Wildfires in Canada are once again impacting our air quality, and while the AQI is not nearly as bad as it was two years ago, when it peaked in the mid-400s, caution is still warranted. This is an update to a post I made in 2023 with links to useful information and resources.
Overall, an AQI of 150 is the tipping point. Below that, things are generally OK for most people, whereas above it, everyone should reduce “long or intense activities” and take more breaks.
The most official site to check is the EPA’s AirNow.gov, which also usually includes a forecast for the next day and a smoke map that shows unofficial sensors. There’s also an app version for iOS and Android. AirNow also offers information about using the maps during wildfires and an essential activity guide.
Those unofficial sensors come from PurpleAir, which collects local particulate readings from people with their own sensors. I find its crowd-sourced conditions map extremely helpful.
Although not much is said about masking, an AirNow page about dealing with smoke in the air does support the wearing of N95 masks. The Tompkins County Health Department has also recommended N95 masks, though running in one might be too annoying.
If you are healthy, you’re usually not at a major risk from short-term exposures to smoke. Still, it’s a good idea to avoid breathing smoke if you can help it. Everyone should take the steps below when wildfires are present.
Use common sense. If it looks smoky outside, it’s probably not a good time to mow the lawn or go for a run. And it’s probably not a good time for your children to play outdoors.
Stay healthy out there, and if you’ve run across other useful guidance for athletes, please share!
One note about the PurpleAir map. By default, it shows uncorrected numbers that are often higher than the EPA numbers. The EPA has a correction factor that can be applied to the PurpleAir map in its settings that tweak the numbers somewhat to account for the fact that the the low-cost PurpleAir sensors have some known issues with particulate readings, particularly at higher levels.
I would like to add to bullet #3–as someone who studies wildfire smoke, I only recommend https://fire.airnow.gov/ for crowd-sourced data for two reasons:
Off-the-shelf sensors (e.g., PurpleAir) can become extremely inaccurate at high concentrations of wildfire smoke. EPA’s fire.airnow.gov incorporates the same sensors, but has an empirical correction factor so the values should be closer to reality.
EPA’s fire.airnow.gov shows all the PurpleAir sensors plus all the sensors from some other companies and all the official air monitoring stations. The official air monitoring stations (big dots on the map) represent the most reliable measurements (these give the official values, accessible at https://www.airnow.gov/)), which can help double-check that the off-the-shelf sensors (small dots) are accurate.
In other words, https://fire.airnow.gov/ has more data & more reliable data than any other site I’ve come across so it remains the only one I recommend and use personally.
Thanks @stairmaster—great to have someone who actually knows about this stuff. Perhaps you can answer some questions I’ve had about the sensors.
I saw the bit about the off-the-shelf sensors becoming inaccurate “at high concentrations of wildfire smoke” too, which was why I added the comment about the EPA correction factor. But what counts as “high” in that phrase? And how successful is the EPA correction factor at taking that into account?
I ask that in part because the official EPA sensors are quite far away: roughly 45 miles from both the one in Addison and the one in Syracuse. Given that the unofficial monitors show that there can be significant differences between relatively close areas at air quality boundaries (and they don’t necessarily move all that quickly), I’ve preferred the hyperlocal sensors for getting a feel for what our local air quality is, rather than that of Addison or Syracuse. Does that make sense?
The EPA AirNow iPhone app’s reading seems to use Syracuse as our “local” number, which seems wrong in general, given that it’s northeast of us, whereas Addison is southwest. I’d expect winds and thus smoke to come from the west. Any idea why this, or if we should prefer the Addison sensor?
The AirNow app’s forecast is actually provided by the NYS DEC—is it using the EPA sensors, or something else?
Overall, I prefer the PurpleAir map (or actually the Paku app on iPhone, which remembers location properly) because it shows the numbers for each sensor, whereas the EPA’s smoke map shows only the color and you have to click each sensor to find out the number behind it. Given that there’s quite a range between 101 and 149, but both would show up as yellow, I find the numeric values more helpful.
While riding the ElliptiGO and wearing a mask because of yesterday’s 120 AQI, I went by a guy in his yard who was smoking a cigarette. I immediately wondered what the AQI of being near a smoker is. Unsurprisingly, it’s off the charts—easily 500+.
Although that’s horrible, in some ways, it’s also reassuring. We’re worrying about what to do with levels around the 100-150 range for an hour or two over a few days, but as much as smoking is undeniably bad in every imaginable way, smokers don’t keel over right away even with living in insane AQI conditions. So while it’s always best to avoid the bad air, it’s not as dire as it might seem.
FYI, campfires are also essentially bad AQI generators—100-500 AQI depending on the conditions and how close you are.
Thanks Adam, your “natural experiment” is indeed great context!
Here is my recommended use cases, more specifics on your questions below:
First, the caveat that this isn’t medical advice, so if your doctor is saying something, do that!
Off-the-shelf sensors (e.g., PurpleAir) are best for getting real-time, hyperlocal trends (e.g., should I run right now or wait). While the specific number might be wrong, it’s going to be consistently wrong so you can see if the smoke is heavier or lighter right now. I personally generally look at AQI color rather than numbers because I feel that the specific numbers hide nuance (see details).
Official monitors can give real-time data, but are often far away. In a regional smokewave, usually all the sensors inside a smoke plume have similar readings, so that can often be good enough, especially when compared to more local off-the-shelf. This can break down out West where there are deeper valleys that trap air differently, but in NY in my experience in smoke waves I find the local sensors typically agree with the official monitors, at least in AQI color (the main thing I look for).
Note that technically, AQI means 24-hour average values, so if something is “real-time” it is probably a “Nowcast AQI”. The 24-hour average (e.g., if a day is assigned a value) averages over that whole time and can be really wrong (either high or low) for a specific time period on that day. This is mostly sorted out now, but occasionally a value is not what someone is expecting due to this difference.
Forecast is going to be 24-hour average, and right now we only get that ~the afternoon the day before. I generally interpret it as am I inside or outside the smoke wave, again noting that specific times of day can be much higher or lower than the daily average.
What I currently personally do for decision-making about outside activities:
Look at https://fire.airnow.gov/, evaluating the local sensors (small dots) and further away official monitors (big dots). If anything is showing orange or higher, I’m probably not running. This is based on personal experience–I just don’t feel good exercising in those conditions, and I feel I get the best data by combining the strengths and weaknesses of all the local sensors.
There are several components of AQI–fine particles and ozone being the key. For wildfire smoke, the main concern is the concentration (mass per volume) of particles with diameters <2.5 microns.
“Official” monitors are actually measuring that concentration. This is expensive and is mainly done by EPA and its partners (shoutout to NYS DEC!). “Off-the-shelf” sensors usually use scattered laser light to estimate the concentration empirically. Therefore, if particles are lighter/darker than expected or have odd shapes (both true with wildfire smoke) then the laser estimation can be off. This link describes the research on correction factors and how bad the estimates get: AirNow Fire and Smoke Map: Extension of the U.S.-Wide Correction for PurpleAir PM2.5 Sensors Webinar Archive | US EPA. Seems to show off by 2x at low concentrations, and worse at higher concentrations and with higher humidity.
A key reason that I look at AQI color rather than number is that those were developed for general air pollution (think car/factory exhaust) and wildfire smoke can be something like 10x deadlier for a given concentration. It really matters what is in the smoke–if it’s burning houses/cars then you might be breathing toxic organic and metallic compounds; if it’s “just” burning vegetation then the health impacts might not be as severe.
AirNow just assigns the closes monitor “as the crow flies” (I think). Again, I’m usually looking at all available data in my area to make the best decision rather than relying on any one sensor.
Ooo, you are my new best friend for the day, @stairmaster. Wonderfully helpful detail about a topic that has been difficult to wrap one’s head around.
I now understand the need for the EPA correction much better, and I see why you’re glossing over the numbers in favor of color. Previously, I had thought the colors were just a way of dumbing down the data for an audience that wouldn’t relate to numbers, but I get that it’s actually a better way to combine multiple types of data now. I hadn’t thought about the different types of smoke, either, but it makes sense that it’s mostly trees and vegetation, that’s less nasty than if it’s houses and cars.