Introduction to Air Quality Modeling: Emissions Modeling
The process of preparing emissions for input into the photochemical model is referred to as “Emissions Modeling." With few exceptions emissions are not measured directly but are estimated using a variety of models ranging in complexity from a simple look-up table to the sophisticated and complex models used to estimate on-road vehicle and biogenic emissions. Emissions data in the air quality modeling process are divided into six source categories:
- Point Source Emissions Data
- On-Road Mobile Source Emissions Data
- Area Source Emissions Data
- Non-road and Off-road Mobile Source Emissions Data
- Biogenic Source Emissions Data
- Boundary Conditions
Point Source Emissions Data
The TCEQ processes industrial point source emissions for use in photochemical modeling in several steps. The first step is to acquire a point source emissions inventory for the year being modeled. Point source emissions are retrieved from the agency's database, the State of Texas Air Reporting System (STARS). STARS data extracted include reported daily average emission rates, location coordinates, stack parameters, chemical species, standard industrial classification (SIC), source classification code (SCC), and other data needed to model each source. Location coordinates (for example, longitude and latitude) allow the emissions to be placed at the appropriate location in the modeling grid. Depending on stack parameters (stack height, discharge velocity, temperature, etc.), the emissions may also be placed directly into elevated layers of the three-dimensional grid. Emissions from areas outside of Texas are obtained from a variety of sources, including the National Emissions Inventory (NEI). The TCEQ exchanges emissions information with other states and gathers reports and data for the Gulf of Mexico, Canada, and Mexico.
Because the composition of VOC emissions is critically important to accurately simulating ozone formation, the TCEQ asks industries to provide detailed breakdowns of the hydrocarbon species emitted at each reported emission point. In cases where this information is unavailable or incomplete, default speciation profiles are used to complete the speciation of each point based on its reported SCC.
To determine the time when pollutants are emitted, the TCEQ uses a variety of information. Most continuous-process point sources report operating 24 hours, seven days a week but some report more variable operating schedules. Where available, hourly emissions data for sources are used. Nationwide, most large electricity generating facilities report hourly emissions of NOx, SO2, and CO to the EPA’s Clean Air Markets Program. TCEQ occasionally conducts special inventory surveys to obtain hourly speciated emissions from specific sources. The TCEQ conducted such a survey during the Second Texas Air Quality Study (TexAQS II) intensive period, collecting hourly emissions from major point sources in East Texas from August 15 through September 15, 2006. A 2011 survey of certain flare operations in the Houston-Galveston-Brazoria area was recently conducted as well.
On-Road Mobile Source Emission Data
For ozone nonattainment areas within Texas, on-road mobile source emission inventories are typically based on vehicle miles traveled (VMT) estimates that are output from local travel demand models (TDMs) for various roadway segments or "links." Hourly VMT estimates for each link are multiplied by emission rates calculated with EPA's Motor Vehicle Emission Simulator (MOVES) model . Emission rates are developed separately for freeway and arterial links and matched to the hourly VMT based on average hourly operating speed. Hourly temperature and humidity inputs based on locally observed data from specific time periods are used as MOVES inputs to more accurately characterize the emissions on the ozone episode days chosen for modeling.
For attainment and/or non-metropolitan areas within Texas that do not have TDMs, Highway Performance Monitoring System (HPMS) data are used to determine hourly VMT by roadway type for each county. Similar to the approach described above, hourly VMT estimates by roadway type are multiplied by emissions rates from MOVES that vary as a function of speed, temperature, humidity, and drive cycle (i.e., high-speed freeway driving versus stop-and-go arterial driving).
Whether the VMT data are developed with TDMs or HPMS data sets, Weekday, Friday, Saturday, and Sunday "day type" on-road inventories are developed that differ in both the magnitude and hourly distribution of both VMT and estimated emissions. The TCEQ uses Version 3 of the Emissions Preprocessor System (EPS3) to convert the on-road inventory data into a gridded format appropriate for photochemical model input. For the TDM-based inventories, grid cell allocation is based on the X-Y locations of the link endpoints. For the HPMS-based inventories, grid cell allocation is based on spatial surrogates specific to each county and roadway type. For example, if a single grid cell contains 15% of the interstate highway miles in a specific county, then 15% of the interstate highway emissions are assigned to that grid cell.
In addition to gridding the hourly emissions, EPS3 assigns speciation profiles to appropriately group the exhaust and evaporative hydrocarbon emissions estimates based on reactivity for ozone formation. EPS3 is also used to make necessary emission adjustments by county and/or vehicle type. Sometimes these adjustments are needed to test out various control strategy scenarios.
For all non-Texas areas contained within the modeling domain, EPA's MOVES model is run in default mode to develop daily emission estimates by county for an average Summer Weekday. These emissions are processed with EPS3 and adjustments are applied to develop Friday, Saturday, and Sunday day type inventories based on pollutant-specific ratios from the Texas on-road inventories for Friday/Weekday, Saturday/Weekday, and Sunday/Weekday. In addition, the hourly distributions of the Texas on-road inventories by both pollutant and day type are applied to the non-Texas portions of the modeling domain.
For the Mexico portions of the modeling domain, the on-road portion of the 1999 Mexican National Emissions Inventory (NEI) is projected to specific years using a combination of the MOBILE6-Mexico model and an assumed annual VMT growth rate of 2%. In a similar way, the 2006 Canadian National Pollutant Release Inventory (NPRI) is used and projected with MOBILE6-Canada and a 2% annual VMT growth rate assumption. The end result of this process is a gridded and speciated inventory for photochemical model input with relatively high spatial and temporal resolution of on-road emissions.
Area Source Emission Data
Area source emissions come from of a variety of anthropogenic (man-made) sources that are too small, too abundant, or too dispersed geographically to inventory individually. Examples of these sources include dry cleaning, vehicle refueling, cooking, and solvent usage. Area sources are modeled using a top-down approach, meaning emission totals are estimated for large geographic regions, usually states or counties. When possible, the TCEQ will conduct special studies to develop bottom-up inventories for area source emissions of concern. For example, a comprehensive oil and gas production study was completed in 2010. This study produced county-level emissions for many processes involved in oil and gas production, such as compressors, pneumatics and gas dehydrators. The TCEQ houses Texas county emission totals in the Texas Air Emissions Repository (TexAER), and obtains data for other areas from the NEI and other sources.
Before the photochemical model can be run, the geographic location of air pollution emissions must be identified. It must also be determined what time of day pollutants were emitted and what particular VOCs were present. Timing is important in photochemical modeling, as the sun plays a large role in the chemical reactions that create ozone. So emissions occurring midday will react differently from emissions at midnight. Identifying the correct VOCs present (or chemical speciation) is important because some VOCs are far more reactive than others.
The TCEQ uses spatial surrogates, diurnal profiles, and chemical profiles to accomplish these goals. A surrogate is a readily available geographic substitute that can be used to help locate area source emissions spatially. A good example is emissions from personal care products, which should be highly correlated spatially with population. Using census tract population data from the U.S. Census Bureau, county-level emissions from these sources can be reasonably distributed geographically within the counties. The TCEQ frequently produces custom surrogates, such as well-head density from Railroad Commission of Texas data, to spatially allocate oil and gas production emissions.
Determining the time of day that emissions are released is accomplished by using diurnal profiles. For example, vehicle refueling activity is related to driving activity, so refueling emissions are distributed through the day in proportion to the amount of traffic. Modeled emissions also vary by day-of-week and by month through the application of activity profiles.
Chemical speciation is simply assigning the correct proportion of different chemicals to different activities. For example, emissions from a dry cleaner will differ chemically from those of a print shop, so each category is assigned its own speciation profile.
Non-road and Off-road Mobile Source Emission Data
Like area sources, emissions from non-road mobile come from of a variety of anthropogenic (man-made) sources that are too small, too abundant, or too dispersed geographically to inventory individually. Examples of non-road mobile sources include construction, recreational boating, lawn care, and logging. For all non-road mobile categories except aircraft, locomotives, drilling rigs, and commercial marine vessels, the emissions are calculated using the TCEQ-developed Texas-specific NONROAD (TexN) model, which utilizes the EPA's latest NONROAD MODEL 2005. Although operating the EPA's model with all of the default surrogates is acceptable, the EPA encourages states to update the model with local, county-level data based on surveys and other relevant information. The TexN model is a software tool for developing emissions estimates for non-road mobile sources in Texas using county-specific activity data. The model allows the TCEQ to replace the EPA's default data with local data. Local, county-level data is incorporated into the TexN model as it becomes available to the TCEQ.
In 2010 the TCEQ developed a comprehensive drilling rig inventory to cover this important category for Texas. Surveys, well depths, and rig horsepower and loads were included to get an accurate snapshot of 2008 activity and emissions. Custom surrogates of new well locations and county-level drilling trends give good spatial distribution and projections to required model years.
For areas outside Texas, emissions are calculated using EPA's NMIM or are obtained from other sources. The TCEQ also adjusts emissions of NOx from diesel equipment to account for humidity for modeling.
Emissions from ships, locomotives, and aircraft are not available through NMIM or the NONROAD model. These sources are sometimes referred to as off-road mobile sources to distinguish them from other non-road sources. Texas emissions for off-road sources are obtained from TexAER, except emissions from ships in the HGB and BPA areas. For these areas, where shipping contributes a considerable fraction of daily emissions, ship emissions are estimated based on studies conducted in 2000 (BPA) and 2007 (HGB). Outside Texas, emissions for off-road sources are obtained from the NEI and from other sources like the Gulfwide Emissions Inventory (GWEI) , Waterway Network Ship Traffic, Energy and Environment Model (STEEM) , and EPA near port study .
Spatial allocation of non- and off-road emissions is analogous to that described for area sources above, except that off-road sources use special surrogates tied to their specific characteristics (rail lines, airports, and shipping lanes). In addition, because large ships emit their exhaust from elevated stacks and along shipping lanes, TCEQ allocates ship emissions similar to mobile source links but in the model layer corresponding to their stack height.
Similar to area source emissions, modeled non-road emissions vary by hour, day-of-week, and month through application of activity profiles. Chemical speciation is also accomplished by using profiles.
Biogenic Source Emissions Data
In the 1980s, air pollution researchers discovered that in order to properly simulate ozone formation in their computer models, they needed to take into account the natural emissions of various chemicals. These natural emissions are referred to as biogenic emissions, and they can play an important role in the atmospheric chemistry of ozone formation. The most important sources of biogenic emissions are trees, which can emit a number of compounds. Pine trees, for example, produce compounds called pinenes, which have the familiar resinous aroma of pine needles. There are many factors affecting the biogenic emissions. In order to keep track of the effects of the many different factors, the TCEQ uses computer models such as MEGAN to estimate the biogenic emissions. These models incorporate the latest surveys of local land cover and vegetation type, and up-to-date scientific knowledge of biogenic emissions. Since 1997, the TCEQ has commissioned several special studies on the biogenic emissions of Texas vegetation, in order to ensure that the biogenic emissions models will work well for our unique state.
The following types of data are needed to estimate biogenic emissions:
- Land use/land cover (LULC) map- Based upon satellite data, field surveys, and aerial imagery, researchers assign the appropriate land cover category to each mapping unit in the area of interest. Examples of land cover categories include "herbaceous cultivated" or "urban needle-leaf evergreen forest." The resulting land cover map forms the basis of the biogenic emissions model.
- General and species composition- The relative abundance of different types of vegetation in the area to be modeled is particularly important because different types of vegetation emit different types and vastly different quantities of VOC. For example, forests dominated by oak trees emit great amounts of isoprene, while forests dominated by pine trees emit substantial quantities of terpenes. Genera and species composition is ascertained primarily from field surveys and remote sensing data.
- Leaf biomass density- Emissions are directly related to the leaf biomass - all other factors being equal, more leaf biomass means more emissions. Leaf biomass density is determined by field study and by remote sensing. Surveyors first identify the tree and then measure the height of the tree and diameter of the trunk. From those data, they calculate the crown area and total leaf biomass of a typical tree of that type, based on equations published in the scientific literature. If there are areas that cannot be surveyed on foot, the researchers rely upon aerial imagery, satellite imagery, and LiDAR (Light Detection and Ranging).
- Meteorological variables - Emissions depend strongly upon the temperature and solar radiation to which the leaves are exposed. TCEQ uses spatially interpolated hourly temperature observations from the National Oceanic and Atmospheric Administration (NOAA) and/or the meteorological model as input to the biogenic emission model. To create hourly maps of solar radiation intensity, the TCEQ has acquired solar radiation data from satellite retrievals.
Although our photochemical modeling is focused upon Texas, the modeling must account for emissions and the transport of pollutants from outside of Texas and United States. On the edges of our largest modeling domain, the concentrations of pollutants are defined as boundary conditions. The boundary conditions vary by hour for the specific modeling episode(s) and are set vertically from the surface to the highest model layer many kilometers into the atmosphere. The TCEQ obtains boundary conditions from the output of global-scale chemical transport models such as:
- Goddard Earth Observing System Chemistry model (GEOS-Chem), maintained by Harvard and NASA
- Model for Ozone and Related chemical Tracers (MOZART), maintained by the National Center of Atmospheric Research