During a recent period of excellent air quality I noticed that with the US EPA conversion, the Purple Air map was showing some AQI values of negative one (-1). That seems like an error in the conversion formula. I wonder if the same error accounts for the difference between the EPA Air Now AQI values and the Purple Air map AQI values.
Thank you for your observation. This is an issue that we are aware of with the US EPA conversion. When at a very low AQI score, and using the US EPA conversion, the map can display negative numbers. This is obviously not correct. We are looking to implement a fix in the future.
Additionally, this would probably not explain differences between data on the PurpleAir Map and the AirNow map. It is important to first note that the EPA does display PurpleAir data on their Fire and Smoke map, however the base AirNow map only uses official FEM/FRM sensors. The core difference between the two maps is that the readings are coming from different sensors.
Thanks Kieran. I see my sensor on the Purple Air map as well as the AirNow Fire and Smoke map (https://fire.airnow.gov/#).
If you are looking at your sensor on both the PurpleAir map and the Fire and Smoke map, the core difference is that the Fire and Smoke map automatically applies the US EPA conversion. If you want the two maps to match more closely, we recommend using the US EPA conversion on the PurpleAir map. For more information on the differences between maps, check out this community article.
Sorry…should have mentioned that I use the Purple Air EPA conversion.
The values are fairly close.
Purple Air: 28
Air Now: 33
I think there is an impact from a different averaging or perhaps the timing of the sample.
At low values, the EPA conversion can give a small negative value because of their RH correction. No real reason to do anything about it; 0 = -1 =excellent air quality.
I DO notice that, if you use the Real Time US EPA for PM2.5, those numbers tend to be lower than the AirNow values for US EPA PM2.5. Keep in mind that the Air NOW is updated hourly, and is an average of ALL the 1-minute readings taken at that government sensor. Since it is an AVERAGE, it is likely that the hourly average won’t be negative, unless there are multiple 1-minute readings in that average, making the total hourly average negative. Negative AQI should be rare. I HAVE seen real-time AQIs on Purple Air’s map show zero on those rare clean air days, but not negative. I have yet to see a negative sensor reading. If I see negative readings on the map, I will certainly post the fact I saw them. It would be interesting if I saw them. If you see negative readings, switch to the 10-minute real-time average, which is still updated every 2 minutes; but that’s an average of the last 10 minutes prior to the current time. It is a “rolling” average of 10-minute readings updated every 2 minutes.