Transit users and STB readers use – and like to use – route maps. Routes are tangible, and are how we think when we actually use transit. This post uses some geographic analysis – focused on population and walking access – to show more of the big picture.

In April Metro recommended several phases of service cuts, for a total of about 16% of current service hours, corresponding to the estimated shortfall revenue following the defeat of county Proposition 1. It estimated loss of existing riders (10.8M rides – about 10%). This is indeed a key metric but I’ve not found quantitative answers to another seemingly basic question: what effect would the full set of recommended Metro cuts have on residents’ access to service (measured by 1/4-mile walking distance to service stops)?

So I’ve built some tools marrying Metro service details with a geographic information system (GIS).  Here’s a sample map showing areas of 15-minute frequency service during commute hours, after the cuts (green) overlaid on current (orange). Compare this to route-oriented maps from Metro and Oran Viriyincy’s remarkable before-and-after route frequency map.
peak ge4

Translating this information to population impact for Seattle (county-wide to come soon):

Service hours cut Loss of served population
30-minute freq. 15-minute freq.
Peak 16% 2.3% 6.4%
Off-peak 16% 6.4% 12.6%
Night 16% 9.1% 11.9%

On these results, Metro’s planning, following the Service Guidelines, looks to be quite effective, including restructures that increase the population that can be served with limited revenue.

What’s next? This is a start and there are many ways we can extend this information:

  • Evaluate possible outcomes from the County Council’s decision to defer the final 200,000 hours in cuts (those after February 2015).
  • Evaluate the impact to low-income population
  • Results for the complete Metro service area (not just Seattle)

Other ideas in the hopper:

  • Additional service measures: bus crowding and trip duration
  • Evaluate service separately to key destinations: Downtown, UW, Bellevue Transit Center, etc.
  • Include further demographics: employment, land-use indicators, and others.

Please comment to suggest further improvements!  More maps are below.

Additional Maps

The first 2 maps show all the areas within ¼ mile of bus stops that currently have at least every 30 minutes – with and without the stops. Then the area served is calculated from the orange region, and the population served is calculated from population densities (right map) using 2010 census data.

Area with 30-minute peak-hour frequency – with and without stops shown Population density
current peak ge 2 with seattle stops current peak ge 2 population density - census block

The maps below show service today, and after all recommended cuts to meet 16% reduction of service hours:

Key: Green – after cuts – overlaid on Orange – current
30-minute frequency 15-minute frequency
Peak peak ge2 peak ge4
Off-peak off-peak ge2 off-peak ge4
Night night ge2 night ge4


  • Current service details, stop locations, routes, and schedule are largely derived from Metro’s published General Transit Feed System data
  • Post-cut service details are outlined in some detail here. Many of the restructures amount to splicing together parts of existing routes, so I could project the likely stops on restructured routes accordingly. This left a small amount of “new” routing, which was estimated manually using existing stop locations.
  • Service analysis is around stops. The schedule at each stop is constructed from the service details (above), including the frequency of service at different periods of the day. Routes can be analyzed for availability, frequency, and (scheduled) trip duration of service to downtown, the UW, and other areas. These analyses use a handful of small programs. No doubt similar analysis is done for tools such as Walkscore’s transit analysis.
  • Map development is done with Q(uantum)GIS, an open-source graphical information system compatible with ArcGIS (the industry standard). Information from the Service Analysis (above) is provided as a table keyed to bus-stops. For a particular stop, the area served by that stop is then assumed to be a ¼-mile radius circle. Population analysis is based on 2010 census block data. Further demographic analysis can be done as additional GIS-oriented data are incorporated.

As no doubt clear already, this is NOT a professionally developed nor commercial project (by contrast with Walkscore or OneBusAway, for example) – my aim is to conceive and demonstrate measures for policy decisions. If you are aware of similar or overlapping tools and analysis, please share – I and other readers would likely find value!

35 Replies to “Transit Geography Visualization”

  1. This is a really interesting way to look at this. Thank you for putting this together.

  2. Very cool — would be great to see the map with 1/4 mile replaced with the corresponding walking isochrone, with extra bonus if this accounted for hills.

    1. In this vein, I think the map is very useful (Thank you Jim!), but it only tells us where one or more frequent services exist and will exist with the planned cuts. It does not show how many frequent services will be cut or where those services could have taken you. It does not show the impact of cuts in areas where most service is getting cut back from 15 minutes to 20 or 30 minutes as long as one service nearby going somewhere (not necessarily the same destination) is available.

      For example, you might have had two 15 minute frequency lines near you (a downtown and a crosstown service) but now only have one frequent line (the downtown service). The crosstown service reduction doesn’t show up on the map as long as there are other frequent downtown lines every 1/2 mile along the crosstown service street. That would, however, show up on a series of isochrone maps.

      1. Great point, Gabe. Taking it further, what we might want is a collection of heat maps where — at a given time (weekday, rush, weeknight, Saturday, Sunday) the color at any location is assigned by the proportion of the city (measured on area, jobs, wealth, residents, whatever) from which you can reach the location, by bus+foot, in time less than (constant)*(car travel time + parking penalty, maybe).

        For example, downtown might fare pretty well in weekday AM peak.

        (actually, this idea might not have enough nuance to be useful. I’ll throw it out there anyway so folks can tinker with it.)

    2. z7, if you’re still following, can you point me to walking-distance tools? I’m using QGIS, I’m afraid, but even if it’s something in the ArcGIS world, that would still be useful to know.

    1. Wish that were true, John. Biggest or not, real damage will be the continuing loss of service that should be there already, but isn’t. On Magnolia and many other places. For instance, regular service between Magnolia and Ballard would definitely carry passengers- like 24 extended through Ballard and along 17’s old route. Making sure absolutely nobody misses the 61.


      1. I agree Mark. Plus it becomes a vicious circle. Someone in Magnolia gets tired of taking the bus because it comes so infrequently. This causes further reduction in usage. When Metro decides to redo their routes, they notice the low usage, and cut the area further.

      2. There are a number of factors that work against transit ridership in neighborhoods like Magnolia and Laurelhurst.

        That said both neighborhoods are suffering from the classic transit service death spiral as well.

        Certain corridors like Yesler and Summit have been suffering a death spiral as well. In both cases demographics and density says bus service should do well on both corridors. However they are both too close to other much higher demand corridors.

    2. When I visit Seattle, Magnolia is where I stay. I am actually reasonably impressed with how busy the 33 and 24 actually are. Equivalent wealthy neighborhoods here in Portland have nowhere near that ridership. I have yet to take a 33, even at 10 in the morning, that didn’t stop at about 2/3 of the stops.

      Sometimes the 31 is reasonably busy too, and sometimes not.

      1. Domestic servants and nannies have to have some way of getting to their jobs.

        But maybe now their mistresses and masters will have to give them enough spare change for an Uber ride.

      2. One surprising thing is that a significant number of 24 riders go from one part of Magnolia to another.

      3. A few weeks ago, I observed people needing to stand on the 33 on a Saturday afternoon. So, yes, people do ride the bus in Magnolia, and not just during rush hour.

  3. And question of the morning: how much of our cut service will be replaced by Uber, Lyft, and their species? If they’re real capitalists, maybe they’ll create their own low-income fare too, cell phones and all- in turn creating permanent market that will care a great deal less what Metro cuts or doesn’t.

    Remember, when miners and loggers could finally afford cars and not have to ride trains, it put an end to the company towns and their soul owning stores Tennessee Ernie Ford sang about. Also, if TNC put poles on their roofs, like Russian delivery trucks- good bye to all those neighborhood runs Bruce doesn’t like.



    1. Maybe it’s time to reconsider the plan I proposed in 1993 — subsidized computer routed cabs.

      1. Consider if taxi scrip could be used for rideshares, *and* could provide a larger reimbursement to certified alternative-fuel-vehicle drivers.

      2. Computer dispatched cabs is what you have with the elderly/ handicapped service, and those cost several times what bus service costs.

        One driver and one passenger mean no economy of scale.

        Flexible route transit might work in a few areas maybe.

    2. If they’re real capitalists, maybe they’ll create their own low-income fare too, cell phones and all

      I’m not sure I follow. Very few private, for-profit businesses offer goods and services at a reduced rate for low-income customers, for obvious and easily understood reasons. The exception is generally those companies incentivized to do so by the government. Are the people who run Walmart, Costco, and (in the realm of transportation) Greyhound and Delta not “real capitalists”?

  4. This is really cool, one thing that I’d be really interested in seeing is how this compares if you do a half-mile walk radius. I know that quarter-mile has generally been the standard for bus service, but (especially for 15 minute service) I think looking at the population within a 10 min walk is more telling. I know a lot of people who ride the bus, and many of them do walk more than a quarter-mile to reach a bus stop.

    1. I’m not gonna walk 10 minutes for 15-minute service when I could wait an extra 5 for half-hourly service.

      1. It depends. If I see a bus coming and know that the next bus comes in 15 minutes, I’ll take the first bus even if it means extra walking. These are all rough estimates anyway — a ten minute walk seems like a lot, but if the difference is only five minutes of walking (ten minutes versus five minutes) then it doesn’t seem that far at all to me.

        I’m with Stephen, showing a view with a half mile radius would be interesting

      2. The distance people are willing to walk is directly proportional to the quality of service they are walking to. Fast, reliable, and frequent service is much more of a draw than slow, unreliable, or infrequent service.

        At least personally I optimize for total trip time door to door which means walking to more frequent nearby service is usually a win unless there is something unpleasant about the walk (weather, hills, carrying heavy packages, poor pedestrian facilities, etc.).

  5. Great map. I really appreciate the work that went into this.

    This is a bit off topic, but does anyone know where to get a map of Seattle showing employment? I know of several maps showing population density, but don’t know about the other side of the coin (where people work). Even maps showing office, retail or warehouse space would give me a rough idea. For example, Fremont has a fair amount of white collar jobs, but Ballard doesn’t. On the other hand, Ballard has plenty of hospitals and clinics, while they both have a fair amount of retail. It would be nice to see this information on maps, just so we can see how well we are doing in connecting people to their jobs.

    1. +1 on the employment density map. That would be very useful.

      Employment density by employment type would be particularly fantastic input to an employment (weighted at 1.5 maybe?) + population density map of Seattle.

    2. A quick search for Seattle employment density map led me back to STB (the first non-missing image). I assume there’s a higher quality version in the Transit Master Plan.

  6. Awesome map!

    South Park should actually be orange, and more of it. The 60 is South Park’s only 15-minute service during peak, and is proposed to be reduced to 30-minute minimum headway. Also, the 60 stop pair closest to Highway 509 adds more square blocks that have frequent service.

    You correctly ignored the typo on Metro’s page about the 60 reorganization, incorrectly claiming 20-minute peak headway. Under the post-bridge-opening schedule, the 60 has a period of 10-minute headway at peak of peak.

  7. If these maps are only showing Metro service, then Link would not be included (correct?). Most of the Link route is shadowed by very frequent Metro service during the peak and midday periods; but at night, more of Rainier Valley would be colored green if Link is included.

  8. A couple of answers to questions / suggestions in comments so far:
    1. Link and ST Express bus service ARE included as available service in this work. I’ll check on night-service comment and other possible errors.
    2. Part of the overall Metro service reduction plan included terminating some routes at Link stations rather than running them all the way downtown. We’re likely to see more of this as U-Link goes into service, and of course longer-term when East and North Link lines open.
    3. Agree about including major destination service in the analysis (downtown, UW, Bellevue TC, + ??) – work in progress. Thanks for the other suggestions and error-checks too – I’ll look carefully and see what I can do.

  9. The first big map shows NOAA as currently having 15 minute service (it’s orange). That’s not the case- we have route 30 every half hour. Am I misunderstanding what you are representing on that map?

    Also I don’t really have a better suggestion of a way to represent this on a map, but the fact that NOAA (where I work) only has half-hourly service means that 15-minute service from my house going elsewhere does me no good. I live on 45th Street, which will probably always have frequent service no matter what cuts they make, but effectively for me it’s already half-hourly because I need to transfer to a half-hourly bus, and it’ll get way worse with the cuts. My point is that this map makes things look better than they are.

    1. Thanks for alerting me to possible error – I can Google NOAA but best if you can give specific location – address or at least intersection

      1. 7600 Sand Point Way- it is the blob sticking out into Lake Washington just north of Magnuson.

  10. Jim, this is fantastic work! First attempt I’ve seen to visualize the impact of the cuts to transit coverage. Any chance you could replicate this work for the entire county? This would be very useful to the work I do with the King County Mobility Coalition in assessing transportation needs and gaps for special needs populations.

  11. Nice map.

    At first glance, the largest loss of service is to Magnolia. Anything to do with the resistance of Magnolia residents to rational route improvements? Hmm.

Comments are closed.