The Documentation of the USDA, Economic Research Service’s (ERS) Frontier and Remote area (FAR) codes data product is organized as follows:
- Background
- Scope/Coverage of Data
- Data Sources
- Methods
- Strengths and Limitations
- Additional Resources
- Recommended Citations
Background
Small population size and geographic remoteness bestow highly cherished benefits for residents and visitors alike, but those same characteristics often create economic and social challenges. Job creation, population retention, and provision of services (such as groceries, health care, clothing, household appliances, and other consumer items) require increased efforts in very rural, remote communities, all things being equal.
The Frontier and Remote area (FAR) codes categorize sub-county locations based on their remoteness from cities and towns. The FAR codes are comprised of four increasing levels that measure how “frontier and remote” a location is based on its geographic isolation from urban areas of varying sizes. This results in a delineation that is both geographically detailed and adjustable with reasonable ranges.
Most rural-urban definitions treat rural locations as uniform. For example, the U.S Office of Management and Budget’s (OMB) Core-Based Statistical Areas divide counties into one of three categories: metropolitan, micropolitan, and noncore. The U.S. Department of Health and Human Service’s National Center for Health Statistics expanded upon the OMB’s definition to further differentiate between large metropolitan, medium metropolitan, and small metropolitan counties. ERS’s Rural-Urban Commuting Area (RUCA) codes group all rural census tracts together while differentiating between different sizes of urban cores.
To assist in providing policy-relevant information about the conditions across a variety of sparsely-settled, geographically isolated areas of the United States to public officials, researchers, and the general public, the FAR codes were developed in the early 2010s. They used ZIP Codes to classify locations because they split counties (and associated county equivalents) into multiple pieces, acknowledging that counties are often a combination of urban, suburban, exurban, and rural locations, and allowing the degree of remoteness to vary within a county.
To develop measures for varying degrees of geographic isolation, the concept of urban hierarchy was used. Urban hierarchy relates the size of a city or town to the types of goods and services it provides. The larger the urban area, the more specialized (i.e., high order) the goods and services available, while the smaller the urban area, the more basic (i.e., low order) the goods and services available. Therefore, people will drive farther distances to reach larger urban areas to purchase more specialized goods and services less frequently but need more consistent access to basic goods and services. (Note: Cities with high order goods and services also have intermediate and low order goods and services.) The four FAR code levels are based on different population thresholds that are meant to reflect likely access to high order services (Level 1), low order services (Level 4), and intermediate order services (Level 2 and Level 3).
Scope/Coverage of Data
The 2020 FAR codes are available for census tracts and ZIP Codes within the 50 States and Washington, DC. The census tract boundaries were delineated for the 2020 decennial census and include identifiers for 2020 and 2023 to account for the change from counties to planning regions in Connecticut. The ZIP Code area boundaries are from June 2024 and cover the entirety of the United States.
The 2010 FAR codes are available for census tracts and ZIP Codes within the 50 States and Washington, DC. The ZIP Code area boundaries used for the 2010 FAR codes are from 2014. The 2000 FAR codes are available for the contiguous United States. The ZIP Code area boundaries used for the 2000 FAR codes are from 2010.
New vintages of the FAR codes are released about every 10 years, following the tabulation and release of data from the decennial census.
Data Sources
As the FAR codes are essentially measures of geographic isolation from urban locations where goods and services can be obtained, three types of data are needed to create them: origins, destinations, and connectivity. These data inputs are used to calculate grid cell FAR code datasets that are then aggregated to census tracts and ZIP Code areas. The data sources listed here are for the vintage 2020 FAR codes.
The origins data are 3-arc-second (approximately 90 meter) nighttime population grid cell data for 2020 from Oak Ridge National Laboratory’s LandScan USA. These data use information about the built infrastructure (e.g., buildings and roads), natural features (e.g., rivers and mountains), and socioeconomic characteristics (e.g., journey-to-work) of a location to estimate the spatial distribution of the residential population within each census block. This finding results in a more detailed understanding of where people live than can be obtained from using census statistical units alone. This result is particularly true in rural areas of the United States where census blocks can be so large that the blocks have the same boundaries as their associated county.
The destination data are the 2020 urban areas from the U.S. Department of Commerce, Bureau of the Census (Census Bureau). Urban areas are densely developed locations that include residential, commercial, and other non-residential and uses. They are defined using population density, size, and built environment criteria and often include several adjacent communities. Because “high order” goods and services are generally only available in larger urban areas, while “low order” goods and services are available in urban areas of all sizes, the population size of the urban areas is another important element of the destinations data.
The connectivity data model the barriers and catalysts to movement between locations. Catalysts to movement occur when travel between locations is easier due to roads or ferries. The network of roads and ferries, as well as associated speed limits, was obtained from HERE Technologies’ North America Roads and Routing data set. Version 13.0 (2024 Q1) was used for 10 States (NH, NJ, NM, NY, NC, ND, RI, SC, SD, WV), while Version 13.2 (2024 Q3) was used for the other 40 States and Washington, DC. Any location without a road was assigned a base speed of 5 kilometers per hour, which is the approximate usual walking speed for healthy adults according to Murtagh, et al. (2021). Barriers to movement occur when travel between locations is impossible due to rivers and lakes. The location of water features was obtained from the U.S. Geological Survey’s 2024 National Hydrography Dataset. Travel over these water features was possible via bridges and ferries.
Finally, our geographic units of interest are census tracts and ZIP Codes. The 2020 census tract boundaries are from the Census Bureau. The ZIP Code areas (accessed April 2025) were obtained from ESRI using June 2024 data from TomTom North America, Incorporated and the U.S. Postal Service.
Methods
There are three main steps to creating the FAR codes. First, travel times to urban areas are calculated for grid cells throughout the United States. Then, locations are classified as “FAR” or “Not FAR” for each of the levels of remoteness. Finally, the grid cell data are aggregated to units that can be used more easily by the public, policymakers, and researchers.
Step 1: Calculating the travel time to urban areas
To measure travel times to urban areas, grid cell data are used. While the population data that comprise the trip origins were already represented by grid cells, the urban areas that comprise the destinations, as well as the road and water networks that comprise the connectivity data, were not. These three datasets were converted to grid cells using the population data as a template for grid cell size and location.
Once all the input data were represented by grid cells, the Distance Accumulation tool from version 3.1 of the ArcGIS ArcPy package was used to calculate travel times from each population grid cell to the edge of the nearest urban area of a certain size category. This tool uses the maximum travel speed within a grid cell (either driving the speed limit or, if there is no road present, walking) and the location of waterways and waterbodies (which prevent movement) to estimate the minimum travel time. The travel time tool was run 4 times, once each to calculate the travel time to the edge of the nearest urban area with population of: (i) 50,000 or more; (ii) 25,000 to 49,999; (iii) 10,000 to 24,999; and (iv) 5,000 to 10,000. Calculating travel time this way was a methodological innovation when it was first implemented for the vintage 2000 FAR codes and allows for longer travel-time bands around larger urban areas. An illustration of travel times near Billings, Montana is below.
Step 2: Classifying grid cells as FAR
The travel time calculations are used to determine whether a grid cell is a frontier and remote area. Frontier and remote areas are defined and adjusted along two dimensions:
- A population size dimension: Frontier and remote areas only include urban areas up to a certain size.
- A distance dimension: Rural areas and smaller urban areas will be counted as frontier and remote only if they are located beyond defined bands of proximity to larger urban areas.
Urban areas are defined by the Census Bureau (see Data Sources ), and rural areas include all territory outside those urban areas. The travel time beyond which areas are considered frontier and remote should be longer around larger urban areas because people tend to travel farther and less frequently for high-order goods and services. The travel-time bands to different urban area size groups are listed in the table below.
| UA size category | Travel time band for defining FAR area |
|---|---|
| 5,000–10,000 | 15 minutes |
| 10,000–24,999 | 30 minutes |
| 25,000–49,999 | 45 minutes |
| 50,000 or more | 60 minutes |
| UA = urban area; FAR = frontier and remote. Source: USDA, Economic Research Service. |
|
With the travel time thresholds in mind, each grid cell was classified as either frontier and remote or not for each of the four nested FAR code levels. The criteria used are:
- Level 1—Rural areas and urban areas of up to 50,000 people may be classified as frontier and remote (FAR). Level 1 FAR areas are 60 minutes or more from an urban area of 50,000 or more people.
- Level 2—Rural areas and urban areas up to 25,000 people may be classified as frontier and remote (FAR). Level 2 FAR areas are 45 minutes or more from an urban area of 25,000–49,999 people, and 60 minutes or more from an urban area of 50,000 or more people.
- Level 3—Rural areas and urban areas up to 10,000 people may be classified as frontier and remote (FAR). Level 3 FAR areas are 30 minutes or more from an urban area of 10,000–24,999 people, 45 minutes or more from an urban area of 25,000–49,999 people, and 60 minutes or more from an urban area of 50,000 or more people.
- Level 4—Rural areas may be classified as frontier and remote (FAR). Level 4 FAR areas that are 15 minutes or more from an urban area of 5,000–10,000 people, 30 minutes or more from an urban area of 10,000–24,999 people, 45 minutes or more from an urban area of 25,000–49,999 people, and 60 minutes or more from an urban area of 50,000 or more people.
Step 3: Aggregating grid cells to census tracts and ZIP Codes
Once frontier locations were delineated for the four FAR categories at the grid-cell level, the results were aggregated to larger geographic units: census tracts and ZIP Codes. For each census tract and ZIP Code, we calculated the percentage of the population classified as frontier for each level. If at least 50 percent of the census tract or ZIP Code area population resided in a frontier location, that entire unit was classified as FAR for that level.
Strengths and Limitations
The Frontier and Remote area (FAR) codes were created to provide a sub-county classification of U.S. territory at varying degrees of geographic isolation. This process results in a geographically detailed and adjustable delineation with reasonable ranges, so that the FAR codes can be usefully applied in diverse research and policy contexts. It also allows large political units like counties to be split into multiple pieces, capturing differing degrees of geographic isolation throughout the unit.
Another advantage to the FAR codes is the methodology used to create them. First, it allows the travel time to reach urban areas to differ based on whether high, intermediate, or low order goods and services are available. Additionally, conducting the majority of the analysis at the grid cell level and then aggregating the data to units that are more useful for research and policy results in a data product that is relatively robust to the unit of aggregation (e.g., census tract or ZIP Code) used. Finally, the data inputs used to create the FAR codes are the best available, resulting in a high-quality statistical product.
Though the data inputs are the best available, they may contain inaccuracies due to data collection errors. For example, the Census Bureau has difficulty counting some groups of people when collecting the data to produce the decennial census because they are hard to locate, contact, persuade, or interview. Therefore, a post-enumeration survey is used to evaluate the error in the data collected during the decennial census. The decennial census data are subsequently used to create the urban areas and grid-cell population estimates used in creating the FAR codes. Another example is the frequent change in possible routes to urban areas due to changes to the road and ferry networks.
The release of census-tract level and grid-cell level FAR codes in this 2020 version allows users to avoid the inherent limitations of analysis using ZIP Code level data, which was the only geographic unit available in previous releases of this product. As part of the latest update for the FAR codes, we have also released a census tract version of the vintage 2010 data. There is a wide range of socioeconomic data available at the census-tract level for policy and research use. Whereas the use of ZIP Codes as a primary geographic unit of aggregation is limited for socioeconomic or demographic analysis.
Census tract boundaries were created for statistical purposes and only change once a decade. ZIP Codes were created to efficiently process and deliver mail, and there is no official geographic delineation of their boundaries. All versions of ZIP Code boundaries were compiled by a third party, such as the Census Bureau or ERSI/TomTom. This means that the boundaries can vary significantly between sources.
For example, the ZIP Code Tabulation Areas (ZCTAs) created by the Census Bureau are created from census blocks and do not cover the entirety of U.S. territory. This means that if a census block contains more than one ZIP Code, it will comprise the area of only one of the ZIP Codes. Additionally, large portions of land and water area are excluded from the ZCTAs, especially in sparsely populated portions of the country. The ZCTAs also sometimes represent large volume postal customers with dedicated ZIP Codes (e.g., State capitol complexes or large businesses) as a very small area rather than a point.
Conversely, the ZIP Code areas used for this data product are provided by ERSI/TomTom and include all areas of the United States, even those without postal service. To do this, ERSI/TomTom uses their professional judgment to extend boundaries and create areas related to special uses, such as national parks. They also separate ZIP Codes into two different data sets. ZIP Code “points” represent large volume customers and post offices, while “areas” represent more physical postal routes.
ZIP Codes can also change at any time, leading to significant variation over time, even within the same data source. This makes it very difficult to know exactly which locations were used to create the ZIP Code versions of the data. Despite the geographic issues related to using ZIP Codes for research and policy, they are still convenient and easy location measures. To ensure users have an accurate record of the boundaries used to create the ZIP Code version of the FAR codes, geospatial versions of the data are available for the vintage 2020 FAR codes data product.
Additional Resources
All vintages of the downloadable Excel data files include a worksheet for the FAR codes, as well as worksheets with documentation and a codebook. The vintage 2020 FAR codes are available in geospatial format from the ERS Geospatial Resources page. The grid-cell-level (raster) data for each FAR code level is available there as well.
Geographic identifiers are provided, which can be used for mapping applications and to explore various spatial categories. To determine a census tract for a particular location within the United States, please see the Federal Financial Institutions Examination Council's FFIEC Geocoding/Mapping System.
Reference
Murtagh, E. M., Mair, J. L., Aguiar, E., Tudor-Locke, C., & Murphy, M. H. (2021). Outdoor walking speeds of apparently healthy adults: A systematic review and meta-analysis. Sports Medicine, 51(1), 125-141.
Recommended Citations
U.S. Department of Agriculture, Economic Research Service. (2026, March). 2020 Frontier and Remote Area Codes.
U.S. Department of Agriculture, Economic Research Service. (2015, April). 2010 Frontier and Remote Area Codes.
U.S. Department of Agriculture, Economic Research Service. (2012, June). 2000 Frontier and Remote Area Codes.