Most people checking their air quality app today are looking at a number that represents yesterday's average pollution level smoothed, delayed, and aggregated from monitoring stations that may be miles from where they actually are.
That is not a minor technical detail. It is the difference between useful safety information and misleading reassurance.
Here is what is actually going on behind the AQI number and why understanding it matters if you work with environmental data or build tools that use it.
How the AQI is calculated and where the methodology breaks down
The Air Quality Index in the United States is calculated from five pollutants: PM2.5, PM10, ozone, carbon monoxide, sulphur dioxide, and nitrogen dioxide. The index runs from 0 to 500. Above 100 is unhealthy for sensitive groups. Above 150 is unhealthy for everyone.
Sounds straightforward. Here is where it gets complicated:
The 24-hour rolling average problem the EPA's standard AQI calculation for PM2.5 uses a 24-hour rolling average. A wildfire that dumps heavy smoke on a city for three hours will barely move the AQI reading on your app until the averaging window catches up. Meanwhile people are outside breathing air that is genuinely hazardous right now.
NowCast was designed to fix this it is an EPA algorithm that weights more recent hours more heavily, making the AQI more responsive to rapid changes. But not all apps use NowCast, and most do not make it clear which calculation method they are using.
Sparse sensor network coverage the US has roughly 4,000 EPA reference grade monitoring stations for 330 million people across 3.8 million square miles. The spatial resolution is genuinely coarse. Air quality two blocks from an industrial site can be dramatically different from the nearest official station reading.
Low-cost sensor networks like PurpleAir supplement this with hyperlocal data but low-cost sensors have known accuracy limitations, particularly at high PM2.5 concentrations, and require correction algorithms to be comparable with reference grade data.
Pollutant-specific blind spots the standard AQI does not capture VOCs, ultrafine particles below PM2.5, or many industrial chemical emissions. The number on your phone is a partial picture of air quality, not a complete one.
What good air quality data infrastructure actually looks like reference grade instruments at strategic locations, supplemented by calibrated low-cost sensors for spatial coverage, feeding into systems that apply appropriate averaging algorithms and present uncertainty ranges alongside point estimates.
Enviro Testers builds reference-grade air quality monitoring instruments used by environmental agencies and industrial operators across North America — the kind of hardware that generates data worth trusting.
If you build on environmental data, the instrument layer is where accuracy starts.
Understand the data. Then build on it.
Top comments (0)