How each method works
An optical (light-scattering) sensor draws a sample of air past a laser or LED beam. Particles crossing the beam scatter light, and a photodetector measures the intensity of that scatter. Because larger and more numerous particles scatter more light, the sensor converts the signal into a particle count and, via an assumed particle density and size distribution, into a mass concentration in µg/m³. The whole cycle takes a second or less, so an optical monitor produces a continuous, real-time trace.
A gravimetric sampler takes the opposite approach: it pulls a precisely metered volume of air through a pre-weighed filter for a fixed period, typically 24 hours, then the filter is conditioned and weighed again on a microbalance. The difference in mass divided by the volume sampled gives the concentration directly, with no assumptions about what the particles are. Gravimetric analysis is the reference method behind most regulatory PM standards precisely because it measures mass rather than inferring it.
Where they diverge
The trade-offs follow from the physics. Optical sensors assume a particle density and refractive index; when the real aerosol differs (sea salt, wildfire smoke, or particles that have absorbed water at high humidity), the inferred mass drifts from the true mass. Humidity is the single biggest source of optical error, because swollen droplets scatter far more light than the dry particle would. Gravimetric measurement has none of these assumptions: it weighs whatever is on the filter, so it is unaffected by composition or optical properties.
Gravimetric's weakness is everything optical is good at. A 24-hour filter gives one number per day, with no timing information, several hours after the fact, and only after manual handling in a laboratory. It cannot alert you to a dust event, attribute a spike to a passing truck, or drive a live dashboard. It is also labor-intensive and unsuited to dense sensor networks.
Reconciling the two
In practice the methods are complementary rather than competing. Reference-grade programs use gravimetric (or beta-attenuation, which is continuous and traces gravimetric closely) for compliance, and deploy optical sensors for spatial coverage, real-time alerting, and trend detection. The standard way to make optical data defensible is co-location: run an optical monitor alongside a reference instrument for a period, derive a correction factor for the local aerosol and humidity, and apply it. A corrected optical network then delivers regulatory-adjacent accuracy at a fraction of the cost per point.
The practical rule is to match the method to the decision. If the output feeds a legal or compliance judgment, anchor it to a gravimetric or beta-attenuation reference. If the output drives operations (ventilation control, worker alerts, fence-line awareness, mapping where a problem is worst), optical is the right tool, and its real-time continuity is worth far more than a once-a-day filter weight.
Frequently asked questions
Which method is more accurate?
Gravimetric is the reference for absolute mass, because it weighs particles directly with no assumptions about their composition. Optical sensors are accurate for trends and relative comparisons and can be made accurate in absolute terms through co-location and correction, but their raw reading depends on assumptions about particle density and is sensitive to humidity.
Why do my optical sensor readings spike in humid or foggy conditions?
Water condenses onto hygroscopic particles and makes them physically larger, so they scatter much more light and the sensor infers a higher mass than the dry particle mass. Better optical monitors mitigate this with a heated inlet or a humidity-correction algorithm; if yours does not, treat high-humidity readings with caution and validate against a reference.
Can optical PM readings be used for regulatory reporting?
Not on their own. Regulatory PM limits are defined against gravimetric or federal-equivalent methods such as beta-attenuation. An optical sensor can support a regulatory program once it has been co-located with a reference instrument and its readings corrected, but the reference method remains the compliance anchor.
Should I choose one method or run both?
For most monitoring programs the answer is both, in proportion: a small number of reference-grade instruments for compliance and calibration, and a larger number of optical sensors for real-time coverage and alerting. The reference instruments keep the optical network honest, and the optical network gives the timing and spatial detail the reference method cannot.