As cities across North America move towards implementing networks of protected cycling lanes, they are often hamstrung by limited budgets. Therefore for those cities trying to kick off the process of deciding where to install the first protected bike lanes can be like taking a shot in the dark. A good phrase that comes to mind is that it is difficult to judge the need for the people by amount of people swimming across the river. In this sense it is difficult to determine which bike lane locations will generate the highest usage when there are limited people cycling on the roads because of conditions that are not safe nor comfortable enough to invite people to cycle in the first place.
So how do you start deciding where to place bike lanes? The best strategy is to build bike lanes where people are already cycling. Usually there will be people cycling in your city along certain corridors regardless of the lackluster state of cycling infrastructure. This may include streets like 102 Avenue in Edmonton or 10th Avenue in Vancouver. These are often called desire lines, which represent a disconnect between how we design our cities and how people actually want to use them. The best way to understand a desire line is to go to your local park and look for the diagonal dirt lines cutting through a grassy area. People will always opt to take the shortest route regardless of how you want them to behave. Besides the small sampling of corridors that people may be using, how else can we find these desire lines?
Luckily insurance companies and police often keep detailed records of the vehicle collisions with people walking and cycling. Take for example the City of Edmonton’s 1,070 vehicle collisions (2009-2014) with people cycling that is now available thanks due to the hard work of the Paths for People advocacy group (unfortunately the city uses an arcane in-house geocoding format for the collision locations and fortunately we have went through and manually entered each of the collisions into a map so that no one ever has to do this again. The geocoded data can be downloaded here).
Typically collision data is used for safety improvement projects. However mapping this data can serve as a proxy for cycling volumes to reveal patterns about where people are cycling. These patterns often materialize themselves along corridors demonstrating latent demand for safe, convenient and comfortable protected cycling lanes. For example with Edmonton it is very clear that people want to be cycling on 76 Avenue, 82 Avenue, 100 Avenue, 103 Avenue, 104 Avenue, 107 Avenue, 109 Street and so on. Of course you have to take the road dimensions into account when considering bike lanes on these streets, and the political willingness to tackle the public’s perceptions.
Often these desire lines form on the main streets that were the original street car corridors from long ago that still foster the exciting fine grain retail environment that people still love today. This can also been seen from a recent cycling safety report from the city of Vancouver. This makes sense since people want to be where the action is. They want to see and be seen. These routes are usually also the most direct and offer the most convenience in terms of being able to see the businesses and stop spontaneously. The only caveat is to ensure that these collisions are not the result from trying to cross these corridors.
Collision data can also give you an indication of when people are cycling the most. From the chart above it is clear that people start to pick up cycling again in March, until it peaks in July and declines through the winter months.
Of course you can also do more sophisticated analysis as the example above using the ArcGIS Hot Spot tool. Based on this map the city would get the highest return on investment with the highest reduction in collisions per dollar spent in the downtown and Old Strathcona areas. Incidentally this will also increase the rate of cycling and therefore improving cycling safety even further in a virtuous cycle.
For those cities and advocacy groups on a tight budget, collision data is a low hanging fruit that is often already being collected. Some data cleaning and geocoding may be necessary to transform into a usable format. Once ready it can be a inexpensive way to determine the desire lines for cycling.