By Brendan Murphy, Lead Researcher, Accessibility Observatory, University of Minnesota
As the rate of bicycling continues to increase in North American cities, partly in accordance with placement of better bicycling facilities, it becomes all the more important to better understand to what destinations cyclists are traveling, and the specific routes they are using to get there. Properly measuring bicycle accessibility—a measure of how many jobs you can reach, by bike, in a certain amount of time—requires methodology distinctly different from what we use to measure accessibility by car, transit, or even walking.
Cars typically have few, if any, restrictions on where they may be driven, and while drivers do not always use the perfectly shortest path, transportation networks available to cars are considerably more robust and redundant than those afforded to bicycles. Transit networks are more similar, in that a limited number of (usually) fixed routes are available, but the user is still at the mercy of schedules. Walking as a travel mode is, while slow, thoroughly route-unrestricted aside from limited-access facilities such as interstates, so long as there is a suitable sidewalk. Choosing a route when bicycling is a much more sensitive affair—the shortest and quickest route may be legally bikeable, but often isn’t safe, and many cyclists would opt for a longer and more circuitous route if it were considerably safer. Calculating access to destinations by bicycle must account for these considerations, or else we are simply calculating accessibility by slow-moving car.
To account for these fundamental differences, we are currently incorporating Level of Traffic Stress (LTS) methodology into how we construct the networks on which we calculate bicycle accessibility. The LTS framework, popularized by Peter Furth’s research group, attempts to classify and rank roadways by how bicycle-friendly (or unfriendly) they are, based on physical roadway design attributes such as number and widths of lanes, speed limits, the presence of a bicycle facility, and measured auto traffic volumes (where available). The rankings vary from “1,” indicating fully protected and/or off-street facilities suitable for children, to “4,” corresponding to higher-speed and/or higher-traffic roadways with no bicycle facilities, such as 4-lane arterials. Once a city’s street network has been categorized using this methodology, you can then construct the low-stress bicycle network—that is, the subset of roadways that qualify as a “1” or “2” under the LTS framework—and perform accessibility analysis on such a reduced network.
The below map shows the low-stress bicycle network in the Twin Cities area, determined from using a set of heuristics and assumptions on OpenStreetMap tag data1. Streets classified as LTS 1 are shown in green, and LTS 2 streets are shown in blue. The vast majority of LTS 1 streets are residential neighborhood streets, some of which are bicycle boulevards; there are visible gaps in the blue LTS 2 network.
The next map shows the current bicycle network in the Twin Cities area, composed of the low-stress facilities previously pictured, plus LTS 3 facilities (orange), representing the typical North American standard prior to the current shift toward more protected facilities.
Accessibility can then be compared across the different LTS levels; we define an “accessibility gap” as the average accessibility loss experienced by a cyclist when they are unwilling to travel on roads of rank LTS 3, compared to the low-stress network of ranks 1 and 2. Why LTS 3 as the cutoff? Although the standards in North America are gradually changing for the better, “door zone bike lanes” and similarly unsafe facilities still comprise the backbones of many of our bicycle networks, and classify as LTS 3. Alternatively, this can be interpreted as the accessibility gain that a user would experience if all of a city’s LTS 3 facilities were upgraded to low-stress LTS 2 facilities. The following map shows this “accessibility gap” between LTS levels 2 and 3 for the Twin Cities area.
Using this “access gap” metric, we can then compare different cities and metropolitan areas, or even neighborhoods within a given city, to gauge how extensive cities’ low-stress networks are, and whether certain neighborhoods are adequately or inadequately served by low-stress bikeways, relative to other neighborhoods. For example, based on some preliminary2 work the Accessibility Observatory performed earlier this year, we found that Minneapolis neighborhoods such as Bryn-Mawr and Northeast Park experienced the highest access gaps, with reductions of approximately 80% in access to jobs by bicycle, within 20 minutes. Neighborhoods such as East Harriet and CARAG close to the Midtown Greenway, a major low-stress bicycle facility, experienced the lowest reductions in access around only 13%, when bicycle travel was restricted to the low-stress network.
The accessibility analysis framework that the Observatory employs is both flexible (can be used on multiple travel modes and different scenarios) and scalable (we apply it nationally at the U.S. Census block level for our ongoing Access Across America report series). Additionally, such analysis is very useful in alternatives analysis for planning future transportation investments. Coupled with more accurate measurements of bicycle access via using the LTS framework, we can get good estimates on the accessibility gains associated with building specific bicycle facilities actually experienced by real-world, route-conscious people on bikes. Additionally, looking at the “access gap” metric can illuminate areas of the bicycle network that are particularly lacking in low-stress options, providing prioritization for planners deciding where to invest money in bike infrastructure.
We’re moving in the right direction regarding the types of bicycle facilities being constructed, and where they are being placed, but with limited amounts of public dollars available for infrastructure, it’s all the more important to be careful and precise. Marrying LTS labeling with measuring access to economic opportunities—and perhaps other typical cycling destinations as well—gives us a framework to not only measure bicycle access in a way that reflects the realities of bicycle routing, but also evaluate potential bicycle network investments in an effort to prioritize them based on their actual benefits to users.
1 OpenStreetMap data are crowdsourced, and while the data quality is overall relatively high, it can take time for the OSM community to encode new bicycle facilities as they are built. Gaps in the OSM network data may exist where a new bicycle facility was recently installed—you can learn more about OSM and how to contribute to the project here.
2 The data and maps presented in this article are preliminary results from methodology tests, and not to be taken as ground truth. Work to improve and finalize our bicycle LTS accessibility analysis framework is ongoing in 2018, and will be implemented within the Observatory’s National Accessibility Evaluation project later this year.