Zombie Fires: A Novel Detection Method That Is Environmentally Sustainable and Cost Effective

Known as one of the world's most undetectable fires, zombie fires (also known as holdover and overwintering fires) have become more and more frequent over the last few years in Northern Canada.

In 2023, my family and I were evacuated from Yellowknife, NT due to severe smoke and wildfires in the area. Because of this I learned the struggles and uncertainties of being forced to evacuate from your home without having any control over the circumstances. Ever since then I have had an interest in wildfires but not known what to do with it.

So, when I found out what zombie fires were, I was hooked. If you are unaware of what zombie fires are, they are forest fires that stay alive under the snow during the winter and remerge in the spring sometimes even worse than before. They typically occur due to dryness in the area. These fires are also harmful to the environment, so if I could make something to detect these, that could me my part in helping protect the environment.

With climate change getting worse, early detection for fires as dangerous as these could be crucial in the future. Not just for preventing people from being evacuated, but also to help protect the animals and the ecosystem. While forest fires are getting acknowledged for the destruction they cause, not many people are aware of what zombie fires can do and the people who are aware often overlook just how dangerous they could become.

To make sure I knew how to create the detector, I emailed one of the regional climate change officers about doing an interview. After she agreed, we sent a few emails back and forth with questions and answers. With a mix of the knowledge she gave me and some reliably sourced journals and websites, I finally knew enough to start building the detector.

After attempting to make many smoke detector models that didn't work, I finally found one model that looked promising. It was one that I could re code to work how I wanted it to. Unfortunately, the model had some parts that I didn't have. Luckily my school's robotics club lent me some of the parts I needed, but they didn't have the part that actually sent the alerts.

Because I live in an isolated area, I had to order the part. While it was coming in, I looked for different aspects of the detector to complete. Because the detector would have to withstand the cold for long periods of time, I did an insulation experiment to see which insulator could keep the heat inside for the longest. I tested items such as spruce needles, insulation foam, and window insulation.

After the parts came in, I began to design the prototype. I found a recyclable, cold proof, container that I decided to use as the outer shell of my detector.

After creating the new code, I put all the parts together and inserted it into my model. Next, I placed the parts in the shell, I put them in spots that were visible so I could see if they were working. To finish it off, I wrapped the insulator all around the outside of the shell. After the detector was complete, I was ready to test it.

The first test I ran was to make sure that the detector could detect smoke and send alerts. To do this, I put a match directly below the detector and watched the serial monitor to see if it detected anything. Luckily it did! An alert did send to my phone, but unfortunately it took around 5 minutes for me to actually receive it.

Since I tested the detector while it was plugged into a computer, my next step was to see if it could work when it was not plugged in. To test if it could still work, I ran the same test from my front yard. This not only tested if it would work unplugged, but also if it could still work in colder climates. Thankfully the detector did work, and the alert sent a little faster than the first time!

Because my detector is supposed to be used in isolated areas, I next had to test to see how far from my phone it could send an alert from. To test this, I programed the detector to send an alert even when it detects the tiniest amount of smoke. This meant if an alert didn't send, the SMS system would be the problem. So, I tested the detector from my front yard, my back yard, 100 meters out of town, 500 meters out of town, and 1 km out of town. Unfortunately, it only worked 100% from my front yard, and 50% of the time from my back yard. I didn't get any of the alerts from the other spots which was a little discouraging. But now I know to look into more data-based SMSs in the future.

My next test was to see how far away the smoke has to be from the detector to be detected. I ran the test inside my house to make sure there was no wind to consider as a variable. I started testing it at 5 inches and kept going down by one until it worked. Unfortunately, it only detects from 1 inch away. That might not be how I wanted it to work, but it's important to realize that there will always be wind pushing the smoke. Another important thing to consider is I only used one match for most of these experiments so that I could simulate the tiniest zombie fire, so depending on how much smoke it produces, my detector may be able to detect from further away.

When I started this project, I imagined the detector being on the ground, but after running those other tests I realized that smoke doesn't go outwards it goes upwards. So decided that I would run one last test, finding out what distance off the ground is best for detecting smoke. Starting at 17 inches, I did 3 tests for each distance using 5 matches each time. After testing 6 different distances, I found that 55 inches off the ground worked the greatest number of times.

My project's end goal was to make a reliable, sustainable, cost affective zombie fire detector, and I did! Although, there are still many parts of it that need improvement before it can become a real product. For example, the SMS system doesn't work from out of town, or the detector only detected smoke consistently if it was at a distance of 51 vertical centimeters. However, some tests worked really well, such as it works in the cold, the shell and insulation are very effective, as well as it is able to detect minimal amounts of smoke. Plus, the majority of the detector can be recycled once it doesn't work anymore.

It's important to know that 2024 had the most zombie fires ever detected in NWT history with a total of 24. That number may seem small, but with climate change getting worse every year, it's safe to say that number will continue to rise in the future. I have shown that a new approach to monitoring zombie fires - using smoke to detect them - is a sustainable and cost-effective possibility for the future. Finding a way for the detector to communicate and send a message from a remote area was my barrier.

Even making a first model to try to stop zombie fires is an important first step towards my goal of helping the environment.  My project isn't just about making a zombie fire detector; it's also about spreading awareness and inspiring others to think outside the box to make solutions.

During my testing phase, I reflected on what improvements my detector will still need in the future. The most important improvement is getting an alert system that works in remote areas and is able to contact the organizations that fight wildfires. There are a few other minor improvements too, such as looking into a more sensitive smoke detector and potentially finding a new software that allows me to create a more complex code.

I know that if I put in enough work, my detector can become a real product that doesn't just help northern communities, but also all of Canada.

I would like to thank the East Three Secondary School's robotics club for giving me some of the essential part for my detector.

Another person I would like to thank is the climate change officer who did the interview. They gave me some of the key information that helped move my project forward.

I would also like to thank Emilie Leneveu, Shawn Feener, and Jenn Parrott for guiding me through this process, it would have been way more difficult without their help.

The last people I would like the thank are my parents, I'm so thankful for all the support they gave me throughout this project.

Arduino LLC. (2015). Arduino projects book. Arduino LLC.

BC Wildfire Service. (2024, June 11). Overwintering fires. Province of British Columbia Wildfire Service. https://blog.gov.bc.ca/bcwildfire/overwintering-fires/

CBC News. (2024, February 21). Scientists track ‘zombie fires’ to predict where they’ll rise from the earth. CBC. https://www.cbc.ca/news/science/zombie-fires-1.6032452

CBC News. (2023, December 19). ‘Zombie fires’ in Canada: How wildfires smouldering through the winter could make fire season worse. CBC. https://www.cbc.ca/news/climate/zombie-fires-canada-wildfires-1.7207765

EBSCO. (2024). Holdover fire (zombie fire). EBSCO Research Starters. Retrieved from EBSCOhost. Holdover fire (zombie fire) | Earth and Atmospheric Sciences | Research Starters | EBSCO Research

G. Rein, X. Huang (2021) Smouldering Wildfires in Peatlands, Forests and the Arctic: Challenges and Perspectives,

Current Opinion in Environmental Science & Health, 100296. https://doi.org/10.1016/j.coesh.2021.100296

Greene, J. (2024). Holdover fire (zombie fire). In Earth and Atmospheric Sciences: Research Starters. EBSCO. https://www.ebsco.com/research-starters/earth-and-atmospheric-sciences/holdover-fire-zombie-fire

Leverhulme Wildfires Centre. (n.d.). Smouldering peat fire in the Arctic. https://centreforwildfires.org/projects/smouldering-peat-fire-in-the-arctic/

North, T. D., Markle, C. E., Fallas, R. Y., Moore, P. A., & Waddington, J. M. (2024). Initial impacts of wildfire on overwintering conditions for a species‑at‑risk snake. Geo‑Ecology and Conservation Journal.

SunFounder STEAM Education. (n.d.). Arduino X MQ2 gas/smoke module | Detecting gas with Arduino! [Video]. YouTube. https://www.youtube.com/shorts/nZ9b8ysTmzg

[Stock Picture of map of the Northwest Territories] (n.d.). Pixel. Retrieved 4/30/2026

[Stock Picture of wildfire in snow] (n.d.). Pixel. Retrieved 4/30/2026

[Stock Picture of wildfire] (n.d.). Pixel. Retrieved 4/30/2026

[Stock Picture of map of winter forest] (n.d.). Pixel. Retrieved 4/30/2026

Williams, O. (2025, March 26). Overwintering ‘zombie fires’ aren’t doing what we thought they’d do. Cabin Radio. https://cabinradio.ca/229566/news/environment/wildfires/overwintering-zombie-fires-arent-doing-what-we-thought-theyd-do/

Working Forest. (2025, May 20). Understanding overwintering wildfires in northern forests. The Working Forest. https://workingforest.com/understanding-overwintering-wildfires/

Yousif, N. (2024, February 17). ’Zombie fires’ burning at an alarming rate in Canada. BBC News. https://www.bbc.com/news/world-us-canada-68228943

Arduino. (n.d.). Arduino Cloud | Build, control, and monitor your IoT projects [Website]. Arduino. https://cloud.arduino.cc/