How We Created the Most Detailed Map Ever of Cancer-Causing Industrial Air Pollution
ProPublica spent two years analyzing billions of rows of EPA data to visualize the spread of toxic air pollution from industrial facilities in the U.S. The result is the most detailed map of cancer-causing industrial air pollution ever published.
The analysis underlying our map uses data from an EPA model called Risk-Screening Environmental Indicators, or RSEI. This model takes a variety of inputs, including emissions data, weather modeling, and facility specific information, and puts out estimated concentrations of toxic chemicals in the air around industrial facilities. In this methodology, we explain how we used the EPA’s database to compute cancer risk estimates for air releases and discuss the advantages and shortcomings of using EPA data to map industrial cancer risk.
Cancer is the second-leading cause of death in the U.S., after heart disease. A variety of factors may contribute to an individual’s risk of developing cancer, including age, family history, access to health care, diet and exercise. Our map should not be used to try to establish with any certainty that an individual case of cancer was the result of toxic releases from specific industrial facilities. Rather, it provides a starting point for readers to learn more about the potential sources of industrial cancer risk near them.
What Our Map Does (and Does Not) Include
Cancer-Causing Air Toxics
Air toxics, also known as hazardous air pollutants, are a class of 187 chemicals that are known or suspected to cause cancer and other serious health effects, according to the EPA. The purpose of ProPublica’s map is to visualize the cumulative cancer risk from toxic industrial air pollution. When the EPA undertakes a risk assessment, it does so by considering certain types of facilities and equipment in isolation, which underestimates the true amount of pollution in the air for people who live near multiple industrial facilities. While the EPA might deem the risk of an individual facility “acceptable” or “safe” by itself, our analysis shows that the harm can be more substantial when risks from facilities are considered in aggregate.
Our multiyear analysis computes estimates of cancer risk generated by many of the large industrial facilities in an area. Unlike the risks from other conditions linked to toxic air pollution, such as asthma and diabetes, cancer risk from EPA-regulated hazardous air pollutants is additive. Therefore, our analysis only includes pollutants for which the EPA has sufficient evidence to produce a cancer risk value. For example, lead, a dangerous neurotoxin that has been linked to developmental deficiencies in children, is not included in our map.
Not all known carcinogens are classified as hazardous air pollutants. Just because the EPA does not consider an air toxic to be a known human carcinogen does not mean that it cannot cause cancer — it simply means that the EPA’s body of science has not yet identified an association. It is important to note that air toxics are linked to other health effects as well, such as headaches, asthma, seizures and poor birth outcomes.
Our analysis excludes the six “criteria pollutants” that are common in our environment: carbon monoxide, lead, ground-level ozone, particulate matter, nitrogen dioxide and sulfur dioxide. While some of these chemicals have been linked to adverse health effects, they are regulated differently than other air toxics, and are not reported to the EPA’s Toxics Release Inventory. While the EPA classifies particulate matter in diesel exhaust as a likely human carcinogen, the agency has not yet assigned it a cancer potency value, and RSEI does not model it.
Selection of Facilities in Our Analysis
The EPA’s RSEI model that powers our map uses emissions data from facilities that report to the agency’s Toxics Release Inventory database, or TRI. Not all industrial facilities are required to report to TRI. Only facilities that fall under certain industry codes, have ten or more full-time employees, and manufacture, process or use certain amounts of specific chemicals are required to submit their data annually to TRI. For this reason, toxic hot spots may be present in many places that are not indicated on our map, which only visualizes pollution from certain kinds of large industrial sources.
Other Sources of Cancer-Causing Air Pollution
Our map and analysis focus on cancer-causing pollution generated from stationary sources. This type of air pollution acutely affects Americans who live on or near the fence lines of industrial facilities. To be sure, industrial facilities are not the only sources of cancer-causing air pollution in our society. Mobile sources — such as cars, diesel trucks and buses and ships at ports — represent significant sources of air pollution, but they are not included in TRI or in our analysis. Wildfires, toxic waste at Superfund sites or the like and use of pesticides and herbicides, to name just a few other examples, may also increase a community’s risk of cancer. Given the myriad sources of cancer-causing air pollution, our analysis is likely an underestimate.
Using Air Modeling to Understand Cancer Risk
ProPublica’s map does not show an individual’s actual exposure to toxic compounds; rather, it uses data from a sophisticated computer model that predicts where air toxics are concentrated around facilities based on emissions data that those facilities report. While air models provide conservative estimates of real-world conditions, they are often the best starting point for understanding how much any one facility or chemical might be contributing to a community’s cancer risk in an industrialized area. State environmental agencies and the EPA frequently use air modeling to screen for potential risk and make critical decisions regarding a community’s health.
Estimating Cumulative Cancer Risk with the EPA’s RSEI Model
The power of RSEI lies in its granularity: The model divides the entire country into squares of land, or “grid cells,” that are less than a mile wide — 810 meters wide, to be exact. For each grid cell, the model estimates concentrations of toxic chemicals in micrograms per cubic meter of air. For example, RSEI can provide a modeled concentration of how much ethylene oxide a particular 810-meter by 810-meter chunk of Houston contains as the result of a particular facility’s reported emissions.
We analyzed nearly 7 billion rows of RSEI data to estimate industrial cancer risks for the entire country. To turn the concentration estimates that RSEI outputs into cancer risks, we first removed all of the chemicals in the database for which there was insufficient evidence to develop a cancer potency value. We…