are not known. Contributions to NO_X, CO and VOCs from wild fires and prescribed fires are considered as precursors to background O₃ formation.

Historically, two approaches to estimating North American background concentrations (previously referred to as PRB) have been considered in previous O₃ assessments. In the 1996 and earlier O₃ AQCDs, measurements from remote monitoring sites were used. In the 2006 O₃ AQCD, the use of chemistry-transport models was adopted, because as noted in Section 3.9 of the 2006 O₃ AQC D, estimates of background concentrations cannot be obtained directly by examining measurements of O₃ obtained at relatively remote monitoring sites in the U.S. because of the long-range transport from anthropogenic source regions within North America. The 2006 O₃ AQCD also noted that it is impossible to determine sources of O₃ without ancillary data that could be used as tracers of sources or to calculate photochemical production and loss rates. As further noted by Reid et al., the use of monitoring data for estimating background concentrations is essentially limited to the edges of the domain of interest. This is because background O₃ entering from outside North America can only be destroyed over North America either through chemical reactions or by deposition to the surface. Within North America, background O₃ is only produced by interactions between natural sources and between North American natural sources and precursors from other continents. The current definition of North American background implies that only CTMs [chemistry transport models – editor] (see Section 3.3 for description and associated uncertainties) can be used to estimate the range of background concentrations. An advantage to using models is that the entire range of O₃ concentrations measured in different environments can be used to evaluate model performance. In this regard, data from the relatively small number of monitoring sites, at which large contributions to background are expected, are best used to evaluate model predictions.

Page 3-45 line 23 (ARP 188)

In general, the GEOS-Chem predictions tend to show smaller disagreement with observations at the high-altitude sites than at the low-altitude sites. Overall agreement between model results for the base case and measurements is within a few ppb for spring-summer means in the Northeast (see Figure 3-49 in Section 3.8) and the Southeast (see Figure 3-50 in Section 3.8), except in and around Florida where the base case over predicts O₃ by 10 ppb at one site, at least. In the Upper Midwest (see Figure 3-51 in Section 3.8), the model predictions are within 5 ppb of measurements, the same is true for sites in the intermountain West (see Figures 3-52 and 3-53) and at lower elevations sites in the West (see Figure 3-54 in Section 3.8) including California (see Figure 3-55 in Section 3.8). However, the model under predicts O₃ by 10 ppb at the Yosemite site. These results suggest that the model is capable of calculating March to August mean O₃ to within ~ 5 ppb at most (26 out of 28) sites chosen. Currently, there are no simulations of North American background concentrations available in the literature apart from those using GEOS-Chem alone. However, as noted in the 2006 O₃ AQCD, an ensemble approach as is done in many other applications of atmospheric models is to be preferred.

The GEOS-Chem calculations presented here represent the latest results documented in the literature. However, all models undergo continuous updating of inputs, parameterizations of physical and chemical processes, and inputs and improvements in model resolution. Inputs that might be considered most relevant include emissions inventories, chemical reactions and meteorological fields. This leads to uncertainty in model predictions in part because there is typically a lag between updated information for these above inputs, as outlined in Section 3.2 for chemical processes and emissions and in Section 3.3 for model construction, and their implementation in CTMs including GEOS-Chem. Examples might include updated emissions for year specific shipping, wildfires and updates to the 2005 NEI; updates to the chemistry of isoprene and multi-phase processes, including those affecting the abundance of halogens; and updates to species such as methane. To the extent that results from an updated model become available, they will be presented and used to help inform NAAQS setting.