|
The widespread conversion of rural lands to urban and
built-up uses is an issue that has drawn attention at
all levels of government. Given its importance, it is
no surprise that this land use issue has given rise to
numerous sets of statistics that attempt to measure the
extent of conversion and its effects on agriculture. In
general, statistics concerning rural-to-urban conversion
are of two types. The first set of statistics uses survey
information to measure the absolute amount of land that
has been converted from rural to urban or developed use,
or to measure the conversion rate (acres per year). Two
basic sources are the U.S. Bureau of the Census (see U.S.
Census, Geographic
Areas Reference Manual) and the USDA
National Resources Inventory (NRI).
The second set of statistics uses survey and other information
to identify (and measure) areas where remaining farmland
is subject to the effects of interaction with urban-related
population. In such areas, nonfarm-related activities
(residential, commercial, industrial) made conversion
of farmland to nonfarm uses more probable, change the
value of farmland and other open land, and alter the resource
allocation decisions of farm operators and landowners.
Such areas are often called "urban influence" and "population
interaction zones."
Measures of Urban Development
Measures of urban development provide a look at development
that has already occurred. The Bureau of the Census and
USDA's National Resources Inventory (NRI) (USDA/NRCS 2000)
measure trends in the amount of urban and developed land
in the United States, each using different procedures,
data, and concepts/definitions. The range of estimates
provided from these sources varies depending partly on
how "developed" and "urbanized" land categories are defined;
whether the basis of comparison is private land, farmland,
or total land area; and whether the comparison includes
Alaska and Federal land (Vesterby
and Krupa, 2001). Interrelationships among key geographic
entities embedded in these statistics are described below,
providing a brief review of the geographical concepts
and terminology associated with definitions of urban,
developed, metropolitan, and rural land.
U.S. Census Bureau
Rural/urban—The Census Bureau classifies
all U.S. land as either rural or urban. The rural/urban
designation begins with classification of urban areas
(see the Measuring Rurality
Briefing Room). The Census Bureau defines "rural"
as all territory, population, and housing units located
outside of urban areas. In general, rural land consists
of unsettled land and population concentrations with
less than 2,500 persons. The U.S. rural population was
59 million (21 percent) in 2000.
Urban areas consist of population and territory designated
as urbanized area (UA)or urban clusters (UC).
UA consist of an urban nucleus of 50,000 or more persons
with a population density of 1,000 persons per square
mile and associated area with a population density of
at least 500 persons per square mile. In 2000, 68 percent
of Americans lived in 452 urbanized areas. UC meet the
same criteria, but have populations of 2,500 to 49,999.
In 2000, 11 percent of the U.S. population lived in
3,158 urban clusters.
Metropolitan/nonmetropolitan areas—The
Office of Management and Budget (OMB) defines counties
as metropolitan or nonmetropolitan. Metropolitan area
counties (in 2003) were defined as central counties
with one or more urbanized areas, plus outlying counties
that are economically and socially integrated with the
central counties as measured by daily work commuting.
Nonmetro counties, which are those outside the boundaries
of metro areas, are subdivided into two types. Micropolitan
counties are centered on urban clusters of 10,000
or more persons. All remaining nonmetro counties are
labeled "noncore" counties. Most metropolitan or nonmetropolitan
counties contain a combination of urban and rural populations.
National Resources Inventory (NRI)
Urban and built-up areas consist of residential,
industrial, commercial, and institutional land; construction
and public administrative sites; railroad yards, cemeteries,
airports, golf courses, sanitary landfills, sewage plants,
water control structures, small parks, and transportation
facilities within urban areas.
Large urban and built-up areas include developed tracts
of 10 acres and more.
Small built-up areas include developed tracts of
0.25 to 10 acres, which do not meet the definition of
urban area, but are completely surrounded by urban and
built-up land.
Developed land in the National Resources Inventory
consists of urban and built-up areas and land devoted
to rural transportation.
Rural transportation land includes highways,
roads, railroads, and rights-of-way outside of urban
and built-up areas.
The Census urban area series runs from 1950, whereas
the NRI started providing a consistent series in 1982.
Prior to the 1982 NRI, Census urban area was the only
reliable national source of urban area data (Heimlich
and Anderson, p. 10-11).
Measures of Population Interaction
"Population interaction" is not urban development—It
is important that researchers be able to identify and
measure areas subject to the effects of urban-related
population interaction because the economic and social
environment in which agriculture operates under such conditions
is different than the environment in a rural area.
First, demand increases for farmland to develop into
nonagricultural uses. Urban-fringe farmland becomes valuable
for commercial, industrial, and residential uses. Characteristics
unrelated to agricultural production become important
determinants of its value. For most land parcels in population
interaction zones, crop and livestock production generate
less in net returns per acre than do nonagricultural uses.
Consequently, the price of farmland in population interaction
zones inevitably rises above the price at which it is
economical for sustained use in agricultural production.
In regions where farmland is in great demand for conversion
to urban use, relatively large proportions of the market
value of farmland is attributable to urban-related demand
(Barnard, 2000).
Second, population interaction affects the economic
costs and returns associated with urban-fringe farm operations
(Nehring and Barnard, 2003). In some cases, the rural/nonfarm
conflicts that arise from the increased interspersion
of nonfarm and agricultural activities, plus rising property
taxes, induce farmers to sell farmland for nonfarm development.
In other cases, farmers may adopt types and intensities
of agricultural production that enable land near population
concentrations to remain longer in agricultural production.
For instance, Gardner (1994), Heimlich and Barnard (1992),
and Vesterby and Krupa (1993) cite the increased production
of high-value fruit and vegetable crops as urbanization
advances.
ERS county-level urban influence codes—For some time,
ERS has provided the research community with county-level
classifications (codes) that delineate rurality, classifying
counties into ordinal categories according to the amount
of urban influence to which the counties are subject (e.g.,
ERS county-level urban
influence codes and rural-urban
continuum codes). These classification schemes are
implemented using alternative methods to categorize counties
according to size of metropolitan core-county population,
adjacency to metro core counties, and commuting patterns
(e.g., Measuring Rurality
Briefing Room). An advantage of county-level measures
of urban influence is that they are compatible with the
many U.S. statistical data series that provide data at
the county level. More recently, ERS has developed rural-urban
commuting area codes (RUCAs), which are based on subcounty
units called census tracts. The RUCAs are especially useful
in applications where geographic units smaller than counties
of interest.
ERS population interaction
codes—There is growing demand for a subcounty system
that is capable of classifying spatial data according
to the amount of urban-related interaction to which they
are subject. Two newly available coding schemes attempt
to satisfy the need for increased geographic specificity
in measuring population interaction while providing a
bridge to county-level measures. Although only a few U.S.
statistical series currently provide nationwide subcounty
data, Geographic Information Systems (GIS) software improvements
make subcounty spatial analysis increasingly feasible.
Increasing availability of GIS and Global Positioning
System (GPS) technologies increases the opportunity for
data-generating agencies to provide subcounty data. Farm
Service Agency (FSA) and other USDA agencies are in the
process of implementing GIS and GPS technology to better
serve both their agencies and their clients. The Natural
Resource and Conservation Service (NRCS) has established
the National Cartography and Geospatial Center (NCG) as
the main distribution source for most GIS data sets maintained
by NRCS, including NRI and soils data.
A population interaction index (PII)—One alternative
for developing a sub-county measure of population interaction
is to create an index number for geographic points or
grid areas, with the index indicating the potential influence
of nearby population concentrations. Creation of the index
begins with use of common GIS software to assign an index
number to each of many small (5-kilometer) grid cells
spanning the contiguous 48 States. The index numbers vary
according to the size and proximity of population concentrations.
These indexes, which we have called "population interaction
indexes" (PII), provide a cardinal measure of interaction
between agricultural land and nearby population concentrations.
(By a cardinal measure, we mean that the codes effectively
rank each location or area on a continuous scale.) The
population interaction indexes are based on the regional
economist's or geographer's concept of a "gravity" model,
which provides measures of accessibility to population
concentrations (Shi et al., 1997). This model measures
population interaction by accounting for the size of all
populations near a given location and the intervening
distance. For a particular region, the PII for individual
locations are mathematically aggregated. The PII increases
as population increases and/or as distance from the parcel
to nearby population decreases.
The concept of a "gravity" model evolved from marketing
analysis, where it was first used to assess the attraction
of consumers to retail markets (as described in Shi et
al.). Recently, the concept has been applied in the agricultural
and resource economics literature. Shi et al. describe
the gravity model as a "parsimonious method for capturing
urban influence in a single variable that combines [population]
size and distance [from urban concentrations]."
This is a stylized depiction of the construction of the index—with
the first three frames representing the process of creating
a population interaction index from population data.
This is a map displaying the index across the 48 contiguous
States. A technical description of the PII data is available.
Designating population interaction zones for agriculture
(PIZA)—The continuous population interaction index,
however, does not identify which grid cells are rural
and which are subject to the effects of urban-related
population interaction. To take that step and classify
grid cells into either "population interaction zones"
or "rural" zones, we set thresholds for the population
interaction index (PII) described above. Thresholds were
established that distinguish between background or rural
levels of the index and levels that indicate potential
interaction between urban-related population and agricultural
production activities. Index numbers below the thresholds
account for background levels of population (or density)
that would likely exist in the absence of urban-related
population interaction. The background level includes
population that supports an active commercial farming
industry, including employees of input and output industries
that support production agriculture. That background level
can be expected to vary regionally due to differences
in the productivity of farmland. (Consequently, we established
thresholds for all 20 USDA Land
Resource Regions (LRRs)). Index levels above the thresholds
were assigned to the "population interaction zone."
To establish separate thresholds for each LRR, we
examined the distribution of PII in areas of each LRR
that clearly have not been subject to urban-related
population interaction. Cromartie (2001) and Cromartie
and Swanson (1996) identify Census tracts that are "totally
rural" based on 1990 commuting data and U.S. Census Bureau
geographic definitions. Totally rural means that the tract
does not contain any part of a town of 2,500 or more residents
and that the primary commuting pattern was to sites within
the tract. (In essence, these are category 10 in the RUCA
codes.) Assuming that the most rural census tracts defined
by Cromartie (2001) and Cromartie and Swanson (1996) provide
a suitable reference or background level for the index,
any grid cell whose index exceeds the background level
calculated for these rural tracts is classified into a
population interaction zone. Thresholds for individual
LRRs were established at the 95th percentile of the distribution
of index numbers for 5-kilometer grid cells in the set
of totally rural tracts in the region. Grid cells were
classified into a "population interaction zone" if the
cell's population interaction index exceeded the associated
regional threshold.
Grid cells initially classified into a population interaction
zone are further classified into three categories representing
increasingly higher levels of population interaction.
This yields a system of codes defined by four zones of
population interaction: rural (little or no urban-related
population interaction) and low, medium, and high population
interaction.
The map displays the four zones for the 48 contiguous States.
GIS software, however, is needed to retrieve the population
interaction zone codes for any given geographic point
(latitude/longitude) or 5-kilometer grid cell. A technical
description of the PIZA data is available.
Although there is no definitive way to "ground truth" this classification system, we can provide some verification
by visually comparing these classifications to alternative
data related to population interaction.The map below shows the correspondence between the population
interaction zone codes and changes in "developed and built
up" areas derived from the National
Resources Inventory. Clearly, there is close correspondence
between newly developed areas (as identified by NRCS in
the NRI) and the outer edges of areas we coded as subject
to urban-related population interaction.
Corresponding county-level classification—It
is also possible to convert the location- or grid-specific
indexes into county-level indexes by calculating population
interaction indexes for county centroids. If an index
value for the 5-kilometer grid cell containing a county
centroid is above previously established rural thresholds,
then the county is assigned to one of the population interaction
zones. The county classifications, by FIPS code, can be
obtained from files linked in the Data Availability section
below.
When the indexes are converted to county classifications,
they retain a distinct advantage. Each county classified
as a population interaction zone retains a cardinal measure
of the intensity of the population interaction (within
its LRR). In other words, even though the county is coded
into a rural zone or one of three population interaction
zones, the population interaction index associated with
the county centroid remains a cardinal measure of the
level of population interaction to which the county is
subject. In this manner, each county retains a cardinal
measure of population interaction (within its LRR), in
contrast to the ordinal measure available from other ERS
classifications. The classifications based on population
interaction indexes are not subject to definitional or
procedural changes associated with OMB's designation of
counties as metropolitan.
However, county-level population interaction zones created
in this manner suffer from the same weakness as other
county-level classifiers: namely, U.S. counties are not
of uniform size and shape, leading to unsatisfactory classifications
for a number of counties.
Studies using population interaction zones for agriculture
(PIZA) and indexes (PII)—The four-zone codes (PIZA)
based on population interaction indexes (PII) have been
used in several studies at ERS to represent areas or locations
subject to population interaction. An initial analysis
estimated that 17 percent of U.S. farmland lies within
a population interaction zone. That analysis was based
on June Agricultural Survey (JAS) data for 1994-96 and
population interaction indexes derived from the 1990 Census
of Population. This same analysis determined that urban-related
population interaction has increased the value of U.S.
farmland by 25 percent.
Further analysis became possible with the availability
of the 2000 Census of Population. Using the 2000 Census
of Population data, we calculated new population interaction
indexes and re-categorized all 5-km grid cells as rural,
low, medium, or high. With that done, GIS tools permitted
comparison of population interaction zone codes based
alternatively on the 1990 and 2000 Census of Population. The map below displays the difference between the two categorizations,
showing where grid cells were re-categorized from a "rural"
zone to a "population interaction zone." The areas where
grid cells were recategorized from the "rural"
zone to a "population interaction" zone signal
that probabilities of development have risen.
Data sets related to the Population Interaction Indexes
(PII) and Population Interaction Zones (PIZA) are available
in an ERS
data product. County-level codes
are also available.
|
