Air Quality Forecasting of Fine Particulate Matter with Purple Air Monitors: Monitor Calibration and Model Validation

During the winter months many New Hampshire residents burn wood to heat their homes creating particulate matter in the air. Fine particulate matter (2.5 micron diameter) can bypass the human body’s natural defenses and can be harmful to human health. Breathing in this fine particulate matter (hereafter PM2.5) is irritating for the lungs and can be related to increased risk of asthma, bronchitis, and pneumonia.

In this poster we will present particulate matter observations that were obtained through using inexpensive Purple Air sensors around the Keene NH Valley. The use of these inexpensive sensors is gaining traction in communities that want to monitor particulate matter in a finer spatial scale. Monitors can be stationed in various locations throughout a community and be connected to local Wi-Fi for the data collection. Before the sensors can be used to collect particulate matter, they must be calibrated to the governmental standard BAM dataset. In the past, this has been done with a straight linear or polynomial fit to describe the correction. However, it is seen that the Purple Air sensors do well at low values of particulate matter but tend to over predict the higher values seen during temperature inversions. The same temporal structure is observed, but the higher PM2.5 values are exaggerated. We will show that the Purple Air Sensors are sensitive to local weather conditions, and that the inclusion of the weather conditions provides a better, tighter, calibration with less spread and lower error.

Normally wood burning smoke will rise into the atmosphere, though when temperature inversions occur the smoke becomes trapped. A temperature inversion is a reversal of the normal behavior of temperature in the troposphere in which a layer of warmer air sits on top of a layer of cooler air on the Earth’s surface. This effectively traps the wood smoke generated PM2.5, increasing levels that humans will breathe. Normally these temperature inversions can be predicted by large weather forecasting models. However, Keene, New Hampshire is located in a valley that is surrounded by large hills and Mount Monadnock, which create small scale, low altitude inversions that are not captured by the larger models. We can use the elevated PM2.5 to show when temperature inversions occur because there will be a large and sustained increase of PM2.5. Once we have identified the temperature inversions in the PM2.5 we will use the local weather conditions to build an empirical model of when temperature inversions occur. This empirical model can then be used to predict future temperature inversions. In addition, we know that Keene, NH is not the only small valley that exists and would be subject to temperature inversions trapping PM2.5. These valleys could follow the methodology we outline here to predict when temperature inversions will occur to improve their local air quality forecasts.