STATE-OF-THE-SCIENCE PAPER ON STATUS OF EMISSIONS INVENTORY METHODS FOR PM-2.5 DRAFT REPORT OF THE PM-2.5 COMMITTEE OF THE EMISSIONS INVENTORY IMPROVEMENT PROGRAM Prepared by Pacific Environmental Services, Inc. 5001 S. Miami Blvd. Research Triangle Park, NC 27709-2077 SEPTEMBER 1998 ------- DISCLAIMER This draft report was furnished to the Emission Inventory Improvement Program and the U.S. Environmental Protection Agency by Pacific Environmental Services, Inc., Research Triangle Park, NC. The report is a draft for review and comment and should not be cited. The draft report has been reviewed by the EIIP PM-2.5 Committee and approved for further review and comment, however, this draft report has not been accepted as a final report by the Committee or the U.S. EPA. 1 ------- ACNOWLEDGMENTS This report was prepared with significant input and advice from the PM-2.5 Committee assembled as part of the Emissions Inventory Improvement Program. The Committee members listed below were instrumental in the development of this report. Pacific Environmental Services also wishes to recognize significant input and technical advice offered by Mr. John Core of Core Environmental Engineering, and Mr. William Bernard of E.H. Pechan and Associates. Mr. Core and Mr. Bernard made significant contributions to this report and the technical knowledge and expertise offered by them is appreciated. PM-2.5 Committee of the Emissions Inventory Improvement Program Sheila Holman, Committee Co-chair Tom Pace, Committee Co-chair Mohammed Mazeed Bob Betterton Duane Ono Patrick Gafifney Sam Wells Brett Jacobs Mike George Dick Forbes Marc Deslaurier State of North Carolina U.S. EPA State of Delaware State of South Carolina Great Basin Air Quality Management District State of California State of Texas Lane County, Oregon State of Arizona State of Illinois Environment Canada 11 ------- CONTENTS Page Disclaimer i Acknowledgments ii Contents iii Tables v Figures v List of Acronyms vi INTRODUCTION 1 Purpose 1 Background 1 Nature of the PM-2.5 Problem 1 Origins of PM-2.5 2 Relationship of PM-2.5 to PM-10 and Visibility 3 Organization of the Report 4 SOURCE CONTRIBUTIONS TO PM-2.5 IN AMBIENT AIR 5 Precursors to Sulfate 5 Precursors to Nitrate 5 Carbonaceous Particles 6 Geological (Soil) Particles 7 Relationship Between Source Magnitude and Ambient Concentration 7 OVERVIEW OF EMISSION FACTORS AND INVENTORY METHODS 8 Inventory Development Procedures 8 Direct Measurements and Indirect Estimation Approaches 8 CEM Data Versus Application of Emission Factors 10 Area and Mobile Source Estimation Methods 10 Types of Inventories and Inventory Applications 11 Level of Detail in Activity Data, Emission Factors And Projection Factors 13 Inventory Development Tools 14 Summary of the NET Inventory 14 Development of the Area Source Model 18 PM-2.5 INVENTORY ISSUES IN FUTURE PLANNING 19 Spatial Influences 21 iii ------- Temporal Influences 23 Assessing Priorities 24 CONTENTS (continued) Page TECHNIQUES FOR INVENTORY VALIDATION 27 Speciated Linear Rollback and Simple Regression Models 28 Data Attribute Rating System (DARS) 29 Approach 29 Applications 30 Use of CMB and Dispersion Modeling 31 Development of Source Profiles for CMB Modeling 32 SUMMARY AND RECOMMENDATIONS 33 On-going National-Level Activities 33 Recommendations for State Activities 34 REFERENCES 36 APPENDIX A. Pie Charts of Speciated Ambient PM-2.5 Data A-1 APPENDIX B. Status of PM-2.5 Emissions Estimation Tools B-l iv ------- TABLES Page TABLE 1. 1996 NET INVENTORY DEVELOMENT METHODOLOGY 16 TABLE 2. SPATIAL INFLUENCES IN PM-2.5 PLANNING 22 TABLE 3. TEMPORAL INFLUENCES ON PM-2.5 SIP INVENTORY DEVELOPMENT 25 TABLE 4. DARS SCORING BOX 30 TABLE B-l. FACTORS AND ACTIVITY DATA FOR PM-2.5 PLANNING B-5 FIGURES Page Figure A-1. PM-2.5 Composition in the Eastern United States A-2 Figure A-2. PM-2.5 Composition in the Western United States A-3 v ------- LIST OF ACRONYMS ASEM Area Source Emissions Model CAA Clean Air Act CEM Continuous Emission Monitor CMB Chemical Mass Balance Model DOA Department of Agriculture EIIP Emissions Inventory Improvement Program FRM Federal Reference Method g grams GCVTC Grand Canyon Visibility Transport Commission IMPROVE Interagency Monitoring of Protected Visibility Environments |iin Micro meters (10"6 meters) NAAQS National Ambient Air Quality Standard NESCAUM Northeast States Consortium of Air Use Managers NET National Emissions Trends Inventory NH3 ammonia N02 nitrogen dioxide NOx oxides of nitrogen OAQPS Office of Air Quality Planning and Standards OTAG Ozone Transport Assessment Group OMS Office of Mobile Sources PART5 Mobile Source particulate matter emission factor model PM-2.5 particulate matter with mass median diameter less than 2.5 |im PM-10 particulate matter with mass median diameter less than 10 jam SMP Smoke Management Plan S02 sulfur dioxide SO A secondary organic aerosol SPECIATE species profile database maintained by EPA TCM Transportation Control Measures U.S. EPA United States Environmental Protection Agency VMT vehicle miles travelled VOC volatile organic compounds WESTAR Western States Air Resources vi ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 INTRODUCTION Purpose The purpose of this document is to provide a concise overview of the state-of-the-science related to future needs for the development of emission inventories to support planning to reduce high ambient concentrations of PM-2.5, particulate matter with mass median diameter less than 2.5 micrometers (|im). The primary objective is to address issues that are important in the development of emission inventories for PM-2.5 and its precursors. This paper addresses emission factor and activity data development, spatial and temporal variability of PM-2.5, and priorities for methodology and emission factor development on national, regional, and nonattainment area scales. This document is not intended to replace the emissions inventory guidance for PM-2.5 currently being developed by the U.S. EPA. Background Nature of the PM-2.5 Problem During the Environmental Protection Agency's last review and update of the air quality criteria documents, clinical and epidemiological evidence indicated that fine particles contribute to public health problems. The review indicated that exposure to PM-2.5 continued to cause health effects even after ambient concentrations of PM-10 (particulate matter with mass median diameter less than 10 |im) approached or achieved the National Ambient Air Quality Standard (NAAQS) for PM-10. PM-10, which has been retained as a NAAQS, was the sole air quality indicator for particulate matter prior to the promulgation of the PM-2.5 standard. Exposures to PM-2.5 affects the most sensitive individuals in the population, including young children, the elderly and those with asthma or other chronic respiratory conditions. Faced with this strong evidence indicating a causal relationship between PM-2.5 concentrations and hospital visits, missed work or school days, and, in severe cases, death, EPA decided to promulgate a new NAAQS for PM-2.5, which is also referred to as fine particulate. PM-2.5 is a new indicator for air quality, and as such, planning tools are limited and experience in planning for implementation of such a standard has only just begun. Based on the limited monitoring data that is available, PM-2.5 is known to be present in essentially all parts of the country. Examination of the available data reveals some interesting features of the PM-2.5 problem. The relative contributions of broad source categories to observed ambient concentrations are not consistent throughout the country. The differences seem to result in part from the mix of source categories, geographical influences and atmospheric processes. These differences 1 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 have a significant influence on control strategies that would be effective in those different areas. It is important to note that the available ambient data has not been collected using an EPA Reference Method, and, therefore, these data have not been used to determine the locations and boundaries of areas that exceed the concentration level of the PM-2.5 NAAQS at this time. Origins of PM-2.5 There are three primary origins of PM-2.5: primary particulate matter that is emitted directly in the solid phase, condensable particulate that can be emitted at high temperature in the gas phase but which condenses into the solid phase upon dilution and cooling in the plume, and secondary particulate that is formed through atmospheric reactions of gaseous precursor emissions. Control strategies for the primary and condensable components can employ source specific control technology and abatement activities. Control strategies for the secondary components involve complex chemical and physical interactions among the precursor emissions that are emitted from various source types. Control of the secondary components will require appropriate controls on the sources of the various precursors. The atmospheric reactions and transport processes that result in secondary particulate formation extend over large distances from the sources of the emissions. Similarly, directly emitted PM-2.5 can also be transported over large distances. Therefore, regional control programs may be needed to address all three forms of PM-2.5. Primary PM-2.5 particulate results largely from combustion of fossil fuels or biomass, although selected industrial processes can also be significant in some areas. The sources of PM-2.5 include, but are not limited to, gasoline and diesel exhaust, wood stoves and fireplaces, agricultural, land clearing, and wildland prescribed burning, and wild fires. Smaller sources of primary particulate include fugitive emissions from paved and unpaved roads, dust from ore processing and refining, and to a lesser extent, crustal material from construction activities, agricultural tilling and wind erosion. The condensable components are largely made up of semi-volatile organic compounds that condense at ambient temperature to form aerosol. Secondary PM-2.5 forms through heterogeneous (gas to particle) chemical reactions that convert some common gaseous pollutants into very small particles. The observed secondary PM-2.5 is dominated by sulfur and nitrogen species in most locations, however there can also be significant contributions from secondary organic aerosol in some locations. Sulfate aerosol chemistry can involve either gas-phase or liquid-phase oxidation processes. Gas-phase conversion involves the oxidation of sulfur dioxide (S02) by hydroxyl radicals (OH), which are formed through the photodecomposition of ozone. The reaction results in sulfuric acid aerosol as an 2 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 airborne particle with a typical diameter of less than 0.3 |im. The sulfuric acid particles are then neutralized by ammonia and molecular oxygen to form ammonium bisulfate and ammonium sulfate. Liquid-phase reactions involve S02 and oxidants (H202, 03 and O) along with catalysts, such as manganese and iron, that are dissolved in cloud or fog droplets to form sulfuric acid aerosols with diameters in the 0.5 to 1.0 |im size range. Nitrate aerosols can be produced by several pathways. The most important pathway is the gas-phase reaction of N02 with hydroxyl radicals to produce nitric acid, which is then neutralized by gaseous ammonia. When the levels of ammonia and nitric acid are sufficiently high, ammonium nitrate can be formed within the PM-2.5 size range. Secondary organic aerosols (SOA) are formed by more complex mechanisms involving organic gas-phase particle precursors in the presence of ozone and hydroxyl radicals. The precursors of SOA arise from both anthropogenic and natural sources. Studies have shown that a significant fraction of the a- and P-pinenes which react with ozone and hydroxyl radicals in the atmosphere can lead to organic particle formation which then becomes a component of the organic carbon fraction. For example, in poorly ventilated urban areas such as Los Angeles, which has extensive brush covered hills, natural emissions of gaseous hydrocarbons may contribute up to 50% of the secondary organic particles.1 Currently, the details of SOA formation are not well known, and the implications for needs related to the development of emission factors and other emissions estimation tools are uncertain. Large carbon number organic compounds that have an affinity to stick together may contribute significantly to these processes. These processes may also be catalyzed by metals or other trace components. Future development efforts may need to be directed to expand VOC speciation profiles to include compounds that have not been of interest in ozone chemistry, and to improve the methods for characterizing sources of trace metals. These conversion mechanisms from gas to particle forms are thought to be one of the primary atmospheric removal mechanisms for gaseous pollutants. Once in the particle phase, those pollutants are efficiently removed, under the right conditions, by acting as cloud nuclei, or through wet and dry deposition to the surface. Prior to removal, however, they can contribute to ambient concentrations that near or exceed the level of the PM-2.5 NAAQS. Relationship of PM-2.5 to PM-10 and Visibility PM-10 emission inventories are typically dominated by fugitive dust sources. These sources were estimated to contribute approximately 89% of the total PM-10 emissions on a national basis in the 3 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 1996 National Emissions Trends (NET) inventory. Within the fugitive dust category the main sources were unpaved roads (33%), natural sources (20%), agricultural crops (17%), construction (14%), paved roads (9%) and all other sources of fugitive PM-10 (2%).2 In all cases, these sources contribute primary particle emissions. Overall, the contribution of condensable and secondary particle origins to PM-10 are an important but usually small fraction of the mass unless the airshed is influenced by emissions from woodstoves or other forms of biomass burning. The available ambient measurements of PM-2.5 suggest that anthropogenic combustion sources, fire and other emitters of condensable and secondary origins contribute a large percentage of the overall PM-2.5 emissions. Therefore, existing control programs for PM-10 are expected to benefit but not to alleviate PM-2.5 problems. Planning for PM-2.5 will need to be directed to the principal sources contributing to the problems in specific areas, and those programs may extend to different sources and source categories than those previously targeted for PM-10 programs. Particles with diameter less than 2,5 (am scatter sunlight more efficiently than the larger particles included in the PM-10 mass, and as such contribute significantly to the regional haze issue. Recent studies suggest that regional haze in the east is caused primarily from sulfate, nitrate and secondary organic aerosol, or the secondary particulate origin of PM-2.5. In the west secondary particles are still of importance but there appears to be a larger contribution from primary emissions resulting from wood smoke and other combustion sources.3 Therefore, it is expected that programs that address high PM-2.5 concentrations will have beneficial effects on regional haze as well. Organization of the Report The remaining sections of this report provide a preliminary assessment of some of the important issues that need to be addressed and understood before starting activities to develop an inventory for PM-2.5. The next section discusses the primary sources of PM-2.5 based on available air quality data. This assessment is based primarily on data collected to support analyses of visibility in national parks and recreation areas, and details on specific urban areas are not well represented. A review of the approaches used to develop emission factors and emissions inventories is provided along with a summary of some specific inventory development tools that are available for PM-2.5 programs. The following section summarizes some of the important considerations that might affect PM-2.5 inventory development activities in future planning exercises. Some selected techniques that can be applied to evaluate the quality or validity of inventory data are presented. The report concludes with some general recommendations that are meant to provide some insights to help in planning and prioritizing efforts in the future. Pie charts showing measured ambient concentrations are shown in Appendix A. Appendix B presents a concise summary of data and tools that are available and remaining needs related to the 4 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 primary components of PM-2.5. This report is not meant to be a definitive guide on inventory development or to identify all of the specific issues that might affect planning in any given area. 5 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 SOURCE CONTRIBUTIONS TO PM-2.5 IN AMBIENT AIR PM-2.5 is a new air quality indicator, and consequently, there is a relative paucity of information on the sources of PM-2.5 and emission factors for estimating PM-2.5 emissions. There is, however, a fairly robust ambient monitoring database that includes information on the composition of ambient PM-2.5. That database provides important insights on the source contributions to ambient PM-2.5. These data are taken largely from the Interagency Monitoring of Protected Visual Environments (IMPROVE) project. IMPROVE was designed to explore the current status and potential causes of visibility impairment in National Parks and other Class I areas and, therefore, that database is most useful to explore the regional distribution of PM-2.5. A limited set of data that is more representative of urban conditions is also available and those data are useful to infer information about the differences between rural concentrations and urban area concentrations. Since the urban data represent only a select few urban locations, it is not possible to present a comprehensive assessment of urban PM-2.5 distributions. These data have been discussed in detail elsewhere.4 Summary pie charts showing the composition of measured ambient PM-2.5 are included in Appendix A to this report. Similar trends in sulfate concentrations between east and west, and relative magnitudes of sulfate and nitrate between urban and rural locations have also been observed in monitoring data collected across Canada.5 Precursors to Sulfate Review of the figures in Appendix A reveals that sulfate is a significant component of PM-2.5 in the east and is less prevalent in the west. This is not unexpected, since the east is strongly influenced by major sources of S02 from coal burning utilities in the Ohio Valley, Tennessee Valley and along the east coast. Other industrial sources of S02 are also common in the east. There are also large sources of ammonia arising from major livestock production and fertilizer application throughout the Midwest, gulf coast, mid-Atlantic and southeastern States, in addition to the sources of ammonia associated with human activities. In addition, water vapor and radicals from photochemical systems are frequently found in the east. These conditions combine to produce the observed large contributions of sulfate, on a percentage basis, at the eastern monitoring locations. In general, the sulfate fraction measured at eastern monitors ranges between 30% to 60% of the total PM-2.5 mass. For comparison purposes, the sulfate fraction measured at western monitors, that are affected by fewer S02 sources, rarely exceeds 20% of the total PM-2.5 mass. Precursors to Nitrate 6 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 The principal sources of NOx, motor vehicles and all fossil fuel combustion, are much more ubiquitous across the country and, as a result, there is no recognizable gradient in observed ammonium nitrate concentrations from east to west. The nitrate fraction, as a percentage of observed PM-2.5 mass, is also generally lower than or approximately the same as the sulfate component, although in some areas in the west the nitrate fraction can be greater than the sulfate fraction. One exception, illustrated in Appendix A, is for data collected in the San Joaquin Valley where agricultural sources of both NOx and ammonia from fertilizer use might combine to increase the amount of ammonium nitrate. Based on the data summarized in Appendix A, the fraction of nitrate rarely exceeds 20% of the total PM-2.5 mass at any location. Carbonaceous Particles Carbonaceous particulate matter represents a significant fraction of the observed PM-2.5 in many locations. The data summarized in Appendix A show that in the east approximately 25% to 40% of the PM-2.5 is carbonaceous, and in the west carbonaceous particulate contributes between 50% and 75%) of the mass of PM-2.5. A distinction has been made between elemental carbon and organic carbon to refine the identification of the sources of total ambient PM-2.5 carbon. These two forms of carbon can arise from fundamentally different types of processes; organic carbon emissions are associated with low temperature combustion processes such as biomass burning, while elemental carbon emissions result mainly from high temperature combustion, primarily diesel engines. Tracking the two forms of carbon particles separately enhances the resolving power of source apportionment techniques relative to those based solely on the total carbon fraction. The distinction between elemental and organic carbon is based on the laboratory measurement methods commonly used to analyze PM-2.5 ambient air particulate filters for carbon. Although a variety of analytical schemes have been developed to distinguish between organic and elemental carbon (soot), all of the methods expose a portion of the filter to a carrier gas stream within a heated oven where the particulate carbon on the filter is converted to a gas (carbon dioxide or methane), which is then measured by a detector. Since organic carbon is released at a lower oven temperature than elemental carbon, the temperature dependence can be used to distinguish between the two classes of carbon.6 Nearly one hundred percent of the elemental carbon observed in ambient samples results from primary particulate emissions from fossil fuel and some biomass combustion processes that achieve high temperatures. These particles are produced primarily in the size range of less than one |im in diameter. Recent source apportionment studies indicate that motor vehicle sources dominate the elemental carbon observed on ambient samples taken in urban settings, with emissions from diesel exhaust contributing 7 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 between 50% and 70% of the elemental carbon mass concentration.7 Gasoline powered vehicles contribute around 25% of the total and much of that results from automobiles that may not be maintained at optimum performance. Small contributions associated with road dust may include resuspended tire wear particles. Organic carbon represents the remainder of carbon containing particulate matter. This fraction includes contributions from primary particle emissions, condensed particulate and secondary particulate. Motor vehicle sources contribute up to 80% of the organic carbon mass in urban areas, although meat cooking and wood burning can also represent significant contributions in many western locations. Geological (Soil) Particles Geological or soil particles become airborne when natural soils are mechanically disturbed. While soil sources contribute up to 85% of observed PM-10 concentrations, analysis of ambient samples reveals only small contributions (10% to 15%) of soil sources to the total PM-2.5 in most locations. That trend toward small contributions from soils is also seen in the data collected in Canada.5 The primary sources of geological particles are agricultural tilling, construction activities, road dust from both unpaved and paved roads, and windblown dust. Soil particles contain common metals including Al, Si, K, Fe, Ca, and other trace metals. It is not expected that soil particles will be significant in terms of control strategies for PM-2.5 in most areas. Relationship Between Source Magnitude and Ambient Concentration Estimates of the emissions magnitude for some sources of PM-2.5 are sometimes inconsistent with the resulting ambient concentrations measured at nearby monitoring sites. This effect is observed most clearly with respect to surface fugitive dust sources of primary emissions. The causes for these discrepancies are not well established at this time. One cause appears to be related to the source measurement methods that have been applied in some recent monitoring programs of area and line sources (e.g., agricultural tilling, unpaved roads, etc.). Frequently, the source emissions rate is estimated using a procedure that measures the horizontal flux of emissions through a vertical plane in the downwind direction very near to the source activity. That measurement is then compared to a similar estimate immediately upwind of the activity and the difference is used to calculate the emission rate. While this technique is thought to provide an accurate estimate of the mass emissions rate from the source, the approach may not accurately account for other influences that can serve to remove a portion of that mass before it is entrained into the transport layer. For example, nearby vegetation or other physical structures may filter some of the mass. Additional mass could be removed by adhering 8 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 to larger particles that are deposited close to the source through gravitational settling. One hypothesis is that the portion of the emissions mass that can be transported to monitor sites is limited to the mass that rises above surface features and becomes entrained into the local and/or regional wind flow pattern. This problem seems to be confined primarily to near surface sources of fugitive dust. Emissions from most point sources released from stacks, and other surface combustion sources may be affected by heat induced buoyancy that serves to elevate these emissions into the transport layer. EPA is currently coordinating with the Department of Agriculture (DOA), the Forest Service, and other experts with experience in these processes to refine the emissions estimation methods for these sources. Currently, the mechanisms that produce fugitive emissions and affect the transport of those emissions to potential receptor sites are not well represented in the emissions estimation methods. Therefore, the confidence in PM-2.5 emissions estimates from fugitive dust sources is low. A similar removal mechanism may also affect near surface releases of NH3 and other noncombustion related emissions of the gaseous precursors to secondary PM-2.5.8 OVERVIEW OF EMISSION FACTORS AND INVENTORY METHODS Inventory Development Procedures Emission inventories are the basis of essentially all air quality management activities. Inventory data can be applied to a variety of analyses and for several different purposes. Similarly, inventory data can be prepared using different methodologies, with specificity and detail consistent with that required for the particular application. Emission factors are often applied to common activity data to develop emission estimates. Emission factors can also be estimated using a variety of techniques, each with its own specificity and accuracy for different applications. In other cases, emission rates depend on some other variable(s) (e.g., temperature, wind speed, soil moisture, or process parameters). In those cases, an algorithm is applied to estimate emissions for the particular set of conditions. Under certain circumstances, emission factors and emissions estimation techniques are applied to sources and/or for analyses other than those for which they were developed. This situation arises most frequently in analyses of emerging air quality issues that are based on an immature foundation. Currently, this situation can affect some sources important in PM-2.5 planning activities. The following discussion will serve as a brief introduction to the various approaches used to develop emission factors and other inventory estimation methods. The discussion is provided to assist readers in understanding the strengths and weaknesses of these various approaches, and to prioritize those activities that could improve the understanding of PM-2.5 issues in specific situations and locations. 9 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 Direct Measurements and Indirect Estimation Approaches Generally, direct measurements of emissions from specific sources provide the most accurate emission factors and emissions estimates. Unfortunately, the source testing required to obtain these direct measurements is expensive and even direct measurements can produce misleading results for sources that have variable operating characteristics. In addition, EPA promotes the use of standard methods for testing to facilitate consistency among different sources and to increase the reliability of emission factors when they are applied to other untested sources with the same operating characteristics. Sometimes the method itself can produce, either by design or inadvertently, a biased or incomplete result. For example, the Modified Method 5 sampling train for particulate matter, adapted for PM-10, specifies that the sample line be heated to ensure the collection of only the filterable fraction. This particular situation does not present a significant problem for PM-10 analyses, since in most cases, the contribution of the condensable fraction, mostly less than one |im in diameter, is often minimal in terms of the total PM-10 mass. The application of the Method 5 sampling train, with modifications to produce a size cutoff for PM-2.5, however, would not represent accurately the mass of the condensible fraction of PM-2.5. For PM-2.5, this approach can create a considerable bias, particularly for some combustion sources. For example, the emission factor for the condensable fraction of PM-2.5 from natural gas combustion is three times the factor for the filterable fraction.9 Currently, the Federal Reference Method (FRM) for condensable material is Method 202. Method 202 yields only the mass in a solvent extraction and the mass in an aqueous extraction after the sample is drawn through a water impinger. This method, therefore, precludes any further speciation to assist in understanding the nature of the condensable material. Resolution of the component species in the condensable mass is required for use in receptor modeling studies. Direct measurements will still provide the most reliable estimates of emission rates for PM-2.5, but different methods will be needed to accurately capture the total (filterable plus condensible) PM-2.5 mass. EPA is working on the evaluation of methods for this purpose, but it will be some time before a FRM is published for PM-2.5. Indirect measurements are also often used in inventory development for sources and source categories that do not lend themselves to direct source sampling. Indirect measurement methods can be most effective for source categories that consist of a large number of individual sources that are highly consistent in their emissions characteristics or for sources that are spread out over large areas. Emission factors based on a representative sample of the total population of sources can then be applied with a high degree of reliability to the activity associated with the collection of all individual sources. An example is emissions from residential natural gas combustion sources. These units all operate similarly and an emission factor based on a representative sample provides a very reliable estimate of the emissions from all residential natural gas combustion. Therefore, indirect monitoring of 10 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 the amount of natural gas consumed, allowing for theft and leakage, can be used to prepare the emissions estimates for these sources. Similarly, emissions for either agricultural burning or prescribed burning can be estimated from an equation that relates the emissions to parameters that describe the type and quantity of fuel burned. Indirect monitoring of the nature of the site to be burned can result in a reasonable estimate of total emissions. CEM Data versus Application of Emission Factors For some specific applications continuous emission monitors (CEMs) can be used. CEMs are monitoring techniques that measure emissions in a stack and record the concentration data to electronic storage media on a continuous basis. The best example of the use of CEMs for inventory purposes is found in the acid rain program (Title IV of the CAA). As part of the acid rain program, all affected utility and industrial combustion sources are required to install and operate CEMs to track total emissions of S02 and NOx on a continuous basis. CEM type monitoring has the advantage of reflecting actual in-use emissions, including upsets and other unusual events. The drawback is that large amounts of data are generated and converting all of these data into useable formats takes some effort and time with obvious cost implications. CEMs for application to PM-2.5 source testing will not be available until there is a FRM. An emission factor is usually developed from a set of direct source emission tests. Frequently, the test series is designed to represent the typical range of size and operating conditions for that source type. If the average mass emission rate, expressed as a function of some readily available process related parameter, is constant, the average factor can be applied to all sources that fit the conditions of the test series. Often different factors are required for different size ranges or categories of operating conditions, but the average factors developed in this way are always applied universally to all other sources in the category whether or not they were tested directly. These factors can then be applied to estimate emissions based on more easily measured activity data. For example, emissions of S02 from coal combustion can be estimated reliably based on measurements of the amount of coal burned and the average sulfur content of the coal. Emission factors provide emissions estimates with a high degree of confidence as long as the source matches the operating conditions of the sources that were tested to develop the emission factor. The reliability of emission factors decreases when only a few source tests are used to the develop the factor. Emission factors based on a small number of tests may not reflect operational variability and application of these factors could introduce bias in emissions estimates when the operating conditions vary. Area and Mobile Source Estimation Methods 11 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 Not only is it difficult to measure emissions directly from most area and mobile sources, but is it often difficult to measure spatially and temporally resolved activity data as well. Often measures of the relevant activity can be estimated at the State-level or national-level through economic or other indicators that represent a particular activity. For some applications acceptable estimates of emissions can be developed by applying an assumption to represent the distribution of that total to smaller spatial scales. In other cases, this approach can introduce unknown bias. For example, PM-2.5 emission estimates for construction activities in the National Emissions Trends (NET) inventory are based on activity estimates that are derived from total annual dollars spent on construction activity at the EPA region-level. These estimates are then allocated to counties using a procedure that depends on construction costs and estimates of acres under construction in each county. Obviously, this technique will result in approximations that will match actual county-level construction related emissions at varying degrees of accuracy. Other area source emissions estimates are based on State-level activity data that is then allocated through population, land use or some other surrogate distribution factor. Fires are important sources of PM-2.5 emissions in many locations. Agricultural field burning, prescribed burning (planned burning of natural areas), slash burning (land clearing), and wildfires can all contribute to ambient PM-2.5 loadings. While it is possible to measure the acres burned or biomass burned, in the case of slash burning, during these activities other estimates about the fuel loading, moisture content of the fuel, and estimates of the specific wood types burned need to be estimated. This can be done by applying measurements that have been made in other nearby or similar land use types. The uncertainty associated with this kind of emission estimation procedure is related to how well the assumptions on fuel type and fuel loading match the conditions of the area burned. Similar issues arise when estimating emissions from mobile sources. Most mobile source emissions are calculated by applying an average emission factor expressed in terms of grams per vehicle mile traveled (g/VMT). Those factors are developed from a subset of actual in-use vehicles and are frequently based on controlled tests using a dynamometer. Those factors are then applied to an estimate of the total VMT at county resolution. Differences in the mix of vehicles, especially for heavy- duty diesel vehicles, or temporal activity patterns in any given area can introduce unwanted bias into the emissions estimate. Types of Inventories and Inventory Applications Planning for the development of an emission inventory depends on the type of air quality planning analysis it is intended to support. For example, approaches based on national-level or State-level activity data, and subsequent allocation to county or nonattainment area scales, can be useful for tracking trends or for evaluating air quality management programs over time. Those 12 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 approaches are often referred to as top-down methodologies. Inventories at that level of detail are often inadequate for other more rigorous air quality management activities. For example, a very detailed inventory is needed to support an attainment demonstration and the development of a control strategy. That inventory needs to be source specific, spatially allocated to the correct scale, and representative of temporal variability that can affect the outcome of modeling exercises. A discussion of the various types of inventories that can be developed and the types of analyses they support can be found in another document prepared by the Emission Inventory Improvement Program (EIIP).10 The level of detail needed in the emissions data to support the activities indicated decreases from Level 1 to Level 4. The categories of inventories discussed in that report are listed below: • Level 1 - Source specific, used for permit and compliance programs, • Level 2 - Urban scale, used for State Implementation Plan (SIP) and other large scale planning activities, • Level 3 - Industry wide, applications that do not drive regulatory issues, and • Level 4 - National and international Greenhouse Gas (GHG) issues. Inventory issues in PM-2.5 planning will include: developing a baseline understanding of local and regional influences, reasonable further progress (RFP) planning and demonstration, modeling attainment demonstration, periodic inventories, emission statements, VMT reduction and transportation control measures (TCM) planning, regulatory development and cost analyses, etc. These examples are not inclusive of all of the applications of inventories in air quality management. Many of these analyses can build off of different levels of specificity and detail in terms of both specific emissions estimates and temporal or spatial resolution. The specific types of inventories required will be dependent on the types of analyses that need to be completed and the specifics of the particular areas. Some selected examples of specific types of analyses and the inventory needs to support those analyses are listed below: • Emission control development and attainment modeling activities need source specific inventories with process level resolution. PM-2.5 programs need these data for a base-year and future-year controlled case and need estimates of primary, condensable and secondary contributions, representative temporal distributions, spatial allocation of nonpoint sources, and speciation profiles. • Inventories with similar detail are needed for related air quality management activities, including source identification and prioritization, rule development, cost/benefit analyses, and compliance monitoring. 13 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 • RFP analyses can be based on inventories at lower resolution to track overall emission reductions or specific reductions from sources that have been controlled combined with other broader growth estimates for the remaining parts of inventories. • Emission statements can be prepared at industry level resolution, but may not need the species and temporal resolution required for air quality modeling, to validate adoption and effects of emissions control rules. The definition of a base-year has significant implications in PM-2.5 planning. First it will be necessary to define one base-year for all parts of the country to promote consistency in regional modeling analyses. The base-year needs to have a sufficient amount of appropriate monitoring data available to ensure that the specific problems that need to be addressed are identified. Specific areas will have to evaluate the occurance of natural events, unusual meteorological conditions and any other external factors that could be affecting that base-year. Level of Detail in Activity Data, Emissions Factors and Projection Factors As mentioned earlier, many emission estimation methodologies for area and mobile sources depend on activity data that is itself hard to monitor directly. VMT for highway mobile sources, hours in operation for many off-highway mobile sources, and fuel use in residential wood heating are examples. This poses particular problems for PM-2.5 planning efforts, since area sources represent such a large percentage of the overall emissions totals. This situation is similar to the problems encountered in developing inventories for VOC and NOx to support ozone planning. Estimating emissions from both highway and off-highway diesel vehicles is a good example of this problem. Diesel vehicles will contribute a significant portion of elemental carbon found in most locations. A small number of dirty vehicles can contribute a large percentage of the overall emissions from mobile sources. The available methodologies for estimating VMT for highway sources and hours in operation for heavy diesel off-highway sources do not consider specific types of vehicles and assume that all vehicles are operated at conditions that are near optimum efficiency. Concerning highway vehicles, this problem is compounded by the fact that the VMT estimation method is targeted primarily to count gasoline-powered vehicles (passenger cars) and may introduce bias when applied to area wide estimates of the number of heavy-duty diesel vehicles. Emission factors and speciation profiles are often developed for specific sources but then applied, as is, or with minor modifications based on engineering assumptions, to other related sources. It is possible that surrogate factors and/or size and speciation profiles will be used in early PM-2.5 planning activities, simply because factors and profiles will not be available for all sources. Many of 14 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 those factors and distribution profiles that are available were developed several years ago, and may not be representative of current operating and production conditions. Projection factors are almost always based on broad economic forecasts of industry sector growth. This technique does not consider changes in technology, productivity, and other issues that can influence the emissions relative to the projected growth assumptions. Growth factors are important because they are used to determine the emissions inventory in the attainment year for the purposes of developing the control strategy and attainment demonstration. Although the industry average growth assumptions can serve as a good starting point, the assumptions and factors should be modified to reflect local conditions. Per employee factors, per capita factors, and per unit area factors are particularly difficult to apply in this context. Inventory Development Tools EPA is developing a set of tools to assist the States in the preliminary evaluation and planning for PM-2.5 to support future air quality management activities. Currently, EPA has compiled an inventory of PM-2.5 emissions for each county in each State based on the current understanding of emissions processes and availability of emission factors. That inventory is included as part of the 1996 NET inventory database and is intended to support the States in preliminary assessments of important issues, and to help prioritize future planning and development efforts. The 1996 NET inventory and updates to the NET inventory that will be developed over the next two years, will serve as the preliminary basis for future planning purposes. These planning activities will also support the development of improved national inventories that are needed to support regional modeling efforts. EPA is also supporting efforts to bring all of the various methods for estimating area source emissions together into a single shell program to assist in the development and improvement of PM-2.5 emissions data. It is anticipated that the computer program, referred to as the Area Source Emissions Model (ASEM) will be available for State use sometime in the Fall of 1999. Some of the features of the 1996 NET inventory and the ASEM are discussed in this section. Summary of the NET Inventory The National Emissions Trends (NET) inventory for 1996 represents EPA's most recent and most complete national-level air emissions inventory.2'11 The 1996 NET inventory has evolved from the 1990 Interim Emissions Inventory, although the database does include several more detailed and specific estimates. The 1996 NET inventory includes estimates of annual and summer day emissions for VOC, NOx, CO, S02, PM-10, PM-2.5, and NH3. Emissions are estimated for all States and the 15 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 District of Columbia. Point source emissions are reported at the AIRS segment/SCC-level, and area and mobile sources are reported at the county/SCC-level. The 1996 NET inventory is the result of one of EPA's standing commitments to distribute an assessment of air quality and emissions trends over time. These trends inventories have historically been useful for tracking progress of air quality management programs and in demonstrating a relationship between control programs and improvements in air quality. The 1996 NET inventory was developed with significantly more process-level detail than many previous national-level inventories that included some top-down estimates. One reason for improving the detail is simply to take advantage of other programs that have developed more complete and specific inventory data. Another reason is to provide a national-level inventory with sufficient detail and coverage so that it can be used by States as the basis for their preparation of the 1996 Periodic Emissions Inventory (PEI). As improvements are made to the NET inventory development methodology, previous year estimates are adjusted to be consistent with those changes in methodology. A version of the NET inventory for 1997 is currently in development. Updated CEM and VMT data are available for application to utility and on-road emissions respectively for application to the 1997 data. While these data are being compiled, the 1996 data will be adjusted to be representative of the updated activity data. Once the 1997 NET inventory is complete, the 1996 data will also be updated. State-specific data developed to support the Ozone Transport Assessment Group (OTAG) analyses in the eastern United States, and similar data developed by the Grand Canyon Visibility Transport Commission (GCVTC) in the west were used to overwrite emissions developed by EPA. These data were compiled in State supported collaborative efforts and are believed to be more representative and detailed than any data that can be prepared at the national-level. The exception is for on-road mobile categories for which State-specific data were not used, even if they were submitted. In addition, some S02 and NOx emissions for utilities taken directly from continuous emission monitors (CEMs) from the acid rain program have been used. The methodologies applied to prepare the 1996 NET inventory for emissions for States in other regions, and for source categories that were not of interest in OTAG and/or GCVTC, are summarized in Table 1. A more detailed discussion of the inventory preparation methodologies applied to the 1996 NET inventory can be found in a document titled "National Air Pollutant Emissions Trends Procedures Document 1900 - 1996,"10 which can be found at the following Internet site: http://www. epa. gov/oar/ oaqps/ efig/ ei 16 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 The nature of the PM-2.5 problem requires both regional and local analyses for directly emitted fine particles, condensable fine particles and those formed as secondary pollutants through chemical reaction in the atmosphere. Emissions data are required for NOx, S02, and NH3 to evaluate the regional formation of sulfate and nitrate secondary particles, VOC to evaluate the formation of secondary organic aerosol, as well as PM-2.5 that is directly emitted in both solid form and condensable form. As a result, the effort that will be required to develop the emission estimates to support this complex air quality planning effort may be, in many ways, more demanding than those completed in the past for other criteria pollutants. The 1996 NET inventory provides an excellent starting point for this effort. States are encouraged to review the TABLE 1.1996 NET INVENTORY DEVELOPMENT METHODOLOGY Source Category/Sector Methodology Utilities S02 and NOx VOC, CO, PM nh3 1995 activity data grown to 1996 using DOE data (1995 data is CEM data from acid rain program where available with DOE/AP-42 used to replace data gaps) DOE data (throughput, controls, and fuel characteristics) with AP-42 emission factors. None estimated. On-Road Mobile CO, VOC, NOx PM and S02 nh3 MOBILE5b; 1995 HPMS VMT data projected to 1996; State provided MOBILE inputs including 1990 registration distributions, I/M programs, and summer RVP data, supplemented with national 1996 registration distribution, OMS I/M program data, AAMA survey RVP data; OMS control program data for RFG and oxyfuels; State-level/monthly temperature data; national vehicle/road type speed data. PART5; same VMT and registration data as used for CO, VOC and NOx Emission factors based on Volkswagen data applied to 1996 fleet; same VMT as for other pollutants. NH3 emissions for subsequent on-road mobile categories will be based on factors derived by OMS NH, Agricultural Sources 1990 emissions grown to 1996 using BEA earnings data. PM Fueitive Dust Sources Paved and Unpaved Roads Agricultural Tilling Construction Wind Erosion Same as the on-road emissions. A correction factor for to account for precipitation was added to the unpaved road estimates. Controls were applied. AP-42 EF's, 1996 CTIC tilling data and tilling practices. New Emission Factors with default EPA correction parameters projected with 1996 Bureau of Census data. Controls are applied. Modified/simplified 1985 NAPAP methodology. 17 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 Open Burning Wildfires Prescribed Burning Agricultural Burning AP-42 emission factors applied to acres burned and fuel loading data from DOI and the Forest Service Based on 1989 Forest Service inventory for prescribed burning with specific emissions based on the ratio to 1985 NAPAP base 1985 NAPAP inventory grown with BEA growth assumptions Other Sources Non-utility point sources and all other area sources Non-road Mobile Sources 1990 emissions grown with VOC and N0X point and area source 1990 CAA controls applied. For selected States data were grown from 1995 AIRS/AFS submittals. NONROAD national emissions model developed by OMS. Source: National Emissions Inventory Documentation Attachment A http://www.epa.gov/oar/oaqps/efig/ei 18 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 data, and the description of the methodologies used to develop the data, and use that process to help assess the apparent strengths and weaknesses of the inventory and to prioritize future inventory development efforts. As States develop improved estimates for specific source categories, they will be encouraged to share those data with EPA for application to an improved base line national inventory of PM-2.5 for the 1999 base-year. General data quality assessments of the emissions of PM-2.5 are provided below, however, individual States are encouraged to make their own assessments based on the nature of the PM-2.5 that affects the State (e.g., transport, locally generated, secondary), and the source mix that is suspected of contributing to local high PM-2.5 concentrations. NOx In general, the quality of the NOx data is good. Data from large utilities are available from CEMs through the acid rain program. Data from other nonutility point sources has been supplied through OTAG and GCVTC. States are encouraged to review the data and make any additional adjustments based on local conditions and data sources. VOC Nonutility point source data were supplied through OTAG and GCVTC. Other solvent use data are not as reliable. Some of the estimates are based on old studies and grown to represent 1996 activity levels, and others are based on national mass balance and are allocated to county-level by surrogate distribution factors. States can significantly improve these estimates based on more detailed and specific information. S02, CO Emissions estimates for utilities and on-road mobile sources are considered to be representative and accurate. Estimates for other sources are primarily based on the older databases with growth factors applied. State review and replacement will improve these estimates considerably. PM-2.5/-10 Much of the mass of PM-2.5 directly emitted emissions in the current inventory is attributed to fugitive dust sources. Many of these estimates are based on the application of a scaling factor that is derived from size distribution functions that were available from analyses that supported PM-10 planning efforts. While these emissions estimates are likely to be representative of the source strength in many cases, they are not necessarily consistent with ambient air quality speciation data which implies that a portion of those emissions ultimately are removed by some other mechanical deposition process. Differences between the spatial and temporal nature of the inventories and air quality data are also likely to contribute to discrepancies. Therefore, emission estimates for PM-2.5 from fugitive sources are suspect and in need of improvement. While EPA is working to improve the understanding of these processes, States are encouraged to 19 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 review these data and provide any additional data or supporting databases that could be used to improve these estimates. NH3 NH3 is the least understood of all of the precursor species for PM-2.5 formation. States are encouraged to review the data in the NET inventory to evaluate the relative importance of NH3 in each State, and to begin to collect and assemble any data that would be useful to develop improved and more specific emissions estimates. Specifically, States can begin to collect activity data representative of suspected major NH3 source categories. Development of the Area Source Model EPA is currently conducting an effort to assemble the principal computer-based numerical tools that are used to develop and/or estimate area source emissions and link them together under an umbrella shell. This collection of tools is referred to as the Area Source Emissions Model (ASEM) and when finished it will be made available to States to assist in the development, evaluation, and quality assurance of PM-2.5 emissions estimates. The ASEM can also be used by the States to prioritize development efforts related to improving both the basic activity data for area source categories of emissions of PM-2.5 and the precursors to PM-2.5, and the emission factors, size distribution functions, or assumptions related to emission rates for various sources. The ASEM and the users manuals will be developed to be consistent with the methodologies used to prepare the draft 1996 NET inventory, other inventory development guidance and inventory requirements documents that are being developed by EPA. The ASEM will include area source emissions estimation procedures for PM-2.5, S02, NOx, and NH3. To the extent possible, the ASEM will use input activity data that is already collected for other inventory applications. For some particular source categories, additional types of activity data will be necessary as well. The program will contain default activity data for most, if not all, source categories and emissions of PM-2.5 could be generated using these default data. States will also have the opportunity to substitute more detailed and complete locally generated activity data for nearly all categories. In many cases, the default activity data included in ASEM can be improved by substituting local data that is more representative. A review of the nature of the source magnitudes, the quality of default emission factors and other considerations associated with PM-2.5 planning efforts will help States to prioritize their preliminary planning needs. States should focus limited resources on efforts to improve emissions estimates for the most important sources and source categories affecting the planning area. Once the activity data are developed, the ASEM will apply recommended emission factors and calculation procedures to develop a county-level emissions inventory for PM-2.5. The ASEM is expected to be available in the fall of 1999. 20 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 PM-2.5 INVENTORY ISSUES IN FUTURE PLANNING Air quality management requires a detailed understanding of the local and regional processes that result in high ambient concentrations of the target pollutants. That understanding serves as the foundation for a comprehensive plan to reduce the relevant emissions magnitudes to levels that will result in acceptable ambient air quality. In nearly all cases, that planning effort is based on modeling analyses. While there are many technical considerations that affect the design of modeling scenarios and the interpretation of modeling results, emissions data are one important input to those models. An accurate and comprehensive baseline emissions data set is needed to evaluate model results against current and historical ambient air quality measurements. Meteorological data representative of the appropriate conditions that give rise to high ambient concentrations are also critical for these analyses. The emissions data must be resolved to be consistent with the controlling meteorological scenarios. Future-year inventories are required to test the net effects of population growth, industrial growth and source-specific control measures. In most locations, ambient measurements of PM-2.5 include a significant mass fraction of secondary particulate. In the east, ammonium sulfate is the primary component of secondary particulate. The S02 precursor for sulfate formation comes primarily from power plants and other coal-fired and oil-fired boilers, and the ammonia precursor is emitted largely from agricultural activities. The sulfate precursors interact under typical large scale meteorological conditions found in the east to produce a relatively uniform regional contribution to PM-2.5. The nitrate aerosol precursor N02 arises primarily from transportation sources and also produce a fairly uniform contribution on regional scales. The nitrate components represent a somewhat larger fraction of the ambient PM-2.5 in the west, than in the east. Carbonaceous particles from motor vehicles, biomass burning and industrial sources are likely to make up the bulk of the controllable PM-2.5 in many areas. The effect of emissions from those sources are more likely to be dependent on local influences of source mix, meteorology, and terrain features than are the major secondary components. Source apportionment models and dispersion models can be used to identify and prioritize the most important sources affecting an area and to determine the relative contributions of those sources to the ambient mass loadings. Once the major source contributions are known, an effective control strategy can be developed. Source apportionment and dispersion modeling studies will be most useful to evaluate contributions and to identify control options for local sources. It is likely that regional modeling analyses will also be necessary to track the regional scale transport processes and chemical transformations that result in secondary aerosol. Regional modeling is also necessary to provide boundary conditions for modeling local sources in subregional level planning. The regional modeling studies will provide an understanding of the interrelationships among different primary sources and sources of precursors to secondary particle formation. That level of understanding 21 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 is necessary to develop effective control strategies for PM-2.5 in most areas. Emissions data need to be spatially resolved into a grid cell system, temporally allocated to hourly values, and chemically allocated into the individual species that interact in the controlling chemistry before they can be used as input for these air quality models. One approach that has been used to evaluate large scale regional air quality issues involves nested modeling. In this approach, the modeling spatial resolution begins with a relatively coarse grid size of up to 80 km on a side. The spatial resolution is increased in steps for subregions that surround the planning area. The highest level of resolution might be a grid system as small as 2 km on a side to describe the urban area. Each level of resolution is used to provide boundary conditions for the smaller scales and to account for the main large scale emissions and meteorological influences that effect the local area. The region included in the most coarse grid system can be over 1,000 km and can include several States. These analyses will be similar in nature to investigations of the large scale influences of small particles on regional haze on the Colorado Plateau that were completed by the Grand Canyon Visibility Transport Commission (GCVTC). For States east of the Mississippi River that were not part of the GCVTC, the efforts to evaluate the effect of regional NOx emissions on ozone concentrations that were completed for the Ozone Transport Assessment Group (OTAG) is an example of this kind of analysis. Discussions of these programs that describe the planning that went into them and the conclusions that were developed can be found at the following Internet sites: http ://www.epa.gov/ttn/otag/ http://www.nmia.com/gcvtc/ Air quality management plans are composed of many different parts. While each of these parts is linked to emissions data, inventories that differ in detail in terms of source specificity, pollutant specificity, spatial coverage, and timing can be applied to different planning activities. This section discusses some of these issues from the perspective of inventory development and gives some ideas of the principal issues that might influence inventory preparation. Whenever possible regional or geographical influences will be discussed to assist in identifying specific planning issues that will be important in different parts of the country. The following discussion addresses spatial influences and temporal influences separately. This organization is used only to simplify the presentation of the important issues, and the reader should understand that these two influences act together to control the emissions characteristics in myriad ways. This is particularly true in PM-2.5 planning, since certain spatial influences operate within specific temporal regimes. The combined effects of the spatial and temporal influences can have quite different results when planning in response to high 24-hour concentrations or for high annual concentrations. 22 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 Spatial Influences Assessment of the limited PM-2.5 ambient monitoring data leads to the observation that PM- 2.5 air quality in different regions often results from different combinations of influences. There are certainly contributions to PM-2.5 air quality that are ubiquitous (e.g., carbonaceous components from diesel engines), but there are also significant contributions to PM-2.5 that vary on local and regional scales. An obvious example of a local influence is residential wood burning. For example, wood burning for residential heating will affect ambient PM-2.5 in many locations, but will not be a concern in New York City. An example of a regional influence is the observed contribution of ammonium sulfate that contributes roughly 50% of the ambient PM-2.5 in most samples in the east, but only 20% or less throughout the west. States should review all of the data that is available to understand the origins of PM-2.5 in their areas during periods of high ambient concentrations. Then States can begin a preliminary assessment of the sources of both primary PM-2.5 and the precursors to secondary PM- 2.5 that are important in those areas. Some of the spatial influences that need to be considered when preparing a plan to develop emissions data for PM-2.5 are summarized in Table 2. A critical concern relative to spatial influences is to prepare emissions estimates with spatial resolution that is consistent with the air quality problem(s) affecting the area. For example, some areas may experience high 24-hour concentrations that are affected by residential wood burning. This type of problem can often be confined to limited spatial scales within a mountain valley. The inventory to support analyses of such a problem needs to be resolved within the valley. An emissions inventory technique that calculates a county-wide emissions estimate will not provide the information necessary to evaluate the conditions within the valley. This is particularly true in the west where counties often cover very large areas. In this example, the regional contributions to the problems can be handled with coarse resolution, but local influences will have to be characterized at finer spatial detail. In some cases, emissions data may need to be prepared at varying spatial resolution to develop a complete understanding of the relative contributions of the different causes of PM-2.5. The analyses to understand these complex issues may rely on more than one modeling approach. Inventories with fine spatial resolution can often be aggregated to represent coarser spatial resolution for input into larger scale modeling exercises without introducing significant bias. The reverse process will almost always introduce bias and uncertainty when inventories with coarse spatial resolution are used to derive a more spatially resolved inventory using surrogate distribution factors such as population or land use factors. Some of the activity data associated with PM-2.5 sources are related to land use or other geographic features that are unrelated to county or other geopolitical boundaries. Examples include acreage in various types of agricultural use, forested areas, and animal husbandry 23 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 TABLE 2. SPATIAL INFLUENCES IN PM-2.5 PLANNING Source Category Type of Emissions Spatial Concerns Modeling Terrain Transport Fossil Fuel Combustion - Stationary Sources Primary and condensable emissions, and precursors to secondary PM-2.5 Regional modeling applications; Primary PM-2.5 may be used in local modeling Terrain not important in regional applications, may be important in local applications Secondary precursors can be transported over considerable distances, emissions mainly released from tall stacks Industrial Sources Primary and condensable emissions Regional modeling applications; Primary PM-2.5 may be used in local modeling Terrain not important in regional applications, may be important in local applications Not generally as important for long range transport as are utility sources; Emission are sometimes transported down valleys with typical diurnal valley flow regimes Agricultural Sources Primary emissions and precursors to secondary PM-2.5 Important for Source apportionment modeling, field burning also for regional modeling Terrain and surface features can prevent emissions release to transport layers, local feature concentrate pollutants Field burning emissions can transport over large distances, fire, dust and other emissions likely to be local influence; deposition is important issue that needs additional study Mobile Sources Primary and condensable emissions and precursors to secondary PM-2.5 Important on both regional and local scales Urban canyon effects, and inversions concentrate soot Important in regional modeling, and secondary aerosol formation,local sources can be concentrated in urban areas Fugitive Dust Primary emissions Mainly of concern for local modeling Terrain and surface features can prevent emissions release to transport layers, local feature concentrate pollutants Mostly of concern in local transport; deposition is important issue that needs additional study Other Combustion Primary and condensable emissions and precursors to secondary PM-2.5 Important on both regional and local scales Terrain effects can be significant Emissions from open fires can transport over very large distances, but can also be very significant on local scales, deposition is important issue that needs additional study ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 activities. Often it is convenient to organize and manipulate this type of activity data using geographic information systems (GIS). When such data are already available in GIS format, States should consider ways to use those data in their planning efforts. It is usually relatively simple to aggregate such data to be consistent with more standard geopolitical features to facilitate other analyses associated with air quality management programs. Several of the sources listed in Table 2 are identified as significant local sources of primary and condensable components of PM-2.5. Most of these local sources result in emissions that are released close to the surface. In all cases, when these types of sources are located in river valleys, or in basins surrounded by mountains, those primary PM-2.5 emissions and surface emissions of secondary precursors will concentrate under conditions with limited mixing volumes caused by temperature inversions. Once the inversions break and the area is ventilated those trapped emissions can be transported downwind to contribute to regional problems. In any given area, any measured high concentration of PM-2.5 will likely be the result of the combined contributions from local sources and regional processes. The important issues for each specific area must be determined from analyses of all available information. The types of information that will be useful to begin this assessment include the preliminary inventory of primary emissions of PM-2.5 for each county that was developed as part of the periodic emissions inventory process (NET 1996), available ambient monitoring data, and characterizations of visibility reducing conditions that affect any given area. Temporal Influences Temporal factors can influence planning efforts in several ways. Perhaps the most important temporal consideration is whether analyses are completed to address 24-hour concentrations, annual concentrations, or both. In most urban locations, it is almost certain that multiple emissions scenarios will be responsible for high annual concentrations. For example, there could be different causes of high PM-2.5 concentrations in summer and winter months, and both seasons contribute to annual average concentrations. It is also possible for high 24-hour concentrations to occur at different times of the year, and be related to quite different origins of PM-2.5. In these cases, the planning effort will have to address each of the potential sources contributing to high concentrations. Planning efforts to address high 24-hour concentrations will require inventory data that are resolved to hourly levels for those conditions that result in the high concentrations. The hourly inventory might need to also reflect a particular seasonal condition or a specific set of meteorological conditions (e.g., hot, dry and stagnant meteorology). Planning for high annual concentrations will likely require inventory resolution that reflects seasonal variability. Monthly resolution is recommended for these applications. 25 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 Frequently, strong persistent low-level temperature inversions are accompanied by low surface winds and minimal advection. In many locations those conditions that can increase the contribution of local sources are common during the winter months. On the other hand, most of the sources of secondary particle precursors will have a greater affect during the summer months when meteorological conditions favor deep mixing and photochemical activity. In addition, there are conditions that vary seasonally and diurnally that simply favor higher than normal emission rates of PM-2.5 and important precursors to PM-2.5. For example, emissions from unpaved roads and construction activities will increase during periods that are both dry and windy relative to emissions under wet and calm conditions. All of these considerations must be included in the preliminary planning for inventory development. Some of the temporal influences that might affect inventory development are listed in Table 3. Most of the issues that influence 24-hour concentrations are related to local sources while planning analyses of annual average concentrations must also consider the regional influences of secondary particle formation. Assessing Priorities The discussion above has pointed out some of the interrelated issues that will affect planning activities to address PM-2.5. The variable origins of PM-2.5 will present challenges for effective planning to address the range of sources that influence PM-2.5 concentrations. While some of the sources of PM-2.5 can be treated with refinements to the approaches historically used in PM-10 planning, there are additional area sources of PM-2.5, and the precursors to PM-2.5 that have not been important in any planning activities completed previously. For example, PM-2.5 planning will require comprehensive estimates of emissions of NH; as a precursor for secondary formation. It will also be useful to differentiate between elemental and organic carbon from primary PM sources. Initially, States will have to prioritize efforts and focus on the causes of high PM-2.5 concentrations in their State and other affected nearby States, and the sources that can be controlled. EPA anticipates that the majority of planning activities will focus on the causes of and control of sources that result in high annual concentrations of PM-2.5. Emissions estimates may need to be developed with monthly resolution to address the strong temporal influences that affect sources of PM- 2.5 and precursors to PM-2.5. In local areas, additional processes that vary over shorter time scales might also be important and, therefore, additional temporal resolution may be required in some areas. 26 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 While it is expected that there will be common contributions to the PM-2.5 problems in all areas, it is almost certain that different States will identify different basic scenarios that result in the high measured concentrations of PM-2.5. Planning to address these problems will be 27 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 TABLE 3. TEMPORAL INFLUENCES ON PM-2.5 SIP INVENTORY DEVELOPMENT Source Category NAAQS Temporal Scale Temporal Issues Fossil Fuel Combustion- Stationary Source Primarily annual, some 24-hour Seasonal to Annual - Hot, humid conditions with deep and active mixing layers, related to ozone season - Potential for emission rates to vary with seasonal electric demand. Local Industrial Sources Primarily 24-hour Diurnal - Local sources of primary and condensable PM-2.5 dependent on actual wind speed and direction - Some local sources could be dependent on temperature - Potential to concentrate under tight surface temperature inversions. Agricultural Sources of NH3 Both annual and 24-hour Primarily Seasonal - NH3 resulting from fertilizer application dependent on season and soil moisture content - NH3 emissions from livestock production increase with temperature - NH3 emissions may be deposited near the source by nearby vegetation when in full leaf Mobile Sources Both annual and 24-hour Primarily Diurnal - Primary emissions follow traffic density patterns - There are differences between daily emissions (week day, weekend day, etc.) Fugitive Dust (including agncultural fugitive dust sources) Emissions appear to be overstated for these sources using current methodologies. Both annual and 24-hour Seasonal - Some paved road dust dependent on road sanding practices - Road dust dependent on soil moisture content - Agricultural sources dependent on growing cycle activities - Construction activities dependent on seasonal weather conditions Other Combustion Primarily annual Seasonal - Residential wood burning is confined to winter months - Biomass combustion dependent on growing cycles and other seasonal considerations ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 facilitated by continuing a partnership between EPA and the States and among the States. EPA will be providing tools to assist in the analyses of sources and emissions that are ubiquitous. Each State will have to develop its own tools to address the specific local influences. States should explore opportunities to share the results of their work with each other, so that each State will not have to expend limited resources on all of the possible problems. To that end, each State should focus on developing tools and improving the activity data, emission factors and/or emissions estimation methodologies for the important sources in their local planning areas that can be controlled to help reduce ambient concentrations. The application of the national-level tools being developed by EPA, other methodolgies and emission estimation tools developed by other States and any specific tools developed within each State will provide resources to begin the planning process. It is recognized that as the program develops and the collective experience in PM-2.5 inventories matures, additional refinements and improvements in the inventory development will be available. For sources that rank low on the priority scale, initial emission estimates based on national-level methods and activity data available through the area source model will suffice. Similarly, for low priority point sources simple size distribution functions included in AP-42 (Compilation of Emission Factors) can be used to relate PM-2.5 emissions to PM-10 emissions. Other estimation methods for low priority sources could include the application of per capita, or per employee factors or land use factors that are based on national-level activity data. Estimates based on that level of detail may be sufficient to treat the sources that are not large emitters and can not be controlled to levels that would be of use in the control strategy. Therefore, resources should be concentrated on developing reliable activity data, and in improving the emission factors and estimation methods for those sources that contribute significant emissions and can be controlled with identifiable technology at acceptable costs. Some of the steps that can be taken to develop priorities are discussed below. • Review all available ambient monitoring data including speciated samples to get an understanding of the principal causes of PM-2.5 that affect the expected planning areas. For example, rank likely local sources resulting from transportation, open burning, residential wood burning, fugitive dust, industrial sources, etc., relative to the contribution from regional secondary aerosol. The inventory development plan may require coordination with nearby States in addition to collection of activity data, and the development of emission factors for the local sources. • Review the draft inventory data for counties in the expected planning area and establish a ranked list of the main contributions of primary PM-2.5. • Rank the sources of the main precursors of PM-2.5 represented in the draft inventory in a similar way. 29 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 • Identify important sources that affect the area that need new or improved emission factors, size distribution factors, new or improved sources of activity data, etc., to support the planning effort. • In the near term, States can begin to identify contacts, locate pertinent sources of activity data, and start developing an inventory preparation plan with schedules. • Develop a plan and the instruments necessary to conduct surveys of the activity data for relevant priority sources (e.g., surveys of wood use for residential heating, typical agricultural or prescribed burning activities, or construction activities). • Review available data on the size distributions and source speciation for the major contributors to PM-2.5 in the expected planning area (size distribution and source composition profiles in SPECIATE, previous chemical mass balance (CMB) or other receptor modeling analyses, and in published research papers). SPECIATE is a collection of source profile data and CMB is a source receptor modeling approach. Both of these tools can be found on the EPA Internet site. • Compare source data and ambient data to derive intuitive source/receptor relationships (e.g., speciated linear rollback models). • Apply receptor modeling and other dispersion modeling to any identified fingerprint or tracking species. • Evaluate the potential amount of emissions control possible (known control technologies or pollution prevention methods and estimated control cost). Once the priorities have been established, identify possible assistance from the affected sources or trade associations that represent the affected sources. New or refined approaches for developing emissions estimates should take advantage of all readily available data related to specific processes and activity levels. Surveys are a preferred method to develop activity data for several of the source categories that are important in PM-2.5 planning. For example, sources such as residential wood burning should rely on surveys of households, while sources such as fertilizer application or prescribed burning can rely on surveys of fertilizer distributors, or public and private land managers, respectively. Details on conducting surveys will vary on a case-by-case basis. Recent EIIP documents and other EIIP documents that are available for review at the Internet site below, provide useful information on how to conduct surveys and the type of information that can be collected through surveys.12 http://www.epa.gov/ttn/chief/eiip/techrep.htm TECHNIQUES FOR INVENTORY VALIDATION 30 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 Depending on the nature of specific problems, any of the available tools and data sources discussed in this report may introduce uncertainty or bias when applied to individual local situations. States are encouraged to evaluate the reliability, representativeness and uncertainty of these techniques to understand the impacts of any potential uncertainty or bias for local area planning efforts. The following discussion presents a selection of techniques that States might find useful to complete this type of evaluation. The techniques discussed here are not the only methods that can be used to validate emissions inventories, but are discussed as examples of the kind of analyses that can be applied. States are encouraged to develop additional methods and techniques to assist in the evaluation of inventory data quality. Speciated Linear Rollback and Simple Regression Models The concept of linear rollback was first applied in early efforts by EPA to understand and prioritize the relative effects of various emissions sources on ambient air quality problems. This technique is based on the assumption that reductions in the amount of emissions from sources and/or source categories that affect a particular air quality problem will result in proportional reductions in the air quality measurement. This assumption does have some validity, although most air quality problems, including PM-2.5, do not exhibit a one-to-one relationship between emissions and air quality. The speciated linear rollback model is based on the assumption that the mass of each type of particle (e.g., sulfate, nitrate, elemental carbon, etc.) is related to the spatially averaged emissions of the relevant pollutants. The requirements for the application of this approach are a speciated inventory covering the major contributors within the region, and a speciated ambient data set that represents the temporal averaging of the primary sources. The model will identify the contributing sources or source categories based on the resulting ambient air quality. The model can be used to complete preliminary tests of the adequacy of the regional inventory and to help prioritize efforts to find sources that may not be adequately represented or those that may not be represented at all. In this regard, it is a simplified receptor model. Regression models use empirical relationships derived from the source and ambient data to apportion ambient particulate data among the distribution of sources in the region. The technique requires that data are available over a range of conditions and over a reasonable expanse of time. If sufficient information is available, these models can attribute source contributions for the precursors to secondary particulate, as well as the direct emissions of primary particulate. Both of these techniques can be applied with limited information in a screening mode to identify serious weaknesses in the underlying emissions data. Take as an example a case in which these techniques indicate that there must be a major source or collection of sources that contribute organic 31 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 carbon. Upon review of the inventory, no sources of organic carbon with a sufficient magnitude are found. This result would lead to an investigation of the possible sources of the missing organic carbon. Details of the application of the speciated linear rollback model and regression models can be found elsewhere.1 Data Attribute Rating System (DARS) Approach The Data Attribute Rating System (DARS) was developed to assist in evaluating data associated with emission inventories. DARS provides an overall confidence rating for emissions inventories, that in most cases is less subjective than the usual qualitative letter grade rating procedures (i.e., A through E) used to characterize data quality. DARS was originally developed as a research tool for rating national and global greenhouse gas inventories. State agency personnel used DARS to rate their base-year State Implementation Plan (SIP) ozone precursor inventories. In addition, particulate matter (PM-10) inventories (state and national levels) were evaluated by inventory developers trained in the use of DARS. More detail on the development and application of DARS can be found in EIIP documents available on the EIIP Internet site shown previously.13 The proposed applications of DARS include: • to validate emissions estimates to identify the weakest areas of an existing inventory for further research and improvement, • to quickly compare and rank different inventories, • to rank alternative emission estimation methods, and • to set Data Quality Objective (DQO) targets during inventory planning and for future inventories. The DARS score is based on the perceived quality of both the emission factor and activity data. Numerical scores are assigned to four data attributes: measurement/method, source specificity, spatial congruity, and temporal congruity. These scores, which range from 1 to 10, reflect the confidence that the user associates with each of the attributes. A key feature of DARS is that these attributes are orthogonal, that is, they are independent of each other, and therefore, the score for each attribute is independent of the other scores. However, the emission factor and activity scores for a given attribute are not necessarily independent. This is because the choice of one is usually limited by the selection of the other. Table 4 shows a DARS scoring box. The procedures for filling in the scores for emission factors and activity are briefly described below. The emissions scores for each attribute (i.e., the right- 32 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 hand column of the box) are computed by first dividing each score by 10, and then multiplying the factor score times the activity score. The composite score for factor, activity, and emissions (i.e., the bottom row of the box) are computed by averaging the scores in a column. TABLE 4. DARS SCORING BOX Attribute Factor Activity Emissions Measurement/Method ei ai ei*ai Source Specificity e2 a2 e2*a2 Spatial Congruity e3 a3 e3*a3 Temporal Congruity e4 a4 e4*a4 Composite 4 Z ei i=l 4 I a; i=l L (ei * a;) i=l 4 4 4 Applications Although DARS was originally conceived as an application for global inventories developed using area source inventory methods, not as an application for point sources, the method can be applied to inventories at any scale. Scoring an entire point source inventory, however, will nearly always require more effort than required for the area and mobile source inventories. DARS scores can be applied to groups of sources rather than to individual sources. The scorer may want to spend more time on sources where indirect or nonstandard methods are used. Also, keep in mind that some approximation methods, like mass balance used at an individual facility, may be more accurate than when used to develop a national estimate. If other losses (i.e., non-air releases) are accurately accounted for, the DARS activity measurement score could reach a 9 (and possibly a 10 if well supported). Keeping the above comments in mind, the general guidance given for area sources will apply to point sources. For a given source type (e.g., industrial fuel consumption), if AP-42 emission factors are 33 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 used, the measurement attribute score will be the same as for area sources. The scores for the other attributes will usually be different. One of the strengths of the DARS approach is that the significant factors that relate to the application of both emission factors and activity data represented by the attributes are considered independently. Analyses using DARS can be applied on varying spatial and temporal scales for purposes of evaluating emissions estimates. Comparisons of the results of DARS at national- versus local-levels, for instance, can be very instructive to help establish the inventory development priorities. Use of CMB and Dispersion Modeling Dispersion and receptor modeling are powerful tools for identifying and assessing various sources of primary emissions. The utility of these tools increases for applications to chemically inert species, but, in many cases, these tools can be used to evaluate sources of reactive components as well. One receptor modeling approach based on the Chemical Mass Balance Model (CMB), has been used effectively in many PM-10 planning efforts. The results of CMB modeling often can point to the major sources of the primary particulate and even to give estimates of relative source magnitudes. Similarly, dispersion modeling of chemically inert components can provide important information on the relative accuracy of emissions estimates. In both of these applications ambient concentration data are required. While total mass concentration can be applied to some analyses, the power of these tools increases dramatically when applied to data that represent a full range of chemical speciation in the ambient data set. Currently, there are no ambient data for PM-2.5 collected using the FRM. Monitors consistent with the FRM are being developed and the first of these monitors will be installed in the near future. The current monitoring plan requires that a subset of these monitors will collect samples for analyses of the chemical composition. These data will provide a wealth of information to use in CMB and dispersion modeling analyses to verify important sources, evaluate emissions estimates, and to prioritize emissions inventory improvement efforts. Of course, the use of these approaches also requires detailed source composition data. Currently, the only chemically speciated source data for PM-2.5 have been developed in research field studies. Some of these data include the condensable fraction and others do not. Many of the studies were completed prior to the promulgation of the NAAQS with the specification of the 2.5 |im cut off and some available profiles may be based on other size cutoff" limits. Even with the difficulties presented with the existing data, it may still be possible to apply either CMB or dispersion modeling to help States evaluate the specific sources and source combinations that are important in particular areas. 34 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 This type of analysis will be most useful when applied across seasons, emissions scenarios, and other temporally variable scales that can affect the emissions from specific sources or the mix of emissions from the collection of sources in any given area. In the majority of cases, emissions data are expected to be applied to analyses of annual ambient concentrations. The causes of high ambient concentrations might change, however, with season or other temporally cyclical conditions. Repeated comparisons of the composition of the ambient samples to source components can be extremely informative in terms of identifying the principal causes of high measured concentration and, therefore, ultimately in developing effective control plans. Development of Source Profiles for CMB Modeling Most local sources of primary PM-2.5 will be combustion sources, and the emissions will include a significant fraction of carbon. Direct sampling data and/or estimates of emissions from sources that are similar to other measured sources will benefit significantly if the elemental and organic carbon fractions are differentiated. This is such a critical requirement that States are cautioned that sampling of sources is not expected to be cost effective if results are not available for both elemental and organic carbon. Speciation of other components of the source mix that can serve as unique markers for specific sources will also improve source apportionment analyses. Developing the source profiles can be an expensive part of any source apportionment project and the testing phase should be planned carefully to develop all of the data necessary. Other issues of concern in developing source profiles are listed below: • Ensure that plans are made to collect data with sufficient chemical speciation to allow CMB to distinguish specific contributions. • Develop profiles for all of the important sources affecting the receptor. Keep in mind that profiles that are specific for the actual sources of interest improve the confidence in the results relative to analyses that use source average or surrogate profiles. • Complete screening tests to determine if two important sources have characteristics that are so similar that CMB can't distinguish between them. If that appears to be the case, consider adding a chemical marker to one or both of the sources to make it unique. • Test the sources under the typical operating conditions that are believed to result in the air quality effect. • Be sure to collect data that is representative of the spatial scale and the temporal scenario under which the air quality effect is observed. SUMMARY AND RECOMMENDATIONS 35 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 Although the development of PM-2.5 emissions inventories will represent a new planning activity for States, there is a considerable amount of information available to begin gathering basic information on the sources of PM-2.5. EPA is currently working to bring the available information and inventory development tools together to make them more accessible to States. Many of these tools and data sources are national in scope and may not represent sufficient detail for all local planning applications. These information sources do, however, provide a firm basis upon which future national and State planning efforts will build. The task ahead is challenging, and will be facilitated by coordinated efforts involving the EPA and States. The regional nature of PM-2.5 will require the development of regional approaches, combined with appropriate local control measures. It is recognized that mandated national-level planning and control activities will not always provide the most efficient solutions for all States. States are encouraged to form regional planning organizations and to use existing organizations like Western States Air Resources (WESTAR) and North East States Consortium for Air Use Management (NESCAUM) to collectively address some of these regionally specific problems. In the meantime, EPA will continue to work with the States to provide guidance on significant issues that are important on the national-level, and to help States coordinate efforts and to share information. On-going National-Level Activities 1. EPA will continue to develop guidance documents for PM-2.5 inventory development. The guidance documents will provide information on what is expected of the States, including schedules, and will direct States to data sources and other relevant information on how to collect and assemble activity data, and to apply emission factors. 2. EPA will review the information that is already available and prioritize PM-2.5 source categories that are important on the national scale. The prioritization will be based on expected emissions magnitude, potential for emissions control, and assessments of the credibility and reliability of existing activity data and emission factors. 3. Based on the results of that review, EPA will begin to address those issues that are of high priority on a national scale (e.g., evaluate and improve outdated or low quality emission factors, and address area source methods for sources of national importance). 4. EPA will continue to coordinate with the DO A, DOI, and other organizations to refine methods for calculating emissions from fugitive dust and open burning sources. 5. EPA will continue to support the development of national-level emissions estimates in the form of updates to the draft 1996 NET PM-2.5 inventory. 6. EPA will continue to develop estimation tools like the ASEM, FIRE, and SPECIATE. 7. EPA Emissions Measurement Center will continue to assess candidate methods for a reliable source test to distinguish between filterable and condensable PM-2.5. 36 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 8. EPA will continue to oversee and support the deployment of the PM-2.5 ambient monitoring network and to assist States in the collection and evaluation of ambient mass and speciated PM-2.5 data. Recommendations for State Activities 1. States should develop a preliminary inventory preparation plan. This should include an initial assessment of the priority issues and which of those can benefit from national-level activities and which will be more appropriately handled at the State- and regional-level. The plan should include a strategy and a schedule for collecting the appropriate information and merging that information with existing data. The plan should address all relevant spatial scale issues, and temporal considerations. 2. States should become actively involved in coordinating with EPA and with one another to ensure that the results of all development efforts are applied in a timely way and that efforts are not duplicated among States. One effective opportunity is to participate in EIIP activities and use that forum to address key issues of concern. 3. States should review all available EIIP documentation on methods and approaches for estimating PM-2.5 emissions from area sources. These documents will provide useful information on preferred methods for obtaining and managing activity data for important area sources of PM-2.5, open burning sources, residential wood combustion, etc. The document on residential wood is available and others are near completion. 4. States should review the existing inventory data, emission factors in AP-42 and supplements to AP-42, and all ambient monitoring data as they become available, to develop an understanding of the type and scope of the problems that affect them. 5. States should begin to identify local sources of activity data for area source and point source categories that can be used to replace national-level default values that will be provided in the ASEM. This activity should include plans for conducting surveys to specify local activity rates, timing of significant fire events, and other significant features related to source strength for appropriate categories. Guidance on conducting surveys for several biomass burning categories is included in EIIP documents. 6. States should begin coordinating with other nearby States to develop cooperative agreements for sharing the burden of data collection, and in developing regional plans to address PM-2.5 issues. States should contact representatives from GCVTC States and other existing State compacts to get advice on and to facilitate that type of cooperation. 7. States are encouraged to become active in the EIIP process to make contacts, identify joint projects and to share information. 8. States will need to develop a list of priorities that are of importance in their local areas and complete decision making processes on how best to address those priorities and 37 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 how to commit resources to maximize their understanding of the processes that result in high PM-2.5 concentrations. The list of priorities for any particular area should consider EPA list of national priorities and focus on those issues that will not be covered by the national effort. 9. States should begin to review existing approaches for source apportionment, and other models that are useful to discriminate the relative contributions of various sources to observed PM-2.5 concentrations. 10. States should review policy guidance on prescribed burning and the application of smoke management plans and the relationship of those activities to PM-2.5 planning. States should begin a dialog with Federal Land Managers (FLM), who have familiarity with the policy and information on how fires are managed in terms of acres burned and timing of fires in each local area. 38 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 REFERENCES 1. National Research Council, Protecting Visibility in National Parks and Wilderness Areas, National Academy Press, Washington, DC 1993. 2. U.S. Environmental Protection Agency, National Air Pollutant Emissions Trends, 1900-1996, EPA-454/R-97-011, December 1997. 3. U.S. Environmental Protection Agency, National Air Quality and Emissions Trends Report, 1996, Chapter 3: Visibility Trends, EPA 454/R-97-013, January 1998. 4. Pace III, Thompson G., and Kuykendal, William B., Planning Tools for PM2.5 Emission Factors and Inventories, Presented at the 91st Annual Meeting and Exhibition of the Air and Waste Management Association, San Diego, CA, June 14 - 18, 1998. 5. Brook, Jeffrey R., Dann, Rom F., and Burnett, Richard T., The Relationship Among TSP, PM10, PM2.5, and Inorganic Constituents of Atmospheric Particulate Matter at Multiple Canadian Locations, Journal of the Air and Waste Management Association, Volume 47, January 1997, pp. 2-19. 6. Definition of elemental and organic carbon is included in the draft of the Speciation Guidance Document (found under "Speciation" section of the AMTIC Internet Home Page). The analytical method (NIOSH Method 5040) can be found in Birch, M.E., Analyst, Vol. 123, pp 851-857, May 1998. 7. Watson, J.G., Fujita, E.M., Chow, J.C., Zielinska, B., Richards, L.W., Neff, W., and Dietrich, D. Northern Front Range Air Quality Study (NFRAQS), Final Report Executive Summary, DRI Document No. 6580-685-8750.1F2, June 30, 1998. 8. Duyzer, J. Dry Deposition of Ammonia and Ammonium Aerosols Over Heathland, Journal of Geophysical Research, Vol. 99, No. D9, ppl8,757 - 18,763, September 20, 1994. 9. U.S. Environmental Protection Agency, Compilation of Air Pollutant Emission Factors Volume 1: Stationary Point Sources, Fifth Edition, AP-42, Supplement D, Section 1.4 Natural Gas Combustion, 1998. 10. Emission Inventory Improvement Program, Volume 1: Introduction to the Emission Inventory Improvement Program, Prepared by Eastern Research Group, July 1997. 39 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 11. U.S. Environmental Protection Agency, National Air Pollutant Emissions Trends, Procedures Document 1900-1996 (Section 4: National Criteria Pollutant Estimates, 1985 - 1996 Methodology), EPA-454/R-98-xxx, 1998. 12. Emission Inventory Improvement Program, Volume 3: Area Source, Chapter 2, Residential Wood Combustion, prepared by Eastern Research Group, September 1997. 13. Emission Inventory Improvement Program, Volume 6: Quality Assurance Procedures, Appendix F. EIIP Recommended Approach to Using the Data Attribute Rating System (DARS), prepared by Eastern Research Group, July 1996. 40 ------- APPENDIX A PIE CHARTS OF SPECIATED AMBIENT PM-2.5 NOTE: The pie charts are best viewed on the screen in color or after printing to a color printer. Printing these pie charts on a black and white printer may result in gray scale gradations that will be difficult to distinguish. A-l ------- Figure A-1. PM-2.5 Composition irt the Eastern United States Nonurban locations Urbar Boundary Waters (5.1 ug/m3) New England Average (9.5 ug/m3) Appalachian & Mid-Atlantic (11.35 ug/m3) Washin (19.2 Mid-South (12.1 ug/m3) Note: PM-2.5 mass concentrations are determined on at least 1 year of monitoring at each location reference methods. They should not be used to determine compliance with the PM-2.5 NAAQS. A-2 ------- Figure A-2. PM-2.5 Composition in the Western United States Urban locations Nonurb: Spokane (11.0 ug/m3) San Joaquin Valley (Avg - 30 ug/m3) South Coast (Basin Avg - 28 ug/m3) vv. Knoenix (13.5 ug/m3) Sonoran Desert (4.3 ug/m3) Note: PM-2.5 mass concentrations are determined on at least 1 year of monitoring at each locati reference methods. They should not be used to determine compliance with the PM-2.5 NAAQS. A-3 ------- A-4 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 APPENDIX B. STATUS OF PM-2.5 EMISSIONS ESTIMATION TOOLS This Appendix provides a summary of the status of available emission factors, emissions estimation algorithms and activity data that could support the preparation of future PM-2.5 emissions inventories. The Appendix is not meant to be a comprehensive list of all factors and approaches, but rather a guide concerning the perceived strengths and weaknesses of the methods and tools that are available at this time. The specific information that is available can be obtained through readily accessible sources. Those information sources are identified in this Appendix. A summary of the available data and further needs is provided in tabular form in Table B-l. Stationary Combustion Sources Emission factors are available for PM-2.5 and its precursors for most stationary combustion sources. Many of the factors are based on size distribution functions. Most, if not all, of those size distribution functions were obtained through source testing using some form of size separation sampling device (e.g., cascade impactor, multi-cyclone separator). In some summaries, the factors for PM-2.5 may be presented as a percentage of PM-10 or total particulate. The reader is advised that in most cases these percentage factors are based on sampling data, and are not simply an assumed fraction. These factors apply to S02, NOx, VOC and PM-2.5 for most stationary combustion sources. Many of the factors that are applied to area sources may have been developed specifically for point sources. These combustion sources should not differ widely, however, in terms of emissions per unit of fuel consumed. The PM-2.5 factors are largely representative of the filterable fraction, although in some cases a condensable fraction or total PM-2.5 is represented. These factors can be found in AP-42 - Fifth Edition, the supplements to the fifth edition, and in the procedures document for the 1996 NET inventory. There are three remaining weaknesses related to PM-2.5 from stationary combustion sources. As indicated above the condensable component of PM-2.5 is not well characterized for many of these sources. Work is now underway to add these factors. Recent updates to AP-42, including Supplement D to the Fifth Edition, do include some specific factors for condensable PM-2.5 from selected source categories. The processes that result in the formation of condensable PM-2.5 are not well understood at this time. Although VOC emissions from most stationary combustion sources are typically minimal, there is the potential for some of these emissions from low temperature firing applications to participate in the formation of SOA. Currently, there is an incomplete understanding of the contribution and role of VOCs from combustion sources in the formation of SOA. B-l ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 There is also a considerable lack of understanding of NH3 emissions from point sources. The tools to adequately represent NH3 emissions from stationary combustion sources are limited. Stationary combustion sources are not thought to be a significant source of NFL; and, therefore, this weakness is not expected to have serious consequences in terms of national-level planning in the short term. Improvements in these factors will be needed to promote higher accuracy predictions of secondary particle formation in future studies. There are likely specific processes that will be important sources of NH3 in some local areas. Further efforts will be necessary to address these local sources. The NH3 factors that are available are discussed in the procedures document for the 1996 NET inventory. Activity data for stationary combustion sources are the same as those used for these sources in the preparation of other more traditional inventories. There is adequate guidance available to assist States in the collection and organization of these data. There are no significant weaknesses related to the development of activity data for sources that are of national importance. Some weaknesses may exist in the tools and data sources for developing activity data for selected sources that are important in a limited number of local applications.. Open Burning Sources The data sources available for estimating PM-2.5 from open burning sources arise largely from size distribution functions applied to data that had been developed to support PM-10 planning. There is a significant amount of information related to filterable PM-2.5 emissions, although similar confidence can not be attributed to the condensable fraction from all of the origins of PM-2.5 from open burning. Emissions of NOx and VOC are also released from these sources. These sources tend to burn their fuel at lower temperatures than do boilers and other combustion related point sources. One result is an increased potential for emissions of organic carbon whether emitted directly in the solid phase or in the condensed phase. Characteristics related to NOx and VOC emissions can also be quite different relative to high-temperature fuel combustion in boilers. The information that is available is summarized in the procedures document for the 1996 NET inventory, and additional information can be found in recently published and soon to be published EIIP documents (Open Burning Sources Chapter V, Volume 16). This guidance does not yet include information related to the agricultural or forest related sources of open burning. EPA is continuing to coordinate with organizations through the DOA and the U.S. Forest Service to develop information and resources to assist in calculating emissions from these sources. Reports will be prepared in the near future to provide information useful for application to these kinds of sources. EPA has developed a policy on prescribed burning to promote the sound management of forest ecosystems, to maintain habitat for threatened wildlife species, and to limit the potential for destruction from natural fires. States in regions where prescribed burning is practiced for any of these reasons should become familiar with this policy and develop Smoke Management Plans (SMP) if applicable. B-2 ------- DRAFT FOR REVIEW AND COMMENT Do not cite September 21, 1998 Activity data to support emissions estimates for these sources vary considerably. Details on available information and how to improve that information from local perspectives is provided in the EIIP documents. Additional efforts are needed to prepare adequate estimates of activity data for these sources. Mobile Sources Emissions of VOC and NOx from mobile sources are routinely estimated using well established approaches and data collection techniques. Guidance from EPA's Office of Mobile Sources (OMS) and the Office of Air Quality Planning and Standards (OAQPS) is available to assist in the collection and application of both emission factors and activity data. The data available for PM-2.5 emitted both as solid particles and as condensable material is less established. OMS has developed an emissions estimation model for PM called PART5. The current version of PART5 operates with input similar in nature to the input files used to support other mobile source emission factor models prepared by OMS. The model uses measurement data to calculate an aggregate emission factor for PM-2.5. Some issues remain concerning the application of the model. First, in its current form, the model produces a fleet average emission factor and can not separate output factors by vehicle class. This condition makes it difficult to separate out the competing influences from various types of on-road or off-road heavy-duty diesel vehicles. Overall emission factors will also include emissions from tire wear and brake wear. OMS has also recently completed an assessment of NH3 emissions from various types of mobile source activities. These data will be reflected in subsequent national inventories (e.g., 1997 NET inventory). All emission factors are presented in terms of VMT for on-road vehicles and in terms of hours-in- operation for off-road vehicles. Additional work is needed to improve the understanding of and data to support the development of emissions of PM-2.5 from mobile sources. Fugitive Dust Sources While it is generally recognized that most fugitive dust is generated or quickly aggregates into forms that are in the coarse range (2.5 |im to 10 |im), there is still the potential for a contribution of PM-2.5 from fugitive sources in selected areas. Data on source strengths based on size distribution functions suggest that there is indeed a large amount of mass included in the emissions of particles under 2.5 |im. The ambient data collected, usually at more than kilometer distances away from these sources, suggest that the amount of mass that is entrained into the prevailing transport regime, and ultimately to be collected at ambient samplers, is much lower than that emitted. There are several potential physical processes that could remove or alter the fine particulate before it can reach the transport layer. Some candidate mechanisms, including impaction on nearby vegetation and structures, and aggregation into B-3 ------- larger particles, are being evaluated. Once these investigations are completed, additional information will be made available to assist in the development of fugitive dust emissions of PM-2.5. In the meantime, estimates based on current factors and estimation methods that have been made available at the national-level are suspect. Agricultural Sources Fugitive dust sources from agricultural activities are similar in nature to the sources discussed above. Additional work will be required to adequately address these sources. It is not thought that fugitive sources of PM-2.5 related to agricultural tilling will be significant in terms of future planning activities. Other agricultural activities do contribute significantly to the burden of NH3 in many areas. Both animal husbandry and the application of nitrogen-based fertilizers can result in NFL; emissions. These sources are not well understood and continued research is required to develop more reliable estimates of emissions of NH3. The results of a significant amount of research on NH3 sources from agricultural activities completed in Europe are available. These studies represent most of the basic understanding used in estimates prepared at the national-level. These studies are well conceived and conducted and the information is reliable and applicable. Further understanding of the specifics of animal management (conditions in feedlots, dairies, etc.) and waste management activities (lagoons, land application, confinement and treatment, etc.) is necessary to ensure that these studies are representative of actual conditions in the U.S. Other Sources Emissions from noncombustion industrial sources can potentially be significant in some areas. Dust sources from minerals mining and processing, emissions from wood products industries, and bulk storage of materials can all contribute to PM-2.5 loadings. It is possible that there are other specific types of sources that might contribute to PM-2.5 emissions. These sources, however, are not well understood and are likely to be confined to specific locations. States will be encouraged to explore these types of sources and develop methods and tools to estimate emissions. TABLE B-l. FACTORS AND ACTIVITY DATA FOR PM-2.5 PLANNING Source Contribution to Ambient PM-2.51 Emission Factors Activity Available Information Remaining Needs Available Information Sulfur Precursors to Secondary Aerosol Formation Point Sources High quality emission factors are available for nearly all major combustion sources. CEM data available for many utility sources. Emission factors for smelters are reliable. Data source AP- 42. National-level analyses for industrial combustion sources based on average industry factors. Updates to reflect changes in technology could improve inventories. Fuel use and sulfur content is routinely and reliably monitored for utilities and large industrial boilers. ------- Source Contribution to Ambient PM-2.51 Emission Factors Activity Data Available Information Remaining Needs Available Information Remaining Needs Mobile Sources Emission factors for sulfur and PM from mobile sources is from PART5, and is linked to VMT estimates for fleet average. Off-road estimates based on national analyses. Improved methods for allocation among heavy diesel on-road and off-road categories are needed. Relative contributions are small. On-road estimates related to VMT. Off-road estimates related to national analyses of heavy diesel equipment, and other non-transportation activities. Emissions estimates could benefit from area specific surveys of construction, and other heavy diesel equipment. Area Sources Point source factors applied to area source emissions for wide spread small combustion sources (e.g., small diesel generators, etc.) Data source AP-42. Improved emission factors for open burning sources with specific studies. Relative contributions from industrial sources are small. Fuel use and allocation are primarily dependent on growth and earlier inventory assumptions in National analyses. Emissions at local levels based on specific activity data will be much improved over national methods. Data can be readily obtained. Nitrogen Precursors to Secondary Aerosol Formation Mobile Sources Emission factors for on-road sources are from MOBILE models developed by OMS. Off-road sources are based on similar on-road engines. Emission factors for off-road vehicles might be improved through specific testing. VMT collected as part of normal ozone inventory development. National-level estimates are available. Significant guidance exists. State data needs to be filled in for all counties that have not been subject to specific planning requirements in the past. Point Sources Emission factors available for most large combustion sources, CEM data available from utilities. Data Source AP-42. Updates to reflect changes in technology could improve inventories for industrial sources. Fuel use and firing type is reliable for most major combustion sources. Specific dat for fuel use and spatial/temporal allocation for industrial boilers could improve emissions estimates. TABLE B-l. FACTORS AND ACTIVITY DATA FOR PM-2.5 PLANNING (continued) Source Contribution to Ambient PM-2.5 Emission Factors Activity Data Available Information Remaining Needs Available Information Remaining Needs Nitrogen Precursors to Secondary Aerosol Formation (continued) ------- Source Contribution to Ambient PM-2.5 Emission Factors Activity Data Available Information Remaining Needs Available Information Remaining Needs Area Sources Emission factors for industrial combustion sources are reliable. Data based on AP-42 point source factors. Factors under development for many open burning sources. Factor for small repeated units could be improved with better understanding of operating conditions. Need better understanding of wild fires and prescribed fires. Fuel use data is routinely and reliably monitored for industrial boilers. Fuel loading for open fires needs local development and updates. Fuel loading factors need to be factored into smoke management plans. Organic Precursors to Secondary Aerosol Formation Mobile Sources Emission factors from MOBILE models by OMS and Off-Road model for non road vehicles. Factors are reliable for ozone precursor type species. Improved factors and/or speciation profiles to represent higher carbon number compounds and semi-volatile compounds. Activity from VMT estimates and estimates of hours in use for off road sources. National defaults are available. Improvements can be achieved with locally derived data particularly for off-road sources. Could require surveys. Area Sources For many sources of solvent use assumption of 100% air emissions can be made. Factors are dependent more on capture and control efficiency. F actors for organic emissions including condensable from combustion and open burning (low temp, combustion) are needed. Activity data from typical industrial sources is readily available as applied in other inventory efforts. Guidance is readily available. Methods are needed to develop activity data for open burning and some other unique area sources. Point Sources Typically, point sources contribute a minimal mass of VOC emissions. Emission factors are available for most point sources in AP-42. Speciation factors need to be reviewed to determine if low or semi volatile VOC compounds are adequate. Activity data to support point sources emissions are well developed and guidance to develop these data is readily available. Needs related to development of activity data for point sources generally is a low priority issue relative to PM-2.5 emissions development. B-6 ------- TABLE B-l. FACTORS AND ACTIVITY DATA FOR PM-2.5 PLANNING (continued) Source Contribution to Ambient PM-2.5 Emission Factors Activity Data Available Information Remaining Needs Available Information Remaining Needs Ammonia Precursors to Secondary Aerosol Formation Area Sources Emission factors for agricultural area sources have been developed in European studies. Limited data are available for other types of activity. Studies to determine the conditions and applicability of European results and further development of factors for other industrial activities are needed. Default national data on animals in production and acres treated with specific fertilizers is available. Specific analyses of actual activity data for local areas should be developed and compared to national default estimates to promote specificity and accuracy. Mobile Sources Emission factors for mobile sources have been developed by Volkswagen and recently some additional data are available through OMS. In general, emission factor development is necessary for most if not all mobile source activity. This is a weakness of current inventory development methods. It is expected that factors will be developed to be consistent with existing methods for estimating VMT and hours-in-operation. Additional needs for activity data for mobile sources is not expected to be a serious problem. Point Sources There are limited data available on NH3 emission factors from point sources. No estimates have been included in the national draft inventory. Emission factors need to be developed for all relevant point sources. Relative contributions are small. Activity data for most point sources will be readily available. Activity data development for point sources is not expected to be a problem. Emissions of Catalysts All Sources and Source Categories In general, emission factors for specific metal and alkaline catalysts are limited to data from speciation profiles and some HAP emissions development guidance from Locating and Estimating Documents. EPA does not recommend the application of data from speciation profiles as emission factors. Additional work is needed to characterize important sources and to specify adequate emission factors. In general, it is expected that most estimation methods for catalysts could be related to activity data that is developed for other more direct emissions sources. In some applications for some particular sources additional activity data may be needed. These needs are not well understood at this time. This need represents a remaining unknown. 1 Major source category classes are arranged in order of importance relative to expected emissions magnitude. ------- |