Logging a consumer PM2.5 sensor: what to record beside each reading and what the number cannot say

What it is

The EPA defines PM2.5 as fine inhalable particles with diameters generally 2.5 micrometres and smaller. A consumer sensor reports an estimate of their concentration. The EPA's sensor toolbox states that such sensors come with lower accuracy and higher uncertainty in the data they produce, and that an effective way to assess one is collocation: running it beside a federal reference or equivalent method instrument and comparing the series, from which a correction can be developed. The EPA's siting guide adds three points for a household: a correction developed from an outdoor collocation may not always apply indoors; evaluating a sensor by comparing it with nearby monitors does not apply to indoor sensors; and windows, doors and ventilation ducts can introduce rapidly changing temperature and relative humidity, which may adversely impact some sensors. The same guide asks users to document location, height and installation date and to keep notes on temporary activities such as cleaning or cooking for as long as the sensor is in use.

Why it matters

A number on a display invites comparison with official values. Without the sensor model, the averaging, whether a correction is applied, and what was happening in the room, it cannot be compared with anything, including the same sensor a month later.

How to apply

• Header per sensor: model, firmware or app version, whether the displayed value is raw or corrected (and which correction), the averaging interval, and the room, height and distance from windows, doors and vents, with a photo as the guide suggests.
• Log at fixed times, or export the device's series, and keep the raw column untouched; derived hourly means go in a separate column.
• Beside each reading, log relative humidity and temperature from the same or a nearby instrument.
• Keep an event column with times: cooking, candles, vacuuming, windows opened or closed, visitors, a nearby fire or fireworks.
• For an outdoor sensor, record at the same times the nearest official monitor's value, station name and distance, noting that the official value is usually an hourly or daily mean. For an indoor sensor, record the outdoor value as context, not as a reference.
• Once, if practical, run the sensor beside a second unit of the same model for a few days and log both, to see how far two units agree.

Pitfalls

Reading an AQI category as a concentration or the reverse. Comparing a one-minute indoor value with an hourly outdoor mean. Switching a correction on or off mid-log without a row saying so. Treating agreement between two identical units as accuracy. Drawing health conclusions from the log; it records numbers and their context, nothing more.