The Poisoned Places series relied on analysis of four datasets relating to sources of air pollution regulated by the U.S. Environmental Protection Agency: the Clean Air Act watch list, the Air Facility System (AFS), the Toxics Release Inventory (TRI) and the Risk Screening Environmental Indicators model (RSEI).
The Clean Air Act watch list
The Center for Public Integrity’s iWatch News and NPR obtained the “watch list” through a Freedom of Information Act request to the EPA. Two versions of the list were obtained: one current as of July 2011, the other as of September 2011.
While these facilities are regulated by the states and the EPA, not all facilities report to the EPA’s Toxics Release Inventory (TRI); certain criteria must be met.
Further research indicated that two of these facilities are under construction, two are temporarily closed and nine are permanently closed. Additionally, not all were flagged in the data as high priority violators (HPVs) as of August 2011. iWatch News and NPR placed watch list facilities into industry categories and used the primary four-digit Standard Industrial Code; data entry for the more current North American Industry Code System was not as consistent.
The Air Facility System
AFS tracks permit, enforcement and compliance information for sources of air pollution. All major sources and some minor sources are required under the Clean Air Act to obtain operating permits that stipulate what they must do to control air pollution. The data contain information about inspections, enforcement actions, penalties and compliance, including HPV status. The HPV flag is activated when a facility has a high priority violation, as defined by criteria established in a 1998 EPA memo. It is deactivated when the violation is fully resolved. In some cases an HPV flag can remain after a violation has been resolved.
State or local agencies are required to report data to the EPA on a regular basis. However, because of some technical complications or lack of diligence, data are not always entered in a timely manner. Therefore the data do not always present a complete picture of enforcement or compliance for a particular facility. States’ comments on data inaccuracies can be found on the EPA’s website.
iWatch News and NPR downloaded AFS data from the EPA Web site in October, and the data are current as of August 2011. To compensate for incomplete data, iWatch News and NPR contacted multiple state and local agencies, EPA regional offices and the national office in Washington, D.C., to try to corroborate information gleaned from AFS.
iWatch News also obtained a detailed subset of the AFS database through the Freedom of Information Act. It details the steps regulators have taken to address concerns involving what the agency considers "high priority violators" across the country.
Read more about AFS.
The Toxics Release Inventory
The TRI, authorized under the Emergency Planning and Community Right-to-Know Act of 1986 (EPCRA), houses emissions data reported yearly by polluting facilities. More than 20,000 facilities reported emissions of some 600 chemicals in 2009, including most of the 187 hazardous air pollutants that the EPA is required to control, as defined in the 1990 amendments to the Clean Air Act.
Not all toxic chemicals or sources of pollution are included in the TRI, however. Smaller sources (such as dry cleaners and gas stations) and mobile sources are not required to report. Only facilities in certain industry sectors, with a minimum level of production and number of employees, must report each year.
In addition, some quantities reported are estimated rather than monitored, and facilities use different estimation methodologies that could result in slightly different amounts. It is widely acknowledged that the TRI also contains some reporting errors, and in some instances facilities underreport.
The EPA also notes that no health risk conclusions can be drawn from the TRI alone, since there is no information on toxicity or distribution of chemicals.
iWatch News and NPR analyzed the most recent complete version of data through 2009. Only fugitive and stack (on-site) air releases were considered when analyzing emissions. Completed 2010 reports became available from the EPA 11 days before publication — after iWatch News and NPR finished their analysis. iWatch News and NPR plan to use the 2010 data as the series continues.
Read more about the TRI.
The Risk Screening Environmental Indicators
RSEI was created by the EPA to assess chronic human health risks from chemical releases reported in the TRI. For each release, the model takes into account the toxicity of the chemical, its fate and transport through the environment, the exposure pathway (air or water) and the number of people potentially affected. The model produces a relative risk score for each release. The latest version of the model (version 2.3.0) is based on the 2007 version of the TRI.
The screening model does not produce a measurement of actual risk, nor does it address acute (or immediate) risk from toxic air releases. It produces relative results for comparison, and its primary use is for identifying chemicals, industries or localities that require further investigation.
The model makes some necessary assumptions that should be considered when looking at its risk scores. Chemicals are assigned toxicity weights, but in some cases where chemicals are grouped together, the entire group is assigned the toxicity of its most toxic member. Also, if some facility information (such as stack height) is unavailable, the model will plug in alternate information (such as the national median stack height).
Since the model is based on TRI reports, it is subject to the same caveats. If a facility reported incorrectly in 2007 and later amended the report, the incorrect quantity will be reflected in the RSEI score, which was calculated using the 2007 version of the TRI.
For the interactive map, iWatch News and NPR used the latest version of RSEI to place facilities in one of five risk categories. Before doing this, several RSEI experts were consulted, both inside and outside of the EPA. To place the facilities in a risk category, iWatch News and NPR first averaged the risk screening scores associated with the facilities over the five-year period, from 2003 to 2007 (the latest year of data available).
Because the initial analysis revealed the distribution of those averages to be skewed, iWatch News and NPR applied a Log10 data transformation to place the data in a normal distribution. Once transformed, the data were grouped into five categories based on the transformed quantities, from lowest to highest.
Read more about RSEI.