I have noticed some concerning inaccuracies in the outdoor Purple Air sensors the last couple of months. I live in Salt Lake City, which traditionally has poor air quality in the winter with the “inversion,” the summer with high ozone levels, and sporadically in the spring and fall with dusty wind storms. On the last couple of poor air quality days, the Purple Air sensors still show the PM2.5 as low, whereas the official and calibrated EPA sensors show elevated PM2.5. Visually, there is definitely an elevated level of particulate matter in the air, so I am inclined to believe the expensive and calibrated EPA sensors versus Purple Air. The PA sensors used to overestimate PM2.5. So when the EPA sensors were showing, for example, a raw value of 12, the Purple Air sensors might show 15. This is a well documented phenomenon and Purple Air has published many conversion factors to try and correct this problem. Was there a firmware update that went out to all the sensors, and now Purple Air is vastly underreporting the actual PM2.5 in the air? This is a huge problem because many schools, community centers, churches, and other rely on various sources when determining level of outdoor activity. Many of those sources pull data from Purple Air. You have to know which sensors are the EPA sensors near you (Rose Park and Hawthorne for me) for comparison. Look at the data from today’s dust storm and see just how underreported the outdoor Purple Air sensors have become. I fully understand there are differences in when readings are taken, and raw value vs AQI, but the difference today (and other days in prior months) is worrying.
This is a screen shot from AirNow showing a PM2.5 raw value of 37 (based on the EPA sensor at Hawthorne)
Can someone from Purple Air please explain what is going on? Something is wrong and this community deserves to know before we make outdoor decisions based on this inaccurate data.
I am not sure if this is applicable, but if an old sensor was switched out with another one, see this post below (posted by Adrian on March 9th), it may address some of your questions.
We believe the answer is to do with the PM10 and it being a dust storm. The Plantower PMS5003 is not as good with larger particles like PM10 and to some extent smaller particles towards PM2.5 sizes. Since this event was dominated by PM10, the official sensors were picking up larger 2.5um to 10um particles and our sensors were missing some of those. If it was smoke for instance that has particles ranging from sub-micron up to 2.5, the Plantower fares better.
We do have a new version of the Plantower being used in the PA-II-FLEX and there is a possibility it will do better but that is to be shown as time goes on.
In summary, these sensors are not as good with a particle size distribution that is larger but are better with smaller particles.
I hope this helps and please let me know your thoughts?
Hi Adrian, thank you for the response. I understand what you are saying, but I do not recall there being such a disparity between the official sensors and the Purple Air sensors in the past. We have been using Purple Air for about 3 years and the Purple Air sensors almost always seem to overestimate PM2.5. I understand there could be a difference when there is a high level of PM10 and the Purple Air sensors may not be as good at picking up those larger particle sizes. I would be very interested in looking back at the data prior to 2022 to compare the official sensors during high levels of PM10 with Purple Air sensors. Furthermore, wouldn’t it be true that the PA sensors in service before the Plantower device show slightly different data than the newer ones?
I have attached a few pictures from today. This time there does not seem to be a PM10 event, and yet, the Purple Air sensors are way understating the PM2.5 compared to the official EPA sensors. In the picture, the numerous green squares are the PA sensors, and the larger, yellow circles are the EPA sensors. The circles are all showing elevated PM2.5 while the PA sensors show much lower values.
(Edit- the community website is not uploading the picture, but I will email it to you)
The particle data was incorrect again today, so I took a couple of new screenshots. Again, the circular government sensors are orange due to PM2.5, while the PA sensors stay green. What really concerns me is that decisions about school outdoor times can be made based on PA data because the data is so widely populated across various websites. But as of right now, I cannot really trust the PM2.5 readings PA is posting.
