Measuring metro areas by 20-mile radius

[hotdogPi]

Since official metro area definitions follow county borders, it is not a perfect count of metro area population. I used this website to determine populations within a 20-mile radius of each major city, and here is what I found (third column is my own data):

Rank	City proper	Official metro	20-mile radius
1	New York	New York	New York
2	Los Angeles	Los Angeles	Los Angeles
3	Chicago	Chicago	Chicago
4	Houston	DFW	Houston
5	Phoenix	Houston	Philadelphia
6	Philadelphia	Washington DC	DFW
7	San Antonio	Philadelphia	Washington DC
8	San Diego	Miami	Phoenix
9	Dallas	Atlanta	Detroit*
10	San Jose	Boston	San Diego*
11	Austin	SF-Oakland	SF-Oakland
12	Jacksonville	Phoenix	Miami
13	San Francisco	Inland Empire	Boston
14	Columbus	Detroit	Atlanta
15	Indianapolis	Seattle	Denver
16	Fort Worth	Twin Cities	Twin Cities
17	Charlotte	San Diego	San Jose
18	Seattle	Tampa-St. Pete	Seattle
19	Denver	Denver	Las Vegas
20	El Paso	St. Louis	El Paso-Juárez*
21	Washington DC	Baltimore	Baltimore
22	Boston	Charlotte	Tampa-St. Pete
23	Detroit	Orlando	Portland
24	Nashville	San Antonio	San Antonio
25	Memphis	Portland	Sacramento
26	Portland	Pittsburgh	Orlando
27	Oklahoma City	Sacramento	St. Louis
28	Las Vegas	Cincinnati	Cleveland-Akron
29	Louisville	Las Vegas	Kansas City
30	Baltimore	Kansas City	McAllen-Reynosa*

*Goes into Canada or Mexico

Notes: Some of the 20-mile circles I drew may be a bit suboptimal. I didn't include a rank for the Inland Empire as I don't think there's a maximum there; keep going west and it will keep increasing until you reach LA's metro area. Sometimes the center of the circle isn't anywhere near downtown. This is common where the city borders the ocean or one of the Great Lakes (especially Cleveland, where the optimal circle includes Akron), but Nashville's circle also includes Murfreesboro, which would not have happened if it was centered. Buffalo kept malfunctioning, which I estimate to barely miss the top 30.

What do you think of this algorithm? How can it be improved?

[jwolfer]

Generally good.. but for Miami/South Florida it's very linear..   a few miles west of Miami and it's Everglades with absolutely no one..

In Florida for the other big metros your method works well, Jacksonville, Orlando and Tampa it works.

Z981

[jeffandnicole]

Generally good.. but for Miami/South Florida it's very linear..   a few miles west of Miami and it's Everglades with absolutely no one..

In Florida for the other big metros your method works well, Jacksonville, Orlando and Tampa it works.

Z981

Wouldn't the Everglades condition be similiar to any metro area with an ocean by it, including Miami and Tampa?

NYC is still #1 even though a portion of its 20 Mike radius would be uninhabited due to the Atlantic Ocean.

[jwolfer]

Generally good.. but for Miami/South Florida it's very linear..   a few miles west of Miami and it's Everglades with absolutely no one..

In Florida for the other big metros your method works well, Jacksonville, Orlando and Tampa it works.

Z981

Wouldn't the Everglades condition be similiar to any metro area with an ocean by it, including Miami and Tampa?

NYC is still #1 even though a portion of its 20 Mike radius would be uninhabited due to the Atlantic Ocean.

He did mention coastal cities the center may not be right at CBD.. Miami has ocean on one side and Everglades on the other

Z981

[webny99]

Generally speaking, I like it. I'd like to see some actual population figures associated with each metro, but I'm not sure how challenging those would be to access and share.

Also, how did you decide on 20 miles? Would changing it, say to 25 or 30 miles, influence the results?

[hotdogPi]

Generally speaking, I like it. I'd like to see some actual population figures associated with each metro, but I'm not sure how challenging those would be to access and share.

Also, how did you decide on 20 miles? Would changing it, say to 25 or 30 miles, influence the results?

The numbers I have for the 20-mile radius (in millions):

New York	12.5
Los Angeles	8.44
Chicago	5.42
Houston	4.27
Philadelphia	3.98
DFW	3.98
Washington DC	3.78
Phoenix	3.34
Detroit	3.34
San Diego	3.33
SF-Oakland	3.19
Miami	3.19
Boston	3.01
Atlanta	2.71
Denver	2.58
Twin Cities	2.55
San Jose	2.40
Seattle	2.36
Las Vegas	2.27
El Paso-Juárez	2.23
Baltimore	2.21
Tampa-St. Pete	2.20
Portland	2.08
San Antonio	1.95
Sacramento	1.94
Orlando	1.93
St. Louis	1.91
Cleveland-Akron	1.88
Kansas City	1.69
McAllen-Reynosa	1.69

[Bruce]

Generally, it's a bad idea to use radii when measuring land that may or may not include a lot of water.

For example, the Seattle metro area is very linear (skewed north to south) because of the considerable amount of water, so a 20-mile radius around the city center would exclude Tacoma and Everett, two of the principal/anchor suburbs.



(Used this tool to generate the circle)

[DTComposer]

I didn't include a rank for the Inland Empire as I don't think there's a maximum there; keep going west and it will keep increasing until you reach LA's metro area.

I'm not sure what you mean here. Draw a 20-mile radius from downtown Riverside and you get 2.768 million people and barely, if at all, touch either Los Angeles or Orange Counties, let alone come near a 20-mile raidus from downtown L.A.

[US 89]

Generally, it's a bad idea to use radii when measuring land that may or may not include a lot of water.

For example, the Seattle metro area is very linear (skewed north to south) because of the considerable amount of water, so a 20-mile radius around the city center would exclude Tacoma and Everett, two of the principal/anchor suburbs.

The same thing happens for Salt Lake City. The metropolitan corridor is skewed north-south along the I-15 corridor. A 20 mile radius around downtown Salt Lake City includes large areas of mountain ranges, the Great Salt Lake, and surrounding wetlands, but misses everything north of Kaysville and south of Draper, including Ogden and Provo. In fact, I'd bet more than half of that circle is completely uninhabited. (link to generate circle)

[Henry]

As shown here, a 20-mile radius is way too small for a place like Chicago, as it barely reaches the I-80/I-294 concurrency to the south, clips a very small piece of Northwest IN, and completely misses the I-94/I-294 connector to the north and I-355 to the west.

[froggie]

Generally, it's a bad idea to use radii when measuring land that may or may not include a lot of water.

Correct.  Which is why the Census Bureau and local/regional jurisdictions have maps showing the "Urbanized Area", basically a measure based on population density and other development parameters.  THAT would be a far better method of comparison than some arbitrary radius from a city center.

[jeffandnicole]

That all said though, another thread on here asked about people's max commuting.  Even though some cities may be linear and the metro area extends further out, a 30 mile ride for some people would be out of the question, even if it mostly all highway.  So the 20 (or whatever) mile radius does have some benefits for people if they want to live a certain distance to a city.  Of course, 20 miles away from downtown without convenient highway/transit access is a much different 20 miles if you live right off a direct highway or train line into the city.

[J N Winkler]

A different way of getting a total population measure for a metro area that avoids the arbitrariness of county lines is to sum the populations of all contiguous areas that have population density above a certain level.  For most metro areas this will be somewhat less than the total metro area population because it will exclude people living in low-density exurban areas, but I suspect the two figures will be within 10% of each other.

[US 89]

A different way of getting a total population measure for a metro area that avoids the arbitrariness of county lines is to sum the populations of all contiguous areas that have population density above a certain level.  For most metro areas this will be somewhat less than the total metro area population because it will exclude people living in low-density exurban areas, but I suspect the two figures will be within 10% of each other.

The Salt Lake City area would be an interesting place to see this method. The census bureau considers Salt Lake City, Provo, and Ogden to be three separate metropolitan areas, but they have all grown into each other across county lines to create one large urbanized area, the Wasatch Front. The official SLC metro population is 1.1 million, but counting the urban area in this way would likely put that figure in the range of 2-2.5 million. The census bureau's CSA would fit the idea of the total metropolitan region, since that also includes outlying areas like Tooele and Park City.

[DTComposer]

A different way of getting a total population measure for a metro area that avoids the arbitrariness of county lines is to sum the populations of all contiguous areas that have population density above a certain level.  For most metro areas this will be somewhat less than the total metro area population because it will exclude people living in low-density exurban areas, but I suspect the two figures will be within 10% of each other.

The Census Bureau already does this with Urban Areas and Urban Clusters:

https://www.census.gov/geo/reference/ua/urban-rural-2010.html

However, they introduce other data (namely commuting patterns) that split contiguously-developed areas into two or more separate urban areas (in California, notably splitting San Francisco and San Jose, and splitting Los Angeles from the Inland Empire).

[ET21]

As shown here, a 20-mile radius is way too small for a place like Chicago, as it barely reaches the I-80/I-294 concurrency to the south, clips a very small piece of Northwest IN, and completely misses the I-94/I-294 connector to the north and I-355 to the west.

And misses a good chunk of the mid and outer rim suburbs

[doorknob60]

A different way of getting a total population measure for a metro area that avoids the arbitrariness of county lines is to sum the populations of all contiguous areas that have population density above a certain level.  For most metro areas this will be somewhat less than the total metro area population because it will exclude people living in low-density exurban areas, but I suspect the two figures will be within 10% of each other.

The Salt Lake City area would be an interesting place to see this method. The census bureau considers Salt Lake City, Provo, and Ogden to be three separate metropolitan areas, but they have all grown into each other across county lines to create one large urbanized area, the Wasatch Front. The official SLC metro population is 1.1 million, but counting the urban area in this way would likely put that figure in the range of 2-2.5 million. The census bureau's CSA would fit the idea of the total metropolitan region, since that also includes outlying areas like Tooele and Park City.

Yeah I find it ridiculous that North Salt Lake, UT and Salt Lake City, UT are technically in different MSAs. And yes CSAs exist but often those are too big, like it's hard to argue that Salem, OR and Portland, OR should be in the same metro area (they are in the same CSA, not MSA), even though some people commute between them; they're quite separated. And you mentioned outlying areas like Park City, also not really a part of the metro area. And Boise's CSA stretches from Ontario and Vale, OR all the way to Mountain Home (whereas the MSA is more reasonable, though really I think it should maybe be only Ada and Canyon counties).

If you need to keep the Ogden and SLC in different MSAs (I'm not convinced you do, but I can see why they're split), then it would have to be around Farmington where you split it to make sense to me. I could accept that. But of course, since MSAs are defined by counties (and Ogden, Farmington, and North Salt Lake all reside in Davis County), that just can't happen. I guess ultimately it doesn't really matter though, haha.

[US 89]

A different way of getting a total population measure for a metro area that avoids the arbitrariness of county lines is to sum the populations of all contiguous areas that have population density above a certain level.  For most metro areas this will be somewhat less than the total metro area population because it will exclude people living in low-density exurban areas, but I suspect the two figures will be within 10% of each other.

The Salt Lake City area would be an interesting place to see this method. The census bureau considers Salt Lake City, Provo, and Ogden to be three separate metropolitan areas, but they have all grown into each other across county lines to create one large urbanized area, the Wasatch Front. The official SLC metro population is 1.1 million, but counting the urban area in this way would likely put that figure in the range of 2-2.5 million. The census bureau’s CSA would fit the idea of the total metropolitan region, since that also includes outlying areas like Tooele and Park City.

Yeah I find it ridiculous that North Salt Lake, UT and Salt Lake City, UT are technically in different MSAs. And yes CSAs exist but often those are too big, like it's hard to argue that Salem, OR and Portland, OR should be in the same metro area (they are in the same CSA, not MSA), even though some people commute between them; they're quite separated. And you mentioned outlying areas like Park City, also not really a part of the metro area. And Boise's CSA stretches from Ontario and Vale, OR all the way to Mountain Home (whereas the MSA is more reasonable, though really I think it should maybe be only Ada and Canyon counties).

If you need to keep the Ogden and SLC in different MSAs (I'm not convinced you do, but I can see why they're split), then it would have to be around Farmington where you split it to make sense to me. I could accept that. But of course, since MSAs are defined by counties (and Ogden, Farmington, and North Salt Lake all reside in Davis County), that just can't happen. I guess ultimately it doesn't really matter though, haha.

Actually, Ogden and SLC used to be one metro area, but they were split up in 2005.

And I agree, the Salt Lake City CSA is really too big to get an accurate representation of the population of the area. The north half of I-15 in Utah (200 miles) is in the Salt Lake City CSA, and I-80 and I-84 have all their Utah mileage within the CSA. But most of that is probably because of the large counties.

I feel like SLC gets screwed by these definitions and people think it's a lot smaller than it is. The population of the city proper is "only" 190,000 and the MSA is 1.1 million, but as I said above, the Wasatch Front urban corridor actually has a population of a little more than 2 million.

[bing101]

That's interesting though if the 20 mile radius was used for the Bay Area there's 2 sets of challenges cities like San Mateo, Hayward, Fremont and Foster City could have parts of their city count as San Francisco Bay area and the other section of their respective cities get counted as San Jose if the census went with this method.

[english si]

The numbers I have for the 20-mile radius (in millions):

Philadelphia	3.98
DFW	3.98
Washington DC	3.78
Phoenix	3.34
Detroit	3.34
San Diego	3.33
SF-Oakland	3.19
Miami	3.19
Boston	3.01

Some really similar figures there.

Incidentally, doing a 30km circle (can't get miles on that site for some reason) around my ~20k pop town gives a population of 3.52 million: a similar density to those large metros, though upping it to 32.14km (as close as I can get to 32.19km that's 20 miles) that increases to 4.06 million as there's a couple more large towns, plus an extra arc of London in there. While those figures includes about a fifth of London, filling a sixth of the circle, the rest of the circle is rural dotted with towns - about 50% is urban by my reckoning.

I can get 638k for a 16 km/10 miles ring (Oxford only gets 335k off 10 miles!), which doesn't take in any of London, but 8km/5 miles around my town only contains about 100k people, and a 6km circle gives something nearly at the 50k that my urban area actually is. I'm able to grab several nearby large towns with my arc.

A 32.20km circle around London gives 10.29 million. 30km: 9.92m. 25km: 9.16km. 35km: 10.80m. I wasn't expecting that pattern to be that linear - the 25km is about where the Green Belt starts, the 30km one has got most of the inner bits of the Green Belt and the 35km circle covers the first set of commuter belt towns beyond the Green Belt, but the increments are roughly equal - I think the reason is due to London not being circular and so the ring of less population is spread through all three. A 50km (roughly 30 miles) circle has 13.42m that's roughly the Metropolitan Area: it's got to sort of where there isn't Green Belt in between towns anymore.

A 160km (~100 miles) circle around London has 29.06m people in it, and the same around NYC has 29.83m. Broadly the two areas are as-dense (OK, NYC's circle might be a little more aquatic and has that little bit more population), but with a rather different urbanisation pattern: London's being much more patchy - with discrete towns and cities.