Where the numbers come from. CHATTER does not measure anything itself. Every value on this site is computed from public monitoring records kept by two agencies.
The cancer-risk and other-health-risk pollutants come from the Louisiana Department of Environmental Quality’s (LDEQ) canister network: at each site a 24-hour air sample is collected every sixth day and analyzed in a laboratory for a fixed list of volatile organic compounds. Results through 2023 are taken from EPA’s Ambient Monitoring Archive for Hazardous Air Pollutants, the national compilation LDEQ reports to; results from 2024 onward are downloaded directly from LDEQ’s canister-data portal, because the national archive is published with a lag of a year or more. The two sources report the same samples in the same units, and CHATTER stitches them into one record per monitor.
The EPA criteria pollutants — ozone, fine particles, nitrogen dioxide and sulfur dioxide — come from the continuous regulatory monitors LDEQ operates across the state, whose hourly and daily readings are retrieved from EPA’s Air Quality System (AQS). These monitors report within weeks, so the criteria tab usually runs several months later than the canister tabs.
How a heatmap cell is calculated. For a canister pollutant, CHATTER takes every valid sample in the period (a calendar year, the year to date, the last 365 days, or the most recent complete calendar quarter), averages them, and divides that mean by the pollutant’s EPA screening level — the concentration below which long-term exposure is not expected to raise health concerns. A cell of 1.00x means the average sat exactly at the screening level. Cancer-risk cells use the cancer screening level; other-health-risk cells use the non-cancer screening level for the same compound, and a health-effect total adds up the ratios of every pollutant linked to that effect.
For a criteria pollutant the cell is a single-period design value — the particular statistic EPA compares against each National Ambient Air Quality Standard, such as the fourth-highest daily maximum eight-hour ozone concentration or the 98th percentile of daily fine-particle readings — divided by that standard. EPA’s official design values average three years; CHATTER’s are for one period at a time, so they signal where a closer look is warranted rather than declare an area in or out of attainment.
Two marks show when a number rests on thinner ground. An asterisk means that only 60–75 % of the scheduled samples were available for that period (below 60 % the cell is left blank). A double asterisk means that most of the samples behind the number were below the laboratory’s detection limit; such samples are counted at half the detection limit, so the true value is uncertain and could be anywhere from near zero up to that limit.
Every value on this site, in one file. CHATTER_Louisiana_values.csv holds one row for every heatmap cell — monitor, pollutant (or health effect, or total), and period — together with the summary statistics it was computed from. It is regenerated from the same code that draws the pages, so a row in the file cannot disagree with a cell on the site. It opens in Excel, R or Python; about 3,500 rows.
Each row carries the group (Cancer Risk, Other Health Risks, EPA Criteria Pollutants), the monitor and its AQS site number, the parish, the period and its calendar bounds, and value_shown, the number in the cell. For canister pollutants it adds the mean concentration in ppb, the screening level it was divided by, the number of valid samples and the number scheduled, the resulting coverage and whether the cell carries the asterisk, and the share of the readings that were detection-limit substitutes and whether the cell carries the double asterisk. For criteria pollutants it adds the design value in the standard’s own units, the standard, the statistic used, and the number of daily values behind it. Totals are marked as such and carry only the displayed sum.
How each statistic is defined. CHATTER_statistics_documentation.xlsx is the reference sheet: every statistic that appears on the site, its formula, and where in the code it is computed.
Screening levels. EPA Regional Screening Levels, Resident Air table, November 2023 edition, converted from µg/m³ to ppb at 24.45 L/mol; cancer values at a 10⁻⁶ target risk, non-cancer values at a hazard quotient of 1. Two compounds added for Louisiana (trichloroethylene and vinyl chloride) have not yet been assigned a non-cancer health-effect category and appear under “Health effect not yet assigned” on the other-health-risk pages.
Sampling schedule and coverage. Canister monitors follow a one-in-six-day schedule. Scheduled samples for a period are counted from each monitor’s first sample date in six-day steps; coverage is valid samples divided by scheduled samples. Periods below 60 % coverage are not shown; 60–75 % are shown with an asterisk.
Periods. Calendar years, the year to date, a rolling 365 days, and the most recent complete calendar quarter, all measured back from the data cutoff. For the canister tabs the cutoff is the latest day on which any monitor has a sample, so a monitor whose results arrive later shows reduced coverage in its recent periods rather than holding the whole site back. The criteria tab runs to its own last day of data and says so in its column headings.
Non-detects. A result below the laboratory’s method detection limit is counted at half that limit. For 2024-onward portal downloads, which report non-detects as zero without a detection limit, the most recent detection limit reported for that monitor and compound in the archive is used. When at least half of the summed readings behind a cell are such substitutes, the cell carries the double-asterisk mark.
Statistic displayed. The arithmetic mean of the period’s samples. A 95 % upper confidence limit of the mean (Student’s t) is also computed and kept in the working tables but is not displayed.
Criteria pollutants. Daily values from AQS: the 24-hour PM₂.₅ sample, the daily maximum 8-hour ozone average, and the daily 1-hour maxima for NO₂ and SO₂, primary monitors only. Single-period design values: 98th percentile of the daily values (PM₂.₅ against 35 µg/m³; NO₂ against 100 ppb), 99th percentile (SO₂ against 75 ppb), and the fourth-highest daily maximum (ozone against 0.070 ppm). On the front page each parish card takes each pollutant’s row whole from the parish monitor with the highest value in the Last Available Month column, and names that monitor; the criteria tab shows every monitor.
Code. The site is a Quarto project; every table is built by one Python module per pollutant group, and the pages only draw them. The file above is written by the same modules.