Published in


Modeling Bicycle Comfort with Conveyal Analysis (Part 2)

Existing 11th St Bridge and the Anacostia Riverwalk Trail near Washington DC’s Navy Yard (photo by taedc under CC BY-SA 2.0)

[ Editor’s note: This guest post is authored by Juliet Eldred, who recently joined Trillium Solutions as a Project Manager.]

Conveyal makes it easy for transportation and planning professionals to model bicycle trips and sketch changes to street networks. In my previous post, I examined the Level of Traffic Stress (LTS) metric and how you can use Conveyal to create isochrones and analyze connectivity through Single-Point Analysis.

In this post, we will show how to expand your analyses in the following ways:

  1. Create and compare Regional Analyses
  2. Add a new bike link to your street network
  3. Compare changes in regional accessibility between the existing and augmented street networks

As in the previous post, I will demonstrate these features by examining how grocery store access via bicycle is affected by the stress levels of Washington, DC’s street network, and whether a bike bridge would increase grocery store access for residents of Ward 8.

Create and Compare Regional Analyses

Conveyal’s Regional Analysis essentially repeats an accessibility calculation for every location in a rectangular grid. As such, instead of working from a single point of origin, a regional analysis enables the visualization of access from all points across a given region.

To create a regional analysis, first run a single-point analysis with your desired settings, then click the “regional analysis” button. From there, you can select the destination opportunity layers that you’ll be using for analysis — that is, what you’ll be measuring access to. With Conveyal’s newest Routing Engine (v6.2) you can select up to six destination layers, and set up multiple travel time cutoffs. For this analysis I have set my four cutoff times at 15, 20, and 30 minutes. Once the first Regional Analysis is created, repeat this process again for the secondary scenario, with the same destination opportunity layer(s) and time cutoff(s).

In the Regional Analysis interface, we’ll be comparing DC’s low-stress street network (LTS 1) versus its low-to-high stress street network (LTS 1–3), in terms of the number of grocery stores reachable by cycling. To visualize this comparison in the positives rather than negatives — that is, to see how many more grocery stores are accessible to people willing to cycle on higher-stress streets — choose the higher-stress Regional Analysis as the primary one, and then select the lower-stress Regional Analysis as the comparison using the drop-down menus. This will show how many more grocery stores are accessible on the higher-stress network when compared to the lower-stress network.

Regional Analysis comparing grocery store access between low-stress and high-stress bicycle network. Dot density layer shows approximate residential population.

Here, you can see that some of the largest disparities in grocery store access by road stress levels are in the densest areas, where in some places up to 25 more grocery stores are accessible on the higher-stress network than the lower-stress network within 20 minutes. As options for building out a low-stress cycling network are developed, this type of analysis is one way to highlight where missing connections are limiting access to opportunities.

Another option for prioritizing investments is to compare access for different districts or communities of concern, focusing on equity in access to opportunities. The Aggregation Areas feature allows users to upload shapefiles of geographic areas by which they can summarize results. I used a shapefile of Washington DC’s eight Wards in order to aggregate accessibility at the Ward level. We’ll focus on Ward 8, which is located to the east of the Anacostia River and has the highest concentration of food deserts in the District. You can also weight the data in the destination opportunity dataset by an origin population, including pre-loaded block-level data from the Census Bureau’s Longitudinal Origin-Destination Employment Survey (LODES). In the example below, I’ve selected “Workers total” (by home location) to use as a rough proxy for residential population.

Grocery store access by LTS level within 20 minutes in Ward 8. While 86% of the population can access at least 1 grocery store via higher-stress routes via bike within 20 minutes, only 65% can do so on low-stress routes. Dot density layer shows proxy residential population.

Use the “percentile of accessibility” slider to view how access varies across the population. We can see that in Ward 8, 86% of the population can access at least one grocery store using streets with an LTS rating of up to 3 within 20 minutes, but that figure drops to 65% when restricted to only LTS 1 routes.

By contrast, essentially all of Ward 2’s population can access multiple stores within 20 minutes on LTS 1 routes, and that number jumps to at least 12 stores when using routes with LTS 1–3.

Grocery store access within 20 minutes by bike in Ward 2. 99% of the Ward 2 population can access at least 4 grocery stores via bike on low-stress routes within 20 minutes. Dot density layer shows proxy residential population.

If a policy objective is ensuring all residents have access to at least one grocery store within a 20-minute, low-stress bike ride (an objective which recent academic literature on mobility justice would consider to be “sufficientarian”), investment would need to prioritize Ward 8, where 45% of residents lack this level of basic access.

Analyzing the Impact of a new bike link in Single-Point and Regional Analysis

Conveyal Analysis allows users to augment and modify their street and transit networks in order to model how those changes can impact the network, and its accessibility, as a whole. For this analysis, we’ll be using the Modifications tool to assess a new bike/pedestrian connection across the Anacostia River. This car-free bridge would run roughly parallel to the 11th Street Bridge, connecting Martin Luther King Jr. Ave in Ward 8 to the Anacostia Riverwalk Trail by the Navy Yard.

Based on personal experience biking in the area, the default LTS 1 classification of its sidewalk path is inappropriately low; there is little separation from high-speed traffic and the freeway onramp at the south end of the bridge is not “easy to approach and cross.” To represent baseline conditions more accurately, draw a Modify Streets polygon to reset the bike LTS of the existing crossing to 2.

Then, create an Add Streets modification for the new bridge. Use the settings in the menu to adjust the LTS level and whether you want to enable walking, biking, and/or driving. Once you’ve created the modifications, make sure they are active in the scenario(s) which you plan to model and that the new bridge is visible on the map.

Proposed hypothetical bike bridge across the Anacostia River, connecting the Anacostia and Navy Yard neighborhoods

Next, we’ll set up a Single-Point Analysis. However, instead of comparing how bike access varies by LTS levels, we will compare the default street network to that with the new Anacostia Bike Bridge. For this initial analysis, we will use an origin point at Good Hope Road and Minnesota Ave SE in Anacostia. Both the primary blue analysis (with the new bike bridge) and the secondary red analysis (without it) have a maximum LTS level of 1.

The resulting isochrones show a modest increase in area reachable from the origin point. The blue area shows places that become reachable with the new bridge, given the LTS 1 limit and selected travel time cutoff (which varies between 10 and 20 minutes). Links with level of traffic stress greater than 1 can be traversed, but at a slower speed (representing uncomfortable cyclists who stop riding and walk their bike). Note that existing path along the Sousa bridge, northeast of the 11th Street Bridge, provides relatively comfortable access case across the river, north of the origin. The new bridge expands access northwest of the origin, and thanks to its improved connection to the Anacostia Riverwalk Trail, to a sliver along the trail extending northeast.

Animated isochrones from origin (blue marker) at Good Hope Road and Minnesota Avenue SE.

When the Grocery Stores destination opportunity layer is added in, the resulting chart shows that from this example origin, 3 grocery stores are accessible within a 20-minute low-stress bike ride without the new bike bridge, and 5 grocery stores are accessible within a 20-minute low-stress bike ride with it.

The number of grocery stores reachable via a low-stress bike ride from the origin shown above, at various travel time cutoffs, with and without the new bridge.

We will now use these scenarios to create Regional Analyses, using the same steps we previously used to compare access by LTS levels. When using grocery Stores as the Opportunity Dataset, aggregating to Ward 8, and weighting the results by Total Workers (by home location), average accessibility does not increase substantially. A close inspection of the accessibility distribution suggests minimal progress toward the sufficientarian policy objective of ensuring access to at least one grocery store within 20 minutes.

Regional Analysis of bike access to grocery stores with proposed bike bridge in Ward 8.

Most of the access gains are seen by residents who have access to 2–3 grocery stores within a 20-minute low-stress ride in the baseline (note the red bands in the histogram) and who would have access to 4–6 grocery stores with the new bridge (note the blue bands in the histogram). As shown in the user interface notice, “Weighted average accessibility may not be representative” of residents’ experience or policy goals, which is why Conveyal provides options to explore more nuanced indicators based on the distribution of accessibility across the population in any given analysis area. To meet the the example policy metric, alternative bike infrastructure or the addition of new grocery stores within Anacostia, could be explored.

In Summary

These analysis tools have a wide variety of potential applications. LTS and network modification tools can be used in conjunction with Conveyal’s transit analysis tools, and allow the modeling of multimodal access as well. Incorporating multiple opportunity types can provide a broader and more nuanced assessment of how network improvements would affect access to jobs, schools, public transit, medical facilities, and other key amenities at the regional and local levels. Recent Conveyal updates allow further refining accessibility indicators with decay functions, and upcoming releases will make it even easier to specify a combined accessibility index built on different thresholds for multiple opportunity types. Get in touch if you’re interested in trying out these new features.




We enable people and institutions to make smarter decisions by applying technical expertise in open source software and open data to create improved transportation systems, more livable cities, and a better world.

Recommended from Medium

Using Data Decided the Location of a Business (The Battle of the Neighborhoods Project)

What you need to know before taking the Machine Learning course by Stanford

5 Amazing NLP Use-cases to add to your Portfolio

Interview: Mary Cameron, Data Scientist at Tophatter

Classifying Marketplace Inventory at Scale With Machine Learning

On The Future of Life Logging — A Speculative View

Air Jordan 5 Racer Blue Resell Prediction

TimeCluster: Dimension Reduction Applied to Temporal Data For Visual Analytics

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Juliet Eldred

Juliet Eldred

More from Medium

BOSS Events Phase 1: Sensitize through Open Science FAIR Symposium

Sentiment analysis of Rickenmann Twitter mentions, with limitations

Data Management Lists Decision Points (with Schema Details) Mapped for 10 S1000D Issues: 1.9 to 5.0

Agile Data Analysis upcoming courses on Urban Mobility Data Science