Lucy Myles
UEP 232
May 9, 2011
Regional Food Hub Geographic Suitability Analysis
This project seeks to map geographic suitability for new regional food hub locations
in the contiguous United States.
A Regional Food Hub (RFH) is a point of aggregation and distribution for locally and
regionally produced food products. Additionally RFHs can serve in the storing,
processing and/or marketing of food produced for regional consumption. In recent
years there has been growth in the number of RFHs across the country as demand
for locally and regionally produced food has increased. The food hub concept is an
emerging business model that holds a number of opportunities:
1. Stimulate community economic development through the creation of local
jobs
2. Support small and mid-sized producers by marketing the value-added aspect
of regionally produced products and expanding their distribution capacity
3. Increase access to fresh food for consumers, especially in underserved
communities.
Table 1
Data source
US Population Census
Metadata
Food Hub
Collaborative:
Agricultural
Marketing Service
StreetMap USA
Metadata
US Economic Census
Data query tool and
variable
descritptions
Year
2000
Variables
 County polygons
2011

Existing Food Hub zip codes
2002

National highway system
2007














Commercial fishing
Small farms with < $250,000 in annual sales
Refrigerated warehouse and storage
Farm product warehouse and storage (nonrefrigerated)
Local freight trucking
F&V freezing
F&V canning, drying, pickling
Animal slaughter and processing
Seafood prep and packaging
Colleges, universities, and professional
schools
Elementary and secondary schools
Hospitals
# of grocery stores in 2008
Percent organic of total farmland
Percent organic of total producers
% farms using conservation methods
Number of farms with direct sales
% Farms with direct sales
% Farm sales $ direct to consumer
# of farmers markets

Number of farm to school program






US Agricultural
Census
Data query tool
Metadata
2007
USDA Agricultural
2010
Marketing Service
Farmers’ Market
Directory
National Farm to
2010
School Network. Data
accessed through the
Food Environment
Atlas
Data and variable
descriptions
Variables were included in the analysis based on six key suitability concepts:
Regional food production, transportation and storage infrastructure, food
processing infrastructure, environmental services, demand for regional food, and
potential markets for regional food. The table 2 shows each concept with the input
variable.
Table 2
Concept
Variables
Commercial fishing
Food production
Small farms with < $100,000 in
annual sales
Miles of highway
Refrigerated warehouse and storage
Regional transportation
and storage infrastructure Farm product warehouse and storage
and services
(non-refrigerated)
Local freight trucking establishments
Frozen fruit and vegetable
manufacturing
Fruit and vegetable canning, drying,
Regional food processing
infrastructure and services and pickling facilities
Animal slaughter and processing
Seafood processing
Percent organic of total farmland
Value-added:
Number of organic operations
environmental services
Percent organic of total producers
Demand for local/regional
food using direct
marketing as proxy
measure
Potential markets for food
hubs
Number of farm to school program
Number of farms with direct sales
% Farms with direct sales
% Farm sales $ direct to consumer
# of farmers markets in 2010
Colleges, universities, and
professional schools
Elementary and secondary schools
Hospitals
# of grocery stores in 2008
RFH counties
mean values
5.27
168.22
422.76
2.73
0.57
67.80
0.47
2.41
6.76
1.76
0.70
14.82
1.22
0.35
125.92
11.56
2.44
10.78
10.69
50.61
9.00
140.80
Existing food hub counties were identified based on the USDA Agricultural
Marketing Service working Food Hub Database. The mean value for each input
variable was calculated for all counties with existing food hubs (see table 2). This
value was used as a threshold point to identify food hub suitability for each input
variable. A ranking scheme was developed for each concept to assign a suitability
score for each counties based on the number of input variables that met the
determined suitability level. Finally, a composite map was created to represent all
six suitability concepts.
23
Data variables
6
Concept maps
Final suitability
map
List of major steps in data preparation and analysis:
1. Download data from data sources (see table 1 for data sources and appendix
for steps on downloading data)
2. Prepare data in excel for GIS import with matching FIPS code (see appendix
for steps on creating 5 digit FIPS in excel)
3. Import excel spreadsheets into ArcGIS and perform table join with 2000
census county shape file using FIPS as join point
4. Project all data layers to the same coordinate system (US Contiguous Albers
Equal Area Conic 1983)
5. Geocode existing food hubs using 2000 census county zip code shape file
6. Geocode farmers market locations from the AMS Farmers Market Database
using 2002 StreetMap USA
7. Perform a special join with the county data and the farmers market point
locations for a count of farmers markets by county
8. Create a new column to identify counties as food hub or no food hub. Select
by attribute for counties containing a food hub zip code. Add a code for food
hubs to the selected counties. Invert the selection and add a code for no food
hub for the new selection
9. View the mean value for each of the data variables included in this analysis
within food hub counties
10. Create a second column for each of the data variables included in this
analysis. For each variable select by attribute all counties above the mean
value for food hub counties. Assign a 1 in the new column for each county
that meets the threshold for each variable
11. Create new columns for each of the 6 concepts. Add the values for the
variables that contribute to the concept.
12. Create maps for each of the concept variables using quantiles to visualize the
sum of the input variables for each county.
13. Create a new column for the composite analysis. Add the values for all input
variables.
14. Create a composite map using quantiles to visualize the sum of all input
variables used in this analysis
Difficulties:
The major challenge with this project was managing large amounts of data from a
number of sources and successfully merging all data into a single table. I ended up
with many versions as I joined new data tables and had to save each new resulting
data layer. Successful data management required a high degree of organization and
planning to make sure that I was working with the most updated data set.
A second challenge was figuring out how to make composite scores of data input
variables for the different concept maps and for the final map representing all data
inputs. I depended heavily on the ability to make new data columns and selecting by
attribute to make each variable binomial (meets threshold/does not meet
threshold). Again, this required a lot of attention and data management.
Observations:
Based on the variables and threshold values used in this analysis it is clear that the
regions in the U.S. with the highest food hub suitability are the Northeast and the
West Coast. A number of existing food hubs are located in areas that are designated
as highly suitable based on this analysis. However, the largest cluster of existing
food hubs is in the Upper Midwest, which was not ranked as highly based on the
suitability criteria used for this analysis.
Limitations and areas for future investigation:
The threshold criteria for each variable should be explored further to take account
of the range of deviation from the mean value among existing food hub counties and
to test whether the threshold values are significantly different from the mean value
of all U.S. counties. Further it would be interesting to represent more than just the
determined threshold point for each input variable. This could be done by creating
quantiles within each variable and assigning values for the concept based on the
variable quantile range.
Some of the variables specifically reflect structures that are relevant to local and
regional food systems, such as measuring demand for regional food using direct
marketing activity as a proxy measure. However, other variables that were used are
not specific to regional food system. For instance the food processing infrastructure
variables includes facilities that are at a larger scale than producers and distributors
of regional food can access. Looking specifically at small-scale food processing
infrastructure would be more relevant to regional food system mapping, however
these data are not publically available disaggregated to facility size.
There is also room for continued exploration with other concept areas. For example
supporting disadvantaged producers, and increasing food access in underserved
communities are two concepts that many food hubs strive to incorporate in their
models. However, input data for these concepts were not included in this analysis.
Finally, it is important to realize that the variables included in this analysis are only
a few of the many components that contribute to a vibrant regional food system. A
tool such as this could be used to identify potential new food hub counties based on
existing factors However, the most important work in determining food hub
suitability is done on the ground through relationship building in the community.
Annotated references:
The Regional Food Hub Advisory Council has already done some suitability mapping
for California, which is presented in the document “California Network of Regional
Food Hubs”. Through this analysis potential new food hub sites are suggested using
supply and demand indicators. However, the indicators that are used are not
sufficiently sensitive measures.1
The Vermont Sustainable Agriculture Council released a document in 2008 titled
“Mapping Vermont’s Local Food System.” In this document a number of maps are
presented, which use much more refined indicators for both the supply and demand
of locally produced food.2
Gibbs and Bernat examine the benefits of industry clusters on rural economies in
the paper titled “Rural Industry Clusters Raise Local Earnings”. These researchers
use an interesting methodology for identifying industry clusters, which is a Moran
statistic. The Moran statistic measures the value of a given variable for neighboring
counties to the county of interest. This could be a useful methodology to use for
future RFH suitability analysis because it would allow for the analysis to consider
regional designations beyond the county level, yet would not be constrained by state
boundaries.3
In the article “Mapping potential foodsheds in New York State” Peters et al map food
production potential as a way to examine the ability for New York State to feed
itself. This level of analysis will be too detailed for the purpose of the RFH suitability
analysis, however, it could be a useful methodology for future analysis.4
Citations:
1. The Regional Food Hub Advisory Council. California Network of Regional
Food Hubs: A vision statement and strategic implementation plan. September
2010
2. Erickson, Dan. Mapping Vermont’s Local Food System. The Vermont
Sustainable Agriculture Council. October 2008
3. Gibbs, R and Bernat, A. Rural Industry Clusters Raise Local Earnings. Rural
Development Perspectives. Vol. 12. No 3.
4. Peters, C et al. Mapping potential foodsheds in New York State: A spatial
model for evaluating the capacity to localize food production. Renewable
Agriculture and Food Systems. 24(1). 72-84. October 2008.
Appendix:
Downloading and preparing tabular data:
Downloading data from NASS Quick Stats:
Start here: http://quickstats.nass.usda.gov/
1. Select: commodity  Geographic location  Time (the page will update to
provide choices after each selection)
2. Select the “get data” button once everything has been selected
3. View the table to make sure it’s the data you want and then select “spreadsheet”
from the options in the upper right to download in excel format.
4. Note: Ag census data is only available through the Quick Stats tool from 1997
onwards. For older data you have to look up the tables:
http://www.agcensus.usda.gov/
Downloading data from the Food Environment Atlas:
Start here: http://www.ers.usda.gov/foodatlas/
1. Select download data in the upper right tab
2. Select the download button on this page for all of the Food Environment Atlas
data in a single spreadsheet ready for import into ArcGIS
Downloading Economic Census data from American Factfinder:
Start here: http://www.census.gov/cgibin/sssd/naics/naicsrch?chart_code=61&search=2007%20NAICS%20Search
1. Find desired NAICS code(s)
Then go here: http://factfinder.census.gov/servlet/IBQTable?_bm=y&ds_name=CB0700A1
1. From the menu bar at the top, Filter  Filter Rows.
2. On the first line, type in the 6-digit NCAIS code, and hit Search.
3. The Code and the Industry will appear in the box just below. Highlight that line and
hit Add.
4. After the same Code and Industry appear in box below Add, highlight that entry and
hit Show Results.
5. Under Options,  choose “Select Columns” and check off the box marked
Geographic Identifier Code (this is the combined 5-digit FIPS)
6. Click on Update (if you don’t want the number of employees etc, you can de-select
those options under that same check-box place)
7. Under Print/Download on toolbar, select Download.
8. In the popup window, select OK.
Creating 5 digit FIPS in excel for successful GIS table joins
1. Most of the geographic identifyers provide the state FIPS and the county FIPS in
separate columns. However, for the purpose of joining it is necessary to have
them together as a single 5 digit code (2 digit state code and 3 digit county code)
2. Put the state FIPS codes in column A and the county FIPS codes in column B.
3. In column C paste this code: =TEXT(A2,"00")&TEXT(B2,"000")
4. Copy all of column C and select “past special”. Select the “value” option. This will
transfer the FIPS column from a formula to number.