Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 Current and Possible Future Uses of MOBILE in Air-Quality Management AN EMISSIONS-FACTOR MODEL is fundamental for assessing the nature and magnitude of on-road motor vehicle emissions and their impacts on ambi- ent air quality. In the United States, excluding California, the MOBILE model has been the only model used in policy and regulatory settings to simulate actual emissions from automobiles over widely varying scales of resolution. (California uses the Motor Vehicle Emissions Inventory model- ing suite for the assessment of vehicle emissions and their controls, as discussed in Chapter 5.) MOBILE is used in the development of national, regional, and urban emissions inventories; the simulation of regional air chemistry and microscale dispersion of pollutants; the assessment of the effectiveness of control strategies; the documentation of emissions reduc- tions in State Implementation Plans (SIPs); the assessment of air-quality impacts of transportation projects, including the demonstration of confor- mity of transportation and air-quality plans; and the assessment of air- quality impacts of transportation-control measures and projects. Traditionally, the management activities of air-quality regulatory agen- cies at the local, state, and federal level have used MOBILE to estimate vehicle emissions. Increasingly, transportation agencies, principally state departments of transportation and local metropolitan planning organiza- tions ~IPOs), have become more reliant on MOBILE in fulfilling their new obligations under the Clean Air Act Amendments of 1990 (CAAA90) and the Intermodal Surface Transportation Efficiency Act of 1991 (ISTEA). 33
34 MODE[/NG MOB/LE-SOURCE EM/SS/ONS These acts expand requirements for state transportation departments and MPOs to assess the air-quality effects of transportation plans and projects. Also, the automotive and oil industries, consultants, and academic organi- zations use MOBILE in a variety of ways related to air-quality regulation and, more broadly, to develop a better understanding of the dynamics of atmospheric pollutants. FUTURE MOBILE-SOURCE EMISSIONS-MODELING ISSUES Originally, MOBILE was developed to estimate overall emissions levels, trends over time, and the effectiveness of mobile-source emissions-control strategies. The model has undergone significant evolution since then, which is summarized in Chapter 3. Current uses of the model include de- veloping emissions inventories and reductions in SIPs, demonstrating con- formity of transportation and air-quality plans, and providing emissions estimates for dispersion and photochemical air-quality modeling. Thus, the original role of MOBILE has been expanded in ways that now require higher standards of accuracy that incorporate a greater degree of complex- ity. A good example of this evolution is demonstrated by MOBILE's cur- rent use in ozone attainment modeling, which requires precise spatial and temporal estimation of speciated precursor emissions to predict ambient ozone levels for a particular day and region. This application is more diffi- cult and demands greater precision than MOBILEl's use for modeling of exhaust emissions as a function of age or mileage. The need for vehicle emissions models will continue in the future and the demand for more accuracy and versatility will increase. SIP require- ments to meet proposed new fine particulate matter (PM-2.5) and 8-fur ozone standards, regional visibility rules, and increased interest in air toxics from mobile sources will magnify concerns about MOBILE's applica- bility and accuracy. Although MOBILES, the upcoming version of MO- BILE, will address some of these concerns, it will likely fall short of the regulatory burden placed on its use. For instance, MOBlLE6 assumes lower rates of deterioration of vehicle emissions-control systems than earlier versions because of indications that newer technology is more durable. Yet, as discussed in Chapter 3, the ex- tent to which the data supporting this position is fully representative is questionable. A desire for more accurate microscale modeling of localized transportation-control measures and more accurate photochemical model- ing will add to the growing need for instantaneous or modal emissions modeling and better spatial disaggregation of emissions. Disaggregation of certain model components, such as the separation of start emissions from running emissions or the development of facility-specific speed cor-
CURRENT AND POSSIBLE FUTURE USES OF M OBILE 35 rection factors, will provide better resolution, but the activity data re- quired to implement these changes may not be available in all regions. The demand for a more accurate MOBILE model also raises concurrent demands for more-accurate transportation-activity factors such as average vehicle speeds, congestion levels, roadway classification, or miles traveled. Finally, it is conceivable that future users could require a real-time version of a mobile emission model to manage traffic flow in order to minimize emissions in critical air-quality areas. No single model is available to ac- curately address all of these possible uses. MODELING AIR QUALITY: AN INTERDISCIPLINARY ENDEAVOR Efforts to evaluate the air-quality impact of on-road motor vehicles are inherently interdisciplinary, and require the interaction of three different models and related areas of expertise: travel-demand models, emissions models, and air-quality models. Travel-demand models determine the amount of transportation activity occurring in a region based on an under- standing of the daily activities of individuals and employers as well as the resources and transportation infrastructure available to households and individuals when making their activity and travel decisions (Harvey and De akin 1993~. This includes measures such as number of trips, time of day, length of trip, mode of transportation, route or location of trips, aver- age speed of travel, and age of vehicle. The number of transit trips, auto- mobile occupancy, and vehicle miles of travel (VMT) are common perfor- mance measures used to measure transportation activity. The second component corresponds to mobile-source emissions rates. MOBILE estimates emissions rates based on vehicle type, average speed, ambient temperature, and other factors. The product of the transportation activity and the emissions rates from MOBILE results in emissions esti- mates for each modeled pollutant (carbon monoxide (CO), volatile organic compounds (VOCs), and nitrogen oxides (NOW. It is critical that esti- mates of transportation activity and emissions rates be in balance with respect to fidelity, accuracy, and precision to ensure the reasonableness of the emissions estimates. It is impossible to have the same for both (in some cases, the vehicle activity estimates will be more precise, or more accurate, or more refined, and in others the emissions rates will be so). However, transportation and air quality planners should understand the fidelity, accuracy, and precision for each component and take these into account in policy analysis (such as through uncertainty analysis and devel- opment of confidence bounds). The third component of the modeling trilogy is the regional and micro- scale modeling of air quality. These models translate emissions invento-
36 MODE[JNG MOB!LE-SOURCE EMISSIONS ries into ambient pollutant concentrations that vary through space and time. Translating emissions to ambient concentrations can be done di- rectly, for example, by using microscale carbon monoxide modeling. This method estimates concentrations in "hot spot" areas (critical intersections and sites with violations or possible violations of the NAAQS) by simulat- ing the dispersion of the pollutant, using a variety of dispersion parame- ters, such as wind speed and direction. This is also done through urban- scale and regional-scale air-quality models that calculate ozone concentra- tions by simulating both atmospheric chemistry and meteorology. Again, attention to fidelity, accuracy, and precision is needed in each of these three types of models to ensure balance in their integration. Travel-Demand Modeling There are five traditional components of the sequential travel-demand model, namely demographic forecasting and the four-step travel-demand modeling process (Figure 2-1~. Demographic data The location of households and employment (categorized as basic, retail, and service) in small traffic survey zones with- in the urban region. This includes the forecasts of regional economic growth, land use patterns, and future demographic trends. . Trip generation The estimation of the number of trips by zone by time of day and type (both trips originating in a zone, termed trip produc- tion, and trips terminating in a zone, termed trip attraction). Trip distribution The pairing of trip productions with trip attrac- tions resulting in a full spatial pattern of travel by purpose and time of day. . Mode choice The determination of mode of travel, specifically walk, bicycle' drive alone, high occupancy vehicle (HOV), bus, rail, or truck travel. Route assignment or choice Trips are assigned to paths in the transportation infrastructure by minimizing travel times or travel times and costs, and incorporating average speed and other impedance feed- backs. The travel-demand models answer: "Will I travel," "How often and when," "Where," "By which mode," and "By which route"? They provide the MOBILE model with information on average vehicle speeds for each roadway segment that may be aggregated by roadway type or facility (e.g., freeways, arterials, collectors, and freeway ramps). The following points describe some important issues that relate to the use of travel-demand modeling in modeling air quality.
CURRENT AND POSSIBLE FUTURE USES OF MOBILE 37 ,| DEMOGRAPHIC |4 l DATA l TRIP GENERATION BY TIME-OF-DAY ROA DWAY ~l TRIP NETWORK 71 DISTRIBUTION TRANSIT NETWORK r HOV NETWORK MODE CHOICE 1 ~ ~ 1' ROA DW, KY ~ HOV T - NSIT ASSIGNMENT ~1 ASSIGNMENT ASSIGNMENT 1 FIGURE 2-1 Sequential travel-demand forecasting process used in Dallas, Texas. Source: NCTCOG 1999. Seasonal variations Most travel models are calibrated for typical weekdays when primary and secondary schools are in session. Therefore, travel is modeled for a typical weekday in February through May and Sep- tember through November. However, air-quaTity assessments typically do not follow such convenient transportation schedules, because ozone is a frequent summer problem and carbon monoxide is often a winter one. Therefore, the travel-activity information may be modified to mode! the emissions type and season of interest. Adjustments for weekend/weekday Travel models typically sim- ulate weekday traffic, with adjustments for weekend emissions inventories often being required for air-quality assessments. Travel survey data is being collected to assist estimating weekend travel. Duration within dayMost travel forecasts are for a typical 24-hr period. These can be adjusted to simulate peak-hour and peak-period con- ditions with time-of-day factors. Air-quality models typically-need emis- sions for each hour of the day. This task is often performed using travel
38 MODELING MOB`LE-SOURCE EMISSIONS start times by hour and trip type from travel diaries. However, these ad- justments are highly uncertain and rarely validated. Travel by grid Most travel forecasts are conducted for specific transportation facilities or segments. However, air-quality models often need information by "grid," or aggregations of transportation facilities. Geographical information system software is an efficient tool for aggregat- ing travel into grids. Multidimensional Synergistic Impacts from Adjustments to Travel Activity Results As needs for precise estimates of emissions and air quality grow, it is common to use travel activity inputs based on typical weekday travel that may have been adjusted for season, weekend travel, time of day, and grid. Demands for even greater precision might require travel activity estimates for specific vehicle age categories, meteorological conditions, or vehicle types. The committee feels that the level of detail associated with current travel-demand models is insufficient to make these simultaneous adjust- ments without introducing substantial additional uncertainty. A good ex- ample of this problem is the need in air quality modeling to estimate ag- gregate heavy-duty vehicle (HDV) activity and adjust these estimates for time of week and time of day. Because HDVs produce a disproportionate amount of NOx emissions, they greatly impact the ability to model ozone accurately. Yet the multiple adjustments of travel activity results needed to produce estimates of HDV activity by time of day introduces an un- known level of uncertainty to emissions and air quality simulations. Emissions Modeling A primary use of MOBILE is for developing on-road mobile-source emis- sions inventories for use in air-quality planning. Emissions rates devel- oped in MOBILE are combined with average vehicle speeds and travel ac- tivity estimates to develop these inventories. The emissions rates gener- ated by MOBILE require a multitude of input assumptions. For most in- put assumptions, MOBILE provides national default values or users can input TocaDy specific values. Regions are free to choose when to use na- tional defaults or local data. This decision is sometimes made on the basis of whether the national default or local data wiD positively affect their at- tainment demonstration or conformity analysis. MOBILE is particularly sensitive to input assumptions for vehicle age, VMT by vehicle class, aver- age vehicle speeds, and temperature all of which can vary widely from
CURRENT AND POSSIBLE FUTURE USES OF MOBILE 39 region to region. Below is a brief discussion of parameters that are impor- tant in the application of MOBILE to an individual region. Chapter 3 of the report discusses the technical components of MOBILE in more detail. Vehicle Registration MOBILE uses vehicle registration data to determine the percentage of the vehicle fleet for each combination of vehicle type with vehicle age. MOBILE combines this information with average mileage accumulation rates to determine the fraction of overall travel in a region associated with each vehicle type disaggregated by age and average fleet emissions rates. The MOBILE documentation published by the U.S. Environmental Protec- tion Agency (EPA) strongly encourages users of MOBILE to develop locally specific vehicle registration distributions because the default values reflect national averages for 1990. Emissions inventory estimates are affected by assumptions about vehi- cle registration distributions. Within an urban area, the vehicle fleet com- position can vary significantly across subregions, for example, in relation to development patterns and the economic status of the population. Areas with newer development and higher average income levels tend to have newer vehicle fleets, resulting in lower emissions rates than in older areas. The choice of a particular vehicle registration distribution can affect on- road emissions inventories by approximately 5 to 10% (Pollack et al. 1991~. As a result, estimates of on-road mobile-source emissions require accurate vehicle registration distributions at an appropriate level of detail for a par- ticular application. Vehicle Miles of VMT Travel Mix A VMT mix identifies the percentage of VMT that is accumulated by each of the eight vehicle classifications used by the MOBILE model. MO- BILE uses this VMT mix to generate composite vehicle-emissions factors. MOBILE calculates a typical urban area VMT mix based on national data for several variables, including registration distributions, annual mileage accumulation rates, percentage of diesel sales, and number of vehicles. EPA recommends that users develop locally specific estimates of VMT mix for SIP emissions inventories. Policy decisions regarding mobile-source controls are affected by assumptions incorporated into the VMT mix data. National default values based on averages might not accurately represent the VMT mix in any given region. For example, if a particular region has a high percentage of
40 M ODE[/NG M OB/1E-SOURCE EMISSIONS heavy-duty vehicles, the default VMT mix would lead to an underestima- tion of NOx emissions. Average Speed MOBILE uses regional average vehicle speeds estimated by travel de- mand models to develop emissions rates. The model develops base emis- sions rates for various vehicle classes using standard driving cycles such as the Federal Test Procedure (FTP). These base emissions rates are then adjusted to a particular location's average speed using speed correction factors. Speed correction factors are intended to reflect the differences between emissions rates under test conditions and emissions rates under regional driving conditions. More detailed descriptions of the use of these test cycles, speed correction factors, and facility correction factors are con- tained in Chapter 3. It is important to point out that traditional travel-demand models do not estimate average speeds directly, but rather produce average speeds from estimates of traffic volumes. This is important because these speed/ volume relationships are not very accurate and are sometimes adjusted during calibration so that modeled traffic volumes match observed vol- umes. Temperature MOBILE requires locally specific temperature data, in part because no national defaults would be appropriate for temperature. Thus, users must develop average temperature data to develop on-road mobile-source emis- sions inventories. Analysis of the MOBILE model shows that at higher temperatures, a one-degree change in temperature results in a 1% change in emissions factors (Pollack et al 1991), though this premise has not been validated. As a result, temperature variations within a region may signifi- cantly impact emissions inventory estimates. Air-Quality Modeling Air-quality models have become the central tool for analyzing how fu- ture emissions changes, including changes due to new control strategies, will affect air-quality (NRC 1991~. Ozone, for example, is not produced directly from emissions sources, but rather through complex chemical re- actions involving VOCs, NOx, and sunlight (solar radiation). High ozone episodes occur during periods of stable atmospheric conditions that are
CURRENT AND POSSIBLE FUTURE USES OF M OBILE 4 ~ accompanied by high temperature and low winds. Regional photochemical models attempt to predict the formation of ozone for multi-day events us- ing meteorological data, emissions inventory data, and complex air-chem- istry equations. In contrast, CO problems are much more localized in na- ture and require different air-quality models that represents site-specific dispersion of this pollutant. Vehicles emit CO directly, and its local con- centration depends upon the rate that it is emitted and dispersed in the atmosphere. A major problem for air-quality managers is identifying con- trols that will reduce CO, ozone, and PM. On-road mobile-source emis- sions are important contributors to each of these pollutants. For most of the nation, the estimation of on-road emissions data needed for air-quality modeling studies are derived from MOBILE using results from travel-de- mand modeling. In a typical application (e.g., analyzing ozone-controT strategies), an air- quality model will be applied to simulate photochemical pollutant concen- trations during a 3- to 5-day episode, using emissions and meteorology data specific to the period of application. Typical model resolution, and hence the scale of emissions inputs, is approximately 5 kilometers (km). The modeling results are evaluated against observed ambient measure- ments to assess the validity for use in control-strategy assessment. Errors in the emissions and other model inputs are evident from disagreements between observations and model simulations, not only for ozone, but for the precursors as well. Because of nonlinearities in the formation of ozone, it is important that the observations and the simulated values agree rea- sonably weU. Otherwise, the model response to emissions changes will be suspect. With this in mind, EPA has developed model performance guide- lines (EPA l999b). At present, it appears that uncertainties in the emissions, and mobile- source emissions in particular (NRC 1991; Harley et al. 1993a, b), are ma- jor contributors to poor model performance. As described in more detail in Chapter 4, a variety of studies have concluded that mobile-source VOC emissions are significantly underestimated in the models. The perform- ance of air-quality models improves significantly when estimates of mobile-source VOC emissions are increased. However, there are many other factors that contribute to the poor performance of air quality models, including errors in the atmospheric chemical mechanisms and meteorolog- ical inputs within the air quality models as well as uncertainties in the travel activity inputs discussed earlier. After a model achieves acceptable performance, it is used to test control strategies, in particular to identify the set of controls that are likely to lead to attainment. Historically, at- tainment has meant that predicted ozone is less than 0.12 parts per mil- lion (ppm) in each subarea of the region. Thus, the target is an absolute number, and modeling results are used in an absolute sense. The use of MOBILE to develop emissions inventories for ozone modeling
42 M ODEL/NG M OBlLE-SOURCE EMISSIONS (and PM modeling in the future) highlights the evolution of the demands placed upon it. Such applications require estimates at relatively fine spa- tial and temporal resolutions, not simply across a fleet of vehicles in a broad urban area. Topographical features, such as hills, might play a ma- jor role in emissions and atmospheric processes. Further, the varying com- position of the fleet can become important if one area of a city is more like- ly to contain high-emissions vehicles than another. MOBILE was not orig- inally designed to support applications that require a high resolution and accuracy of emissions inventories. It is not apparent to the committee that an emissions-modeling tool designed specifically for use in air-quality mod- eling would have been designed in the same way as MOBILE. The importance of providing finer spatial resolution is highlighted by a recent study ~ackshminarayanan, 1999~. That study used a mobile source inventory for the Atlanta, GA, area, as developed by MOBILE5a, to simulate ozone, nitrogen dioxide and an air toxic (formaldehyde) concen- trations in the region. Next, results from MEASURE (Guensler et al., 1998; see Chapter 5 for details) were used to spatially and temporally real- locate those emissions, thus keeping the same basin-wide mass emissions of each species, but changing the details of the time and location of emis- sions. In this process, the study was also able to develop the emissions at finer grid resolutions than the 4 x 4 km MOBILE inventory. The photo- chemical model was then re-applied using grid resolutions of 1 x 1 km, 2 x 2 km and 4 x 4 km. Peak levels of nitrogen dioxide and formaldehyde were found to be very sensitive to grid size, varying by up to a factor of five or more. Ozone levels were less sensitive. Thus, this study concluded that accurate exposure assessment of primary pollutants, such as air taxies and particulate matter, are likely to require spatially and temporally detailed . . . ~ . emissions Information. In part because of the difficulties in estimating emissions inventories, guidelines for future ozone air-quality modeling might be used in a more relative sense. Thus, rather than ensuring that all concentrations simu- lated by the model are at or below 0.12 ppm (or 0.08 ppm for the new 8-fur standard), a relative reduction from a base scenario could be used to as- sess the adequacy of emissions controls (EPA l999b). For example, if the base calculation led to a maximum ozone concentration of 0.156 ppm, and the control case had a peak of 0.132 ppm, this would represent a reduction of 15%. That 15% relative reduction would then be used to test for attain- ment. If the design value was 15% over the limit, then the modeling test would be passed.1 The use of the model in such a relative sense is believed to be more accommodating of uncertainties, such as those in the emissions 1For a more detailed explanation of this type of attainment demonstration, see EPA (1999b).
CURRENT AND POSSIBLE FUTURE USES OF MOBILE 43 themselves. Thus, as MOBILE has been evolving to provide absolute emis- sions levels (as opposed to relative levels), the regulatory application of air-quality models may be moving towards relative uses, in part, because of problems in validating these models. Ozone modeling is not the only purpose for a model such as MOBILE. Air toxic agents are a societal concern, and automotive emissions contain significant quantities of hazardous air pollutants (HAPs) or air tonics, such as benzene, formaldehyde, and 1,3 butadiene. Unlike ozone, elevated levels from these primary emissions are found near roadways. Concentra- tions in nearby areas might be much reduced by dispersion. A recent Cali- fornia Air Resources Board (CARB) study concluded that concentrations of pollutants inside vehicles can be two-to-seven times greater than concen- trations at air-monitoring stations (CARB 1998~. This has important im- plications in exposure assessments, because it places greater emphasis on knowing the detailed spatial location of emissions in relationship to poten- tially exposed populations. Unlike ozone modeling, which might require emissions with a spatial resolution of 4 km or so, HAP exposure assess- ment might require resolution at a scale of tens or hundreds of meters. MOBTOX (or a version of MOBILE that estimates air toxic agents), in principle, could be applied at this level, but it lacks many features that might be important at such fine spatial scales (e.g., the influence of topo- graphical features and specific traffic-control measures) that get averaged out over larger areas. The use of MOBILE to model the concentrations of HAP s near roadways raises issues similar to those of using MOBILE to model micro-area carbon monoxide. Users of Modeling Components Both the public and private sectors use the transportation, emissions- factor, and air-quality modeling components of air-quaTity planning and regulatory processes. The broad community of model users includes gov- ernment agencies, private consulting firms, public interest groups, and other researchers. The primary purpose for this modeling effort is to fulfill specific transportation and environmental legislative and regulatory re- quirements. Governmental users include agencies at the federal, state, regional, and local levels that use these models to conduct transportation and environ- mental analysis. Private consulting firms often contract with government agencies and industry to conduct specific planning and environmental studies using these modeling tools. Public interest groups, such as envi- ronmental organizations, are often stakeholders in transportation and en- vironmental planning studies that rely on these technical-modeling tools. Universities train future users of these models, and they perform research
44 MODElING MOBI[E-SOURCE EMISSIONS and data-collection activities that support the continued development of these tools. Level of Analysis and Model Uses and Users There are four distinct classes of uses of this compendium of models in air-quality management, as described below and summarized in Figures 2- 2 through 2-5: Level 1 Direct use of the MOBILE model to estimate emissions rates. This is often used by EPA to assess national mobile-source regula- tory strategies (Figure 2-21. Level 2 Use of transportation and MOBILE models to estimate emissions inventories. Such analyses are used by state environmental agencies, transportation consultants, metropolitan planning organizations, and state departments of transportation to support development of SIPs and transportation conformity analysis (Figure 2-31. Such analyses are also used by EPA to estimate the impacts of national mobile-source regula- tory strategies on overall emissions levels. Level 3 Use of transportation, MOBILE, and air-quality models to simulate pollutant concentrations. This is usually done by state environ- mental and transportation agencies and universities to assess attainment of National Ambient Air Quality Standards (NAAQS) (Figure 2-41. Level 4 Use of transportation, MOBILE, air-quality, and exposure models to simulate human exposure and health impacts from air pollut- ants. This is done by health professionals at universities and environmen- tal agencies to help assess mortality and morbidity associated with air pol- lution (Figure 2-51. LEVEL 1: EMISSIONS RATES MOBILE Model Emissions Rates by Speed and Vehicle Type Uses: NationalRegulato~Strategies FIGURE 2-2 Use of MOBILE for esti- mating ~rehicle-emissions factors.
CURRENT AND POSSIBLE FUTURE USES OF MOBILE 45 LEVEL 2: EMISSIONS ESTIMATES Travel -Demand MOBILE Estimation Model Model X Vehicle Miles of Emissions Travel and Rates Average Speeds by Speed and Vehicle Type On-Road Mobile Emissions Estimates L' Volatile Organic Compounds, Carbon Monoxide, Nitrogen Oxides, and Primary Particulate Matter _ National Regulatory Strategies SIP Control Strategies/Rate of Progress Transportation Conformity Transportation Control Measure Effectiveness National Environmental Policy AcVEvaluation of Major Capital Investments FIGURE 2-3 Use of MOBILE for estimating vehicle emissions. Fidelity, Accuracy, and Precision of Each Component Fidelity, accuracy, and precision all relate to the ability of a model to simulate the real world. Table 2-1 provides a brief summary of how each of these modeling fields approaches calibration and validation. It empha- sizes that tra~rel-demand models have standard calibration and validation statistical performance metrics. It is common in travel-demand modeling to compare predicted traffic volumes on a roadway segment or ridership on a transit line with independently observed values. Such standard caTibra- tion and validation procedures and metrics have not been developed for applications of MOBILE. It should be noted, however, that regions rarely validate travel-demand models to ensure that the underlying behavioral assumptions are accurate and that a model predicts the right volumes for the right reasons over time and across the region. Table 2-1 also shows that, although air-quality models are critical for policy implementation, they have greater tolerances for gross error. Be- sides errors inherent to air-quality modeling itself, this is also due to the propagation of errors that occurs through the linkage of models. Errors that occur in the travel-demand estimates propagate and add to the errors
46 MODELING MOBI[E-SOURCE EMISSIONS LEVEL 3: AIR QUALITY ESTIMATES Travel-Demand Estimation Model Vehicle Miles of Travel and Average Speeds On-Road Mobile Emissions Estimates MOBILE Model X EmissionsRates- by Speed and Vehicle Type . . Speciation Model I I Biogenics L | Non-Road Mobile |_ | Area Source 1_ | Stationary Source Meteorology Input Data Dispersion Model Air-Quality Model Emissions by Grid and Time of Day = Volatile Organic Compounds, Carbon Monoxide, Nitrogen Oxides, and Primary Particulate Matter Carbon Monoxide Concentrations Ozone and Particulate Concentration Levels by Time and Location Uses: SIP Demonstration of Attainment National Environmental Policy AcVEvaluation of Major Capital Investments CO, 03, and Particulate-Concentration Evaluation FIGURE 2-4 Use of MOBILE in the estimation of ambient pollutant con- centrations. in the mobile-source emissions model, which in turn, propagate and add to the errors in the air-quality models. USES OF MOBILE IN POLICY DECISION-MAKING The following section summarizes the uses of MOBILE, including a dis- cussion of who is responsible for executing MOBILE and assessing the model's results, modeling issues particular to the application, and future directions EPA should consider. Although the six application areas vary widely in terms of their spatial and temporal scales, the same MOBILE model is used in all settings. These descriptions further illustrate that MOBILE is currently applied in ways for which it was not developed and for uses beyond which it is well suited. EPA has been urged to develop other models more suited to specific purposes. The next version of MOBILE, MOBILES, is intended to narrow this gap, yet the discussion below suggests that the MOBILE structure itself cannot adequately serve such a wide range of uses.
CURRENT AND POSSIBLE FUTURE USES OF M OBILE 47 LEVEL 4: HEALTH IMPACTS ESTIMATES On-Road Mobile Emissions Estimates Volatile Organic Compounds, Carbon Monoxide, Nitrogen Oxides, and Primary Particulate Matter . Biogenics Non-Road Mobile Area Source ~ Stationery Source ~ . _ Meteorology Input Data _ Dispersion Model Air-Quality Model Emissions by Grid and Time of Day Exposure/Response Model Distribution of Sensitivity to Exposure | Uses: Assessment of health impacts | L Carbon Monoxide Concentrations Ozone and Particulate- Concentration Levels by Time and Location ~ ' Exposure Model Population Characteristics by Exposure and Time Health Impacts . . FIGURE 2-5 Use of MOBILE in assessing health impacts. National anct Regional Regulatory Strategies Primary Users and Purpose EPA uses MOBILE to evaluate a variety of national air-quality regula- tory strategies, including new-vehicle emissions standards, fuel-quality specifications, and inspection and maintenance (I/M). EPA also uses MO- BILE for evaluating the contribution of on-road vehicles to the nation's total air-pollution emissions and to inventory and monitor historical trends (EPA 1998a; EPA 1998b). EPA's most significant national vehicle emissions-control regulatory effort may relate to concurrent adoption of Tier 2 new-vehicle emissions standards and new limitations on sulfur in gasoline. Because of the signif- icance of this action, EPA has undertaken a major effort to develop a ver- sion of the mode] that is as close as possible to the expected new MOBlLE6 to assess this major regulatory development (EPA l999c). The version of the model is known as the Tier 2 model, which is a spreadsheet model in- corporating elements of MOBILE5b and MOBILE6. The Tier 2 motor-vehicle emissions standards (EPA 1999a) involve re- ducing new-vehicle emissions standards beginning in 2004. By 2007 pas-
48 MODE1/NG MOBI[E-SOURCE EMISSIONS TABLE2-1 Comparison of Calibration and Validation Standards for Travel-Demand and Air Quality Models with the MOBILE Model Model Structure Evaluation Typical calibration method Travel-Demand Estimation Model MOBILE Air-Quality Model Household travel behavior survey with sample size specification (90% confidence, 5% error Laboratory estimates of emission rates (under- represented sample size, no explicit performance standard) Validity to Field test of passenger Field test of Field test to second date volume estimates emissions levels ozone (e.g., tunnel monitor-selected tests, no explicit episodes (30% performance standard) Laboratory estimates of precursor sensitivity to ozone formation source (traffic counts and transit ridership) by segment (88% correlation, 5% error) gross error +15% bias) - Sources: Pederson amd Samdahl 1982; EPA l999b. senger car and light-duty truck emissions standards will be lowered to an average of .07 grams per mile (g/mi) for NOX. The final rule for the Tier 2/Sulfur Gasoline Program also requires that gasoline produced by refiners or sold by importers meet an average sulfur content of less than or equal to 30 ppm by 2007. Figure 2-6 shows the impact of these new require- ments generated by the Tier 2 version of MOBILE (EPA l999d). In this figure, passenger cars are labeled LDV and trucks are labeled LDT1/2 (light pickups, minivans, and most sport utility vehicles) or LDT3/4 (heavier pickups and sports utility vehicles). This figure shows very large decreases in NOX and PM-2.5 emissions from the combination of Tier 2 and sulfur controls. Emissions of NOX and PM-2.5 in year 2020 are estimated to be cut by two-thirds from projected emissions without these controls. It should be noted, however, that these standards must withstand any legal challenges. Issues and Limitations MOBILE is best suited for national or regional applications because it utilizes an aggregate approach appropriate for wide areas and long time-
49 o o o ~ ~ ._1 ~ O . - ° a, a) Cat is: 5- g _ · ~ o 5_~ U) Cat-- ~ En a' ~ 3 c' CO ~ ~ ° &, Cat Ct ° ~ 5, ~ ~ ¢ .= Z ~ ~ ~ ~ ~ _ ^ ~ 5, 5- ~ o ° ,s ~ c G ° cot ~ ~ EM > a c~ 4- a' ED, - ~ ~ ~ ~ & to tD ~ ~ ~ ~ GO - ~ ~ ~ ~ ~ ~ ~ ~ of v a ~ ~ a ~ ~ >, g ~ ~ ~ 's ~ ~ ~
50 MODELING MOBILE-SOURCE EMISSIONS scales under average conditions. In regional applications, however, signifi- cant error can be introduced using MOBILE's generic national defaults. Policy Implications ant] Future Direction Output from MOBILE for use in national and regional air-quality strat- egies has profound technical and economic effects on the direction of the nation's air-quality management activities. Significant inaccuracies can result in misdirected control strategies that prolong public exposure to health hazards or waste large sums of money. It is not certain what major inaccuracies will exist in the new MOBILES, although clearly the major changes from MOBILE5b indicate that there have been such inaccuracies in the past. Evaluation of Control Strategies, Emissions Inventory, and Rate of Progress Primary Users ant! Purpose State and local governments, MPOs, consultants, research institutions, and others apply MOBILE to develop regional emissions inventories, eval- uate alternative mobile-source emissions-control strategies, and track trends in control-strategy implementation. In particular, MOBILE is used in ozone nonattainment to demonstrate how a region will comply with the Clean Air Act Amendment of 1990 requirement to reduce VOC emissions by 15% *om 1990 to 1996 and 3% annually until attainment. Issues and Limitations Some applications, particularly region-wide ones, warrant an aggregate approach to estimating vehicle emissions. In such applications, it may be adequate to develop representative emissions factors from region-wide pa- rameters, such as average vehicle speeds by facility class (see Chapter 3 for a description of facility correct factors) and registration distribution, that can be combined with VMT to estimate regional emissions. MOBILE is well-suited to these types of applications. However, there are some con- cerns about the inability to see all the underlying assumptions in the model, some of which might have to be altered to fit local observations. The committee feels that such aggregations still introduce inaccuracies in the end result because the assumed average may not truly represent the
CURRENT AND POSSIBLE FUTURE USES OF M OBILE 5 ~ real average in a skewed distribution. Some regional-control strategies cannot be modeled within the current MOBILE structure, including land use changes, vehicle scrappage, and clean-fueled fleet incentive programs. The accuracy of projections used in SIP development depend to a great extent on the accuracy of assumptions about such factors as fleet turnover, effectiveness of control strategies, and future deterioration rates of vehi- cles. Generally, there is concern that national default parameters coded in MOBILES are outdated, and that their use in SIP development is not ap- propriate. Additionally, overall concern is high that the vehicle-operating patterns in the model do not correspond to present and future on-road con- ditions. EPA is addressing some of these concerns through the addition of an "off cycle" factor that accounts for higher average speeds and air condi- tioning use. Improving these aspects of MOBILE will also require im- provement in vehicle-activity. Policy Implications and Future Directions MOBILE users have many questions about its accuracy. For example, MOBILE's ability to correctly evaluate the impacts of air quality improve- ment initiatives, such as vehicle emission inspection and maintenance pro- grams (Harrington et al. 1998) and the use of oxygenates in winter (NRC 1996; NSTC 1997) has been questioned. The development of SIPs requires accuracy in emissions inventories and crediting of emissions reductions from controls, both of which are particularly sensitive to errors. Little has been done to address this issue, and it will undoubtedly become more sig- nificant in the absence of a significant model revision based on better data and science. To date, MOBILE revisions have been infrequent, with the last major update (MOBILES) released in 1993. It is unclear at this point whether the upcoming version of the model, MOBILE6, will increase or decrease regional emissions predictions compared to MOBILES. It is like- ly that at least VOC emissions will increase. This uncertainty as well as delays in the release of MOBILE6 greatly complicate SIP development. A long-range plan is needed to determine the appropriate update fre- quency schedule and whether there should be updates issued more contin- uously. There is significant concern among the committee about the seven- year gap between MOBILES and MOBILE6. This delay results in the use of a version of MOBILE containing information known to be obsolete and incorrect. Users need updates that incorporate the latest findings on the factors that affect emissions and the effectiveness of control strategies so that SIPs can be based on the most accurate information. One possibility may be to allow users access to new information and allow them the flexi- bility of incorporating such information into SIPs and other planning pro-
52 MODELING MOB`LE-SOURCE EMISSIONS cesses. However, there are also problems that might com with more fre- quent or continuous updates, such as inconsistencies between models used in SIP budgets and subsequent conformity determinations. The need for more accurate emissions inventories and assessment of controls requires expanding the capabilities of MOBILE or developing new models. Users desire more disaggregation of model inputs so that local conditions can be better represented. There is a need to assess impacts of some alternative strategies that are not presently incorporated in MOBILE, such as alternative-fueled vehicles. Road grade is an important local factor that will not be incorporated in the current or updated versions of MOBILE. Users will also expect that some form of instantaneous mod- eling will be provided so that both individual projects and larger-area transportation systems can be assessed in a more accurate manner. SIP Demonstration of Attainment Primary Users and Purpose States and, in some cases, metropolitan planning organizations (MPOs) are required to develop a demonstration of attainment for SIPs in ozone and CO nonattainment regions. These must be submitted to and approved by EPA. Such an analysis demonstrates that the proposed emissions-re- duction strategies will attain and maintain ambient ozone and CO concen- trations below the NAAQS. For this, an urban or regional-scale air-quality model is used with on-road mobile-source emissions estimated from MO- BILE and data on other emissions sources (stationary, biogenic, area sources, and non-road mobile sources). Issues ant] Limitations Regional air-quality models are complex and generally require detailed temporal and spatial allocation of emissions. Applications of urban and regional air-quality models usually simulate a 3- to 10-day ozone episode. The model requires hourly "ridded emissions for NOx and VOCs, as well as speciation of VOCs. Additionally, emissions must be disaggregated by emissions mode (e.g., exhaust and evaporative), technology, and emitter group. As discussed in Chapter 4, there is strong evidence that MOBILE emissions, including emissions by mode, are inaccurate and thus inhibit accurate air-quality modeling. Inconsistencies may also result from other emissions sources, due to the lack of temporal and spatial detail needed to model regional air quality. For example, locomotive, marine, and construc- tion emissions might be difficult to accurately characterize because critical data are proprietary.
CURRENT AND POSSIBLE FUTURE USES OF MOBILE 53 Policy Implications and Future Directions One important output from MOBILE is emissions-rate detail for inclu- sion in SIP photochemical modeling. Ultimately, photochemical model ac- curacy depends on the accuracy of the emissions estimates, as well as the methods used to spatially and temporally allocate on-road emissions esti- mates, and to speciate the VOC emissions. Errors in these steps can po- tentially lead to inaccurate conclusions and selection of sub-optimal con- trol strategies. Because of difficulties in accurately determining overall emissions inventories, future guidelines for demonstrating attainment may be based on a relative rather than an absolute reduction in maximum ozone concentrations. This, however, will not eliminate the issues associ- ated with spatially and temporally allocating emissions within a region. In addition, a lower NAAQS standard for ozone will require better spatial and temporal disaggregation over wider regions. Transporlation Conformity and Evaluation of Transportation Impacts in a Nonaltainment Area Primary Users and Purpose The MPO is responsible for performing an air-quality conformity analy- sis for nonattainment areas. Conformity is a determination that emissions from transportation plans, programs, and projects in a nonattainment area do not exceed mobile source emissions budgets established in SIPs (Federal Highways Administration 1992~. The conformity demonstration is intended to show that transportation activities will not cause or contrib- ute to any new violation of air-quality standards, increase the frequency or severity of existing violations, or delay timely attainment of standards (EPA 1993a). The conformity analysis is done for the system of projects contained in a region's transportation improvement program (TIP) and transportation plan (see glossary for definitions of transportation improve- ment program and transportation plan). Regions in CO and PM-10 nonattainment must also conduct project-level conformity analysis ("hot spot" analysis) for critical intersections and sites with violations or possi- ble violations of the NAAQS. Similar to emissions inventories developed for a SIP, the transportation conformity analysis consists of determining emissions estimates as a func- tion of vehicle activity and region-specific emissions factors. However, this must be done for a specific system of projects or programs, and on a small- er scale than for the SIP. In some areas, where planning organizations have limited responsibilities, the state departments of transportation may
54 MODE1/NG MOBI[E-SOURCE EMISSIONS perform the technical analyses. The purpose of an air-quality conformity analysis is to ensure that transportation projects identified in a region's Tong-range transportation plan and short-range TIP are consistent with local air-quality goals and objectives. The conformity analysis is needed to ensure that calculated emissions levels for federally funded and regionally significant projects do not cause emissions to exceed the targets specified in the relevant SIP. Issues and Limitations A critical aspect of the conformity determination is that it is developed for a system of individual projects. Because MOBILE uses top-down pro- cedures to estimate emissions, it is not suited for conformity applications. The need for microscale modeling (the modeling of specific corridors or in- tersections) in the conformity analysis is not adequately supported by the aggregate regional and national data and assumptions used in MOBILE. The conformity analysis also must show consistency among the SIP, trans- portation plan, and TIP. This task can be challenging because the trans- portation plan horizon date is much further out than that of the SIP. The SIP is based on the attainment dates set in the Clean Air Act and the transportation plan horizon is 20 years. Federal conformity regulations require analysis for the "out years", the years beyond the deadline of the SIP but within the deadline of the long-range transportation plan. This creates an inconsistency problem because the maximum growth a region can accept depends on the level of vehicle technology and fuels that are accounted for in the SIP. The current use of MOBILE5b model in conformity analysis is limited by the model's ability to appropriately model emissions estimates for the out years beyond the SIP deadline. Specific out-year technological assump- tions used when MOBILES was developed in 1993 may not accurately rep- resent current assumptions. This could cause an unnecessary strain on regions, as they are forced to meet transportation plan budget tests based on outdated forecasts. For example, prior to the regulations implementing the national low- emissions vehicle (NLEV) standards and new regulations on heavy-duty diesel vehicles (HDDVs) in 2004, a region experiencing rapid growth would find it difficult to pass a transportation plan horizon-year budget test due to vehicle activity outracing vehicle technology. Figures 2-7 and 2-8 dem- onstrate this observation using conformity analysis results from a non- attainment area (North Central Texas Council of Governments 19981. Estimates that include the effects of the new NLEV and HDDV programs in MOBILE, show that NOX and VOC emissions will be reduced by 43%
CURRENT AND POSSIBLE FUTURE USES OF M OBILE 55 400 ~300 - o - a 'n200 ._ Z100 - o AIR-QUALITY CONFORMITY COMPARISON OF EMISSIONS LEVELS NOx · MOBILITY 2020/1998 TIP - ~ MOBILITY 2020/1999 TIP with revised emissions _ 1 1 1 1 1 1 1 1 1 1 1990 1999 2005 2010 2020 Analysis Year FIGURE 2-7 Effect of NLEV program and new HDDV regulations on NO emissions for Dallas metropolitan region. Source: NCTCOG 1998. )x and 32%, respectively, by the year 2020. Since this time, EPA has devel- oped even more dramatic improvements through Tier 2 emissions stan- dards and fuel sulfur reductions. Policy Implication and Future Direction Conformity analysis is often conducted annually for the TIP develop- ment and every three years for the transportation plan. This analysis is most critical to MPOs and state transportation departments because of its potential impacts on constructing transportation projects. In many cases, the conformity determination is made on a system of projects that show relatively small differences in emissions, especially compared to the effects of vehicle technology assumptions. MOBILE's aggregate approach to emissions estimates makes it poorly suited to accurately characterize such relatively small emissions impacts. Additionally, transportation projects often have a complex impact on traffic characteristics, and hence emis- sions, that are difficult to represent in the current linkage of travel-de- mand models to MOBILE. It is important that MOBILE be improved so that it is able to more accurately perform conformity or that EPA develops tools especially for such analysis.
56 MODELING MOBI[E-SOURCE EMISSIONS AIR-QUALITY CONFORMITY COMPARISON OF EMISSIONS LEVELS 400 - . ~5 In 3 0 0 o ._ ~ 200 - ._ LU 100 - o VOC MOBILITY 2020/1998 TIP \ + MOBILITY 2020/1999 TIP with revised emissions I VOC Emission Budget = 165 tons per day 1990 1999 2005 Analysis Year 2010 2020 FIGURE 2-8 Effect of NLEV program and new HDDV regulations on VOC emissions for Dallas metropolitan region. Source: NCTOG 1998. Conformity analysis will require that either MOBILE or an alternate conformity tool be quickly adaptable to vehicle technology advances and future regulatory initiatives. Examples include accounting for the propor- tion of the vehicle fleet to be zero-emission vehicles (ZEV), the Tier 2 emis- sions standard, and limits on sulfur in gasoline. Although these modifica- tions are part of the model used for assessing the Tier 2 proposal (the Tier 2 Model), and will be a part of MOBILES, they cannot be accounted for in MOBILE5b, the model currently used by state and local agencies for SIP development and conformity analysis. Due to the timeline inconsistency between SIP and transportation plans, this is most relevant for a trans- portation conformity analysis, because it is these new technologies and initiatives that have the largest impact on out-year emissions. Transportation Control Measure Effectiveness and CMAQ Eligibility Primary Users and Purpose States and MPOs use both travel-demand models and the MOBILE model to aid in the selection of transportation-control measures (T CM) for
CURRENT AND POSSIBLE FUTURE USES OF M OBILE 57 SIP credit and to aid in the evaluation and selection of appropriate pro- jects for congestion mitigation and air quality (CMAQ) funding. TCMs are projects specifically designed to assist in reducing overall emissions by re- ducing travel demand (VMT and trips), encouraging the use of alternative modes, and smoothing traffic flow. Traditional TCMs include high-occu- pancy vehicle lanes (HOV), signal and intersection improvements, bicycle and pedestrian improvements, light or commuter rail, advanced transpor- tation management technology (e.g., freeway management of incidents and accidents), and travel-demand management strategies (e.g., parking pric- ing). TCMs are evaluated and selected for CMAQ funding often using technical methodologies to estimate their effects on emissions. After eval- uation and commitment by the MPO, these projects can be inventoried and used in emissions-reduction strategies in SIPs. Issues and Limitations One critical issue with TCM and CMAQ projects is the inability to eval- uate their impacts using the traditional travel-demand modeling process outlined in Figure 2-1. Travel-demand models, the backbone of transporta- tion planning, cannot assess small-scare project specifics such as those commonly found with intersection and signal improvements. Therefore, air-quality impacts of most TCM and CMAQ projects are evaluated "off- model" or with post-processing techniques and do not benefit from the many internal travel-model features affecting volumes and speeds region- wide. As with the conformity analysis, this affects the accuracy of the emissions-reduction estimates from MOBILE because MOBILE is used to estimate the effects of a small change in travel parameters on a subset of the overall vehicle fleet. Because no formal blueprint outlines appropriate methodologies at the national level, each region has developed its own approach to evaluating TCM and CMAQ priorities, and these variations may cause important na- tionwide inconsistencies. Consistency in evaluation of TCM and CMAQ projects has been enhanced by post-processing software packages that es- timate several useful measures and assess the likely effects of a particular project. Such models are designed to link directly to the traditional four- step travel-demand modeling process through trip tables, which are passed back and fourth as necessary. Unfortunately, software packages to estimate effects on travel activity or air quality do not exist for all TCM and CMAQ categories, and regions must devise their own methods to eval- uate some TCMs (e.g., intelligent transportation systems for freeway man- agement).
58 MODE[/NG MOB`[E-SOURCE EMISSIONS Policy Implications and Future Directions Nonattainment regions must evaluate TOM and CMAQ project catego- ries, and these areas would benefit from evaluation methods that quantify changes in nontraditional transportation and in air quality. This will im- prove consistency, accuracy, and efficiency, for which national benefits could be inventoried. As with conformity analysis, the current MOBILE model is poorly suited to estimate the emissions-reduction benefits of TCMs. Users need more refined tools that provide a greater resolution of the impact of TCMs on traffic flow and emissions. National Environmental Policy Act and Evaluation of Major Capital Investments (Transit and Highway) Primary Users and Purpose The National Environmental Policy Act (NE PA) requires documentation of the environmental impacts caused by major capital investments that use federal funds, such as the construction of major transit and highway projects. NE PA requires that a project will not result in a violation of air quality standards and that the project be included in a TIP. NE PA also requires planners to provide a relative comparison of the air quality im- pacts of alternatives including the no-build alternative. Many agencies rely on MOBILE results to evaluate air-quality impacts and suggest alter- native transportation investment options. The primary users of the MO- BILE model in evaluating major capital transportation investments and developing environmental impact studies are state departments of trans- portation, federal resource agencies, state resource agencies, MPOs, local governments, consultants (generally working for these government units), and universities. Issues ant! Limitations MOBILE is designed to evaluate emissions impacts on a regional level not at finer levels such as corridors. Because many of the internal defaults in MOBILE are based on national and regionwide estimates, it cannot pro- vide the resolution needed to assess impacts for individual corridor-specific projects. One example to demonstrate this point is that the vehicle regis- tration data used to estimate vehicle age at the county level is not the same in each corridor in the county, especially considering the dramatic variations in income levels within a county.
CURRENT AND POSSIBLE FUTURE USES OF M OBILE 59 Policy Implications and Future Directions Because MOBILE is the recognized national emissions factor model, its results have broad impacts. For example, it may be used to estimate the difference between a four-lane and a six-lane freeway, the effects of a new rail line or a reversible HOV lane, or the impact of a range of TOM strate- gies. As described above, MOBILE is clearly not the most appropriately designed model to carry out such analysis. It is recommended that an- other tool be developed to accurately quantify small-scale impacts. How- ever, EPA must first approve a new model before it is used to evaluate emissions impacts. Thus, not only does there need to be development of improved methods for evaluating such projects, the EPA will also need procedures for vetting, documenting, and evaluating new models and methods for estimating corridor-level and other small-scale emissions im- pacts. SUMMARY OF POLICY IMPLICATIONS AND RECOMMENDATIONS Modeling the air-quality impacts of on-road vehicles is an interdisciplin- ary effort that encompasses the modeling of travel demand, emissions, and air quality. These individual components must be systematically inte- grated to develop analyses with adequate consistency, fidelity, accuracy, and precision. Each of the three types of models has a different focus (i.e., transportation models generally focus on transportation segments, emis- sions models on engine modes, and air-quality models on photochemical reactions and dispersion). Some kinds of uncertainties are inherent within each modeling domain, others occur at the interfaces when output from one model is used as the basis for input into the next. EPA should take steps to improve the linkages among the three models and improve the methods that are used to process MOBILE outputs for use in regional air- quality modeling. MOBILE is a single piece of software with a minimum of six different categories of uses. It is best suited for aggregate analysis and assessment of national and regional regulatory strategies and the development of SIPs for metropolitan areas. It is poorly-suited for analyses of a system of pro- jects or corridor analyses characteristic of conformity applications, assess- ment of TCMs, and environmental-impact assessments. Inconsistencies among these differing categories of uses led the commit- tee to conclude that likely no single model is appropriate for all applica- tions. As described further in Chapter 6, the use of MOBILE should be supplemented with the development and adoption of alternative models specifically designed to better link traffic flow in local settings to emis-
60 MODELING MOBILE-SOURCE EMISSIONS signs. EPA should develop one or more new mobile-source emissions mod- eling processes, perhaps using a Geographic Information System platform, that incorporate localized driving cycles and other conditions that influ- ence emissions. EPA should also consider the development of partnerships for data collection to better characterize emissions rates under a variety of conditions. MOBILE has a critical role in estimating and managing the levels of mobile-source emissions control. Future focus on new emissions stan- dards, the growing concern about air toxic emissions, and the growing cost of control strategies Will increase the demands for accuracy and detail from MOBILE. A strategic and comprehensive long-range plan is needed to better identify emerging needs for the modeling of automotive emis- sions, define the levels of detail and accuracy needed to meet those needs, and set priorities to improve MOBILE. EPA should modify its policy of issuing MOBILE updates on a batch, infrequent basis. Providing updates (known as information fact sheets) for major factors as soon as they are known, such as the adoption of NLEV standards or emissions-reduction credits for oxygenated fuels, would allow users to account for the latest technologies and revisions in SIP develop- ment as soon as possible. The committee recognizes the difficulties of working with a moving target, but concludes that the benefits of up-to- date modeling outweigh the disadvantages of more *equent changes.
