A team has proposed a calibration method for low-cost nitrogen dioxide (NO₂) sensors that adjusts statistical distributions rather than individual data points. The technique aims to correct the drift and dispersion these devices suffer when operating under real-world air quality conditions, where humidity, temperature and interfering gases degrade their accuracy.

Low-cost electrochemical sensors have proliferated in urban monitoring networks because they enable a measurement density impossible with reference stations. Their problem is that factory calibration drifts over time, and classical corrections, based on pairs of measured and reference values, do not properly capture uncertainty or biases that depend on environmental conditions.

The alternative presented consists of calibrating the full distributions of the readings against those of the reference instrument, rather than forcing a one-to-one correspondence between points. In this way, the model absorbs the sensor's intrinsic variability and maintains its reliability across concentration ranges and changing environmental scenarios, according to the authors of the work.

The approach points to a clear practical goal: that dense networks of cheap sensors can be reliably used to estimate population exposure to NO₂, a key urban pollutant linked to respiratory and cardiovascular problems. If distribution-based calibration becomes widespread, it could reduce the need for frequent recalibrations against reference stations and improve comparability between devices.