How AirNow Data Works: From Monitor to AQI

Last updated April 29, 2026 · 9 min read

When you check the AQI on your phone, you're at the end of a data pipeline that starts with a federally certified instrument on a rooftop somewhere, runs through state and federal aggregation services, and is finalized by an algorithm called NowCast. This guide walks the whole path: who measures, what they measure with, how it's quality-controlled, and how it ends up as a single color-coded number on your screen.

What AirNow is

AirNow is the U.S. government's official air-quality data service, run by the EPA in partnership with NOAA, the National Park Service, the U.S. Forest Service, NASA, and tribal, state, and local agencies. It does two distinct jobs:

1. Aggregate real-time observations from thousands of regulatory air-quality monitors across North America and a small number of overseas sites operated by the U.S. State Department.
2. Publish forecasts issued by state and local air-quality forecasters.

Both flows are exposed through public data feeds — the AirNow public website, the AirNow API used by apps like Smog Report, the Fire and Smoke Map, and email/push subscriptions for daily reports.

The monitors themselves

The thousands of stations behind the AirNow data are not consumer-grade gadgets. They are regulatory monitors built around Federal Reference Methods (FRM) or Federal Equivalent Methods (FEM) — specific instrument designs the EPA has certified to meet performance standards under 40 CFR Part 53. Two illustrative examples:

• FRM filter-based PM2.5: ambient air is pulled through a precisely cut size-selective inlet, then through a Teflon filter at exactly 16.7 liters per minute for 24 hours. The filter is removed, conditioned in a lab, and weighed on a microbalance — the gold-standard, but with a 24-hour latency.
• FEM continuous PM2.5: instruments like the BAM-1020 or T640 measure PM2.5 in real time using beta-attenuation or optical methods. Less accurate than the gravimetric FRM but reports hourly.

These instruments cost tens of thousands of dollars, are calibrated and audited on a strict schedule, and are sited per EPA siting rules to represent population exposure rather than worst-case emissions.

Low-cost sensors and the PurpleAir question

The air-quality consumer space has exploded with cheap optical sensors — PurpleAir, AirGradient, IQAir, and a long list of others. They cost a few hundred dollars instead of tens of thousands and can be installed by anyone. They produce useful information, but they're not regulatory monitors. Three things to know about their differences from AirNow:

• Different physics. Low-cost sensors use light-scattering, which is sensitive to particle size, refractive index, and humidity. They tend to overread during smoke events because smoke particles scatter light efficiently; they tend to underread or be unreliable on fog, dust, and fireworks.
• Correction factors are required. The EPA developed a "U.S.-wide correction" for PurpleAir data when used alongside regulatory monitors. AirNow's Fire and Smoke Map ingests PurpleAir but applies that correction. Many consumer apps display raw uncorrected PurpleAir data, which is why their numbers drift higher than AirNow's during smoke.
• Siting is uncontrolled. A PurpleAir on someone's covered porch behind a barbecue grill is going to read very differently than one on a regulatory schedule.

This is why two air-quality apps can show different numbers for the same location: one is using AirNow regulatory data and the other is blending in low-cost-sensor data with or without correction. Smog Report uses AirNow.

The data flow

Here's roughly what happens between a sensor on a roof in your city and the number on your phone:

1. Hourly measurement. The continuous monitor records concentrations for the previous hour and writes them to local storage.
2. Local agency uploads. The state or local air-quality agency that operates the site pushes the data — typically within minutes — to AirNow's central data exchange. Some sites use direct telemetry; some batch upload.
3. AirNow ingests and validates. The data exchange runs basic plausibility checks (impossible negative concentrations, instrument flags, blackout periods) and applies preliminary QA.
4. NowCast is computed. For PM2.5 and PM10, NowCast looks at the most recent 12 hours and weights them based on how much the readings have varied. For ozone, NowCast uses a 12-hour weighted average. The result reflects recent conditions without overreacting to a single noisy hour.
5. AQI translation. The NowCast concentration is fed through the EPA's pollutant-specific breakpoint tables to produce an AQI value.
6. Dominant pollutant selected. Whichever pollutant has the highest AQI wins; that becomes the reported AQI for the location.
7. Publication. The result is exposed through the AirNow public website and the AirNow API.
8. App fetch. When you open Smog Report or refresh a widget, the app queries the AirNow API for the nearest station and displays the result.

Why readings sometimes lag

If the AirNow time stamp on your screen says "9:00 AM" and it's now 10:30 AM, you're not seeing a bug. Several legitimate reasons:

• Hourly cadence. Most regulatory monitors report a "previous-hour" measurement once per hour, with a small delay for the agency upload. The 10 a.m. reading is typically available around 10:15–10:45.
• Instrument flags. If the monitor self-flags a measurement (filter loading, calibration drift), the QA system may suppress the report rather than publish a suspect number.
• Maintenance windows. Monitors are taken offline for routine calibration and audit. Short data gaps on a single station are normal.
• Network outages. Local-agency telemetry occasionally fails. The same station may catch up an hour or two later with backfilled data.
• NowCast smoothing. Even with fresh data, the published AQI weighs the previous several hours. If you compare hourly raw concentration to AQI, the AQI will appear to lag a sharp change.

Forecasts

Alongside observations, AirNow publishes daily forecasts for next-day air quality issued by state and local agencies. These are not produced by AirNow itself; they're produced by trained forecasters at agencies like the South Coast AQMD, Bay Area AQMD, and dozens of others, using a mix of meteorological models, emissions inventories, and human judgment. Quality varies — major metro areas with large agencies put serious science behind their forecasts; smaller jurisdictions sometimes don't issue forecasts at all. The Smog Report forecast view surfaces what AirNow has when it's available.

QA and the difference between "reported" and "final" data

Real-time AirNow data is intentionally fast at the cost of being preliminary. It hasn't been through full QA review. A few times a year, the EPA finalizes the historical record after auditing — comparing FRM filter weights to FEM continuous data, applying corrections, removing flagged data — and publishes the "final" dataset to the EPA Air Quality System (AQS) database. The numbers researchers cite in peer-reviewed papers come from AQS, not real-time AirNow. For consumer use the real-time data is what you want; for trend analysis you want AQS. Both are publicly available.

The takeaway

The AQI you see in Smog Report is the output of a long, well-instrumented pipeline rooted in regulatory-grade monitors. It's not perfect — coverage is uneven, near-real-time data hasn't been through full QA, and the nearest monitor may be miles from your actual location. But it's the most reliable, most rigorously calibrated, and most independent air-quality data available to the public, and it's what most public-health decisions in the United States are actually based on. When in doubt, AirNow is the source of truth.

Primary sources: AirNow.gov · AirNow API documentation · EPA — FRM and FEM · EPA AQI Calculator