Smoke pollution from a local hazard reduction fire

Saturday morning after the hazard reduction burn. The smoke was choking, PM2.5 levels were close to 1000 µg/m3 for 4hrs

After last years terrible fire season I got interested in particulate matter air pollution from smoke and other sources and It’s effect on health. Inspired by this post I ended up buying some parts so that I could put together my own PM2.5 and PM10 detector. There are plenty of ready made detectors you can buy but I figured if I made my own I’d be able to customise it and add functionality that you’d normally have to pay a premium for. It’s also a lot of fun to build these projects and learn along the way! Last Friday there was a planned hazard reduction burn in our area carried out by NSW RFS, Fire and Rescue NSW and the NSW National Parks and Wildlife Service. We got some advance warning from a friend in the RFS so were expecting it. This would be the first proper test of my PM detector! The core of the detector is an old Raspberry Pi2. For the actual pollution measurement I used a SDS011 particulate matter detector. I also had a tiny little OLED display left over from another project that was used for a simple screen. I wrote some Python code to read the pollution values from the detector, display them on the screen and also post them to an online data recording platform. Initially I used code from the the raspberrypi.org post to read values from the detector. It worked but it was simplistic. It had the detector and the fan running around the clock even though I was only taking a reading every 5 minutes. After learning that the SDS011 has a limited operational life of 8000 hours I decided to take a different approach. By issuing commands to the detector it’s possible to only switch it on when making a reading and then switch it off again. The SDS011 is a popular part and there were already several Python libraries out there that handled the low level details, perfect! I went with Ivan Kalchev’s py-sds011 Running the detector, screen and Python code required some setup and configuration of the Raspberry Pi, I automated this using Ansible so the details would be saved and could then quickly be rerun again as needed. Here’s what I did OK, back to the fire. The burn was over a 33ha area of nearby bushland with the closest point only 2 streets away. As it started on the Friday morning huge plumes of smoke appeared, for most of the day air quality at our house actually was fine. All the smoke went straight up into the sky. At around 5pm that changed.