Check out this last of America's largest urban areas ranked by weighted density. You'll see that Los Angeles, despite its reputation, is surprisingly dense. Conversely, transit-friendly Portland isn't especially dense (less so than Houston or Dallas or Las Vegas) which goes to show how much smart policy matters -- if all 23 denser-than-Portland cities on the list were as savvy as Portland about bikes, pedestrians, and transit we'd have a much better environmental situation in the country without constructing any new, denser urban areas.
« A Friend in Need | Main | Hey -- People Put Out New Albums! »
Density Index
26 Mar 2008 12:38 pm
Comments (43)
The problem with that list is that it compares the metro areas, not the actual cities. So any discussion of mass transit needs to take that into account. With few exceptions, mass transit is primarily prevalent within city borders, and not nearly as much in the suburbs.
With few exceptions, mass transit is primarily prevalent within city borders, and not nearly as much in the suburbs.
Isn't that sort of the problem?
Charting anything by city limits is absurd, because city limits are often very arbitrary. Some cities are drawn so that lots of the suburbs are in city limits (LA, NY, Houston, San Antonio), others are drawn so that only the city core is in the city limits (SF, DC, Vegas). You have to use metro areas because it's the only way of comparing apples to apples.
Interesting discussion. I think there are inherent problems with measuring density. If we are to include metro area, where does it stop and start? Chicago's Metra interurban goes into places like Harvard Illinois which is surrounded by farms and Kenosha Wisconsin. Are they included in Chicago's density? Does metropolitan Milwaukee include Pabst farms; fields where they once grew the wheat for PBR?
Thanks for the link.
These figures are for "urbanized areas," which the Census Bureau defines as one or more "central places" (i.e., large, incorporated cities), and contiguous areas with a density of at least 1,000/square mile. Metropolitan areas typically sweep up much more land. One of the nice things about weighted density, though, is that switching to a larger geographic unit has little effect, unless you add a lot of population.
Portland obviously has dense corridors well suited to mass transit. So do most of the other cities above it on the list.
jburnout, large parks have little effect on weighted density; their census tracts typically contain few people, so their density gets a miniscule weight.
Not to get all technical, but I suspect that density-weighted-by-percentage is going to have scaling problems-- particularly with the unequal tract areas-- so comparisons among places with different populations and different areas may be systematically biased in some way. I'd try 1) equalizing the tract areas and 2) raising the result to some exponent (maybe a square root) before believing any comparisons.
Since Pabst Farms is you know 30 miles away from Milwaukee I doubt it.
Metro areas are good if you want to look at total population of an area, but if you're looking at density, they're very misleading, because they're contiguous with county lines. L.A. County (and, hence, its metro area) includes wide swaths of sparsely-populated desert, mountain, and island land. San Bernardino-Riverside, even more so --- much more so. San Berdoo County alone is larger than Massachusetts, Rhode Island, Connecticut, and Delaware combined. Together as a metro area, their size would rank between South Carolina and West Virginia.
If you're looking at density, urbanized area is the way to go. I'm intrigued by this concept of weighting tracts, but I'll have to look at it more closely.
Who knew that Baltimore was more dense than Washington DC? I suspect that's because the DC measurement includes Northern Virginia.
Eastern cities don't tend to get reasonable ratings in these studies because the sphere is so large. New York is a dense city by any measure, but when one considers New York to be everywhere from Trenton to New Haven, and the Hamptons to the foothills of the Catskills, it doesn't come out very dense by math.
Also, what's more important than density is layout. A bunch of subdivisions with homes really close together (as is common in the Southwest) may look good when you divide people by acreage, but it doesn't confer any of the benefits of density in the traditional sense - convenience, community, less environmental impact, etc.
The study is correct, Portland isn't that dense. Most people live in houses or duplexes, and there are very few high-density residential neighborhoods or apartment blocs, though that is changing with the development of the Pearl District.
All my friends own or rent single-family homes - though there are often three or four people living together in these houses.
I used to work in one of the taller buildings in Portland, and when you look out on the city, especially the east side, it looks like a giant park, due to all the trees. Many of the homes people live in aren't even visible from that altitude.
The parks might contribute, but Portland, especially the east side, feels much more like a suburb than a city like, say, New York or Chicago or LA.
I keep looking at this list and then looking out the window and thinking, "Huh? Phoenix is denser than Pittsburgh? Milwaukee is denser than Cleveland?" It occurs to me that these metrics are useful, but do not reflect things like small parks, bridges, rivers, spaces where development is hampered by topography, etc... Also, places like Phoenix and Las Vegas have had significant sprawl growth since 2000 and that colors my credulity a bit.
Interesting stuff nonetheless. I wonder how much population density really impacts public transit use in the first place. Data from the same census (http://www.census.gov/Press-Release/www/2004/pubtrans_place.xls) shows cities further down the density list like Minneapolis, Atlanta and Pittsburgh as having much higher public transit use than denser places like LA, San Diego and so on. I suspect there is a cultural element. Some places stigmatize transit use as being a "poor" phenomenon.
The study is correct, Portland isn't that dense. Most people live in houses or duplexes, and there are very few high-density residential neighborhoods or apartment blocs, though that is changing with the development of the Pearl District.
All my friends own or rent single-family homes - though there are often three or four people living together in these houses.
I used to work in one of the taller buildings in Portland, and when you look out on the city, especially the east side, it looks like a giant park, due to all the trees. Many of the homes people live in aren't even visible from that altitude.
The parks might contribute, but Portland, especially the east side, feels much more like a suburb than a city like, say, New York or Chicago or LA.
Really flawed chart, considering that it can take somewhere between seconds (SanPedro) and over an hour (northwest SFV) to get from "Los Angeles" to LB and then another 15 to 30 minutes from there to SA.
And, does "L.A." mean the whole county, the city itself, the city with some suburbs, or what? If the first, that includes two mountain ranges and one of the world's largest city parks as well as huge undeveloped tracts in the northern part.
See also this. That combines two core themes of L.A.'s wonderful Democratic mayor: a) being in the pocket of builders, and b) trying to obtain power through illegal immigration (youtube.com/watch?v=3jwqQ8DtlPQ).
Excellent post by AC. His approach is designed exactly to deal with most of the problems being raised here -- arbitrary boundaries, the distorting effects of large parks, etc.
Just one quibble -- he presents his table as a jpg, which is the worst possible format for something like that. It's designed for photos and other images with smooth gradations of color. Images with sharp lines and text should be gifs. But why make it an image at all, instead a format that would allow it to be indexed by search engines, copied from, etc.?
I keep looking at this list and then looking out the window and thinking, "Huh? Phoenix is denser than Pittsburgh? Milwaukee is denser than Cleveland?" It occurs to me that these metrics are useful, but do not reflect things like small parks, bridges, rivers, spaces where development is hampered by topography, etc... Also, places like Phoenix and Las Vegas have had significant sprawl growth since 2000 and that colors my credulity a bit.
Interesting stuff nonetheless. I wonder how much population density really impacts public transit use in the first place. Data from the same census (http://www.census.gov/Press-Release/www/2004/pubtrans_place.xls) shows cities further down the density list like Minneapolis, Atlanta and Pittsburgh as having much higher public transit use than denser places like LA, San Diego and so on. I suspect there is a cultural element. Some places stigmatize transit use as being a "poor" phenomenon. Industry demographics might also be a factor.
I guess successful mass transit isn't predicated on density as many detractors would suggest.
I get the impression that some persons commenting above are not quite grokking what AC is up to here. I refer you to the full presentation at AC's blog, but here, I think, is the gist of it:
Rather than measuring the number of people the average unit of ground has living on it -- the usual measure of density -- AC is trying to derive a statistic for the number of people the average person lives "near." Because we are people, not units of land, this statistic is a better measure for making comparisons of perceived density than the usual measure (or so AC argues, and I think convincingly).
Accordingly, the objection that some cities have large areas of land that are unpopulated or sparsely populated doesn't quite work as an objection to AC's calculation. Indeed, that's one of the problems with the usual measure of density, and it's one of the things that AC seems to be trying to correct for in deriving this new statistic. In AC's calculation, those sparsely populated areas drop out of the equation (or, more precisely, are given appropriately little weight).
Also, I agree with lemuel pitkin in that it would be helpful if the chart were posted to Scribd or Google Docs or somewhere like that in a more useable format.
Once again, people, firstly, he's looking at urbanized areas, rather than metropolitan areas, so the vast spaces of San Bernardino County aren't included.
Secondly, by weighting the census tracts, he's taking into account the fact that, for instance, LA's urban area is tightly bound by mountains and deserts, and thus doesn't include much of any area where nobody lives, whereas New York's metro area extends out to very loosely populated suburbs in New Jersey and eastern Long Island and Putnam County. (Thus, the LA Urbanized area actually has a higher unweighted population density than New York, which is obviously silly).
The gradient number is very interesting - the cities with the highest gradients would be the ones with the largest very dense areas, and would probably be the ones people tend to think of as dense, even when they're not actually all that dense. When you order by gradient, the top 4 are New York, Boston, Philadelphia, and Chicago - i.e., these are the urban areas with the most variety within them, in that they all span from very densely populated central cities to very sparsely populated exurbs. On the other hand, places like Atlanta, San Jose, and so forth have low gradients, suggesting that their density is about the same throughout.
It occurs to me that these metrics are useful, but do not reflect things like small parks, bridges, rivers, spaces where development is hampered by topography, etc.
Yes they do. Read the post.
Alkali's summary is exactly right: Usually "density" is used to mean the number of people on an average piece of land. AC is using it instead to mean the amount of land around an average person.
One of the big advantages of the latter approach is that adding plots of land with few or no people on them, whether in the form of undeveloped areas within the urban area or sparse suburbs at its fringe, doesn't change density measured in this way.
Arrrrgh. Has anyone used the term "density queen" in reference to our young master here?
Walking and riding bikes in Southern cities is nowhere near as pleasant as the far NW.
I get the impression that some persons commenting above are not quite grokking what AC is up to here.
You're right. Read it too quickly and filled in the blank spots with assumptions (I'm in an airport - hard to concentrate at times). I suppose topography might still produce some random anomalies here and there, but not enough to really skew the model. It would be interesting to see a progression of this data from, say, 1985 to present and watch how the numbers change as gas prices go up and the housing market sorts itself out over the next few years.
Anyone familiar with streetcar suburbs knows that density does not have to reach Manhattan levels for public transit to work. Indeed, that is what you are seeing in places like Pittsburgh, where the residents of several former streetcar suburbs (now part of the urbanized area) are using busways and light rail for commuting.
On that subject, a poster above wrote:
"I keep looking at this list and then looking out the window and thinking, 'Huh? Phoenix is denser than Pittsburgh? Milwaukee is denser than Cleveland?'"
At least in the first case, I think it is quite useful to look at the last number in the chart, the "density gradient index". Although Pittsburgh has a lower weighted density than Phoenix, it has a higher density gradient index. If I have the right mental picture in mind, that means Pittsburgh's population distribution is a bit more peaky than Phoenix's, even though Phoenix is more dense overall.
And in light of the streetcar suburb insight, that seems right to me. I bet in Pittsburgh this would not actually amount to one big peak Downtown, as it might in Manhattan, but rather a bunch of little ridges and bumps across the urbanized area, corresponding to former streetcar lines and streetcar suburbs. In between you will find intermingled lower density neighborhoods, and for geographic reasons that starts fast in Pittsburgh (meaning you don't have to get too far from Downtown Pittsburgh to find low-density areas, but not so empty as to not register in the weighted calculation).
Anyway, this is a very interesting set of measures, and I agree it is at least getting closer to something that has a natural interpretation than "standard density" alone.
"I guess successful mass transit isn't predicated on density as many detractors would suggest."
No, but extreme civic commitment to public transportation helps. I doubt you could replicate Portland's love of public transit elsewhere.
A little in a hurry, so I can't read the linked article and chime in.
But I make a wild guess, and I will say that the following link could help/ interest a lot of people here.
www.alain-bertaud.com
It doesn't load at the moment, so I can't found the fascinating article there where I got some insight about the way to do such measurements between cities.
By the way, don't worry: IRC, professionals didn't wait for us dilettanti and there are standard rules about parks and rivers...
OK,
try that one:
http://alain-bertaud.com/AB_Files/Spatia_%20Distribution_of_Pop_%2050_%20Cities.pdf
Contrary to what a lot of the commenters are asserting, central city density really DOES matter.
If you look at transit ridership, you will find that the cities highest on the list don't necessarily have high metro area density, but they DO have all high central city density.
The five big cities (over 250,000) where over 25% of commuters use public transit are, in order,
1. NY
2. DC
3. Boston
4. SF
5. Newark
6. Chicago
7. Philly
All of them but DC have over 10,000 people per square mile central city density.
By contrast, cities like Houston have about 3000-4000 per square mile (and granted, that's partially a result of annexation- but I think even if you knocked Houston down to its inner 100 square miles, you'd have a city far less dense and far less transit-oriented than the cities listed above). Having a dense core REALLY MATTERS. Having dense suburbs 20 miles away doesn't matter nearly as much.
Although Pittsburgh has a lower weighted density than Phoenix, it has a higher density gradient index
Don't read too much into the gradient index. Like the contentional density measure, it's sensitive to how the borders of the "urban area" happen to get drawn, so it doesn't tell you anything in particular about the city in question.
Los Angeles really is now very densely populated, with people living in garages, sleeping on couches and floors of apartments, and homes being occupied by extended families with 3 to 5 incomes. It would be nice to wave the magic mass transit wand and solve all of LA's problems, but it's not going to happen -- there's no pattern to the density that can be conveniently served by rail lines. LA is now pervasively dense, so like most cities it will remain dominated by car commuting.
Welcome to the future.
The five big cities (over 250,000) where over 25% of commuters use public transit are, in order...
Interesting point. And many of the next several down the list have tight, pre-war street grids in their urban cores simulating a similar density even if that's not really the case. A city like Detroit would have a similar simulated density, but with a smaller percentage of downtown workers as part of the regional workforce. Maybe workforce density is a bigger factor, or at least a commensurate factor, in determining public transit use.
Apropos of nothing, there are a lot of 50' wide lots in South L.A. that are zoned for two houses. "Immigrants"/immigrants would buy the twin houses, convert (ex-building inspector) the garage into a bedroom and have four families living on one 50' wide lot.
And, a few years ago as a test I tried to get from LosFeliz (north of downtown, northeast of Hollywood) to LAX via public transit. It took four hours, including two buses, two trains, and then finally yet another shuttle bus at the end.
I've lived in LA for over 10 years and couldn't tell you where the hell Santa Ana is. And have never had any reason to go to Long Beach, other than go to the airport.
Does anyone know where Santa Ana is? Without looking at Google Maps?
Wow. Really great stuff.
One point on the observations that you need 1) a dense city-core and 2) a committed populace to have mass transit, one way to go about building this is to do what Houston did in starting with a light-rail in the very dense (with workers) downtown and connecting that to nearby neighborhoods that are beginning / in need of revitalization.
In cities unfamiliar with mass transit and accustomed to car use, you build political commitment by showing people the concrete benefits of mass transit (let's face it, most people would normally prefer to drive - there's a reason that so many rich people, even in New York, take taxis). So, in addition to advocating for transit generally, I suggest that those who would prefer that more transit come to their city also propose a well thought out and yet less expensive 'proof-of-concept' rail line or two to get the ball rolling.
Just my two cents - again, congrats on what seems like some fine analytical work.
OK,
I keep hammering, but with a more amateur-friendly article of that urbanist:
http://alain-bertaud.com/images/AB_Metropolis_Spatial_Organization.pdf
It seems he has now calculations for more cities, but that older article explains it better.
Steve Sailer,
You are aware that LA has a subway, and that it is expanding? The future indeed.
Does anyone know where Santa Ana is?
Um, yes. I'm pretty familiar with just about every corner of Southern California. What exactly is your point? Because *you've* never been to Santa Ana, it doesn't matter?
You should get out more, and get familiar with something beyond your immediate neighborhood. Really, you might learn something.
I said over at AC's blog that a good density measure has to start off with an urban center of mass (which weighted density would be a useful tool for defining and locating), and give greater weight to people who are closer to the center.
Why? Imagine two chessboard cities, with each square of the chessboard being a square mile. Let's give each city 12 squares with 10,000 people each, and 52 squares with 1000 people each.
No matter how you arrange the 12 dense squares on the board, neither the density nor the weighted density changes in the least.
So one chessboard city can have all of the 12 denser squares nicely concentrated in the squares c3, c4, c5, c6, d3, d4, d5, d6, e3, e4, e5, and e6. And the other can have them dispersed over a1, a4, a8, c3, c6, d8, e1, e3, e6, h1, h5, and h8. But they're equally dense by both measures.
This is why a density measure that takes compactness into account is needed. You can use AC's weighted density to help you determine the urban center of mass, but from there, you've got to weight persons closer to that center than you do persons further away, or compactness won't matter.
But in real life, compactness is a major asset of a downtown.
Portland gets a better rep than it deserves. It shifts its sprawl into Washington. The fastest growing county in Washington is Clark County which is home to Vancouver--Portland's very own ugly suburban commuter community.
Portland gets a better rep than it deserves. It shifts its sprawl into Washington. The fastest growing county in Washington is Clark County which is home to Vancouver--Portland's very own ugly suburban commuter community.
Low tech cyclist,
I believe the point you made is exactly what one can find in my last link, with the bonus of nice 3D illustrations for real cities.
Comments closed April 09, 2008.
| 
Heads in the Sand
Not to really argue with a great point, but Portland's density stats might not be taking into account the fact that Portland contains several huge parks, including one that is over 5000 square acres. I suspect that if you excluded the five biggest parks, from the total area (leaving in the 'normal' playground style parks), the density would jump up several slots.
Posted by jburnout | March 26, 2008 12:58 PM