It is relatively easy to find data and visualizations for residential population density. Here is a map of Seattle census tract densities via the City of Seattle, for example. But everyone who commutes to a job knows (sometimes painfully) that a static view of residential density is just a slice of a larger, dynamic landscape. The geographic distribution of people in the city on an average Thursday afternoon is significantly different than it is at midnight on a Sunday. This is especially true for areas with strong, single primary uses like Paine Field (Boeing Factory), Overlake (Microsoft), and Downtown Seattle.

In an effort to understand these daily shifts in our region’s population density, I built a (very) simple model of the home-to-employer population shift on an average weekday for the Seattle metro area. I used two data sources: American Community Survey population estimates and Longitudinal Employer-Household Dynamics data with origin-destination employment statistics. Both sources are published by the Census Bureau and contain data down to the census block group level. Assuming that most* people with outside-the-home employment leave for work in the morning and return in the late afternoon, I produced the population density animation in the embedded video.

A few areas really stand out in the animation: the downtowns of Seattle and Bellevue, UW, the Microsoft headquarters in Overlake, and the Boeing Factory in Everett. If you know your Puget Sound geography, you can also spot the Boeing Factory in Renton, Factoria office parks, SeaTac Airport, and the warehouse district in Kent. There are also subtle decreases in density for heavily residential areas in suburban districts. Of course, these observations are not qualitatively surprising to anyone who knows the Seattle area. More interesting are the estimated population densities in these areas. Parts of downtown Seattle seem to achieve 5200,000+ ppl/sq-mile during some weekday afternoons – literally off of my scale! Good luck serving that kind of density with single occupant vehicles.

I should probably mention some obvious shortcoming of this model. It is built on a simple set of assumptions and cannot account for non-standard commutes, like night-shifts, and non-commute trips which are a certainly a significant portion of trips made**. The model also doesn’t know about the paths that people take between home and work. Still, it is quite striking to see how the region’s population concentrates into half-a-dozen CBDs during the course of a weekday. And I have a renewed appreciation for the economic importance of downtown Seattle to the region.

[*] I assumed that 80% of people with employment outside the home will need to commute on a given weekday. This is not a scientific estimate; it is a guess and nothing more.

[**] Non-commute trips are likely more spread throughout day and night, however, which would dilute their aggregate contribution to shifting population density

29 Replies to “A Day in the Life of Seattle’s Population Density”

  1. Fascinating stuff. Correct me if I’m wrong, but you are basing this on where people live and where people work. Does it include where people go to school? Either way it is interesting to see the UW spike upwards quite a bit during the day.

    As you said, it is relatively easy to find data and visualizations for residential population density. My favorite is this one, as it gets down to the census block level (,47.5018,-121.9355,47.7193). Do you know of any maps that show the employment data in a similar way (or any way for, that matter)? The animation is great, but not detailed enough to make any recommendations. Having data at the census block level makes the case for transit to a particular area much stronger.

    For example, my guess is that Ballard has a fair amount of employment (in the hospitals, mostly). Not nearly as much as downtown (or First Hill) but way more than say, Phinney Ridge. The same is true for Northgate (which has a fair number of medical buildings along with the college on the other side of the freeway). This makes those areas a bit different than other largely residential areas.

      1. Interesting map, but only to find “dead zones” or areas where there are no employment. I can’t use it, for example, to see whether there are a lot more jobs in downtown Seattle versus downtown Bellevue. They look the same, and roughly the same as the north end of Magnolia, since the “one job, one dot” style has a limit. There are places where hundreds of people are employed in one block, but they get the same treatment as a one story shop.

      2. In addition to Ross’ comments, I’m a bit skeptical at how few jobs are reported at UW (Seattle). UW-Bothell appears to have far more!

      3. Al Dimond, all of UW’Seattle’s jobs got squeezed into the little parcel where the UW Admin (former Safeco) building is at 45th & Brooklyn. Yeah, it’s a drawback of the method they used.

    1. “it is quite striking to see how the region’s population concentrates into half-a-dozen CBDs during the course of a weekday.”

      STB had an animation a year or two ago aggregating Metro bus-route movements over a 24-hour period. It showed the same islands of intensity, although as the centers of twice-daily commutes and transfers, and didn’t include car drivers.

      Still, the bright islands show something important: urbanism is succeeding. Downtown Seattle never lost the number of jobs that other cities did. The growth of downtown Bellevue and other satellites is a reversal of the 1970s/80s pattern of isolated office parks. Microsoft’s campus was originally an isolated office park too, but there has been so much growth concentration in and around it that we can call it another bright island. Boeing’s plants have continued on as they always have.

      In many other cities — such as I’ve heard about Dayton, Ohio — the downtown job loss and decentralization was so big that downtown is no more significant than a suburb. So few residents work downtown or have any reason to ever go downtown. We have escaped that fate, as the animation shows.

      1. Great work!

        This strengthens the case for the Paine Field diversion on Everett Link. I am genuinely surprised by the level of employment there. ST must have been looking at similar maps when making the decision to serve Paine Field.

      2. >> I am genuinely surprised by the level of employment there [at Paine Field].

        Me too. Enough so that I’m skeptical of the data or at least the way it is presented. Right now we don’t have any corroborating evidence. As Eric said, it is very easy to find census maps for population — that part is very easy to verify. By I have yet to see an employment density map. I think it is possible, for example, to mistake employment per census block with employment per square mile. The former would make large census block areas seem more dense. Larger census blocks tend to be in rural, suburban or industrial areas, whereas smaller census blocks are in the city. If the map confused the two ideas, then Bellevue would be slightly exaggerated, but Redmond and Boeing employment density would be significantly exaggerated. Given the sprawling nature of the Microsoft campus and modern manufacturing, I fear that is what happened.

        On the other hand, this could all be completely accurate (assuming the census data is) but I would like to see a map with detailed employment density, similar to the ones easily available for residential population. Even just a table would be handy (I could do a little spot checking).

        If it is accurate, it may be that places like Microsoft and Paine Field stand out mainly because of the nature of the way this is presented. Downtown Seattle, Bellevue and the UW max out at over 200,000 people per square mile. But it isn’t clear at all how much more. Plus there are areas (like Fremont and Ballard) that are hidden. Those areas may have just as much employment as Paine Field, but not nearly as much as downtown Seattle, downtown Bellevue or the UW. Both the scale and the geography (and the lack of employment in between) might make those areas stand out more than they should.

      3. Boeing Plan employs 30,000, which is around the same as the Microsoft campus. I think both are large enough job centers to merit light rail.

        Outside of the UW tower and some of the hospital buildings, UW’s buildings are mostly 3~4 stories with large quads between the buildings, i.e the same as Microsoft. And if you work in a large factory, your job is a solid 10~15 minute walk from the parking lot. It’s really not that different than walking 4 or 5 city blocks from a light rail station.

        Are 30 story office towers easier to serve with transit? Certainly. But if the Seattle area wants to have a diverse economy that is more than just tech & professional service workers sitting in cubicles, we need to figure out how to build transit infrastructure that serves people who work in places like corporate campuses and industrial parks, even if those are harder to serve.

        Not everyone will live in a condo tower, and not everyone will work in an office tower.

      4. Boeing is the largest employer in the state. Paine field has also been approved to handle commercial flights, and I believe they might start flying this year.

        In light of all that, I though the criticism of the Paine field light rail stops were pretty rediculous. It feels like some people are just hostile to the areas outside Seattle.

        Another example would be the hostility to park and rides. Seattle won’t build any, and Seattle tax payers won’t be paying for them, but suburban voters were very clear they wanted some of their taxes to go towards expanding the park and rides.

      5. Paine Field is more than just Boeing. There are a ton of suppliers engineering offices around there (even Airbus suppliers). My wife has worked for three different engineering companies up there, plus Boeing.

      6. It all comes down to walkability. I can believe that the Paine Field area has a large number of jobs, but except possibly for the Boeing factory and offices themselves, they’re spread out over a large sprawling area. Some of that sprawl is necessary for the industrial operations and the trucks that support it, but some of it is excess and exists merely because cultural fashions and zoning regulations made it that way. In Brooklyn and throughout Europe, industrial plants exist in ordinary city buildings with a walkable streetfront and normal subway stations. Obviously an airplane assembly plant needs a larger space, but not everything else does.

        So the question becomes, how far are all those jobs from the light rail station? How easy/pleasant/safe is it to walk to them? How will people get to the jobs if they can’t walk? Will the companies provide massive shuttles to the Link station? It’s possible that this Paine Field vision might succeed, but it will require more commitment and cooperation from the governments and companies than merely building the light rail station. What happens outside the transit line is as important as what happens inside it.

        Microsoft has in fact densified, and many of its buildings look as large and dense as UW buildings, at least from the outside. We have to grade suburbia on a curve, and it’s a good start for the outer Eastside. Now if only Crossroads and Overlake Village had that density…

      7. VTA light rail is another example of a possibly-similar area. It has stations for Lockheed Martin, Moffett Park, and Bayshore/NASA. I’ve ridden through on weekends but I don’t know how well the line serves the industrial jobs and facilities. Does anyone know any more about this, and does it offer any suggestions for Paine Field? I don’t mean the entire VTA light rail network, which is a basket case low ridership and bad design, but this specific aspect of it.

      8. I think Paine Field is the exception to the rule of walkability – the scale of those aerospace factories is completely unlike any other urban or suburban environment. My personal vision was Everett Transit would work with the larger employers (Boeing and others) to set up a system of shuttles to get people from the station to the various work sites, plus a bus route to connect the airport to the station. I’d imagine the station will have a large bus bay and plenty of layover space, as the ridership will come in waves during shift changes but be pretty low otherwise.

        “Now if only Crossroads and Overlake Village had that density…” Redmond and Bellevue have ambitious visions, and the development is coming in slow but steady. I’m optimistic about how the area will look in 5~10 years. Take a look at Esterra Park and some of the other developments in the pipeline.

      9. If you look at the Mukilteo Speedway in the vicinity of Boeing going a couple miles each direction (toward Mukilteo and toward Lynnwood), there appear to be a lot of useless open space between the buildings and the road and each other., even after subtracting the active outdoor spaces used for operations.

        I think the Seaview Tansit Center will be at or near the Link station, with the Swift 2 terminus and bays for other local and commuter buses.

      10. Boeing has a pretty large fleet of shuttles that bring people from outlying parking lots to various points near the factory and office buildings, as well as outlying offices in Mukilteo and along Airport Rd. I would assume that those would be restructured to serve the light rail stop.
        I don’t know if Fluke, Jamco, UTAS, or any other employers in the area have a similar setup. I don’t think their employees use Boeing shuttles.

    2. Census OnTheMap provides (very ugly) maps of block-level employment, including job counts and jobs per acre: To get started, search for Seattle, click “Perform Analysis on Selection Area”, and then keep all defaults and click “Go”.

      The data source is LEHD, the same source Eric used for the visualization. This data source has some important caveats. 1) It only includes jobs covered by unemployment insurance. So, self-employed people and military personnel are excluded. 2) The process of turning employer addresses into points on a map isn’t perfect. Furthermore, employers sometimes categorize all of their employees to a HQ location, rather than providing detail about number of employees at each location. For example, I believe this happens with WA public school teachers and Boeing personnel. It looks like something weird may also be happening with Amazon in SLU — employment density is lower than I would have expected.

  2. Striking graphics. So show me two more.

    One, the yawning chasms and sinkholes that should mirror those urban spires representing the percentage of people who make less than two hundred grand a year who can afford to live there. Or get a coffee.

    And two, the correlation of these towering densities with the awesome solid walls that used to be our freeway system, now stretching from Everett to Olympia, and from Puget Sound to Snoqualmie Pass every single work-day rush hour.

    All of them providing the liberty of a soon-to-be-private detention center, subject to a sudden hours-long lock-down caused by a collision forty miles away. Requiring, for those of us who used to live densely and favorably to transit in places like Ballard, a seventy miles to LINK for any serious appointment.

    So now, let’s see a really awesome presentation as to how fast this massive engine of density is going to rebuild the decent and hard-earned lives a lot of us lost in the three years it’s taken to paper-shred thirty years of land-use planning.

    But no sweat about it. What we’ll build will deliver our own fondest wish of staying out of Seattle until it’s once again livable. And incidentally, of you don’t get permanently rid of that green obsenity in the ad-space right now, 9:43 AM, I’m done looking at STB.
    For longer than ’til I move back to Seattle.

    Mark Dublin

    1. I left Seattle (High Point before they tore it down) in 03 and never looked back. The minute I put my house on the market (bought in 1987 for $55k) I was priced out. The smallish town I moved to is now experiencing growing pains thanks to Seattle area techquity locusts using it as a bedroom community.
      At some point maybe we should address overpopulating our habitat, instead of finding more ways to crush what’s left of Cascadia under luxury tenements and stalled freeways.

      1. There’s no overpopulation. Seattle is half as dense as the outer boroughs of New York, or Paris, Boston, Edinburgh center, San Francisco, etc. It’s even less dense than freaking Los Angeles and San Jose by some measures. All Seattle has to do is loosen the zoning in single-family areas, and it will grow like Chicago’s north side. That’s more capacity than demand could fill for a long time — Seattle has twice as much land as San Francisco. So the development wouldn’t be all luxury, the price premiums would be minimal and eventually disappear, and there would still be lots of single-family houses remaining. And to the John Foxes worried about displacement: one house can yield two, three, or ten duplex or apartment units, and the only loss is one house which a working-class person couldn’t afford anyway. And it would do away with the insanity that it’s legal to replace a house with a McMansion but it’s illegal to replace it with a duplex or small apartment.

  3. OK, it’s cars now. Present traffic on I-5 gets one used to them. How much donation do you need to get a new advertiser?


  4. I wonder if it’d be possible to put the Y axis on a log scale so that changes in lower-density areas are more apparent.

  5. Love it. Recommend full screen to get all of the detail. I wonder how much SLU has changed since this data set.

  6. doesn’t use a true measurement of residential density, so if any data sets starts with their numbers to extrapolate, you’re already starting on the wrong foot.

    1. Pablo96: You forgot to mention what’s wrong with its measurements and how it skews the results. Without that it’s an empty claim.

  7. To me, these maps show how ST3 really missed the boat. For instance, ST3 puts one lousy station in Ballard somewhere, completely ignoring the population- and job-heavy corridor between the UW and Ballard. Most of the dense areas of West Seattle are missed. The 522 corridor, with both people and jobs, gets nothing. Meanwhile a light rail is scheduled to open between Bellevue and Issaquah, and a lot of white space gets connected between SeaTac and Federal Way. Madness.

    1. The ST3 alignments and number of stations aren’t decided yet. The ballot measure just had a “representative” example. In Lynnwood Link, alignments as wide as Aurora and Lake City Way were considered, with up to two extra stations (N 130th and 220th SW) and variations between 120th-185th (2-3 stations). That amount of variation is enough to span Real Ballard and Fremont. What the ballot measure implicitly promises is which transit markets (e.g., neighborhoods) will be served at minimum, and the budget maximum. Even with that, ST can ditch a market (e.g., moving the alignment to QA/Fremont) by simply writing a statement justifying the deviation. The most fundamental promise is downtown to Ballard, and secondarily now SLU. There will be additional opportunities to argue for alternatives during the EIS process.

      Ballard-UW and Lake City are separate corridors. A decision was made to do them later. It’s not obvious that these are overwhelmingly more urgent than the incumbant corridors.

      West Seattle is not so much “missed” as that the first phase is short. Later phases will go all the way through West Seattle. If you’re advocating for Delridge instead of the Junction and Westwood Village, the answer is that central WS is growing, and the number of residents isn’t the only factor. The presence of job/commercial centers also plays a role, because that draws additional riders from outside the district. Delridge has no job/commercial centers, is adverse to major upzoning, and the steep hills limit the walkshed east and west.

  8. Regarding VTA light rail and its service to the Mountain View/Sunnyvale area: I’m not seeing station-specific statistics. VTA closed Whisman station for lack of ridership. The ridership in the Sunnyvale-Santa Clara was area such that VTA cut the headway in that segment to 30 minutes. Now it’s recommended that they bring it back to 15 minutes. The great bulk of workers there have free parking. A long, slow ride on light rail is not attractive.

    The big transit success in Silicon Valley has been Caltrain, the commuter rail line between San Francisco and San Jose. Their ridership has gone up steadily and they’ve struggled with capacity problems, and capacity for bikes–there’s an enormous bike ridership on the system. A lot of the Silicon Valley workplaces are outside reasonable walking distance from Caltrain, but an easy bike ride for the disproportionately young workforce.

    Caltrain is fast and pretty pleasant (if you get a seat). It’s got bi-directional ridership. It connects with BART at Millbrae, bus hubs at Redwood City and Palo Alto, and the light rail at Mountain View and San Jose. A lot of corporate shuttles go to Caltrain. Caltrain’s predecessors date from the 19th Century, so most of its stations are in small mixed use downtowns. The big problem is that most of the towns–Redwood City being an honorable exception–won’t allow denser commercial or residential development around the stations.

Comments are closed.