ST4 Should Focus Rail in Densest Areas

Chinatown-International District Station has an open-air design close to the surface. (Photo by Steve Morgan, Wikimedia Commons)

Urban subways serving dense neighborhoods should be the future of Seattle transit

Transit ridership has been falling for years now. Ever since peaking in 2014, more people than ever are turning their backs to transit and toward cars thanks to cheap gas and aging infrastructure. In cities like San Francisco, Chicago, and Los Angeles, transit ridership was dropping by the double digits annually. But not in Seattle. As transit was dying a slow death nationwide, our ridership increased by 4.1%. And even though that number have begun to flatten, we continued to outperform other American metropolitan areas in terms of growth until the pandemic hit.

But growth is just one piece of the puzzle; cities with high growth don’t always have the most ridership overall, and sadly that’s the category we fall into. Nearly twice as many people ride transit in Greater Boston, a region not much larger than ours, though even that pales in comparison to Vancouver BC, whose ridership is triple ours.

Both Boston and Vancouver’s higher transit ridership can largely be traced back to their extensive rail networks. A single 400-foot train can easily carry 800 more people than an articulated bus and can almost always do so faster. More capacity and faster travel times always means higher ridership–service drives demand.

Interestingly, our rail network is notably small for a city our size and even as it begins to take shape it’s more of a suburban commuting tool than a true urban transit network. That’s good, but it’s far from ideal.

Most people simply don’t make the traditional suburb-to-downtown commute anymore so while building transit specifically geared towards it is good, it’s a missed opportunity on a grander scale. Our currently planned Link expansion plans call for lines that skirt freeways and whose stations are few and far between, not much unlike the commuter buses we have now.

Our region has so many neighborhoods like Belltown, Wallingford, and First Hill that are already dense transit hotspots but which are being held back by limited transit capacity and clogged streets. These are the places our rail needs to go, and it’s time we designed a rail system that went there. 

Drawing parallels

Our busiest, densest transit corridors are those that need rail, and Metro’s 8 and 44 buses fit that bill perfectly. Both are among Metro’s most popular buses, but are often plagued with delays; Route 44 is almost reliably late, and Route 8 is sometimes slower than walking.

Doug Trumm recommended a while back for light rail expansion along these bus lines, and Seattle Subway added slightly-modified versions to their Vision Map a bit later. 

I won’t go too in-depth about the proposals here (the original articles are a much better read) but in general they call for two new rail lines, one from Ballard to UW and another from the Seattle Center to the Mount Baker that roughly parallel the 44 and 8 buses, respectively. Let’s focus on a hypothetical 8 Line since there’s a few reasons why it’s a bit more special:

  1. It’s fully underground. The 8 Line won’t interline with existing Link lines and will run on its own dedicated track underground.
  2. It’s going to have blockbuster ridership. The 8 (and 48) bus it replaces already handles thousands of passengers each day and upgrading it to rail could easily multiply ridership by adding capacity, slashing travel times, and accommodating growth.
  3. It’s got a dense, short route. The entire length of the 8 Line will only be about six miles long and many stations would be less than a mile apart. It wouldn’t stray much further south than I-90 or further north than 520.

All of this means that the 8 Line is going to be very different from the Link lines we have right now (ST2 and ST3 included). It wouldn’t run the suburban commute like some other lines do, nor would it mix with street traffic on the ground.

The Link Light Rail (Seattle Southside)

If anything, the 8 Line would be to Link as the London’s Underground (or “Tube”) is to its [under-construction] Crossrail; the two former systems are subways while the two latter ones are more of a mix of that and commuter rail. There are significant design differences between London’s Tube and Crossrail because the two modes have very different purposes–one to be an urban metro and the other to be a suburban commuter–much like the 8 Line and Link.

Both Crossrail and Link have trains that are taller and wider than most subways, and both draw their power from overhead wires rather than a third rail. Both stretch for about the same 60 miles end-to-end, vastly overshadowing their more-urban counterparts–the Victoria Tube line is just 13 miles long.

It might be strange to compare a mid-sized American city to a European metropolis three times the size, but the point is simple: rail isn’t a one-size-fits-all transit mode. Each transit line has its own purpose, and we should build each one with designs that are best fit for it.


Our current Link trains are about nine-foot wide but run through 21-foot wide (diameter) tunnels because of how tall they are. Because tunnels are dug by circular boring machines, there’s no way to make them narrower without also decreasing their height. 

Design of the Beacon Hill Station. (Credit: Sound Transit)

The vertically-stretched proportions of Link trains simply don’t fit well into tunnels–there’s a ton of excess space around the train–and necessitate them to be far larger than necessary.

(By Mark Kobayashi-Hillary from London, United Kingdom – Archway Station – tube approaching Uploaded by oxyman, CC BY 2.0)

The London Underground, one of the busiest transit systems in the world, has deep-level tunnels that are just 12 feet in diameter–almost 40% narrower than ours–yet trains are the same nine-feet wide and have a near-identical passenger capacity. No space is wasted underground either because trains are built to fit the tunnels, not the other way around.

The Victoria Line (By Alex Nevin-Tylee (AL6NT) – Own work, CC BY-SA 4.0)

So why does tunnel diameter matter so much? Simple: it has a dramatic impact on cost. A 10-mile long tunnel is almost ⅓ cheaper to build using 12-foot tunnels rather than 21-foot tunnels. We can save tens of millions of dollars simply by making our tunnels slimmer (if you want to delve deeper into the math behind this, check out the author’s note after the end of the article).

Slimmer tunnels are also quicker to build since they require less dirt to be excavated. 21-foot tunnels might only be about twice as wide as 12-foot tunnels, but their cross-sections are over three times larger, meaning that building 12-foot tunnels will only require workers to dig up one-fourth as much dirt 21-foot ones would require.


The Link stations we have right now are, in a word, extravagant. Not only do they feature cool architectural details and public art (which in and of themselves aren’t necessarily bad things), but they’re also super deep. Some underground Link stations truly remind passengers that the journey is the destination–the journey from the ground to the platform at least (it’s quite a maze).

On the flip side, UW Station cost us over $140 million to build while Los Angeles is getting three stations for less than twice as much they’re are much, much shallower. Take a look at the Capitol Hill Station when it was still being built:

UW Station under construction. (Sound Transit)

Needless to say, it’s very deep. A ton of ground excavation was needed, which cost tens of millions of dollars (though using Capitol Hill Station as an example might be erroneous since it’s naturally deeper due it being located on a hill). Shallower stations, on the other hand, only require less than half as much excavation so they’re quicker to build and cheaper. The ID/Chinatown station is a good example of this (just imagine it fully lidded with a road running on top):

Chinatown-International District Station has an open-air design close to the surface and is now Link trains only but used to include busses. (Photo by Steve Morgan, Wikimedia Commons)

Ostensibly the reason we aren’t building shallower stations is disruption; shallow sub-surface stations need to be built under roads so their construction would require ripping up busy corridors and storefronts for at least a few months. But if we don’t choose to build stations under roads, they’ll have to be deep enough that the connecting tracks won’t hit building foundations. Is it a price that should be paid? Perhaps, but I’m probably not the one to decide.

Interestingly, shallower stations also come with another unique perk: since trains usually cruise at depths much deeper than that of the stations, stations are at the top of a hill in the tracks. Gravitational energy is stored by trains as they decelerate uphill into stations and helps trains accelerate downhill when they have.

Diagram shows how putting stations at higher elevation can save kinetic energy and be more convenient for riders trying to reach the surface. (Credit: Unisouth)

On the Victoria Line these hump-backed stations provide an energy saving of 5% and speed up trains by about 9%. It might not seem like much but in the grand scheme of things these things add up; humped stations allow trains to travel at their top speeds for longer.

And while shallower stations might not be feasible for more drastic elevation changes (from Capitol Hill to Fairview, for example), it’s a good thing to keep in mind for stations in the central area and South Lake Union where there’s flatter terrain.

A good transit line can unleash Seattle’s transit potential and transform, but only if it exists. By choosing to slim out tunnels, we’ll be taking a step to tackle the cost barriers that are preventing quality transit from being built.

In part 2, we’ll explore automation technologies that can help us build a better transit system and visions for our transit future.

We hope you loved this article. If so, please consider subscribing or donating. The Urbanist is a 501(c)(4) nonprofit that depends on donations from readers like you.

Brandon Zuo is a high schooler and enjoys reading about urban planning and transportation. They enjoy exploring the city on the bus and on their bike. They believe that income and racial equality should be at the forefront of urban development. Brandon Zuo formerly wrote under the pseudonym Hyra Zhang.

Inline Feedbacks
View all comments
Fred Mercury

That’s past tense. I don’t know if you’ve noticed but there has been a bit of a problem with a global pandemic, combined with huge increases in crime. With Boeing laying off people as well, stick a fork in it.


I hope The Urbanist strongly supports the shallow cut and cover station in the ID. Tough political decisions will save far more money than clever technology.

Rather than bemoaning how our existing technology is less than ideal for expensive tunnels, perhaps we should leverage the flexibility of light rail to run Link at-grade along various linear corridors in Seattle like 15th Ave or Aurora, rather than insisting we build subways that would be wildly out of proportion for the low density residential neighborhoods that make up most of Seattle north of the Ship Canal. There are certainly corridors that make a good case for a tunnel, like the 8 and 44, but there are many other corridors that can be served at low cost and high quality using the technology we have.

Finally, perhaps we should focus on building dense housing where we have good transit infrastructure, rather than building expensive transit to where we have dense housing. Private development should follow public investment, not the other way around.


I agree with all your points. This is all conventional transit wisdom. Density + Proximity = Ridership. Always has, and always will. Unfortunately, people see bumper to bumper traffic on the freeway, and think otherwise. So yes, light rail should be focused on the densest areas — it always should have. Unfortunately, we are stuck with a line that is largely suburban.

Our stations are extravagant. Some have argued that calling them “stations” contributes to that problem. It implies something grand (like Union Station). In New York City they call them “stops”.

Both of the lines you mention would be better than anything Sound Transit is building. They could have good (urban) stop spacing, with lots of riders at each stop. Both would integrate well with the bus system, but the Ballard to UW line especially so. That would enable the sort of thing that Vancouver has — a very good grid, with high ridership on both the buses and trains.

Unfortunately, I don’t see it happening. We blew our wad on ST3. The goal was never to build a good transit system for Seattle or the region. It was to run a light rail from Tacoma to Everett. There will never be an ST4. We just don’t have the money.

We aren’t alone. BART is almost fifty years old, and there are no plans for expansion within San Fransisco, Oakland or Berkeley (the areas with the most density). Instead, there are plans for an extension, as other agencies push to be part of the system, even if only a handful of residents actually ride the train. I’m not saying that will happen with Link (Everett and Tacoma are far enough, to say the least) but I don’t think there is any regional interest in improving transit within Seattle.


I think there are plenty of regional needs to create consensus for a package large enough to fit in a Ballard-UW for Seattle. IMO we’ll have an ‘ST4’ in 2028.

To sketch together a package
– 3 new bus bases, 1 for each county. Perhaps fund electrification of existing bus bases, which is a big 1-time investment
– ‘Rapid Ride’ improvements to suburban King County corridors. These go for $100M each between station and ROW improvements. A dozen of these is a $1B in YOE
– Capital improvements for 4 planned but unfunded BRT lines in Pierce
– Modest Link extension in Tacoma, 2~3 infill stations for Snohomish, and then Link extensions in Kirkland, Redmond, and maybe Issaquah if needed to absorb excess East King capacity.
– Further investment in frequency & span of service for South Sounder, with perhaps ‘all day’ service as a carrot by the end of the package.
– Stride infill stations in Tukwila and Renton, and perhaps a new Stride line leveraging SR167 connecting to Bellevue.
– Generous station access funds. Could fund all of Leafline’s wish list within ST’s district.

Daniel Thompson

TVR and asdf2 pretty much sum up my thoughts. The two biggest hurdles for this plan are: 1. cost/funding source; and 2. lack of density.

I agree with asdf2 that a regional tax would likely not pass, and ST would not be my first choice for the poster child on any tax increase after I-976 and the car valuation formula. That leaves Seattle to fund this project alone.

The problem with a ST4 is subarea equity, which requires tax revenue raised in one subarea to be spent there. There are five subareas: Pierce Co., Snohomish Co., South King Co., North King Co. (Seattle), and east King Co. Any tax increases under ST for Seattle to fund ST4 have to apply equally to every other subarea.

That was the issue with ST3. The Seattle subarea has huge costs to run a line to Snohomish Co. and its half of the second transit tunnel, and rail to neighborhoods over challenging topography with demands for tunnels that required large tax increases in ST3, which then applied to the other subareas and resulted in the eastside subarea having more ST revenue than it can ever use from ST2 and ST3. Meanwhile Snohomish and Pierce Counties are getting impatient waiting for the benefits of light rail, and a tax increase for a massive project in Seattle would likely lead them to pull out of ST.

That means Seattle would have to fund a subway alone.

The cost of this proposal would be staggering, at a time Seattle has many pressing infrastructure needs, such as repairing the W. Seattle Bridge, finishing the $1.8 billion convention center remodel, the First Ave. bridge repair, and many other bridges. Plus a lot of us are exhausted after all the recent tunneling in downtown Seattle. The mayor right now is cutting transportation funding, and so is Metro, not because she wants to but has to based on declining tax revenue. If you want to tunnel I would suggest concentrating on starting the second transit tunnel under ST3 that is estimated to cost $2.2 billion, with the Seattle subarea paying 1/2, if the Seattle subarea has the money.

Next you have Seattle’s density, or lack of density which TVR discusses. Seattle is not very dense compared to many other cities with subway lines. In fact part of the high cost for ST3 is the need to run rail to distant neighborhoods like Ballard and W. Seattle. (Seattle unlike many other older cities with rail has very hilly terrain because not all of it was flattened, just the area by Puget Sound, which makes stations like Capitol Hill very expensive).

Right now Seattle is losing population as citizens begin to move to the suburbs or buy single family homes. We still don’t know the permanent reduction in commuters into Seattle from working from home, but there will certainly be some reduction. The PSRC 2050 Vision Statement and the Cascadia report — and Urbanism in general — are based on very large future population increases in the region, mostly concentrated in Seattle and maybe Bellevue, but I am not sure that will pan out. In fact I doubt it pretty highly. There just won’t be the population or ridership to support such an expensive subway system. Upzoning can’t fix a lack of population to begin with.

At this time I think Seattle should concentrate on its pressing infrastructure needs, including the W. Seattle and First Ave. bridges, then the second transit tunnel, the convention center remodel, hope the transportation levy passes to refund Metro for basic transit, and wait a few years to see what the permanent effects from Covid-19 and working from home are before proposing huge tax increases for tunnels and subways. If the Seattle economy recovers, and the predicted population growth in downtown Seattle materializes, and the work commuters return, and ST3 is completed, then maybe a massive subway project could be considered.


I didn’t actually talk about density in my post, but briefly touched on volumes (capacity) as a function of trains per hour and how automation can achieve it.

Density certainly drives demand, but Link is also a hybrid system similar to BART and the DC metro (albeit those achieve full grade separation throughout). These systems have metro like frequency through the core due to multiple branches that serve predominately commuters (via transfer stations or communities) merging, similar to East Link and Central Link.

It doesn’t take much to get the density and decrease car dependency in our auto centric suburbs, it takes a decent sub-area plan, zoning tweaks and a basic accommodation of other modes. Also, ‘suburbs’ aren’t exactly a monolith, much of what we consider ‘city’ now are actually old streetcar suburbs. The 1950’s didn’t invent the term, they just created the auto-centric iteration we’re dealing with today. Kirkland and Redmond are doing good work not to mention the Spring District and Northgate’s Thornton Creek when it comes to creating density around transit. Urbanism or being an urbanist in someways is a misnomer, it doesn’t mean city everywhere nor is it necessarily dependent on continued growth. It means making conscious policy choices that improve quality of life. It’s breaking away from the commute dominant patterns we have and moving towards something more like the 15 minute city.

Space and density are not mutually exclusive concepts, Stockholm manages to be half park or water yet maintains the same transit accessibility as Hong Kong. Copenhagen has achieved similar outcomes with their ‘finger plan,’ too.


Link is like BART, not like DC Metro. DC Metro concentrates service within the urban core (like most mass transit systems around the world). Other than Georgetown — an area left out for political reasons — pretty much all of the urban areas are covered. Draw a circle ten miles from downtown Washington D. C. and you include just about all the stations within the D. C. Metro. In contrast, almost all of ST3 is outside that circle. The extension from Lynnwood to Everett is well outside, as is the extension from Kent to Tacoma (even now, Link extends farther — SeaTac is about 11 miles away). Issaquah is also outside that circle, and its train line won’t be connected to downtown Seattle, but will be instead connected to downtown Bellevue.

BART has the same basic flaw. Draw that ten mile circle from downtown San Fransisco and you include all of San Fransisco, all of Berkeley, and most of Oakland. This is where *most* of the ridership comes from, yet it is woefully underserved.

We are building BART, not DC Metro, which is too bad:


“Right now Seattle is losing population as citizens begin to move to the suburbs or buy single family homes.”

That is simply untrue. Not only is Seattle growing, but over the last decade, it has grown faster than the surrounding suburbs in people per square mile (i. e. density). This means that while Seattle was pretty much the only place with significant density a decade ago, the situation has only accelerated.

Citations:comment image. This lists by percentages, which is misleading. For example, in 2010, Redmond was tiny (about 55,000 people). It grew by 31%, which means it added about 17,000. Seattle, in contrast, added about 175,000. Seattle is bigger than Redmond (in physical size) but not ten times bigger. Given that Seattle started out with a lot higher density, this means that the density gap between Seattle and the suburbs has only grown.

Here is a density map as of the last census: As you can see, almost all the density is in Seattle, although Bellevue has a few spots. Areas in Seattle that are a bit dark have become very dark, while there has been a few new spots on the East Side and in Seattle. Almost all the density in the region is in Seattle, with parts of the East Side a distance second. Those spots are actually well served by East Link, while dense areas in Seattle are not well served. The two proposed rail lines would serve those areas well, either directly or indirectly.


Thanks for the article, couple notes about London. There’s a real distinction between the six ‘deep tube’ lines which is characterized by small diameter bored tunnels and the Metropolitan/District/Circle lines that were constructed via cut cover (the ones that employ the humps you referenced). Also I think those humps were originally incorporated for venting exhaust from when the tube was operated by coal powered steam engines… The circumference of deep tube lines were a product of the shield tunneling technology of the era and you correctly state the trains were built to fit the tunnels. However, this small of diameter is not necessarily ideal, and whatever savings in capital costs are offset by having to procure and operate unique systems. While Link perhaps has excess diameter to accommodate centenary lines a better comparison would be to bored 3rd rail metro systems that have undertaken recent expansion. SkyTrain in Vancouver has undertaken both bored and cut and cover projects in recent history. The debate around the Broadway extension being cut and cover then eventually shifted to bored captures well the tradeoffs between the two methods in terms of cost and politics. Blue Line extension in Stockholm is also a good one to look at too, as are the automated lines in Singapore. Topography gets thrown around a lot, but our grade constraints make cut and cover unfeasible for reaching a lot of our neighborhoods.

I think your next article will get to the crux of the issue, the massive cost of rail construction really only makes sense when you can amortize it across massive volumes and drive down your operation costs. Victoria Line, which you talked about, has the lowest headways in London at less than 90 seconds using CBTC technology, Jubilee has modern signaling too. Right now we have a system that can’t run those frequencies or autonomously because of a lack of grade separation.

The biggest issue is likely that it just costs way more to build tunnels in the US than elsewhere for a host of reasons (articles below), although Seattle does do better than other regions in the US, since we’ve been built up institutional capabilities. The amount of institutional knowledge in an organization like Transport for London, and its relative separation from politics just doesn’t exist in our country, but even they still blow through budget and schedule (CrossRail was supposed to open in ’18 for 3 billion pounds less). I did my transport degree in London and about half the students in my program were on work/study with TfL. It’s a fascinating system, but a lot of elements that seem well considered are accidents of history, for example the original Met line was built by a housing speculator in 1863.

Tunneling Costs:


In order for any of things to happen, the state needs to change the law so that cities like Seattle have the right to tax themselves to fund their own mass transit projects.

The Sound Transit model, where Seattle can’t do anything without a region-wide vote, bundled with other transit projects from Bellevue to Lynnwood to Auburn, I fear is not sustainable. Fact is, not only are mass transit projects much more productive in terms of actual ridership per dollar spent in Seattle vs. the surrounding area, but the Seattle electorate has shown a much greater willingness to raise taxes to fund them.

Even though ST2 and ST3 passed, I find it hard to imagine an ST4 even being proposed, let alone getting 50% of the district-wide vote, in the foreseeable future. Realistically, either the state will need to allow Seattle to tax itself to fund additional urban subway lines itself, or it simply won’t happen.

And, of course, future advances driving down the cost of tunneling will help immensely.

Jesse Simpson

Seattle still has the dormant “City Transportation Authority” from the monorail proposal days. This allows it to impose a 2.5% motor vehicle excise tax and a 0.15% property tax, following a citizen petition or council vote and then a majority vote. And “monorail” is only defined as “train cars running on a guideway…not including fixed guideway light rail systems” so driverless subways would easily count.

The main issue is garnering the political support, deciding on the alignment(s), making accurate cost estimates, and deciding how urban Seattle lines would interface with Sound Transit’s system at large.

I do think it’s a while until ST4 could be proposed (probably 2028 at the earliest, maybe more like 2032, which would put the next round of projects out to 2050-2060 openings). Quite simply, the suburbs don’t have that many additional good projects.

The “spine” will be complete after ST3, and I haven’t seen proposals for what else Pierce and Snohomish Counties want. Maybe Pierce County will want a Georgetown bypass line after they see how slow the Tacoma-Seattle ride through the Rainier Valley will be?

East Link will do a pretty good job of connecting the dense centers of the eastside. The ST Vision map’s additional cross-lake and around the lake lines would be horrendously expensive and offer little utility beyond express buses. Maybe a Kirkland and Totem Lake extension would make sense after another few decades of growth?

Meanwhile, Seattle has so many dense and underserved neighborhoods that the challenge is thinking through how to prioritize them.


“train cars running on a guideway…not including fixed guideway light rail systems”

Doesn’t that exclude what we have (a light rail system)?

Jesse Simpson

Yeah, the language was written in the height of the light rail vs monorail rivalry and purposefully excluded funding Sound Transit’s light rail.

My point is that the funding authority could be used to create a supplementary urban metro system along the lines proposed in the article, with smaller bore tunnels and a third rail. So we don’t necessarily need to go back to the state for more taxation authority to build something in Seattle.


Yeah, I get your point, but what I’m saying is that we can’t build light rail. I think every court would rule that Link is light rail. So that means that we can’t connect the lines, even without service. We can’t even use the same trains. We can’t even build bus tunnels, since this has to be rail. That leaves what — heavy rail? That seems crazy (for lots of reasons).

Realistically, we need to go back to the state, even if just means getting rid of that stupid clause.


“Heavy rail” that’s like Vancouver wouldn’t be terrible if it was short trains and high frequency. But yes, would need independent OMF.

If SDOT can find a place for the OMF (perhaps UW’s parking lots??), a Ballard-UW high frequency route would be something I’d support. The lack of interoperability will be silly, but Ballard-UW at 2 minute frequency will be better than Link at 6 minute frequency (identical people throughput).

As the author alludes to at the end of the post, automation will make high frequency much more economically feasible, which tilts the design towards shorter, higher frequency trains. If a different vehicle truly means cheaper tunnels, then it may very well be better to go with a new technology for Ballard-UW, and leave the Link technology for corridors that are best served at-grade or elevated.

Martin Pagel

It’s crazy that Elon Musk’s Boring Machine is putting Teslas in a tunnel, but I am hopeful he will help drive down the cost of small tunnels. I totally support Brandon’s call for a true urban subway along 8/48 line as I bet there is sufficient demand. I also like the hump station design.
Could we even use it for ST3 to Ballard to save money (to avoid covid crunch)? London, Munich, Berlin has plenty of subways running on the surface partially and ST3 only uses dedicated right of way meaning we could rely on 3rd rail rather than overhead. It might make it affordable to run a tunnel directly under the Ballard center.