Winter in Yerevan is a period of temperature inversion. Due to the terrain, colder air sits at ground level while warmer air stays above. This "lid" stops atmospheric mixing, and everything emitted by cars and boilers accumulates in the surface layer. When sensors show 200–300 μg/m³, there's more behind that number than just smoke: it's the result of instrument physics and chemistry happening right in the air.
Gas Heating
Yerevan has hundreds of thousands of apartments with gas boilers. Gas burns with almost no visible smoke—there's little "primary" soot. But the main problem is nitrogen oxides (NOx).
In the atmosphere, NOx converts to nitric acid (HNO3): first through oxidation involving O₃ or OH radicals, then through several intermediate reactions. HNO3 then reacts with ammonia (NH3), which exists in any city's air, producing ammonium nitrate—fine PM2.5 particles. This process happens after the chimney, in the air above the city, and its intensity depends in part on humidity.
Fog
At high humidity (noticeable at RH > 75–80%), liquid-phase chemistry accelerates. Fog droplets dissolve gaseous precursors—NOx, SO2—and speed up their conversion to salts: ammonium nitrate and sulfate. In parallel, hygroscopic growth occurs: particles already present in the air absorb water and increase in mass. All this together leads to rising PM2.5 concentrations—not just on the sensor screen, but in the actual air.
Why Sensors Can "Lie" in Humid Weather
Optical sensors count particles scattering a laser beam. But a water-swollen particle scatters it differently than a dry one: the device "sees" it as heavier. Additionally, ultra-fine particles that aren't detected in dry weather (the visibility threshold of most budget sensors is around 0.3 μm) become larger after absorbing water and enter the field of view.
Yerevan's municipal network operates on Clarity Node-S stations. Inside them are Plantower optical particle counters. For PM2.5, they don't measure mass directly but count and calibrate laser beam scattering, which at high humidity produces systematic upward bias. Clarity compensates for this with cloud-based QA/QC algorithms with corrections for weather factors, but no optical instrument can completely eliminate error.
Window Effect
Many private sensors in Yerevan are mounted on windows. Glass slightly heats the air at the surface—about 3–4 °C relative to outdoor air. With this heating, relative humidity inside the sensor drops: for example, from 70% to ~55%. Particles become slightly drier, and the systematic upward bias is slightly less than for a sensor directly outdoors. This doesn't mean the readings are more accurate—it's just one of the artifacts worth knowing about.
Conclusion
Fog and high humidity don't clean the air. They intensify secondary aerosol chemistry and inflate already existing particles. The pink-gray haze over the city under inversion conditions isn't just poor visibility, but real aerosol load. Even accounting for methodological sensor errors, such conditions mean increased health risk.