[UPDATE: To be absolutely clear, “the limitations of simply bypassing that segment” below refers to the critique that Central Link should have gone straight to the airport via Marginal Way, instead of doing a sharp turn to serve SE Seattle.]

Sound Transit released per-station, per time-of-day boarding data (.pdf) from earlier this year, and it’s a treasure trove of information. Although overall ridership was substantially lower than it is at the moment, there are many interesting trends:

  • About 51% of all trips begin or end between Beacon Hill and Rainier Beach, which illustrates some of the limitations of simply bypassing that segment to go straight to the airport.
  • Westlake and Seatac are the powerhouses of the system, as is usually the case with a terminus.
  • About 11% of trips are contained entirely within the DSTT.
  • University Street and Pioneer Square are weak on weekends compared to the other two tunnel stations, unsurprisingly.
  • The data suggests an average trip length of 7.6 miles, or 149,000 passenger-miles/weekday. That’s exactly half the length of the line, or University-Rainier Beach.
  • The PM Peak Southbound is significantly stronger than the AM peak northbound, but mid-day Northbound is stronger. That suggests that Link’s AM peak is too early.

A pie chart is below the jump. Since each station gets credit for half a trip, the portion of all trips involving a station is actually double what is shown on this chart (i.e., adding up all the true percentages would get you 200%).

Thanks to Sherwin for obtaining this data.

102 Replies to “Link Ridership by Station”

  1. Seatac is interesting – lots more NB than SB. They must be like Seattle enough to cancel their return tickets, and stay awhile.
    ….. or perhaps their still trying to figure out how to get back. Hmmmmm.

    1. A couple possible explanations, one-way cruise passengers who fly back from another airport, hotel airport shuttles offering door-to-door service seems attractive when it’s right in front of you, you can always use Link when your plane arrives but you can’t use Link for 6am flights. But yes, it’s odd.

      1. If you’re comfortable arriving in the terminal no earlier than about 5:20. That’s cutting it pretty close, as they’ll typically board at 5:30.

      2. But you can’t expect cruise-ship visitors to find their way down to Stadium station at any time, much less 4:30 in the morning.

        The airlines release your seat 30 minutes prior to departure if you’re not there, or at least they would if boarding didn’t always take so long by the time they realize you’re gone. Plus you’d have to be extremely lucky (and run) to get from the entrance through security to your gate in fifteen minutes.

      3. If I recall, that 4:40 train from Stadium connects with three night owl trips from the north end (Routes 81, 83, 84).

      4. Again, cruise-ship visitors. You expect visitors to spend an hour pouring through the bus schedules to find those night-owl connections? First, they’d have to know where all the bus schedules are (only at major places like Westlake station and the downtown library).

        If there were a major downtown circulator, e.g., “The Downtown Bus”, going down Third Avenue at all hours to Stadium station, then it would be reasonable to expect visitors to take it to meet Link at Stadium. (But it would be even more reasonable to just open the DSTT an hour or two earlier.)

      5. Generally when you’re going to the airport, you’re loathe to take public transit because you’re worried about getting there on time. On the way back, you’re not pressed for time.

        I live in New York so I have to take some sort of public transit to and from the airport, but my choice of mode is much more biased in favor of speed on the way out than back. It wouldn’t be at all surprising if that were a major factor here.

      6. That’s the problem with public transportation in Seattle: it’ll usually get you there eventually, but if you need to get there reliably, you’ll probably be better off with another mode if you can manage it. And it’s really sad that this holds true even for Link, thanks to the lack of dedicated ROW. I never stress when taking BART to OAK, but Link to SEA stresses me out to no end. [ot]

      7. Actually it’s just the opposite for us. On the way there we’re willing to save money by taking transit (194 and now Link) but on the way back we’re worn out and just want to get home despite the cab fare.

      8. I really like the JFK Link and used it last January. The interface to other subway lines and the LIRR in Jamaica is fantastic.

        But here’s the thing…I was using it to get to my relatives house. Meaning, when I got to the Deer Park Long Island station, my sister was there to pick me up, and then I rented a car.

        A new model might be — have a lot of Flexcars parked in lots where train transit stops. Even better, daily rentals from Enterprise and other car companies.

      9. Absolutely. I should start to count how many times I wanted a car rental at my destination train station. It should just be *standard* for an intercity station — as should a taxi stand — because there’s always *some* arriving passengers going beyond the limits of public transportation. But it isn’t.

      1. I think it’s just one-way Link trips outbound from the airport. Air travelers are probably choosing other modes to get there. I would guess that it’s easier to have a friend drop you off than it is to have them pick you up.

    2. The biggest ridership at airport rail stations nationwide is not folks flying but employees. I strongly suspect the same pattern is true at Seatac.

      The discrepancy in north and southbound riders could reflect shift times at the airport not matching worker times (specifically the early morning shift) Note that ST has 574 coming into the airport at 3:20 a.m. and the first train reaches the airport at 5:10 a.m. 5:10 a.m. is too late for the first shift. TSA begins significantly earlier than 5, but most services in the airport begin opening at 5:00 a.m..

      I take a 6 a.m. flight at least once a month, and I would never consider taking Link to the airport at that time. Other’s are likely making that same choice, but my money on explaining the difference in north and southbound ridership would be on employees getting to the airport through alternative means in the morning.

      1. Ooh. You’re probably right. I wonder if this is a decent argument for early morning runs from downtown to SeaTac? Would those 1000 trips a day pay for the extra running costs (especially if it allowed the elimination of parallel bus runs)? When exactly would it have to run and would it cut into the overnight maintenance window?

  2. Yeah, several of the north vs. south spreads are interesting. Take Beacon Hill, 28% more people alighting from downtown (892) vs. boarding to downtown (695)? Are these people taking the 36 to work and LINK home?

    1. It might because congestion is so much worse in the PM peak. Link becomes more attractive relative to the bus, even though the bus might involve less walking.

      1. Link particularly becomes more attractive than the bus when the bus routes people used to take are eliminated after Link begins operations.

      2. In this case service on the 36 is as frequent as it was before LINK opened. I find everytime I take the 36 it ends up being a poor decision due to its crawling pace through the ID.

    2. When my wife has to drive anyway I’ll catch a ride with her in the morning but take Link home since synchronizing schedules is harder at the end of the work day.

    3. I think some of the Beacon Hill morning to evening discrepancy is due to the topography. I live about equidistant between the Beacon Hill and Mt Baker stations. Due to the steep hill on McClellan, i walk downhill to Mt Baker in the morning, and then downhill from Beacon Hill in the evening. I suspect others might be doing the same.

  3. I find it interesting that more people get off at Mount Baker than get on in the morning peak. Is the area a commuting destination?

    1. High School there and the King County Metro Transit Center is across the street for those transfering to busses…

  4. Are there numbers anywhere on each station’s estimated walkshed population? I’d love to see ridership numbers as a percentage of that, to get a more accurate look at which stations are attracting ridership in their community.

    1. I could do it but I just don’t have time. I know that ST is doing work like this for all of their future stations.

    1. End of the line. Maybe they need to go further north. It’s also uphill from University Street Station; maybe they want a gravity assist for their walk. Maybe they work in retail.

    2. Even before Link opened, Westlake was always busier than the other downtown stations. Terminus stations inherently attract more passengers– people usually want to board the outbound vehicles first.

      1. Was Westlake the “termninus station” for buses in the downtown tunnel? Or was Convention Place station the terminus for those buses?

      2. Convention Place is the terminus for buses, but the boarding and alightings are low there because there is not the same level of shopping, offices, and transit transfers in that area (relative to other tunnel stations).

      3. I don’t know if this was the case before the tunnel renovation, but some buses terminate at Westlake.

    3. Any idea when Orca data will be available? Lot’s of transferees off buses to Link at Westlake, I’m guessing.

    4. Westlake will continue to be the most popular stop in downtown, even after Link is extended. It has little to do with being a terminus – it has everything to do with the other transit connections and the density of all-day destinations (jobs, shopping, etc.) surrounding the station.

    5. Every morning I see people getting off Link at Westlake to transfer to Northbound busses. I always wonder why they don’t get off at ID and get their choice of seating on the bus. I guess it’s just a matter of inertia.

      1. Yes, that is perfectly logical when transferring to surface buses, but I’m talking about people getting off of Link at Westlake to transfer to a bus running in the tunnel.

      2. That is what I do. When I go home to Snohomish County I take a CT bus across from KSS. By the same token, when I go to Tacoma I get off a CT bus on Westlake so I can get the first stop of ST 594.

      3. While I’m in town this trip, I’m staying in the Skyway area and have been taking transit transferring to Link at Rainier Beach and then on occasion transferring to Metro expresses to U-District. I noticed at Pioneer Square station I couldn’t find an ORCA reader to tap off on the platform. I think that’s the same at University Station. In any case, I boarded the bus and hoped ORCA will sort out what happens fare wise.

        Also today, I think I had a snafu with ORCA in that I tapped on at Westlake after exiting the 73X expecting a Link train to come right after that bus. But instead the 106 came, so I tapped off at Westlake and then hopped on that bus. Not sure, but I think the tapping off charged me for another ride. I’ll have to go online and double check. [ OK, I checked they did reverse the charge automatically. ]

    6. U-Street is an easier transfer to… just about anything except tunnel, 4th, and Pine Street buses and the monorail and streetcar. U-Street is directly underneath 3rd, and even though some Westlake entrances are close to 3rd, if you’re coming in to head south you need to cross 3rd to get to them. Even on the 66, which uses Stewart and Olive, I prefer U-Street or Pioneer Square to get to and from the tunnel (it doesn’t help that since moving to 3rd SB, there’s no stop between Stewart/7th and 3rd/Pine). Transfers aren’t the whole story.

    7. University Street has more employers, but Westlake really feels like the center or heart of Seattle. University St might attract more rush hour commuters, but Westlake balances that out by getting a ton of people at all times of the day (and night). It’s cool to be there at, like, 9:30 in the evening on a Tuesday, and still see a bunch of people walking around, it’s vibrant all the time.

  5. I’m not sure I understand the last bullet point. Are you saying we’re missing ridership because Link starts too late in the morning?

      1. Are you taking into account that the pm peak is 3.5 hours long and the am peak is only 2.5 hours long? Maybe they should just lengthen the am peak to 6-9:30, so it would be 3.5 hours long, like the pm peak.

        More likely, they should shorten the pm peak by half an hour, and lengthen the am peak by half an hour. The pm peak could probably easily start at 3:30 instead of 3:00.

      2. I guess I am not sure exactly what you are using for numbers.

        For PM peak southbound, I get 3,596 divided by 3.5 hours = 1,027/hour

        For AM peak northbound, I get 1,950 divided by 2.5 hours = 780/hour

        So, about 33% more per hour peak northbound. That is a significant difference, but not really a problem, I would say.

        1,027 per hour comes to an average of 64 boardings per car in the southbound PM peak. And, at the peak load, which would be between International and Stadium stations, the average passengers per car would be about 46, not exactly a “crush load.” Add 20% for the most recent ridership in July, and that would still be around 55 passengers per car on average during the southbound peak in July.

        If “peak of peak” is 1.5 times average peak, that would be a peak load of about 69 passengers per car, southbound pm peak. Again, not exactly over-capacity for Link cars. Not even one passenger per seat.

      3. My figure is for both directions.

        I agree that as a whole in the peak period we are not near capacity. I won’t speculate as to what “peak of the peak” might be.

        It is interesting how the peak-ridership point on the line shifts throughout the day.

  6. How do the automatic counters tell “alighting” from “boarding” passengers? Can the motion detectors above the doors actually tell which direction a person is moving in, and automatically put boarders in a different “column” from deboarders?

    I just took a quick look at this so far, but it looks about like what I found on my sample trips.

    “•About 51% of all trips begin or end between Beacon Hill and Rainier Beach, which illustrates some of the limitations of simply bypassing that segment to go straight to the airport.”

      1. Do the automatice counters actually count boardings and alightings at Westlake and at SeaTac stations? Previously, someone on this blog said that at those two stations boardings and deboardings were not actually counted.

      2. I never understood why they would not be counted at Westlake or SeaTac.

        But I do know that at SeaTac, the operators change ends of the trains, deboarding out the regular passenger doors and boarding through the regular passenger doors. If the automatic counters are registering this, that would add up to about 250 boardings and 250 alightings per day just at SeaTac from operators changing ends of trains.

        And, at Westlake, one car of each northbound train is boarded and deboarded by one ST employee who just checks that no passengers are left on the train, and the operator also deboards and reboards his car on each train at Westlake to communicate with that ST employee that no passengers are on either car when the train heads north out of Westlake. That would be another 500 boardings and 500 alightings per day counted by the automatic counters if they are used at Westlake Station.

        So that is a total of about 750 boardings per day just from operators and ST employees either changing ends of the trains at SeaTac, or checking for sleeping passengers or “unattended packages” on trains at Westlake, if the automatic counters are being used at those stations.

      3. Unless they’re adjusting for that entirely predictable event. I have no idea if they are or not.

      4. I’m sure the person in charge of estimating ridership has also realized that and made the proper adjustments. You can see in the table above that northbound boardings at Westlake are zero, so they must have subtracted the boardings made by the security inspectors. Also, when the operator switches cabs the train is turned off, so I doubt if the automatic counters would be active.

      5. Exactly. Why can nobody tell us if ST is adjusting for that, or not?

        And is ST adjusting for the 400 to 500 boardings and alightings each day by fare checkers?

        And are the adjusting for the 100’s of boardings and alightings each day by ST security personnel, in addition to those on every northbound train at Westlake (for example, these security personnel regularly ride Link trains just to go from one tunnel station to another during their shifts, and others get on and off Link trains sporadically at any station along the line).

      6. Zed,

        Good points.


        I think it depends on how strictly the fare checkers are scheduled. My guess is that they just sort of roam around so it’d be impossible to say how many boardings there are.

        400 to 500 sounds a bit high to me, though. There are something like 250 trains a day and 4-6 checkers per inspection. I don’t think 40% of trains are being inspected.

      7. I was told by a fare checker who I asked, that they check about 1/3 of all trains, and they were working in teams of 3 at that time. And they checked both cars of each train. So, that would be 3 boardings per each car for 1/3 of all cars, or 500 boardings per day, on average. I don’t think it is 500, either, but it may be around 400 boardings per day by fare checkers.

      8. When the operators switch ends of trains at SeaTac, they do so at the same time as passengers are boarding at SeaTac. So, if the counters are operating at the SeaTac station, they would be counting operators as regular passengers.

        Fair point about boardings at Westlake, but the operator and ST employee both deboard at Westlake, also, at about the same time as the other passengers are deboarding. The trains are not turned off at Westlake station, as far as I can tell.

        Also, I have seen people board northbound trains at Westlake, thinking they could ride it further north, then have to get off when the operator or other ST employee tells them they can’t ride it north out of Westlake Station. So, are those people counted as “alightings” at Westlake?

      9. Really Norman, we’re getting into margin-of-error stufff here, stuff that happens in any transportation system. What if I’m carrying my baby so she’s not picked up by the APC?

      10. What happens to the count for those people standing under a doorway during the trip since there is no room to sit? Does it count movement under the door?

  7. I find it interesting that ridership at rainier valley (and beacon hill) are ~800 people less than Tukwila even though *no* large private TOD projects have opened yet, no P&R have been built and few transfers occur (compared to SeaTac and Tukwila).

    A *very* rough calculations of new travel demand from a TOD project could be 100 units*1.2 people per unit*6 trips per person *20% mode split= 144 trips. So under this very hypothetical example it would take ~555 units of housing to give the valley and Beacon Hill stations the same ridership as Tukwila with has a P&R and a fair number of transfers. That’s about two large developments.

    I think this just give you a bit of perspective about the tradeoffs associated with TOD and P&Rs.

    1. Keep in mind that in addition to the P&R at TIBS you have a fairly large number of housing units within the walkshed, particularly all the apartments you can see from the station.

      It will be interesting to see how ridership evens out over the next 10 years.

      1. Just an informed guesses. Number of trips per person is total number of trips in a day and varies mostly with income and age. For example going to get lunch and then going back to work is 2 trips. I said 20% as a guess. From my experience and understanding TOD projects have heavy self section with people that move in specifically because they want to ride light rail.

        So if the hypothetical person that I created makes 10 trips a day (5 round trips to and from anything) 2 of them would be on link, probably to and from work.

        These are *very* rough numbers and I threw them out just to come up with a ballpark (order of magnitude) guess.

      2. I wonder then if there are numbers that can be derived from neighborhoods in NY and the subway system? I can imagine a trip to/from work, and perhaps a trip to/from lunch as typical – with the balance of trips on foot because of proximity to services – either at home, work or lunch.

        So, I’d tend to land on perhaps four trips a day during the week from that direction.

  8. Come observations if 3 new stations were in filled along the route:

    If ST decides in the future to add a station at Graham St, would that station take away riders from Columbia City or Othello Stations or will it add new riders? I would say ridership at the Graham St Station would be pretty consistent, as it is with the other stations north and south of it along MLK Way.

    If stations were added at Boeing Access Rd and South 133rd, those stations would most likely pick up all new riders, though I would think that the ridership would likely mirror that of Stadium and Sodo Stations being used more during commute hours.

    By the way, just adding these 3 stations would make the average length between LINK stations less than 1 mile (16 stations at 15.7 miles). Add a Convention Place Station and it would be 17 stations for 16 miles. Wishful thinking, but maybe something for the future.

    1. Not necessarily for S 133rd. If a station went in there, the 150 could easily end at the station providing an all-day market from the station.

    2. I live smack-dab between Othello and Columbia City stations, a block off MLK; a prime user of a theoretical Graham St. Station. I tend to use the Columbia City station when possible but my neighbors don’t. The 10- to 15-minute walk to get to the station is prohibitive when they can just drive to work, especially when they’re not used to using transit in the first place.

      So a Graham Station would take me away from CC as a rider … but it would add quite a few from just my little neighborhood, I believe.

  9. This is my favorite post in a while! I really like the graph/table. Nice use of bullets to summarize the data. The engineer in my is happy! That being said I am a Westlake Link user as I live on Capitol Hill and will switch to that station when it opens.

  10. The numbers are interesting. My own observation is that Mt Baker – Rainier Beach stations are underused, yet the numbers show Mt Baker vying with Beacon Hill, and the MLK stations not far behind. And Stadium is surprisingly low: where are all the people I see taking the train to the game? Maybe it’s because stadium events are more like once a week rather than five days a week. But what happened to all those people who went to Stadium but not the other way?

    1. Since there is not a stadium event every day, the non-event days bring the average boardings of that station down and you’re left with people going to/from the Metro bus bases. Saturday boardings are slightly higher than weekdays and Sundays.

    2. Perhaps what’s happening is that you’re looking at the stations at “wrong time”.

      Three stations can have the same northbound boarding numbers, but one has the majority of boardings at 7 AM, one has the majority at 6 AM, and one has boardings distributed evenly throughout the day. The third will seem much less busy if you’re looking during rush hour, but it’s actually an ideal station for mass transit to support. The first two will each look less busy if you’re travelling at the ‘peak time’ for the other station….

      What caused you to perceive the Mt. Baker – Rainer Beach stations as underused? *When* were you observing? Presumably they’re getting a lot of traffic when you’re *not* observing, and that’s interesting.

      1. “*When* were you observing?”

        Mostly in the 5:30-8pm timeframe. I go down to RV after work a few times a month, and come back north a couple hours later.

        Why would one station have peak boarding at a different time than other stations? It’s possible, but there aren’t any factories with 7-3:30pm shifts at any of the stations that would explain a nonstandard peak time.

        My Stadium/SODO observations (besides the above) are mainly on weekends on my way to Costco.

    3. I work downtown and will take the train (University St. Station) to Safeco for a night game. My hubby drives to the stadium, and meets me at our seats. We leave after the game in the car. Maybe situations like this cause the weird numbers.

  11. This information now allows us to calculate Link passengers per weekday “past a point” on the line.

    For example, on the longest segment of Link, and where it parallels I-5 — between Rainier Beach and tukwila, which is around 5 miles long between stations — on the averager weekday, Link carried 4,624 passengers northbound and 4,039 passengers southbound during the period in the study. So, that was an average of 8,663 pasengers in both directions combined past any point on the line between Rainier Beach and SeaTac.

  12. I find it odd that the boardings and alightings match perfectly. I’d expect some margin of error in the counting process.

  13. Martin,

    Of course the Rainier Valley should not have been bypassed in the initial segment. It needed a relatively dense collector area with which to build ridership.

    But in fifteen years, if all goes as planned and the line has reached Federal Way and some degree of development has clustered around the stations, the bypass MUST be built. Each of the other two branches has a long “express” segment with just a couple of stations between its primary developable catchment area (north King County and southern Snohomish for North Link and Bell-Red for East Link). South Link will have the 25 mph slow and unreliable ride down MLK every day for riders south of TIB. Not inviting.

    1. Maybe… after ST2 Link and Westside Link are well underway. But there are several issues:

      (A) If it stops every 1-2 miles, its only advantage will be in the east-west segments (SODO to Mt Baker, and along the Boeing Access Road). That would shave a mere five minutes from the travel time.

      (B) If it’s nonstop, or just one stop in Georgetown… would there be enough public support to build it? It would probably have to be built with stops every 1-2 miles, in which case (A) applies. There are more people clamoring for narrower stop spacing than people clamoring for wider. Second, even with a nonstop segment, it still wouldn’t be fast enough to retire the 57x to Federal Way or 59x to Tacoma. You’d need an express line all the way to Tacoma for that, and that would be a much larger task, and the public would be less willing to fund it.

      1. If you’re gonna build an express from Seattle to Tacoma, make it a high-speed-rail route and you might have a chance of getting funding. After the existing Cascades improvements are finished, naturally.

      2. Yeah I think even with just one stop in Georgetown and one stop at Boeing, it would still only shave maybe 6 minutes or so off the time, not nearly enough to make a big difference percentage-wise for people coming from, say, Federal Way or Tacoma. I think they’ll still need to have the 590 series when South Link is built all the way down there because the buses, not stopping anywhere in between, would probably shave 1/3 off of travel time. But Link near Tacoma would still be heavily used, as there is a fair bit of commuting from South King County to Tacoma, and it would replace the 574.

    2. 35mph.

      You say it’s “obvious” but there are people here that would disagree with you.

      In the list of priorities I’d put hundreds of millions to save a few minutes pretty far down. However, if the capacity of at-grade proves inadequate there’s a pretty strong case to have a separate grade-separated corridor.

      1. Well actually I think even capacity wouldn’t be a problem, because the capacity limits in the DSTT when it combines with East Link restrict headways for South Link probably about as much as the at-grade alignment does.

      2. You’d absolutely need to run it out of the 2nd DSTT, probably through-routed with Aurora Link.

        But that’s all waaaay out.

Comments are closed.