Weighted Density

[mhays]

Originally Posted by memph

Toronto and Vancouver saw not just growth in downtown, but also a lot of growth in suburban nodes and new single family homes being built at higher densities (infill and greenfield).

Are there any US urban areas people think might increase their weighted densities? I expect San Francisco, Los Angeles and Miami will, and maybe New York and Phoenix.

Several cities have been seeing a large volume of infill (some today, many in the last boom) coupled with less outward sprawl. I'd add Seattle and Portland to your list...even while some sprawl still happens, and the volume of infill isn't on the epic Miami in 2006 scale.

[memph]

Originally Posted by mhays

Several cities have been seeing a large volume of infill (some today, many in the last boom) coupled with less outward sprawl. I'd add Seattle and Portland to your list...even while some sprawl still happens, and the volume of infill isn't on the epic Miami in 2006 scale.

I guess that's possible. Seattle and Portland's urban areas have a weighted density of 4500ppsm, which is quite low, so it wouldn't take too much for their densities to increase.

There are a few things to keep in mind when trying to evaluate whether the weighted density will increase.

Regarding sprawl: how dense is the sprawl? Is it much less dense than the average, equally dense or more dense?

Regarding infill, if the effect on Census Tracts is

Very low density becomes low density: may decrease the weighted density since there it involves an increase in the population living at densities below the previous average
Low density becomes medium density: will likely increase the weighted density slightly, but not too much. Many cities might have a lot of apartments being built downtown, but there is still a lot of parking lots, lowrises, offices, etc in the census tracts that contain those new apartments, so this might only represent an increase in density for those CTs from low to medium
Medium becomes very high density: will substantially increase the weighted density, although even in Toronto's condo neighbourhoods, it's mostly been medium to high or high to very high.
High becomes medium density: will substantially decrease the weighted density, especially if on a large scale. This can occur if there is decreasing household sizes in the inner city neighbourhoods but no construction of new units (or rehabilitation of abandonned ones).

When I say low/high density, I mean relative to the average

In the case of Toronto, you had new sprawl at densities roughly comparable to current densities to the average; intensification in areas that were moderate to high density (mostly condo towers); decreasing densities in a few inner city hoods (mostly early 20th century), but other inner city hoods made up for that with small scale intensification; some very low density brownfields became low density since they were only partially redeveloped; many low-mid density suburban areas held their population as teardowns brought larger families and occasional small scale intensification.

[mhays]

Seattle and Portland were originally dominated by bungalows (pre-war), and later (50s-70s especially) by larger lot suburbia. Both are interspersed by a lot of steep hillsides, ravines, wetlands, and other undevelopable land. The result is low baseline density.

But Oregon created a fairly ironclad growth management system. Washington created a system, though a looser one particularly outside King County. This is greatly reducing outward sprawl, particularly in Oregon. It's also helping encourage infill everywhere.

Both are seeing large amounts of infill, particularly Seattle at the moment. The result is apartment districts amidst the houses. It's a veeery slow process to push up even weighted densities, but they should be going up. As for average densities, even with growth management there's still outward expansion, particularly in Seattle, and depending on your methods, the outward growth might still outstrip the infill.

[Suburbanist]

I see a major conceptual problem with this "weighed density" thing: the area unit of measurement.

How do you define which are the units whose average weighed (on population) density will me considered?

There are multiple approaches possible, and they'd yield very different results. An extreme case would be considering each measurement unit as each individual building, which would make the so-called "weighed density" skyrocket.

If defining what constitute a "city area" for purposes of calculating population density is already a very contentious subject, making population comparisons between countries difficult, specifying an uniform criteria to measure weighed density is close to impossible, therefore rendering the utility of the concept as a tool for comparison almost moot.

Let me put other way: how would I calculate my density? Do I consider only the houses on the same block? Only other houses within an officially recognized neighborhood/subdivision name?

[memph]

Originally Posted by Suburbanist

I see a major conceptual problem with this "weighed density" thing: the area unit of measurement.

How do you define which are the units whose average weighed (on population) density will me considered?

There are multiple approaches possible, and they'd yield very different results. An extreme case would be considering each measurement unit as each individual building, which would make the so-called "weighed density" skyrocket.

If defining what constitute a "city area" for purposes of calculating population density is already a very contentious subject, making population comparisons between countries difficult, specifying an uniform criteria to measure weighed density is close to impossible, therefore rendering the utility of the concept as a tool for comparison almost moot.

Let me put other way: how would I calculate my density? Do I consider only the houses on the same block? Only other houses within an officially recognized neighborhood/subdivision name?

Definitely, the smaller the unit of measurement, the higher the weighted density. For the United States and Canada, the unit used was census tracts, which contain on average about 5000 people in all cities of both countries. This does mean that dense census tracts are going to have smaller land areas. I guess you could argue that it would be better to use units of the same land area instead of units of the same population, but to me weighted/perceived density means the density of the average neighbourhood, and in high density areas, neighbourhoods are often considered to be smaller in terms of land area.

In the case of European and Australian cities, they don't have census tracts, so that's why I didn't put them in the same list as the Canadian and American cities.

[memph]

Originally Posted by mhays

Several cities have been seeing a large volume of infill (some today, many in the last boom) coupled with less outward sprawl. I'd add Seattle and Portland to your list...even while some sprawl still happens, and the volume of infill isn't on the epic Miami in 2006 scale.

So I just ran the numbers for 2010 for SF-Oakland and Los Angeles' urban areas. Note that these two don't include the whole urbanized area but the area defined by the US census bureau which is affected by commute patterns. Also note that my 2010 numbers are based off the 2000 urban area definition, not the expected new 2010 urban area changes that would merge a lot of the urban areas.

San Francisco's urban area includes San Francisco, Alameda County (not counting Livermore area), San Mateo County, Contra Costa from El Cerrito to Rodeo and Marin County.

Los Angeles' urban area includes most of LA county except Santa Clarita, Lancaster-Palmdale and part of Malibu, most of Orange County except Lake Forest/Mission-Viejo and San Bernardino County from Fontana Westward.

San Francisco
2000: 15,032 ppsm (2,995,769)
2010: 14,740 ppsm (3,306,927)

Los Angeles
2000: 12,557 ppsm (11,789,487)
2010: 12,543 ppsm (12,238,786)

Maybe there are rather few US urban areas that got denser? I'm still expecting Miami's urban area to get denser though.

[Suburbanist]

In many cities, any "densification" via more residential units in a given area is outdone by the reduction on the average number of people per household.

[Northsider]

Today I leaned Las Vegas, Miami, and Denver have denser urban areas than Chicago. Hmm.

Good discussion though, I like it.

[mhays]

You can certainly densify core areas and scattered spots, while a lot of static areas see smaller household sizes, like suburban housing that hits the 20-year mark and the kids move out.

Those SF and LA numbers show population growth but less density. So they expanded the boundaries. The numbers would be higher otherwise.

[memph]

Originally Posted by mhays

You can certainly densify core areas and scattered spots, while a lot of static areas see smaller household sizes, like suburban housing that hits the 20-year mark and the kids move out.

Those SF and LA numbers show population growth but less density. So they expanded the boundaries. The numbers would be higher otherwise.

Maybe the urban area expanded a little, but I'm pretty sure there was no merging of previously separate urban areas. Also, it's possible for population to increase and weighted density to decrease even if the boundaries are totally static through intensification of very low density areas and decreasing household sizes in dense areas (or at least no change to these dense areas).

Example:
The core has a population of 100,000 in 10 square miles and the outlying areas have a population of 100,000 in 100 square miles. The weighted density is 0.5*10,000 ppsm + 0.5*1,000 ppsm = 5500 ppsm.

Not the outlying/low density areas get denser with 300,000 people in 100 square miles, this means that while the low density areas are denser, they also make up a larger part of the overall population. The weighted density is 0.25*10,000 ppsm + 0.75*3,000 ppsm = 4,750 ppsm.

Originally Posted by Northsider

Today I leaned Las Vegas, Miami, and Denver have denser urban areas than Chicago. Hmm.

Good discussion though, I like it.

Chicago has quite low density suburbs compared to many sunbelt cities, so the gross/overall/standard density of its urban area is lower. However, it's basically a situation where a small portion of the population lives on a large portion of the land, with the average Chicagoan living at relatively high densities. Weighted density takes this into account, so Chicago's weighted density is actually higher than that of those 3 other urban areas.

[memph]

Phoenix
2000: 5,238 ppsm (2,907,049)
2010: 4,716 ppsm (3,833,048)

Miami
2000: 6,810 ppsm (4,919,036)
2010: 7,445 ppsm (5,517,315)

So it seems Miami has gotten significantly denser. I didn't expect that kind of decrease for Phoenix, but it kind of makes sense. If you look at Phoenix, the sprawl is actually not that dense, but not only that, it's often scattered across the desert, with many gaps in development, but the gaps are not large enough to be considered separate from the urban area. Meanwhile, I'm not sure if Miami had such gaps in the past, but I guess because it's hemmed in by the Everglades, it doesn't have many now.

[memph]

Chicago
2000: 10,270 ppsm (8,307,904)
2010: 9,125 ppsm (8,814,602)
2010 including Kenosha: 9,047 ppsm (8,944,424)

Houston
2000: 4,514 ppsm (3,822,509)
2010: 4,589 ppsm (5,153,230)
2010 including Conroe: 4,554 ppsm (5,212,160)

Houston's urban area merged with Conroe, TX and Chicago's merged with Kenosha, WI so I have the numbers before and after. There were probably some smaller mergers as well that are included in the first 2010 values.

[memph]

While I'm slowly chugging away at NYC, Washington and Boston, here's a graph comparing the density distribution of San Francisco-Oakland and Toronto, which are have very similar weighted densities (Toronto just edges out SF).

The first graph show the percent of people living at certain densities and the second shows the total population at those densities. The official definition for San Francisco's urban area doesn't include San Jose, so Toronto's UA population is larger. I think for the 2010 census the urban areas will include all urbanized land in a CSA instead of the 2000 definition of all contiguous urbanized land in a MSA (more or less), so in order to be able to compare to the 2000 values, I followed the 2000 rules for 2010.

[memph]

I did two other urban areas, both of which saw their weighted densities increase. In the case of Boston, the urban area didn't really expand outwards. There was greenfield development, but it was mostly filling in gaps of undeveloped land. Washington had fairly substantial TODs built, so it's not too surprising it's weighted density increased.

Washington
2000: 6,835 ppsm (3,933,920)
2010: 7,405 ppsm (4,652,979)

Boston
2000: 7,711 ppsm (4,032,484)
2010: 8,159 ppsm (4,458,009)

[gabrielbabb]

In Mexico City 34,533 ppsm or 13,333 ppskm, for 20 million in about 1500km2 or 539mi2

[memph]

Originally Posted by gabrielbabb

In Mexico City 34,533 ppsm or 13,333 ppskm, for 20 million in about 1500km2 or 539mi2

Nice!

Did you calculate that yourself? What was the approximate population of the average district/neighbourhood/census tract used?