SkyscraperPage Forum - Population density of cities by neighborhood (census tract)

SkyscraperPage Forum (https://skyscraperpage.com/forum/index.php)

- City Discussions (https://skyscraperpage.com/forum/forumdisplay.php?f=24)

- - Population density of cities by neighborhood (census tract) (https://skyscraperpage.com/forum/showthread.php?t=204523)

Population density of cities by neighborhood (census tract)

I made a series of maps. They show the population density (people per square mile) of census tracts in the central counties of the 20 largest US urbanized areas. The data is from the 2010 US Census. The maps are all to the same geographic scale, and the color scales representing density are also the same on each map (except if you look at the high-end number on each one you can see the peak density for that city).

Anyone can make maps just like these for anywhere in the US, but it is complicated. I had to get a friend who works at the census bureau to help me. Here's the process:

Go to factfinder2.census.gov
Click "advanced search"
Search for topic/table name "GCT-PH1", but do not actually select any of the results.
Once the results are listed on the screen, click the "geographies" button.
Use drop-down menu to select "..... County - 050", then navigate to the county you want.
After you've selected the county, click "close" in upper right corner of the geographies pop-up.
Click the table titled "Population, Housing Units, Area, and Density: 2010 - County - Census Tract"
A new page will load. Click the "Create a map" link.
Find the table column for population, under the desnity per square mile header, and click on the 2nd number from top (do not click the top number).
A pop-up will appear. Click "show map".
When the map loads, click "colors and data classes"
Change transparency to 0%
Change number of classes to 4 (if it is not already).
Change classing method to "user defined".
Set the values to be:
0 - 9999.9
10000 - 19999.9
20000 - 39999.9
40000 - (whatever is the max)
(If you city has a max below 40,000 you will need to adjust the scale, because the webpage will not accept a null result. Just tell us if you do.)
Click "go".
When the map loads, take a screencap.
To go back and make another one for a different place, click "back to advanced search" and select a different geography. You will need to remove the county you don't want anymore from the "your selections" panel in the upper left.

Now here are the 20 cities ranked according to peak density (that is to say, their single densest census tract):

Chicago: 508,697 (This is likely a fluke caused by a prison or something, and will require looking into more deeply.)
New York: 200,764
San Francisco: 161,499
Boston: 110,107
Los Angeles: 94,490
Baltimore: 86,889
Miami: 77,214
Washington: 66,782
Philadelphia: 64,262
Houston: 55,254
Seattle: 51,128
San Diego: 50,077
Dallas: 44,099
Minneapolis: 25,732
Denver: 23,724
Phoenix: 23,431
Atlanta: 21,189
Detroit: 18,404
Tampa: 13,962
Saint Louis: 13,329

Digging into the data a little further, here are maps showing only tracts above 100,000 ppsm. Oddly, I cannot find that Chicago tract, and there don't appear to be any others that are above the 100k threshold in that city. Very weird.

Chicago (none visible):
https://farm9.staticflickr.com/8239/...f7305ce7_b.jpg

New York:
https://farm9.staticflickr.com/8382/...f5535553_b.jpg

San Francisco:
https://farm9.staticflickr.com/8377/...9b200a78_b.jpg

Boston:
https://farm9.staticflickr.com/8532/...926cda9a_b.jpg

I found the Chicago mystery. The 500,000 ppsm tract is super-tiny. It's a couple of very tall residential skyscrapers, and nothing else. Here's an aerial.

https://farm9.staticflickr.com/8511/...02ac8721_b.jpg

It is weird that Edgewater Beach highrises get such a tiny census tract for themselves.

Link

Quote:

Census tracts are small, relatively permanent statistical subdivisions of a county. Census tracts are delineated for most metropolitan areas (MA's) and other densely populated counties by local census statistical areas committees following Census Bureau guidelines (more than 3,000 census tracts have been established in 221 counties outside MA's). Six States (California, Connecticut, Delaware, Hawaii, New Jersey, and Rhode Island) and the District of Columbia are covered entirely by census tracts. Census tracts usually have between 2,500 and 8,000 persons and, when first delineated, are designed to be homogeneous with respect to population characteristics, economic status, and living conditions. Census tracts do not cross county boundaries. The spatial size of census tracts varies widely depending on the density of settlement. Census tract boundaries are delineated with the intention of being maintained over a long time so that statistical comparisons can be made from census to census. However, physical changes in street patterns caused by highway construction, new development, etc., may require occasional revisions; census tracts occasionally are split due to large population growth, or combined as a result of substantial population decline.

Just wondering if the fact that Miami's census tracts span Biscayne Bay for some reason (rather than being clipped at the coast like they seem to be in every other city) throw off the density figures for what should be Miami's densest area - the west coast of South Beach. Clipping the census tracts at the shore line and using that new area to count density would be more accurate.

-edit: It looks like the area values are clipped to the coast.

There are some oddities with how Miami is being shown, but they skew the results in both directions.

Here's Miami zoomed in to the core, and with only tracts above 40,000 ppsm highlighted. Most of the densest tracts already hug the coast, and are quite a bit smaller than the typical neighborhood tract from New York or San Francisco, which skews them to read as "too" dense. On the other hand, there are some tracts that don't hug the coast, most notably Brickell Key and the coastal tracts north of 15th Street; but if they were reduced to hug the coast then they would be too small, so it's a tough problem.

Regardless, I think we can say with authority that Miami's high-rise coast is extremely dense, but that its neighborhoods are moderate.

http://farm9.staticflickr.com/8246/8...8227cef3_b.jpg

The tract directly to the west in Brickell would probably be over 40k if it were cut at the narrow pinch point. Also the tract west of that might too if it didn't include that little peice of commercial area west of I-95. Especially now. Maybe not so much in 2009/2010. Too bad they only do tract level population counts every 10 years. The donut hole tract just west of the Brickell Key tract would be included if they took the census now that the 1800+ unit Icon Brickell went up in that tract (occupancy levels were pretty low until 2011/2012).
The South Beach tract just below the 2 40+k tracts would be included as well if it didn't have that little jaunt to the east for some reason. Even with those less dense blocks included it just misses the cut at 39,900/per square mile.

the only prison/jail that would qualify is Cook County Jail on california south of the courthouse at 26th St, busiest single-site courthouse in the country

It's one of the biggest pretrial detention facilities in the US with a population of 10,000+ most of the time

Quote:

Originally Posted by dave8721 (Post 6061028)

The same could be said of just about every tract in every city in the country. If you set out to define them solely to produce the highest possible density number, everyone would have much higher numbers all throughout.

Quote:

Originally Posted by Cirrus (Post 6061234)

Of course. But what I was doing was actually simplifying things. Basically, all of Brickell: East of I-95, South of the Miami River, north of SW 15th RD (the diagonal street that runs from the bay to I-95) is a 40k+ per square mile neighborhood as well as all of the West Avenue corridor in South Beach.

Great maps, thanks for putting them together.

I'd be interested to see these same maps with change in density for each census tract from 2000 to 2010. I wonder if that's possible from the data available on the census factfinder.

It wouldn't be possible using that process. You'd have to download the data and do it on a spreadsheet, then import the spreadsheet to GIS. But I do have maps from the 2000 census that you can compare.

^Yep. I also poked around the data for Atlanta and the census tracts changed so much from 2000 to 2010 that a comparison isn't really possible. I imagine a lot of cities would have this same problem. Interestingly enough, the most dense tract in Atlanta in 2000 was over 40000/sqmi, and now it's down to around 20000/sqmi. It's due to demolition of public housing in a very small tract.

This is fascinating. Thanks Cirrus.

I suspect they changed the methods for group quarters. The 2010 numbers in Seattle declined significantly in two places: frat row at the UW and the county jail downtown. Seems like gray areas.

The 2000-2010 comparison is interesting. So is the 2010-2013 comparison, and especially "when today's construction projects are open and leased up" comparison.

PS, Los Angeles gets some vindication here. That's a lot of 20,000+ and even 40,000+.

Thanks for putting the maps together, seems very tedious! Some random observations:

*It's easy to see how LA turns out to be the most consistently dense metropolitan area, as high density areas sprawl in all directions within the city and county. I'm sure we'd see the same phenomenon with the entire city of New York, but NYC has a reputation for high density that LA doesn't--but should

*San Diego's densest tract appears to be the campus of UC San Diego, but the map confirms what I've always noticed on the ground--the Hillcrest, University Heights and Normal Heights areas form the main cluster of high population density within the central city

*Minneapolis isn't as dense as I would have expected

*Atlanta is as sparsely populated as I would have expected

*Houston has a relatively high peak density for the South, but it appears to be far outside downtown. Where is that?

*St. Louis really highlights the difference between population density and built density

*Detroit's low population densities are now comparable with the worst of the Sunbelt sprawlers

*The Chicago peak density tract is very odd--there are hundreds of residential towers in New York that could have been assembled similarly--but weren't. Ever. It's odd to define a census tract this way now, all of a sudden and only for one city, to say the least. That said, Chicago maintains huge swathes of high-density cityscape

Valencia:
http://i273.photobucket.com/albums/j...ps5053498b.png

That's over 3/4 of the city's population there in dark green (40k+ ppsm).

http://i273.photobucket.com/albums/j...ps7fa09f44.png
And around 20% of the population at 100k+ ppsm.

Technically these aren't census tracts but neighbourhoods, which are usually a bit bigger.

Is the Chicago census tract new? It would be surprising that we just learned that Chicago had the densest census tract of a half million people per square mile if it has always been there.

The tract must be new, or newly modified to be that small. In 2000 the densest in Chicago was 90,683 ppsm, and the buildings in that 500,000 tract were there in 2000.