2.5 How does this NEI compare to past inventories?

Many similarities exist between the 2020 NEI approaches and past NEI approaches, notably that the data are largely compiled from data submitted by S/L/T agencies for CAPs, and that the HAP emissions are augmented by the EPA to differing degrees depending on geographical jurisdiction because they are a voluntary contribution from the partner agencies. In 2020, S/L/T participation was again somewhat more comprehensive than the previous NEI. The NEI program continues with the 2020 NEI to work towards a complete compilation of the nation’s CAPs and HAPs. The EPA provided feedback to S/L/T agencies during the compilation of the data on critical issues (such as potential outliers, missing SCCs, missing Hg data and coke oven data) as has been done in the past, collected responses from S/L/T agencies to these issues, and improved the inventory for the release based on S/L/T agency feedback. In addition to these similarities, there are some important differences in how the 2020 NEI has been created and the resulting emissions, which are described in the following two subsections.

2.5.1 Differences in approaches

With any new inventory cycle, changes to approaches are made to improve the process of creating the inventory and the methods for estimating emissions. The key changes for the 2020 cycle are highlighted here.

To improve the process, we learned from the prior triennial inventories (for 2008, 2011, 2014, and 2017) compiled with the EIS. We made changes to pollutant, SCC, and NAICS codes, refined quality assurance checks and features that were used to assist in quality assurance but retained the same Nonpoint Survey functionality used in the 2017 NEI (introduced for the 2014 NEI) to assist with S/L/T and EPA data reconciliation for the nonpoint data.

In addition to process changes, we improved emissions estimation methods for all data categories. We summarize the differences in approaches in the following sections.

2.5.1.1 Point data category

For point sources, the only major change for 2020 was our incorporation of the Air Toxics Screening (AirToxScreen) assessment between the draft NEI and this 2020 NEI release. AirToxScreen provided SLTs a review of high-risk air toxic facilities. More information on point source improvements is available in Section 3.

2.5.1.2 Nonpoint data category

We made method improvements for several stationary nonpoint sectors (Section 6). The EPA creates and provides emissions estimation tools for two purposes: 1) as tools for S/L/T agencies to use themselves, and 2) to backfill emissions values where not provided by S/L/T agencies.

As part of the 2017 NEI development process, we introduced “Input Templates” for S/L/Ts to provide activity data for several nonpoint data category tools. By allowing a simple template where S/L/Ts can review the previous year’s data, the data source, and easily update values at a county or state level, that then feeds into EPA’s emissions estimation tools, assures that the calculations and methods are identical. For the 2020 NEI, we centralized the input template download and upload process, and enabled S/L/Ts to directly load their inputs into EPA tools to generate draft emission estimates prior to submittal to the NEI. EPA provided default Input Templates to S/L/T inventory developers for them to modify and return to EPA. We encouraged S/L/Ts to submit inputs rather than direct emission submittals for many nonpoint categories.

We also continued to streamline the Nonpoint Survey (Section 6), first introduced for the 2014 NEI development cycle, to simplify the options and improve transparency. In particular, we added a button on the NP survey that indicates whether an agency submitted an input template. This helped us QA our data twofold: 1) did the agency intend to submit an input template, and 2) did they actually submit a template. By default, all Nonpoint Survey responses were set to “Yes -Supplement my data with EPA Estimates” to ensure complete coverage in the absence of S/L/T feedback.

As discussed in Section 25, for the 2020 NEI, we added default fuel consumption data for nonpoint Industrial and Commercial/Institutional (ICI) fuel combustion, based partially on S/L/T-submitted Point carbon monoxide emissions; this greatly reduced the potential double-counting of ICI fuel consumption estimates for S/L/Ts that did not submit direct nonpoint emissions or an input template. Similar to the 2017, we continue to use estimated point fuel consumption for reconciling the nonpoint component of ICI fuel consumption/emissions -we no longer allow point emissions subtraction. We provided S/L/Ts with cross-references from point inventory facilities to existing U.S. Energy Information Administration (EIA) ICI sector assignments and fuel mapping. We relied on S/L/Ts to provide EPA with these state-level inputs via 4 different Input Template options.

Emissions for residential wood consumption (Section 27) were affected by an updated methodology in the wood consumption estimates obtained from the State Energy Data System (SEDS), which reflected updated national survey data and allocation scheme based on heating degree days which distributed emissions from warmer (southern) states to cooler (northern) states. In addition, we updated to use higher PM emission factors for certified wood stoves as the old emissions were deemed inappropriate for continued use.

The methods used to estimate nonpoint solvent utilization emissions (Section 32) were updated using a new emissions model. This model uses national-level product usage estimates to subsequently estimate speciated emissions, that are further allocated to the county-level using geographically specific sources of data and modulated if the locality reports control mechanisms for select SCCs. In addition, a new SCC (2460030999) was added to this category to reflect emissions from lighter fluids, fuel starters, and other consumer product fuel sources.

Most states saw a significant increase in CO, PM2.5 and VOC from commercial cooking, a result of an improvement in the activity data on the number of restaurants. Large decreases in residential fuel combustion for SO2 is a result of a continued decrease in consumption and more significantly, more widespread inclusion of a lower default sulfur content for distillate fuel oil.

All fires data are now included in the nonpoint data category for the 2020 NEI. This is simply a format issue as the underlying methodology for computing wildland fires (wildfires and prescribed burning) are still developed using satellite data for location and day-specific fires, but for 2020 NEI, are subsequently aggregated to the county-level. Overall, national-level agricultural field burning increased but was mostly offset by corresponding decreases in prescribed fire estimates.

The 2020 NEI introduces (VOC and associated VOC HAPs) from agricultural silage and new asphalt paving processes and methodology. Agricultural fertilizer application (NH3) estimates significantly increased due to several updates: new emission factor measurements, change in how landcover was modeled, improved meteorological data, and an error correction. Oil and gas production increased significantly in the Permian basin; otherwise, most VOC changes result from new Solvents methodology (Section 32), which also includes pesticide application.

For all nonpoint categories, we updated the activity data to use the newest data available, at the time, to represent the 2020 inventory year; in most cases, this is year-2020 activity data. Most emission changes for all nonpoint sources not otherwise discussed in this section resulted from these activity data updates -be they from EPA or new for 2020, provided directly from S/L/Ts.

The Biogenic database incorporated a new version of the Biogenic Emissions Landcover Database (BELD5) and provides updates for all states, including Alaska, Hawaii, Puerto Rico and the U.S. Virgin Islands.

2.5.1.3 Onroad and nonroad data categories

For mobile sources, onroad methodology used an updated version of the MOVES model with updated mobile source activity data such as vehicle miles travelled (VMT), age distributions, and fuel type mix, and improved idling computations; we also received new telematics data from StreetLight Data, Inc. For both onroad and nonroad, we relied on model inputs provided by S/L/T agencies and other sources, except for California and Tribes, who submitted emissions estimates. Sections 5 (nonroad mobile) and 6 (oroad mobile) provide more detail on these improvements.

2.5.2 Differences in emissions between 2020 and 2017 NEI

This section presents a comparison from the 2017 NEI to the 2020 NEI. Table 2 4 compares CAP emissions for the 2020 minus 2017 NEI for seven highly aggregated emission sectors. Table 2 5 compares emissions for select HAPs for the 2020 minus 2017 NEI for the same seven highly aggregated emission sectors. Emissions from the biogenic (natural) sources are excluded, and the wildfire sector is shown separately for CAPs and HAPs. While Pb is a CAP for the purposes of the NAAQS, due to toxic attributes and inclusion in previous national air toxics assessments (NATA) and screenings (Air Toxics Screening) assessements, it is reviewed here with the HAPs. The HAPs selected for comparison are based on their national scope of interest as defined by Air Toxics Screening Assessments. With a couple notable exceptions, CAP emissions are lower overall in 2020 than in 2017. Some specific sector/pollutants increased in 2020 from 2017.

The increases in fuel combustion for most pollutants are primarily a result of increases in residential wood combustion where the underlying source of activity data (fuel consumption) increased significantly via methodology and geographic distribution changes. Conversely, the significant decrease in electric generating unit (EGU) emissions account for the decrease in overall NOX and SO2 fuel combustion. Increases in Miscellaneous CO are from increased prescribed and agricultural field burning. Increases in nonroad gasoline engine lawn and garden and commercial estimates explain the increases in Nonroad Mobile CO. Large increases in agricultural fertilizer application explain the large Miscellaneous NH3 increase. Large Industrial Processes VOC increases are primarily from increased oil and gas activity in the Permian Basin.

As expected, the pandemic contributed to significant decreases in 2020 for all Highway Vehicle pollutants. As discussed in Section 7, there were comparatively more wildfires in 2020 than 2017, explaining the significant increases in wildfire emissions for 2020. Year 2017 was a generally quiet year for such fires.