Access in Seattle

Anson Stewart
Published in
9 min readOct 3, 2019


Downtown Seattle panorama from Columbia Center (photo by SounderBruce under CC BY-SA 2.0)

The network analysis capabilities of Conveyal’s cloud-based software make it a useful tool for exploring the impact of evolving transportation options. To test some new features in Conveyal Analysis, we recently conducted two sets of calculations — access to transit stops with frequent service, and access to jobs — for a study in Seattle. This study focused on how evolving micromobility options (e.g. shared electric 🚲 /🛴) could help cities like Seattle meet accessibility goals.

Progress toward sustainable transportation

The Seattle region is successfully implementing an ambitious sustainable transportation agenda. Drivers of this success include leadership reflecting communities that agencies serve, partnerships between the City of Seattle and King County Metro, and funding to advance a more equitable multi-modal transportation system. In 2014, voters passed additional funding “to improve transit availability and access.”

Among other goals, the Seattle Transportation Benefit District (STBD) vote committed to providing “72% of Seattle households with a 10-minute or shorter walk to 10-minute or better transit service by the year 2025.” Seattle’s progress toward these goals has attracted national attention:

With Light Rail expansions and added frequency on numerous King County Metro routes, progress on the stop catchment metric (“10-minute service within a 10-minute walk”) has indeed been impressive:

From Seattle Department of Transportation 2018 STBD Annual Report

While Seattle is now within 5 percentage points of its 72% goal, the 2018 STBD Annual Report recognizes the reality of diminishing returns:

The first several routes added to the 10-minute network in 2015 and 2016 overlapped very little, allowing for large strides to be taken towards the citywide goal of 72% with relatively minimal investment. Given the interconnectedness of Seattle’s transit network, more recently upgraded routes mostly cover areas that already have access to the 10-minute network, allowing only incremental progress towards the Move Seattle goal.

The report notes that additional frequency in corridors that already have frequent service can have substantial benefits for riders, even if the current stop catchment metric does not capture these benefits. If Seattle were interested in revising or expanding performance metrics for future policies, how could they reflect and guide progress more holistically?

In the sections below, we discuss additional metrics for access to transit stops and access to jobs.

Access to transit stops with frequent service

Seattle’s starting definition for 10-minute or better service is a route with an average of 6 trips per hour between 6 AM and 7 PM, with no individual hour during that period having fewer than 4 trips per hour.

Overlapping routes complicate this definition (though there are ways to account for them systematically). The STBD Annual Report considers King County Metro Routes 3, 4, 13 as a combined route in Queen Anne. Individually, none would be a 10-minute route in this corridor; this adjustment allows multiple stops northwest of Downtown to count toward the “10-minute service within a 10-minute walk” goal. The 3/4/13 route plus other high-frequency routes (8 local bus routes, 3 RapidRide routes, and Link Light Rail) collectively serve hundreds of stops throughout Seattle.

The red area in the figure to the left shows places that can reach at least one of these stops within a 10-minute walk. Conveyal Analysis uses a full multi-modal network, including streets, sidewalks, and bike paths, so this area reflects network-based walking distances, rather than simple circular buffers.

Seattle is designing a shared electric scooter pilot program to complement its current fleet of shared electric bicycles. The prospect of much wider availability of micromobility options raises a question — what percent of Seattle households would be within a 10-minute e-bike or e-scooter trip of these transit stops?

We decided to answer this question by testing new features for simulating waiting time or other pick-up delays at a traveler’s origin, specific to defined areas. As a starting point, we assumed that e-bikes and e-scooters would be widely available within Seattle, and that users starting trips in the city would need 5 minutes to find, reach, unlock, and start riding one, at speeds less than 10 mph.

Assuming a 10-minute total trip (5-minute pick-up wait plus 5-minute ride), approximately 78% of Seattle households would be within a 10-minute e-bike or e-scooter trip of frequent transit — 11 percentage points more than within a 10-minute walk.

In the map above, walking catchments (red) leave notable gaps between the four main frequent-service corridors in the north of the city (15th Ave, Aurora Ave, Roosevelt Way, and 35th Ave). The expanded e-bikes and e-scooter catchments (blue) essentially close many of these gaps.

This type of analysis could help planners design an e-scooter pilot that complements transit, better serving areas not conducive to high-frequency, fixed-route service.

Access to jobs

The preceding analysis gives a sense of how people can access stops near their homes. But what about access to destinations?

Measuring the number of jobs or other opportunities reachable within a certain travel time can be a useful complement to catchment-based metrics. First, when such accessibility measures properly account for waiting time, they naturally account for the benefits of frequent service regardless of whether overlapping routes are combined or not. They also show benefits of investments that reduce in-vehicle travel time, such as bus-only lanes. And ultimately, most commuters want to travel all the way to their jobs, not just to a nearby transit stop along the way.

Consider a prospective commuter named Sam, who just moved to Seattle and is looking for a job within a 45-minute commute of Lake Union.* Sam has frequent transit options within a 10 minute walk, providing connectivity to the north of the city and south past Downtown. Including the initial walk, waiting time, in-vehicle time, and walking to a final destination, Sam could access the area shown in red (on the map below, to the left) within a 45 minute door-to-door morning commute. There are 385 thousand jobs contained within that red area.

With the option to use a shared e-bike or e-scooter in combination with transit, Sam would have access to the expanded area shown in blue, and an additional 34 thousand jobs.

Now consider a different commuter named Victoria, who lives in West Seattle. She lives farther from frequent transit — beyond a 10 minute walk (compare the first map of this blog post). Within 45 minutes, she could not reach Downtown or many other job centers (as shown in red on the map above, to the right); only 67 thousand jobs would be reachable. With the option to use a shared e-bike or e-scooter for the first- and last-mile parts of the trip, even with a five-minute pickup delay, she would have access to an expanded area (shown in blue above) and 261 thousand jobs — nearly 4 times as many.

You can find maps of additional example origins in this report. Of course, these job access counts can be very sensitive to the chosen travel time cutoff and how waiting time is factored in. Conveyal Analysis includes easily adjustable parameters that help users quickly test how results vary with different specifications of accessibility indicators.

Exploring the impact of the Northgate Link Extension by selecting different origins and travel time cutoffs in Conveyal Analysis

Summary of changes in job access across the city

Our cloud-hosted deployment of Conveyal Analysis computes isochrones for a single origin in a fraction of a second, and it scales to handle tens of thousands of origins in a couple of minutes.

So rather than comparing scenarios for only a few origins of interest, we can also calculate the change in access to jobs for every origin in a high-resolution grid of points. For the Seattle region, we used a grid of 66 thousand points to represent origins and destinations.

In the map to the left, the color of each grid cell represents the increase in number of jobs reachable within 45 minutes when commuters can ride e-bikes or e-scooters up to 30 minutes as a first-mile, direct, or last-mile mode of travel. The increases in job access are spread across Seattle and its City Council Districts. This micromobility scenario limited e-bike and e-scooter use to trip segments starting in the City of Seattle, but there are still some access gains for people living outside the city boundary who can use e-bikes and e-scooters as last-mile connections after riding transit into Seattle.

We summarize these fine-grained accessibility results in multiple ways. The image below compares the statistical distributions of access to jobs under these scenarios.

On average, Seattle households can access 289 thousand jobs within 45 minutes in the baseline scenario (transit plus walking️), and about 1/3 more (385 thousand jobs within 45 minutes) in the micromobility scenario.

In the histograms, the baseline scenario is shown in blue and the micromobility scenario is shown in red, with a darker area where the two overlap. The horizontal axis is a scale of number of jobs accessible within 45 minutes. The height of each histogram bin represents the number of households who have the corresponding level of job access. The farther to the right the red distribution is, the higher the job access gains across households.

Comparing scenarios in terms of distribution of job access for households in the City of Seattle

In the baseline, the 10th percentile household (marked by the leftmost vertical line) has access to fewer than 50 thousand jobs, while in the micromobility scenario, the 10th percentile household (the second vertical line) has access to more than 141 thousand jobs. The large increase at the lower end of the distribution suggests that micromobility could help advance more equitable access to jobs.

Conveyal Analysis also makes it easy to compare these distributions across different neighborhoods. The animation below shows histograms for each Seattle City Council District. In every Council District, the distribution shifts toward higher numbers of jobs reachable within 45 minutes.

Comparing scenarios in terms of distribution of job access for households in each Seattle City Council District

If you are interested in additional results (including distributions across workers, rather than households) or methodological details, see the previously referenced report. While our analysis relied on various simplifying assumptions (including a uniform pick-up time citywide), possible extensions include setting neighborhood-specific pick-up times to reflect e-bike and e-scooter availability (e.g. derived empirically from archived GBFS or MDS feeds). The flexibility and performance of Conveyal’s routing engine allow it to handle high degrees of spatial and temporal detail, even at regional or national scales.

On-demand modes in Conveyal’s open-source routing engine

Shared electric bicycles and scooters parked at the entrance to a metro station (photo courtesy of Kuan Butts)

In developing the multi-modal routing engine that powers these analyses (R5), Conveyal focuses on enabling rapid turnaround without compromising detail. R5 incorporates cutting-edge features to enable web-based sketch planning while accounting for uncertainty and variability in public transit, and we’re rolling out new types of network modifications (e.g. pick-up delays and congestion areas) that help planners evaluate multi-modal connections in terms of accessibility.

In addition to numerous micromobility analyses in the US, for example, the new pick-up delay modifications in R5 have also been used to evaluate complementarity between hailed autonomous vehicles and public transit in various European cities. We’re excited to see these features applied even more widely, so please get in touch if you’d like to try them through your own Conveyal Analysis subscription.

* Any resemblance to a movie released 25 years ago is, of course, purely coincidental.



Anson Stewart

Analysis and Research, @conveyal | PhD in Transportation, @MIT | '10 TJ Watson Fellow + @SwatAlum | Californian in exile on East Coast