Blog/

Satellite AOD vs ground-based PM2.5 monitors: what each one actually measures

A city skyline silhouetted behind a thick layer of haze at dusk.

A beta attenuation monitor sits at one point, pulls air through a filter, and tells you the mass concentration of particles at that exact spot, updated hourly. A satellite pass over the same region measures something else entirely: aerosol optical depth, the amount of light scattered and absorbed by the whole column of atmosphere above that pixel. Both get reported as "air quality," and that's where the confusion starts.

If you run a monitoring network, you already know the BAM's limits aren't about accuracy. It's precise where it sits. The problem is everywhere it doesn't sit. A county with three reference monitors and six hundred square miles of area in between is flying blind on everything between those three points, especially during a fast-moving smoke event or a dust plume that forms over farmland twenty miles from the nearest station.

What AOD actually tells you

AOD isn't a PM2.5 number and converting it into one requires a vertical profile assumption, a boundary-layer height, and usually a local calibration that most agencies don't have the staff to maintain. What a coarse-resolution multispectral pass gives you instead is a column measurement: how hazy the atmosphere is over a given cell, refreshed daily, across the whole grid rather than at three or four fixed points.

A BAM answers "what's the concentration right now at this station." A daily AOD raster answers "where is haze building up across my whole region, including the places I don't have a station." The satellite layer fills the space around the reference data, turning a haze event that would otherwise show up as three disconnected point readings into a picture of what's actually happening between them.

Coarse satellite pass vs reference monitor: matching the tool to the question

The pixel size matters here. At 250 meters or coarser, a single grid cell can cover a stretch of highway, a stubble-burn field, and a residential block all at once. That resolution works for regional context: which part of the service area is trending hazier day over day, where a plume is drifting, whether a haze event is forming upwind of a sensitive monitor before the hourly BAM data catches it. Attributing a reading to a specific source or defending an exceedance at a specific address still calls for the ground monitor, and nothing here changes that.

The daily cadence is the other piece. A reference monitor gives you near-continuous data, but only at its own location. A daily AOD grid gives you one look per day, but across every cell in your region, delivered as a GeoTIFF plus a time-series CSV so you can track a given cell over weeks and see whether a pattern is building or a single bad day.

Used together, the two datasets cover each other's blind spots. The BAM network anchors ground truth at known points. The satellite layer shows what's happening in the gaps between stations, flagging where a haze event might be forming before a monitor ever registers it. Neither one is trying to do the other's job.

If you're staring at a network map with more gaps than stations and wondering what's happening in the gray areas between your fixed sites, Air Quality Proxy is built around exactly that gap: a daily AOD raster gridded over your region, delivered alongside the per-cell time series, so you have a between-monitor view without pretending it's a replacement for the monitors themselves.

Next time a haze event shows up on your network map as three disconnected dots, it's worth checking what the satellite layer was doing in the hours before.

Get started

← Back to the blog