This repository contains data and findings for a collaboration between MuckRock and the Chicago Tribune. Reporting and analysis in this collaboration is a follow-up to the multi-newsroom “Smoke, Screened,” project published by The California Newsroom, MuckRock and the Guardian.

"Smoke, Screened" is an investigation into the use of a legal tool called the exceptional events rule of the federal Clean Air Act, which allows local air agencies across the United States to remove bad air days from the regulatory record of the Environmental Protection Agency (EPA), if the data is affected by an "exceptional event," such as a wildfire.

You can find earlier "Smoke, Screened" data and analysis in air-quality-exceptional-events and gao-wildfire-exceptions. Even more data and analysis driving the investigations of MuckRock's news team are catalogued in news-team.

All raw data can be found in data/raw, any data that involved manual work, like data entry, can be found in data/manual. The scripts to clean and organize this data can be found in etl and the output of those scripts can be found in data/processed. Processed data is used for analysis and visualization.

We retrieved historical PurpleAir sensor readings using PurpleAir's "Data Download Tool" for two sensors:

• Woodstock, Ill. sensor, which belongs to Jessica Beverly, a source we interviewed for this story
• Little Village, Chicago sensor, which belongs to the Little Village Environmental Justice Organization

These data are hourly averages and stored in data/raw/purple_air_download.

The EPA's Concentration Plot lookup tool provides daily air quality summary statistics by monitor across the United States. We used this tool to pull data for 8-hour ozone concentrations in several Midwestern cities and compare these to the 2015 EPA standard of .07 parts per million.

These data are stored in data/raw/ozone_8_hour_maxes_2023.

We received this data following an email request to the EPA. Flagging data with a "request exclusion" flag in AQS is one of the first steps in the exceptional events process, though agencies can also add an "informational" flag to data as well.

These flags are not necessarily a step towards the exceptional events rule, but rather a note for the agency as they earmark data they may want to review later. Those reviews can result in the data being excluded through the exceptional events rule.

This data is a snapshot in time and represents data flagged from May 1, 2023, to Aug. 31, 2023, as of Aug. 28, 2023.

These data are stored in data/raw/epa_informational_flags.

We retrieved data on hectares burned this past summer from the Canadian Interagency Fore Fire Centre website and entered the numbers by hand into spreadsheet, then converted to acres.

These are stored in data/manual.

PurpleAir sensor data was adjusted using an Environmental Protection Agency formula, or correction, that improves the accuracy of readings during smoke events.

The data was then cleaned using quality control in three parts: Eliminating hourly data with greater than 68% difference between A and B channels, aggregating the data to daily averages and, lastly, eliminating daily averages that did not contain 90% of the total amount of hourly readings for that day.

The code for this can be found in etl/purple_air_correction.R.