Support from northerners like you keeps our journalism alive. Sign up here.



Wildfire smoke over Yellowknife Bay on August 13, 2023. Emily Blake/Cabin Radio

Q&A: How small sensors transformed NWT air quality monitoring  


Checking the air quality has become routine for many NWT residents after months of wildfire smoke swirling through communities.

But for large swaths of the territory, accessing information on how bad the smoke is outside hasn’t always been so easy.

Until recently, only four NWT communities – Inuvik, Norman Wells, Yellowknife and Fort Smith – had air quality monitoring stations and tracked fine particulate matter, or PM2.5.

A type of air pollution, PM2.5 consists of particles 2.5 micrometres or less in diameter that can travel deep into the lungs and pose health risks.



Last year, the territorial government started an initiative to change the situation, distributing small air quality sensors to communities for free. The easy-to-use sensors, made by a company called PurpleAir, track air quality in real time and display the information on an online map.   

The program was slow to get off the ground but picked up this year when the Department of Health and Social Services got involved, as the CBC previously reported. By early July, sensors had been rolled out in about half of NWT communities – only a matter of weeks before all five regions of the territory came under wildfire smoke advisories.

Now, all NWT communities except Délı̨nę and Ulukhaktok have PurpleAir sensors installed.  

“Never before was I able to track the level of PM 2.5 in small communities,” said Dr Kami Kandola, the NWT’s chief public health officer.



Anna and Joseph Ignacio volunteered to host a PurpleAir sensor (upper left) at their teachers’ residence in Colville Lake. Colin Merz/GNWT
Anna Ignacio and her son Joseph volunteered to host a PurpleAir sensor (upper left) at their teachers’ residence in Colville Lake. Colin Merz/GNWT
Bruce Davidson with a PurpleAir sensor at Prelude Lake Territorial Park. Colin Merz/GNWT
Bruce Davidson with a PurpleAir monitor at Prelude Lake Territorial Park. Colin Merz/GNWT

With exceptionally poor air quality plaguing large parts of the NWT for months on end, the sensors provided invaluable data to residents and officials.

They even outperformed territorial air quality monitoring stations. Having never dealt with such smoky conditions, Kandola said, the regular monitoring stations became blocked up, providing readings that were artificially low.

Information from PurpleAir sensors has made its way onto the territory’s air quality website, where air quality is shown according to the Canadian Air Quality Health Index (AQHI) – a scale that runs from 1 to 10-plus, designed to help people understand what air quality conditions mean for their health.

(The Canadian AQHI scale shouldn’t be confused with the US Air Quality Index, which runs from 0 to 500.)

On Wednesday, Cabin Radio spoke to Kandola about the PurpleAir sensors, how they’ve helped address gaps in air quality monitoring, and how information tracked over 2023’s particularly smoky season will be used in the future.  

This interview was recorded on September 27, 2023. The transcript has been edited for clarity and length. Note that Dr Kandola originally referred to AQI at times but later clarified that the term she meant was AQHI. We have amended this in the transcript to avoid confusion.

Chloe Williams: Last year, the GNWT started distributing small air quality sensors as part of the Community Based Air Monitoring Project, which seems particularly important given the summer that we’ve been experiencing. Could you tell me where the idea came from?

Dr Kami Kandola: The air quality program is under the GNWT Department of Environment and Climate Change. They were federally funded to distribute small air quality sensors, which is what we call the PurpleAir sensors. They started in the fall of 2022, and they were trying to distribute the sensors to communities. But they had very little success. I think, by the time we got involved, there were only six communities that had PurpleAir monitors.



In May of 2023, we already had KFN and Hay River evacuated, which was quite early. We had drought-like conditions in early May, which gave me an indication that we were going to have a very bad wildfire season. Normally, the territory’s air quality program has portable air sensors. I contacted them and said, “Can we borrow your portable air sensors so I can get an idea of what the air quality will be like in Hay River and KFN?” Because the fires were still burning. The machines they typically have are person-dependent and they’re much more complicated. You have to extract the data, it’s very cumbersome. But the air quality program said, “Oh, no, we have something new that’s way better.” And then they introduced me to the PurpleAir sensors.

They’re very small, little units. They almost look like a fire alarm. Basically, through wireless, data can be uploaded every few minutes, and the units can give you an up-to-date, accurate picture of the air quality. I asked the program leaders how many communities have units installed, and they said they had been difficult to roll out. So I said, “Give me the air sensors, because this is going to be a public health issue, and I will get my staff to roll this out in every community.”

As of now, we have sensors located in every single community. We even had one at Prelude campground and one at one of the diamond mines. Two communities are in the process of setting up their air sensors. The remaining two communities are Délı̨nę and Ulukhaktok. In every other community, the sensors have been established and are giving us real-time information.

Never before was I able to track the level of PM 2.5 in small communities. The only places that had air quality sensors or were doing AQHI were Inuvik, Norman Wells, Yellowknife and Fort Smith. So I had four communities, and now I have 31 communities where I can track air quality.

The beauty with PurpleAir is it gives information in real time. It uploads every several minutes. If you go on the PurpleAir map, you can tell exactly at that time what the air quality is and then you can make your decisions.

The wildfire situation this year was one of the worst that we’ve had, even compared to 2014. When the dust settles, I’m hoping to get some dedicated support to look at the data and look at what our exposure was, but that is going to take a few months to actually calculate. It will be something that I’m looking forward to doing in 2024 when I have extra dedicated support.

You touched on this already, but could you tell me about the challenges or gaps in monitoring air quality in the territory before this?

The challenge has always been with small communities. In small communities, I couldn’t get a real-time assessment of the air quality. It’s really important that we have real-time quantification of air quality because there are vulnerable sectors of the communities, like the elderly, pregnant women, infants and young children, people who work outdoors all the time, and then those who have existing illnesses or chronic conditions. They have sensitivities at much lower poor air quality levels than the general public.



For small communities, we had a crude method called the visibility index. We had established in each community a method of looking and seeing if you can notice certain landmarks. That is a very crude method. The air quality division of ECC also had portable air monitors that you could bring to communities, but you needed a person to upload the data and analyze it. So it was very resource-intensive, cumbersome and very costly. PurpleAir sensors are very elegant. They actually call it citizen science because anyone can purchase one and set it up outside their house.

One person wrote to us and they noticed there was a difference between the PM 2.5 measurement from a PurpleAir sensor and an air quality monitoring station, an ECC station. Do you have a sense of what might explain that?

We noticed the same thing, so we flagged that to the air quality division. The explanation was that the permanent air quality sensors had never dealt with this high a load of particulate matter and they got blocked up. So their readings were artificially low. When that happened, they took those sensors offline, and they recently calibrated the Yellowknife one. Its readings are now in line with what we’re seeing with PurpleAir. What I would tell you is, in that situation, go with the PurpleAir monitors, because they are accurately reflecting what’s happening outside. Right now, there shouldn’t be a discrepancy. If there is a discrepancy, flag it to the air quality division and they will calibrate the machines or pull them offline until we fix it.

Could you tell me more about how the PurpleAir data has been used so far this summer?

With the PurpleAir data, we issued several wildfire advisories from my office. Initially, we would rely on Environment Canada to put out an alert and say air quality is poor. But because we had the PurpleAir monitors, we were able to take advantage and issue advisories based on what we were seeing.

We also want to train the public to use information from the PurpleAir monitors. Throughout the summer, practically every community had poor air quality. The best tool we could provide is actually training people to look up readings from their PurpleAir monitors several times a day. In every advisory, we put links to PurpleAir monitoring, and we did a number of interviews around the monitoring.

I think the most important message is that the sensors are in real time and specific to your community. You should be going to that first.

What the public may not have known is that you can play with the parameters, so you can actually change the map to show AQHI. When there was that particularly eerie situation last Saturday, we felt that it was going to be a lot easier if we could convert PM2.5 to the AQHI and then embed that on our webpage. Prior to that, we were giving links to the AQHI, to PurpleAir, to the effects of wildfire smoke on health, but they were all separate links. What I realized is, now and going forward into the future, it’s important to put out the PurpleAir map with AQHI, and below that say what to do with different AQHI readings, who’s at risk, and what they may experience.



We took all those links and consolidated them into one landing page. The information was always there, but it was important to synthesize it and simplify it.

OK, so that’s the update mentioned in the press release that came out on Tuesday? That’s what was changed?

Yes, it was not new information. It was just taking all the information, and instead of having it in three different sources, putting it all on one landing page. It’s more for the future as well, because next year we still have to anticipate a wildfire season like what we’ve seen. Maybe it’s something we have to adapt to, so I’d rather start developing the tools now and have that information now.

We also have established the need for clean air shelters. What you’ll notice in Yellowknife is, when the air quality is particularly bad, they open up the fieldhouse for people who don’t have access to air purifiers or a clean indoor air environment. One expansion we’re looking at, going forward, is for each community to identify a clean air shelter.

Our office has provided outdoor air sensors that can tell you in real time what the quality of air is outdoors. But what we are going to offer communities is an indoor air sensor. They’ll be able to know in real time what the quality of air is in their clean shelter. It may not be perfect, but if you have indoor air that’s at 100 and the outdoor air is at 600, it’s a significant improvement. That’s what we want to roll out this year. And then we want to investigate how to get cheap air purifiers.

You mentioned previously, potentially some larger-scale analysis with the data that’s been collected this past summer. Can you tell me about those plans? How do you think the data will be used in the future?

There was a study done on the 2014 “summer of smoke,” and there’s a nice BMJ article that was published in 2021. They did a study looking at cardiorespiratory impacts of a very severe, prolonged wildfire season in the Northwest Territories. I believe that at that point, they had 3.5 million hectares burned. This summer, we had about 4 million hectares burned, so it’s a little bit higher. This study took a long time to set up and required significant funding, but it is the best study to show the impact.

They noticed ER visits for asthma doubled, and they saw an increase in pneumonia. They also noticed that it was the most vulnerable populations, like children and Indigenous individuals, who were impacted. We anticipate that we have the same issues right now. That study took years, and it required a significant amount of time and data. What I’m hoping to do, what we can do at minimum is, once our wildfire season has abated, we can try to get the number of hours people were exposed to what level of smoke – to at least quantify our exposure during this spring and summer. We have never been able to access that level of information because we didn’t have PurpleAir sensors. The caveat is that we didn’t have PurpleAir sensors in all of the communities until recently. But for the majority of communities that were experiencing severe smoke, we were able to capture the information.



Being able to repeat the study, in terms of pulling in health data, we would need to get access to funding. But at minimum, we need to document and track what the actual exposure was during this period. We need to look at exposure every day, throughout the whole period, including times when we were evacuated. These will be exposures that firefighters or workers could have experienced. We also need to look at the exposure excluding times when there was an evacuation in place, so that would be for the general population.

So the idea would be to hopefully, eventually relate that exposure to health outcomes or health conditions?

First is to track exposures, and I think we should be doing that regularly using PurpleAir monitors. Then if we can get additional funding or if researchers are interested, it would be good to then track that to health outcomes. But that will take a long time. The 2014 study took several years. If we wanted to repeat that, we would need interested researchers and some funding. But at a minimum, what we can do with some dedicated resources is at least quantify the amount of exposure that we had over the spring and summer.

At this point, are there lessons that you think we can draw from the really smoky season we’ve been seeing?

We need to have dedicated clean-air shelters. Something we can do to get prepared for next season is for the communities to identify clean-air shelters that at-risk people can use during the day.

The second area that we want to look into is how to mobilize funding for cheaper air purifiers that everyone can access. During the day, vulnerable people can attend air shelters and breathe better air when conditions are smoky. But at home, people need to be able to access air purifiers to keep their air quality at an improved level.

If we anticipate that wildfires are increasing in intensity and frequency – not just in the Northwest Territories but across Canada, across the world – then how do we better prepare ourselves for the next season? What can we do to improve air quality in our homes or in our communities? How do we provide equitable access? I know, for myself, I had my air purifier going. We had an air conditioner, too, because it was a hot summer. The air quality that was in my own house was at a very acceptable level. But there are vulnerable populations, there are vulnerable communities. It’s important that everyone has the ability to be able to improve their own air quality. That’s something to be looking at next year as we anticipate the next wildfire season.

Is there anything else you wanted to add or that you think people should know?



We’ve been trying for years and years to get air quality information in small communities. It was very costly, both in terms of human resources and equipment. These air sensors are not very expensive, about maybe $350 per machine. Just by having this simple technology, it far exceeded equipment that costs thousands and thousands of dollars. So it does give me hope.

The fact that we could provide people a quantification of what they’re experiencing, I feel like it gives them better control of what they can decide to do because they know exactly what the air quality is. I know so many people this summer who were consulting the PurpleAir monitors several times a day, checking it out even during the evacuation. It was such an excellent tool.

There’s hope that with research and knowledge, we can continue to improve our ways to adapt to wildfire seasons by making things available that would have cost thousands and thousands of dollars but are now somewhat simpler. I’m hoping that this same type of knowledge and expertise can be used for very low-cost air purifiers that everyone can access.