Snow removal in Toronto is a flurry of activity. With thousands of kilometers of streets and sidewalks to clear, the city deploys a fleet of vehicles working around the clock during winter storms.
But where exactly do these snow plows go, and how thoroughly do they cover the city?
The City of Toronto has data on the real-time GPS locations of its snow removal fleet and displays it on an interactive map called PlowTO.
To try to build up a picture of the entire operation, I generated a Python script to download snapshots of this data over a 48 hour period during a recent snow event. The result is a big table with each row representing a vehicle’s location, time, speed, and type of plow (street, sidewalk, etc.). By plotting the location data on to a map, patterns of activity over this period of time are revealed.
Click on the map to view at a higher resolution.
Major arterial roads stand out, even without adding labels to the map. Almost every street and residential neighbourhood has activity. Some of the big white blobs on the map are service and storage facilities, where plows start and end their trips.
There are a few caveats with this data: while I queried only plows that are moving (i.e. speed > 0 km/hour), we do not know whether they are actually clearing snow at any moment in time. Some percent of the data on these maps would be travelling to or from different locations.
Most of the data in the map above pertain to street plows and windrows, but a smaller percent of the data pertain to plows for sidewalks (~25%) and cycling infrastructure (~5%). Let’s take a look at where these types of plows are specifically, relative to their corresponding infrastructure.
Click on the map to view at a higher resolution.
Evident is that the ‘cycle’ plows don’t always go where there are cycle tracks and bike lanes. They could be either miss-classified or simply being used to plow sidewalks or other infrastructure instead of cycling infrastructure. The opposite might be true in some locations, sidewalk – plows may be being used to clear cycle tracks.
As well, one might look at the above maps and think that some neighbourhoods aren’t plowed at all, but it might be the case that sidewalks in these neighbourhoods were plowed shortly after the 48 hour period that I downloaded data for. One interesting direction for this work would be to map over multiple storms which neighbourhoods have recurring patterns of priority over others, and by type.
While comparing where plows go relative to infrastructure is an initial step toward identifying where there might be gaps, it would be more insightful to compare with data on where people travel and the location of important destinations. For example, a well traversed area next to a hospital should have priority for plowing over most other locations.
If you’re interested in this data, or in the technical steps of how it was queried and processed, feel free to reach out to me at jeff.allen@utoronto.ca
