urban v rural

A definitive answer from the Census Bureau Geography Division would be useful, but my understanding is that urbanized areas ("urban") are independent of census tract (and census block group) geographies. This means that any given census tract might be split into an urbanized area portion and a rural (non-urban) portion.

A GIS approach could be to overlay urbanized areas over census tracts, and assign a tract as either urban or rural based on the land area (maybe population) within each tract portion.

Another approach, which I've favored, is to calculate the tract-level (or block group-level) gross population density (total population divided by land area in square miles).

Population densities of 1,000 ppsm or greater are considered urban; less than 1,000 persons per square mile, rural.

In my previous career, we used a six category system:

0 - 500 ppsm = Rural

500 - 1000 = Rural-Suburban

1000 - 6000 = Dispersed Suburb

6000 - 10000 = Dense Suburb

10000 - 20000 = Urban

20000 + = Urban Core

Here, suburban tracts would still be considered "urban" if the dichotomy is merely "urban" vs "rural".... The key value is the 1,0000 persons per square mile density. This value comes up again and again in Census Bureau documentation on urban and urbanized area.

We also used workers at tract-of-work to augment population density. Some census tracts are suburban in nature by their population density (e.g., financial district in San Francisco) but are urban core to the extreme when examining both population and job density. On the other hand, the Golden Gate Park (in SF) census tract is rural as could be, though it's smack dab in the middle of a large urbanized area.

In my analysis of San Francisco Bay Area density from the 2010 Census, I showed 83.9 percent of the land area and 7.8 percent of the population in block groups of less than 1,000 ppsm. Suburban was 10.5 percent of land area; 54.1 percent of total population; and Urban was 2.4 percent of the land area and 37.9 percent of the total population. (I haven't yet updated my analysis for 2020. Retirement means no deadlines.)

My recommendation would be to assign a census tract an urban or rural code based on 1,000 persons per square mile (ppsm) density, and then compare the coverage of those census tracts to available urbanized area boundaries.

Chuck

A definitive answer from the Census Bureau Geography Division would be useful, but my understanding is that urbanized areas ("urban") are independent of census tract (and census block group) geographies. This means that any given census tract might be split into an urbanized area portion and a rural (non-urban) portion.

A GIS approach could be to overlay urbanized areas over census tracts, and assign a tract as either urban or rural based on the land area (maybe population) within each tract portion.

Another approach, which I've favored, is to calculate the tract-level (or block group-level) gross population density (total population divided by land area in square miles).

Population densities of 1,000 ppsm or greater are considered urban; less than 1,000 persons per square mile, rural.

In my previous career, we used a six category system:

0 - 500 ppsm = Rural

500 - 1000 = Rural-Suburban

1000 - 6000 = Dispersed Suburb

6000 - 10000 = Dense Suburb

10000 - 20000 = Urban

20000 + = Urban Core

Here, suburban tracts would still be considered "urban" if the dichotomy is merely "urban" vs "rural".... The key value is the 1,0000 persons per square mile density. This value comes up again and again in Census Bureau documentation on urban and urbanized area.

We also used workers at tract-of-work to augment population density. Some census tracts are suburban in nature by their population density (e.g., financial district in San Francisco) but are urban core to the extreme when examining both population and job density. On the other hand, the Golden Gate Park (in SF) census tract is rural as could be, though it's smack dab in the middle of a large urbanized area.

In my analysis of San Francisco Bay Area density from the 2010 Census, I showed 83.9 percent of the land area and 7.8 percent of the population in block groups of less than 1,000 ppsm. Suburban was 10.5 percent of land area; 54.1 percent of total population; and Urban was 2.4 percent of the land area and 37.9 percent of the total population. (I haven't yet updated my analysis for 2020. Retirement means no deadlines.)

My recommendation would be to assign a census tract an urban or rural code based on 1,000 persons per square mile (ppsm) density, and then compare the coverage of those census tracts to available urbanized area boundaries.

Chuck

Dear Chuck,

Thanks for the thoughtful and thorough reply. Tougher than I thought !  For those of you who are following this thread here are some references for the 2020 census definitions:  Federal Register Notice (3/24/2022):  https://www.govinfo.gov/content/pkg/FR-2022-03-24/pdf/2022-06180.pdf  History of Urban designation: https://www2.census.gov/geo/pdfs/reference/ua/Century_of_Defining_Urban.pdf  Urban and Rural release dates  https://www.census.gov/programs-surveys/geography/guidance/geo-areas/urban-rural.html  FAQ https://www2.census.gov/geo/pdfs/reference/ua/2020_Urban_Areas_FAQs.pdf

Most of the release dates are in the Dec 2022 to Jan 2023 time frame. shapefiles: Dec 2022 

As far as geography goes (from FAQ) Urban areas are defined primarily based on housing unit density measured at the census block level. The basic criteria is 425 housing units per square mile but there are additional criteria used to minimizes the number of noncontiguous pieces of an area.  The census leaves it up to you to decide if their criteria is suitable for your intended use.

Hope this helps those who are interested in the urban/rural designation.

Just as an FYI.  I'm doing some work on CDC data. The CDC urban/rural scheme documentation is here https://www.cdc.gov/nchs/data/series/sr_02/sr02_166.pdf  It seems to be based on a Metropolitan Statistical Area / Urban Cluster scheme.  The reference is based on the 2010 areas and the documentation was released in April 2014 so there should be an update for the 2020 census areas coming along at some point.