Tuesday, June 28, 2011

Density on the Ground: Cities and Building Coverage

Buenos Aires
Alon Levy has a new post up about height and density where he argues that to move beyond a certain density, a city will ultimately have no choice but to build upwards.  Density is not simply a function of the height of buildings, however, but also depends upon the area that they occupy.  A city in which buildings occupy a relatively small proportion of the total city area will have room to grow either vertically or horizontally (the latter of these represented by the process of street narrowing, or other forms of infill).  
Allan Jacobs in his book Great Streets features a section in which he presents painstakingly drawn figure-ground maps of several dozen of the world's best-known cities.  Although he calculates the intersections and number of blocks per square mile, both indicators of connectivity and pedestrian accessibility (see Laurence Aurbach's site for more on this topic), no figures are given for the proportion of city land occupied by buildings and rights-of-way (and other public spaces) for each city. 

Although performing these calculations for all of his chosen cities would be a major undertaking, I've put together some approximate figures, with abundant help from Google maps, for a selected range of cities, both American and not.  Shown are the proportions of each city occupied by 1) buildings or buildable land; 2) rights-of-way, including streets and sidewalks, and 3) other public spaces, including parks and plazas.  Unless otherwise stated, these figures refer to the general downtown areas of each city:

A few quick notes: St. Nicholas Houses is a public housing project in central Harlem composed of cruciform towers on a superblock.  New York's figures include Central Park.  Measuring park and plaza spaces in organic cities like Paris or Vienna is a bit of a challenge: depending on the area examined, for instance, Paris' figure for parks might be somewhat higher.  Fes is just an estimate, as the narrow footpaths of the city's old town are barely visible from above, but in any event it appears to be about the maximum possible built coverage possible without moving to an entirely streetless city.

Hopefully these numbers can provide one additional angle on the density question.  I have some ideas how they might be paired with Jacobs' figures and some other measures to provide an even fuller picture of both density and connectivity, but I'll save that for another post.

Related posts: Charting the Grid


  1. I'm half Taos Pueblo Indian. The village isn't entirely streetless. Check Google Earth and you'll see a few narrow streets and alleys cutting through both the north and south side. There's also ground level courtyards where you'll find adobe ovens. Still, I get what you're saying.

    The cool thing about Taos Pueblo is that you can actually see how it developed as a trade center; almost exactly as Jane Jacobs describes the growth of hypothetical New Obsidian in "The Economy of Cities."

  2. Fascinating! I'm surprised by Georgetown, though...would've thought the building coverage was higher. Chicago's a little surprising, too, but I suppose all those wide-ass streets take their toll.

    - Stephen, http://marketurbanism.com

  3. Thanks for that, Vince. I do wonder if the streets in the Taos Pueblo were somewhat more modern additions (like the ground level entrances). The ancient examples, such as Catal Huyuk, show enclosed courtyards but no streets.

    And great point about "New Obsidian." The ancient pueblos are so startlingly similar in their appearance to Catal Huyuk, and from what I've read their builders were also only partly agricultural. Yet the pueblo is a totally urban form – permanent and built at an extremely high density.

  4. Thanks for calculating these. How are you counting front setbacks in non-superblocks? Are they considered part of the street, or are they parks/plazas?

  5. That's a good question regarding when the streets developed at Taos Pueblo. I'm not entirely sure. The north side and south side have clear division between the older portions on the east ends directly adjacent to the kivas and the newer additions which create the streets in between the older part and the newer. Looking at Anasazi city layouts, I'd suspect that you're right, streets were a more modern addition.

    Taos Pueblo was certainly a regional trade center, even into the late 1800's. Plains Indians came from far and wide. Some families still even have parrot feathers (!!) that came up trade routes to Taos in pre-contact times from Central America. My understanding is that Buffalo meat and skins (game that was unavailable locally) was and still holds important value in the pueblo. This, of course, came from the plains Indians. Exports certainly included pottery and cutting implements. You can't walk in the foothills of Taos Mountain without finding ancient arrowheads. I'm not sure how many people were employed purely as artisans and craftsmen.

    For a good example of that I'd look to the semi permanent settlements of the northwest coast Indians (I'm half Tlingit!) The Indians in that region traded heavily in locally procured native copper, fermented fish oil (liquid gold,) abalone, and finished goods adorned with such materials such as bentwood boxes, carvings, etc. Indians up that way supported an artisan class. The clans up there can still identify their territories and what resources we used to guard jealously back in the day. Poaching another clan's copper would earn you a visit from this guy: http://duanepasco.com/photos/portf.high/hats.high/port.hat.ravwar2h.jpg

    Anyway.... I could go on and on.

  6. Thanks for putting this together. It's very interesting to see. Are you counting sidewalks as street area or plaza area -- I assume street area? I would love to read about your method and what you think it means.

    I would really like to see off-street parking as a separate category instead of including it in buildable area. In America, it exists by legal mandate and is not buildable unless you can afford to put in several stories of underground parking, usually costing multiples of the price of the building itself. It would really reflect a contrast between the old urbanist pattern and American forced suburbia. Also a place like Tokyo with no on-street parking should acknowledge that it shifts that function onto off-street parking, which usually takes much more land per vehicle.

    In Chicago or Tokyo, this may be very consistent from place to place, but in other cities the result really depends on where you take your sample. In Barcelona, you picked Eixample but a measure of Barrio Gotico which has a similar land area would show something like 90-7-3 instead of 73-27-0. Those neighborhoods grew up in very different times and cultural circumstances but they share a culture and a long busy border today.

    Similarly you probably picked someplace like San Telmo -- but with less park land -- in Buenos Aires. The newer parts of the city have much more street area.

    Most of the neighborhoods I know in Mexico City would be roughly 70-25-5 but the exact number depends on whether your sample includes any of the big parks. About 7% of the land in Mexico is on-street parking. That isn't the only kind of neighborhood that exists, though. There are some narrow street places with 85-10-5 and lots of Xochimilco (pop 500,000) is 70-15-5-10 with the final 10% being canals. Some parts are 60-0-10-30. Mexico City has more people living on canals than anywhere else I know, but Venice would present the same confusion.

  7. OK, I'll go on and on....

    Check out Acoma Pueblo on Google Earth. It's West of Albuquerque. The old village of Taos Pueblo didn't fully embrace the automobile, it appears that Acoma did and it really shows in the layout of their community. I've never been there, but now I want to pay a visit.

  8. Eric and Brian: Sidewalks and mandatory front setbacks were both counted as "streets."

    Vince: Go on all you like! I think the history of that area is fascinating, and I'm learning a lot from your comments. I've never heard of the Acoma Pueblo before -- although it does look like there have been concessions to the car, the view from above doesn't seem to show many vehicles (if any) there, which is a hopeful sign.

    Brian: You're absolutely right with your comments. The Barrio Gotico would probably work out similarly to Verona. As for Buenos Aires -- although the outlying areas do have wider streets, as you point out, the blocks themselves are much more intensively developed than, say, a bungalow neighborhood in Chicago, or a streetcar suburb in Portland, which tend to have mandatory setbacks. Buenos Aires' density advantage therefore might actually increase as one moves away from the center.

    As far as parking -- that would be a challenge to calculate, but might be feasible. Certainly measuring the amount of surface parking could be done. I do have more thoughts about these numbers, but they'll have to wait for another post.

  9. I think it would be nice to see more pictures of the areas you are referring to. Saying that a "city" has a certain ratio is vague since a city is so large, and has so many examples of different forms of development within it, but if you refer to a particular form of development as represented in a photo of a real place, that is very illustrative.

    The Buenos Aires example really shows what a place with an 85% building footprint looks like. A photo of the St. Nicholas Houses would be a great example of the idea of tall buildings but very small footprint. These terms like "density" don't really mean anything to most people. What tends to happen is that they associate the word with some form that they are familiar with, which is very often similar to the St. Nicholas Houses form (20th Century Hypertrophism) of tall buildings, wide streets and lots of Green Space. So if you start to talk about "density," in their minds they tend to flip-flop between this 20th Century Hypertrophic model and the Suburban Hell model for "low density."

    However, if we see some real-life examples like Buenos Aires, we can see the great variety of real-world solutions to putting buildings on land. This includes things like setbacks, parking, "green space," and many other topics too burdensome to talk about item by item. This is one reason I like lots of photos and not so many words.

  10. It's an interesting comparison, Charlie - thanks for the compilation. I presume that for American cities you did it by looking at the basic grid pattern, but how did you do it for European cities? Trying to figure out Paris's street/building breakdown sounds hellish to me.

  11. Alon: the American cities were done as you said, by finding the basic "building block" of the grid (a city block with streets on two sides), and adding up the respective areas, which was easy and accurate enough.

    For Paris and Vienna -- it is true that doing the same sort of analysis there seems like the stuff of nightmares. I simply tried to take a representative area (in Paris' case about 15 blocks) that had boulevards and lesser streets in the same proportion as the city as a whole. It may not be exactly the same as the total for the entire city area, whatever that may be, but I do not think it can be very far off.

  12. Is there a relationship between built area and height? Because I note that the places on that list with the highest built percentages are also fairly low-rise. The French Quarter in N.O. is mostly 3-story construction, and old Buenos Aires or Verona probably don't go above 5 floors.

    In other words, is there a trade-off? I doubt that anyone would want to live in a place with 75% built area covered by 10 to 20 story buildings (the old warren of Kowloon comes to mind, before they demolished it). What's the sweet spot for density which is still livable?

  13. Cambias: the related link I have at the base of the post features a chart which tries to express this relationship. Widening streets and building taller does result in a net gain in floor space/acre, but it's not dramatic. A city like New Orleans needs only about 1.8 times the acreage to equal a the same amount of square footage built in the form of 10-story buildings on wide streets (and yet the density of the French Quarter is already amply sufficient for walkable urbanism).

    You can build taller on narrow streets and it will appeal to certain people under certain rare conditions (lower Manhattan), but in general I think a strong argument can be made there's not much benefit to exceeding the 3/4 story/narrow street combination under most circumstances.

  14. Cambias: Ultimately it is a question of aesthetics -- what do you like? Personally, I much prefer a Traditional City layout of narrow streets and 3-6 story height, which can achieve very high densities.

    However, to begin to answer your question, yes, I think you can combine narrow streets with buildings of 5-10 stories. This is something you tend to see in places with very, very high densities. In fact, my own parents lived in an area like this in Ueno, central Tokyo, and the overall effect was quite nice. Here's a photo:


    This is a combination of rather taller buildings and "Traditional City style Really Narrow Streets." Note how there is no automobile traffic on the streets, and people are walking in the middle of the street. The street width here is about 16-20 feet. This is a ten minute walk from Ueno Station, one of the largest train stations in Tokyo.

    It starts to get rather dim at street level. If you go much beyond ten stories, it can get quite dim at street level, unless you make the streets much wider, which of course lowers your density and makes automobile traffic the dominating characteristic of the city.

    I actually think there is a potential hybrid model, of rather tall buildings (30 stories let's say) and what amounts to the traditional "square," or parks. Basically, a tall building tends to need some space around it, but if that space is useless "green space" or very large roadways, the overall result is a failure. However, if we combine tall buildings with a useful space -- basically it would be a park, square, natural area like a beach or river, or a large "boulevard" type street -- then we could have our tall buildings without giving up anything. I don't really know of a good real-life example of what I mean, unfortunately. The basic idea is tall buildings but WITHOUT large roadways, which for some reason everyone thinks you need with tall buildings, and much more detail at the street level.

  15. Nathan and Cambias: I had envisioned something of a hybrid, also, with a city of mainly single-family attached homes at high density, but as Nathan suggests interspersed with public squares that require taller buildings simply to maintain a sense of enclosure. Perhaps not 30 stories, but much taller than what's surrounding it.

    And Nathan: that photo of Ueno seems to indicate there is a setback requirement above 3 stories -- are you aware of such a requirement? I think it produces it wonderful aesthetic effect and probably keeps the streets reasonably bright, but it also makes 6 or 7 stories the effective maximum on small lots.

  16. Esthetically, I think my sweet spots is when the building-to-building street width and the height of the street wall are about the same. It can be off by a factor of about 2 depending on whether the context is meant to be suburban or CBD, but otherwise I feel either that I'm in an alley (if the height-to-width ratio is too high) or that the scale is pedestrian-hostile (if it is too low). Absolute street width is only a problem if it makes the street uncrossable in one cycle, and that's a question of signal phasing more than anything.

    If the street is very wide, then one trick that can work is an elevated train. Consider the 7 el above Queens Boulevard in Sunnyside: it works much better than both the el over Roosevelt (a much narrower street) in Woodside or over 31st in Astoria, and the el-less Queens Boulevard in Central Queens. I've been thinking about it, and it seems to me that the el esthetically breaks QB into two less wide halves, and is about as high as each half is wide. Perhaps this is why QB seems like less of a highway in Sunnyside, and why the impossibility of crossing both halves in one cycle bothers me less than the same on unbroken streets such as in Tel Aviv. (It would be even better if there were retail under the tracks instead of parking).

  17. I think the ideal height to width ratio depends on the location of the city though. If you live in Helsinki, which can get rather cold and is rarely uncomfortably hot, the length of day gets below 6 hours in the winter and the sun is usually quite low (so buildings cast long shadows), you wouldn't want too tall buildings to allow as much sunlight into the streets as possible. In Miami on the other hand, the sun is higher (and stronger), the length of days varies much less between winter and summer and it's never cold but often uncomfortably hot, so having the streets in the shadows might be seen as a good thing.

  18. Perhaps it's too late to reply, but Anonymous 13 Dec 2011 makes a very good point I think. I live in Melbourne, where streets are mostly wide (and outside of the CBD, buildings quite short), but in retail areas there's usually verandahs. There are some quite narrow streets (Centre Place for example), and I've also lived in Europe, with narrow streets and taller buildings as standard.

    On a sunny day, walking down a narrow street with tallish buildings has much the same effect as walking down a wide street with verandahs, and a narrow street without verandahs is far more comfortable than a wide street without verandahs.

    In any case, with streets the widths of Melbourne's, verandahs are the only way to go.

  19. These are actually wonderful some ideas in the blog. You have touched good quality points here. In whatever way continue writing. Bryanne

  20. This comment has been removed by the author.

  21. I was wondering what you counted as Tokyo, or as parks -- Chiyoda looks like it might be 1% of the 23 wards by itself. I just noticed the comment about sampling part of Paris and other complex layouts. Does that make the chart "fabric" numbers, e.g. the NYC 6% is counting lots of small parks, but not Central Park? Or it is counting Central Park, because that was easy, but not necessarily other semi-big parks?