The dependence of PurpleAir sensors on particle size distribution and meteorology and their relationship to integrating nephelometers

Abstract

The National Park Service (NPS) has a need to measure real-time PM2.5 and haze levels across its many park units to inform the public of unhealthy air and track visual degradation of park scenic vistas. To fulfill this need, the NPS deployed PurpleAir monitors (PA) in several park units. PA monitors are inexpensive monitors promoted as particle counters capable of measuring particle size distributions. Each PA is equipped with two Plantower (PMS) particle counters and a temperature, relative humidity (RH) and pressure sensor. In 2022, the NPS initiated a field study to evaluate PA limitations, uncertainties, and how the measurements relate to particle light scattering. The study was conducted in Fort Collins, CO where five nephelometers, particle sizing instruments, semi-continuous ion concentration measurements, and seven PAs were deployed. One nephelometer and two PA's sampled air from an RH controlled chamber with RH ≤ 40%. The PA temperatures were accurate, but the PA RH measurements were systematically underreported RH by ∼28% and it is recommended to scale these data by 1.35. The PMS reported size distributions are shown to be invalid. PA instruments mounted without protection from wind exhibited biases of ±160% for wind speeds exceeding 5 mph. Hourly CH1 data were precise, with errors of ∼100% at low concentrations of 30 #/dl and 15% for concentrations ≥1000 #/dl. Normalizing all PMS sensors to each other reduced the errors to 20% and 8% for low and higher particle concentrations respectively. The reported total particle count, i.e. ≥0.3 μm channel (CH1), was biased low, and the bias decreased linearly with mean scattering diameter (MSD). The ratio of measured ≥0.3 μm to PA CH1 channel increases linearly with MSD. The overall normalized mean deviation (NMD) between the two measurements is 121.8%. The CH1 channel also increases as a function of RH primarily due to the growth of hygroscopic aerosols. On average, the relationship between the atmospheric scattering coefficient (bsp) and the PMS ≥0.3 μm CH1 channel is described by the equation bsp = 0.015 × CH1 for MSD levels less than 0.35 μm. The PMS CH1 response decreases linearly with MSD. The NMD between ambient PMS derived bsp and measured ambient bsp is 27.0%. Recommendations are made for PA network deployment.