Wildfire Smoke Adjustment Factors for Low-Cost and Professional PM2.5 Monitors with Optical Sensors

Published: 22-11-2019| Version 1 | DOI: 10.17632/7symfdcmkc.1
Woody Delp


This data set was used to derive adjustment factors of optical based particulate monitors to specific wildfire events. The data is broken into two broad groups 1) LBNL site analysis, and 2) PurpleAir analysis. The LBNL data is from a lab space connected to outdoors during the 2018 Camp Fire. This data includes reference / professional grade monitors along with several low cost consumer monitors. The reference monitor was a 1405DF TEOM-FDMS, the professional monitors were a ThermoScientific pDR1500, a TSI DustTrak 8530, and a GRIMM mini-WRAS 1.371. The low cost monitors were from PurpleAir (PA-I), Wicked Devices AirQuality Egg2, IQAir AirVisual Pro, and a pre-production unit from eLichens. The PurpleAir set includes publically available data downloaded from the PurpleAir website along with data from the closest available regulatory air quality monitoring station (AQS). The set includes data from three different 2018 fires: the Camp Fire and Car / Mendocino Complex fires in California, and the Pole Creek fire in Utah. From the Camp Fire there are data from 53 monitors near 12 different AQS from Sacramento down to San Jose. The smoke from this fire persisted for nearly 2 weeks. For the Carr / Mendocino Complex fire there was a single PA monitor collocated with a AQS at Red Bluff California. In Utah there were several days of diurnally impacted smoke in Spanish Fork from the Camp Creek fire along with a single nearby PA monitor. The LBNL site data is arranged as hourly average data. The PurpleAir data includes the raw downloaded data (80s) as well as hourly running 4 hour averages. The data allows the exploration of the wildfire smoke time varying adjustment factors for optical PM devices. It is also useful for exploring how general adjustments from one fire can be applied to data from another to reduce the bias inherent with unadjusted optically derived PM data.