Jaclyn Pryll CRP 386: Intro to GIS School of Architecture University of Texas at Austin Fall 2008 Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll
CRP 386: Intro to GIS
School of Architecture
University of Texas at Austin
Fall 2008
Suitable Locations for Grocery Stores in
Underserved Areas, Rochester, NY
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
2
EXECUTIVE SUMMARY
There are areas in the city limits of Rochester, NY, that are underserved by supermarkets
and grocery stores that offer healthy and fresh foods. For several decades, a trend exhibited by
urban grocery stores nationwide is their continuation to close down stores only to relocate or
development new stores in the suburbs. For the purpose of this report, the focus is on medium to
larger retail grocery stores (around 40,000 square feet) to use to identify where there are
locations in urban areas of Rochester, NY, to offer underserved areas better varieties of fresh
foods. This research uses GIS to evaluate the demographic characteristics of census block
groups to identify areas lacking grocery store locations as well as identify areas in need spatially.
The results will provide parcel candidates for new store locations based on a ranking of available
vacant parcels and their locations within certain demographic and spatial characteristics of
Rochester. Demographic characteristics chosen, such as household median income, total
population and racial population distribution, are criteria seemingly most influential in the
grocery gap trend, both directly and indirectly. Adding the demographic data sets with
identifiable grocery industry spatial standards such as location proximity to major streets, current
store locations, and public transit attempt to use standardize criteria to appeal to the practicality
of actually locating a new store in underserved areas. Hopefully, the sites chosen could be
considered good candidates for sites sought by the grocery-chain industry for future
development. The criteria used to locate suitable parcels does not, however, provide concrete
criteria used in every grocery store-chains methodology for what fits individual company goals,
but it does provide basic economic and spatial factors that appeal to any profit-making
corporation. The results provided can be used both by the grocery industry as well as local
policy makers. Grocery store companies can see that success can be possible locating back in
the city limits, and local policy makers can formulate policies to ensure hardships aren’t
increased on those underserved with healthy food choices because of the ineffective policies
currently in place that do not ensure proximity to retail food stores.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
3
INTRODUCTION
The trend exhibited over the past 40 years by grocery stores and supermarkets closing
stores in urban and inner-city areas in favor of locating in suburban neighborhoods is called the
“grocery gap”. This can be considered a by product of “white flight”, the movement of the white
inner-city population to the suburban areas where often the higher paying jobs were locating
and/or followed. The abandonment of the urban areas for lush suburban pastures left many in its
wake such as lower income and minority populations unable to move to or afford a higher cost of
living. Amongst the sprawled environment also came the mega-sized supermarkets that were
able to expand their store footprint from 25,000 square feet to up to 70,000 square feet based
much on available land that also cost less to lease or purchase than many urban properties. The
City of Rochester, NY, has been no exception to the “grocery gap” trend. In the upstate NY
region, several grocery store companies compete for business, but one grocery store chain has
served Rochester for many decades and has also been the largest exhibitor of relocating from the
city to the suburbs. Though this paper isn’t aimed at placing certain companies in the spot light
for their business decisions to not locate in the urban areas of Rochester, their choices to relocate
to the suburbs have left many people underserved by market places offering healthy foods at
competing prices.
Grocery stores, for this paper, are defined as stores ranging from 30,000 to 70,000 square
feet and offering fresh produce, a meat and dairy department, and a grocery area. Wegmans is a
successful grocery store chain that has its birthplace in Rochester and has served Rochesterians
for decades. From its conception, Wegmans has provided Rochesterians with fresh produce,
baked goods, fresh seafood, meat, deli products, and international foods. As the company and its
success grew, so did the store size and its offerings of merchandise of 70,000 items, compared to
40,000 for most supermarkets. Assessing that locating on the city fringe and outside of it has
more benefits than not, Wegmans has closed many inner city stores. The City of Rochester’s
local government and area non-profit organizations (NPOs) have successfully recruited other
grocery retailers to locate and open grocery stores in several urban areas in Rochester, but there
are still areas very underserved and in need of healthy food choices. Currently, Wegmans, Tops
Friendly Markets, and PriceRite Supermarkets are the three grocery store retailers operating in
Rochester. In some of Wegman’s abandoned stores, Tops and PriceRite have opened stores in
the Rochester area.
Inner-city neighborhoods are often times higher concentrations of minorities, lower
income households, and dependant on public transit. Increased distance to retailers of healthy
foods causes a strain on these already struggling households, and lack of healthy choices can
contribute to unhealthy diets and increased risk of dietary diseases. These kinds of consequences
may not be a direct responsibility of the retailer but food is a basic necessity, and as a food
supplier, it seems irresponsible to abandon areas in need of this basic food resource. Rochester is
the county seat for Monroe County in NY state. Within the boundaries of Rochester, there are
seven grocery stores serving 219,026 people (2000 census), and outside the city boundaries but
within Monroe County there are 21 stores serving 516,317 people (2000 census). The ratio of
grocery store per persons in the city is 1:31289, whereas in the suburban and rural areas the ratio
is 1:24,586.
The location of grocery stores by grocery store companies depend on their own market
analysis and business strategies, but the standard industry assumptions for locations depend on
the average volume of traffic on roadways in front of proposed sites, placement at intersections
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
4
of two major thoroughfares, visibility of the site from the road, and the political and business
climate of the community (Nienow, MEMO). Household income becomes a factor when
considering “stores try to locate in areas where they will generate between $350 and $500 of
business annually for every square foot of building space,” (Nienow, MEMO). A 40,000 square
foot building would, then, need to generate an estimated $17 million in sales annually.
PROBLEM STATEMENT
It is necessary to assess the climate and conditions of Rochester to see if there is indeed a
trend exhibited of the “grocery gap”, and if it can be identified, than to provide choices for the
people of Rochester as well as the grocery industry of places to locate a new store in areas in
need. The urban core has become a cliché in rust belt cities such as Rochester of poverty and
physical abandonment. Though Rochester’s population hasn’t increased much over the past
several decades there is still a large population deserving of healthy food choices closer than are
currently available. If it can be shown that there are indeed suitable parcels, then the issue can be
made more of a priority and can also be give more substantiality as an issue needing to be
addressed by the grocery industry as well as local officials.
RESEARCH QUESTIONS
The primary research question is: “Where are suitable locations for new urban stores in
underserved areas in Rochester, NY?” Questions that arise from this primary question are what
areas are underserved, and can grocery stores find locations in underserved areas using their own
standards and criteria? If suitable locations are found, does it support the need for planning and
public policy to better ensure that communities in need of healthy food choices are provided with
closer locations to this basic necessity?
METHODOLOGY
Data Collection
The primary source of spatial data was obtained through the Monroe County Department
of Environmental Services. For $15, a CD was purchased with shapefile data that pertained to
the area of Monroe County (which included information for the City of Rochester). The City of
Rochester didn’t have shapefile information readily available for download, and purchasing the
CD, especially since it had all pertinent layers for my study area, seemed the most viable. The
shapefile layers I used were parcel data, Monroe County boundary limits, City of Rochester
boundary limits, and street centerlines (though there were several additional shapefiles on the CD
that I didn’t think would help with my analysis).
To obtain bus route shapefiles for the city, I emailed a representative with the Genesee
Transportation Council who, after having received a written request for the shapefiles for
licensing purposes, emailed me back the shapefiles for the bus routes. To obtain demographic
information, The United States Census Bureau’s 2000 Census Summary File 3 contained
information regarding total population count, population count based on race, and median
income for households within the Monroe County area. The SF-3 file is given at the census
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
5
block group level. I felt that this level was fine for my project. I joined the census demographic
statistics with TIGER data shapefiles downloaded from on the US Census Bureau website. Since
the Monroe County GIS shapefile data was projected in NY State Plane (NAD83, survey feet)
FIPS 3103, I identified that as my final map projection and projected the TIGER data and the bus
routes in that same projection. I struggled with providing transportation data such as trying to
find the number of owners of cars per households, and I was unable to find a category that
satisfied me within the SF-3 data. I found “mode of transportation used when going to work”,
and “private vehicle occupancy for workers over 16”, but neither seemed suitable for my type of
analysis. I chose to not use census transportation data as I was not focusing on accessibility to
stores as much as locating stores in areas that were without one.
Being from Rochester, NY, I am familiar with the grocery store companies in Rochester
that fit the type of store I was interested in locating parcels for, so to obtain their addresses I used
the internet to search for their store locations on the store’s individual websites. I copied down
the addresses on an excel sheet for all those located within Monroe County. At first I had
thought I would code the different companies individually, but then decided that it wasn’t as
important to highlight which companies located where, but just to look at them as industry
representatives and to code them all the same.
I referenced several articles on the “grocery gap” as well as other articles discussing the
lack of grocery stores in poorer neighborhoods. Data found in articles regarding the need to
locate in more underserved urban areas provided a foundation for my research question and
problem statement. It also provided the “assumed industry standards for store locations” that I
used to chose my data sets for my suitability analysis.
Data formulation and modifications
It was important to clean, “prep”, and formulate new shapefiles in order to run the
analyses. This included a number of processes to be performed:
Geocode List of Current Store Locations Addresses
I geocoded the list of grocery store addresses both automatically and interactively.
After I created an excel sheet of the 28 addresses of all grocery stores in Monroe County,
and then performed a batch conversion, my success rating of matched addresses came
back as 22 matched addresses and 6 unmatched addresses. Three of the addresses needed
the directional moved from the “suffix” box to the “prefix” box. One address needed the
zip code changed. One address needed 4th
to be spelled out as fourth, and the final
address needing a fix was regarding the format of the street name. The input name was
actually two streets that crossed near the store. Following a google search of the exact
street address, I was able to format it correctly. This all produced 100% matches.
Creating Rochester City limits shapefile
Since the City of Rochester was to be used as my study area, it was necessary to
create a shapefile that provided the city boundary for other data to be clipped to. The
Monroe County GIS data included a line and polygon layer called town_villages. I
selected the polygon for Rochester and exported it out as a new layer.
Clipping data to Rochester City Limits shapefile
Using the Rochester city limits shapefile, I clipped other shapefiles to be used for
the analyses to the city limits shapefile. These layers were the TIGER data shapefiles for
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
6
the census information, street centerlines, and parcel shapefiles. I exported out the major
streets in the city of Rochester based on their ROADTYPE classification as divided
highway or state route, and created a seperate shapefile representing major streets.
Joining census tables to TIGER data
My excel documents needed to be in the 97-03 format to be compatible with the
shapefiles. After downloading my census data, I removed the second row that were
column descriptions, renamed the file, and used those renamed files to join to my TIGER
data shapefiles. I downloaded the total population and racial population in one table, and
the median income data in another table.
Creating new TIGER and census data layers
I exported out TIGER data profiling certain census data as separate layers to use
in different maps. For the TIGER shapefiles containing income data, I chose the field
value for median income on the joined census table. I applied the Natural Breaks
classification with 5 classes. The natural breaks method emphasized the lower income
brackets. I felt the information became distorted using the equal interval or quantile
classifications because the data was composed mostly of incomes below $50,000, yet
since there were income brackets reaching $100,000, the data appeared flat, or without
variation as all the lower income were squished into one or two classes. One thing to
note is that when I used the city limits as my extent in creating a median income shapefile
for the city, it only included income levels up to $100,000. When I created a countywide
shapefile with income brackets, though I used the natural breaks with 5 classes, the
highest income level was $130,000. They, therefore, aren’t used to compare to each
other bracket by bracket, but more as an over view of income distribution from two
different scales.
For the total population and racial population shapefiles, I first created a column
in the excel file for acreages per census block group in the TIGER data shapefile. Then, I
joined the census table to the TIGER data. For total population, I used the natural breaks
classification with 5 classes. I normalized it with acre. This created a population
distribution based on census block groups. For the racial population shapefiles, I used the
natural breaks classification with 5 classes as well, but I used the individual racial
population fields and normalized them with the total population field to give a percentage
of the races that are located within the block groups.
Using buffers to show walkability and households with access to stores outside the city
boundary
I created a ¼ mile buffer to place around existing stores to show the capture area
of population that could potentially walk to the stores for their grocery needs as opposed
to relying on cars or public transportation. I also created a ¼ mile buffer around the city
limits to capture any stores outside the city limits that theoretically are within
accessibility of the city population. When applying the city limits buffer to capture
outlying store locations, I found no stores that were located within it that would be
included in the study for the city population. Therefore, I did not include the buffer in
my analysis maps.
Selecting vacant parcels for the suitability analysis
Selecting vacant commercial parcels required using the attribute table associated
with the parcel data. The field PROP_NBR contained numerical coding, and the field
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
7
next to PROP_NBR, PROP_DESC described the properties usage. I obtained a pdf file
of the 2003 zoning map for Rochester, and I georeferenced the pdf (reformatted as a jpg.)
into ArcMap using the parcel layer shapefile that was clipped to the city limits boundary
as my guide. I noted that the parcel PROP_NBR field was in a format where the first
digit more or less correlated with a zoning category. I used the “select by attribute”
function to distinguish the land use categories so I could create the parcel land use map as
well as find the vacant commercial parcels. I also selected parcels that were 2.75 acres or
more. That size lot could support a 40,000 square foot store, a parking lot, and any other
room needed to comply with local building regulations such as impervious cover
requirements. I exported out the PROP_NBR value for vacant commercial parcels and
the acreage as a separate layer to use in the suitability analysis.
Setting up the Suitability Analysis
The suitability analysis required some critical thinking as far as criteria and data sets used
to perform the analysis. I decided I wanted to show outcomes for suitable parcel rankings based
on two different perspectives: the grocery store perspective and the perspective of those in need
such as lower income and dependent on public transit. Both perspectives used the same data sets
taken from the industry standard assumptions mentioned in the MEMO written about store
locations in Cary, NC as well as my own data sets based on other articles I felt relevant to the
study. They were: vacant commercial parcels, major streets, bus routes, existing store locations,
and median income levels. The reclassifying of income data was catered for each perspective as
well as different weights were given for each perspective. All classifications were using equal
interval and were split into 10 classes so they could be compared to each other.
To perform the suitability analysis, the data sets (major streets, bus routes, other stores,
and income) vectors were formatted into rasters, the rasters were then reclassified, and then they
were ranked and formatted back into vectors. Distance in close proximity to major streets, bus
routes, and distance away from other stores remained consistent within the reclassifications. The
income reclassifications were modified for each perspective, though. Weights were catered to the
perspective of the analysis. The vacant commercial parcel layer was intersected with the ranked
data sets to produce the ranked parcels.
Suitability analysis according to grocery store criteria
Reclassification of Income was determined by looking at the locations of stores outside
the city boundary and where they were located with income brackets. I chose this
method to determine reclassification since that’s where stores were mostly located.
Income Bracket number of stores
outside city limits
0 - $13,000 0
$14,000 - $26,000 0
$27,000 - $38,000 1
$39,000 - $51,000 6
$52,000 - $64,000 9
$65,000 - $77,000 4
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
8
$78,000 - $89,000 1
$90,000 - $100,000 0
$110,000 - $120,000 0
$130,000 0
Income Reclassification (10 being the most favorable)
Income bracket ($) reclassification
0 - 10146.1 1
10146.1 - 20292.2 3
20292.2 - 30438.3 5
30438.3 - 40584.4 8
40584.4 - 50730.5 9
50730.5 - 60876.6 10
60876.6 - 71022.7 7
71022.7 - 81168.8 6
81168.8 - 91314.9 4
91314.9 – 101461 2
Weight given to data sets:
Data Set Weight
location within income
bracket area
45%
distance to streets 35%
distance from other stores 10%
distance to bus routes 10%
Suitability analysis according to lower income and public transit perspective
I wanted to reclassify the income brackets differently with more emphasis on
lower income households as well as place more emphasis on bus routes as poorer
households rely more on public transit. I felt locating closer to public transit routes
would help service those underserved areas by making the stores more accessible.
Income Reclassification (10 being the most favorable)
Income bracket ($) reclassification
0 - 10146.1 9
10146.1 - 20292.2 10
20292.2 - 30438.3 8
30438.3 - 40584.4 7
40584.4 - 50730.5 6
50730.5 - 60876.6 5
60876.6 - 71022.7 4
71022.7 - 81168.8 3
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
9
81168.8 - 91314.9 2
91314.9 – 101461 1
Weight given to data sets:
Data Set Weight
location within income
bracket area
45%
distance to streets 10%
distance from other stores 15%
distance to bus routes 30%
FINDINGS (Refer to following maps)
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
10
This map was created to show the regional perceptions in identifying the locations of
grocery stores in Monroe County in comparison to the household income and population
distribution of the county area. Our focus area is Rochester, but this is to show the regional
perspective surrounding the city.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
11
This map was created to illustrate the locations of the seven grocery stores within
Rochester’s city limits. It highlights store location in relation to total population distribution, and
it also shows a ¼ mile buffered distance around each store to show the population captured in a
walkable area from store proximity.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
12
This map was created to show the locations of current grocery store locations in relation
to the median income earned per household within the city limits.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
13
This map was created to show the different racial population distributions in relation to
current grocery store locations within Rochester’s city limits.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
14
This map was created to show the current land uses for the city landscape. It is not a
zoning map, but it largely coincides with the zoning for the city.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
15
This map shows the five criteria used to perform a suitability analysis for detecting
suitable parcels for new store locations. These data sets were chosen based on industry standards
and were weighted and ranked to create the following two suitability analysis maps.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
16
This map was created using the data sets in the map Criteria Used for Suitability
Analysis. It is a ranking of suitable parcels based on weights assumed to most favor the grocery
chain-stores perceptions of successful sites.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
17
This map was created using the data sets in the map Criteria Used for Suitability
Analysis. It is a ranking of suitable parcels based on weights assumed to most favor the lower
income households and households in proximity to public transit routes.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
18
ANALYSIS
Showing Rochester exhibits the presence of “the grocery gap” in underserved areas
This was achieved through the County Population and Income and Store Locations, City
Grocery Store Locations and Total Population Distribution, and Median Income map layouts.
County Population and Income and Store Locations set up the regional setting for the City of
Rochester and tried to show the existence of suburban grocery stores exceeding the number of
inner city stores even though the highest concentration of the counties population lies within the
city limits. The median income brackets for the county compared to the total population
distribution show that the largest concentration of people are also in the lower income brackets.
Even though there are twice as many people outside the city limits, there are three times as many
stores. The stores themselves appear to cluster around higher income brackets except within the
city limits. Even though income brackets for this map were used with the same classifications as
the income brackets for maps at the city limits scale, the inclusion of the highest median income
is different per map scale. The county map includes areas where earnings are of higher median
incomes, so the city center appears very poor compared to the rest of the county.
The City Grocery Store Locations and Total Population Distribution map layouts also
emphasize a clustering of grocery stores around lower population distributions as areas of high
concentrations of population are left without a store. Only one area of high population
distribution shows two store locations. Six of the seven stores over all, though, formulate a
linear pattern from the NW corner of the city towards the SE corner. Along this linear path also
shows a linear pattern of population distributions per census block group of 5500 or less. The
SW corner of the city and the NE corners are physically lacking a grocery store to serve them.
The Median Income map layout shows the grocery store locations in relation to the
median income brackets of the city. Three of the seven stores are located in lower income
census block groups, and the remaining four are located close to the census block groups with
the highest median incomes. Three of the four located within the highest median income areas
are also within ½ a mile of the each other. They are clustered in what appears to be the wealthy
part of town based on income. One misleading factor to note is the presence of a high income
area where the land is actually parkland or forest. That is in the upper most northeast area of the
city limits. There are a very few amount of people living in this area, though it appears to be a
largely wealthy area. The size of this census block group is one of the largest, so the median
income for this block group seems rather distorted when looking at the rest of the area.
The Racial Demographics map layout of the city needed more maps or data to
substantiate a claim that the stores are not locating in minority areas, an assumption that is a part
of the criteria of underserved areas. It is important to note, though, that a majority of the stores
are located in highly white populated areas, and the largest percentage of black population within
the city is without a store. Stores are also located near the high concentration of Hispanic
populations, though. Further maps and analysis would need to be performed, though, to fulfill
the notion that minority neighborhoods as being underserved to take my suitability analysis
towards the socioeconomic justice perspective.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
19
Suitability Analysis – Finding suitable parcels for new stores
I was able to locate suitable parcels for new store locations according to criteria used as
“industry standards” as well as those data sets for areas in lower income that are not considered a
priority with the grocery store industry. I made two suitability analysis maps to show the
difference (if there was to be any) between the suitable location of a new store in Rochester
according to industry standards, and the suitable location of a new store based on standards used
to categorize a neighborhood that is underserved: lower income and public transit users. I did
lack information regarding the areas that are dependent on public transit, and I think if I included
this information, I may have discovered different suitable parcels. As it stood, though, the bus
routes for Rochester appear quite extensive, and if a grocery store were to locate along one of the
routes and there is no current stop, a stop could be created.
Suitable Parcels Based on Industry Standards with Emphasis on High Income and Major
Streets shows the most favorable parcels on the edge of the city, no doubt near higher income
brackets. One of the criteria for reclassifying was that higher income was more favorable, so it
confirms that these areas are those with higher incomes. One parcel is actually right next to the
cluster of existing stores (shown on other maps) that surround the SE part of the city, or the
wealthy area. Locating close to other stores doesn’t appear to be as big a threat as one would
think, either. It seems that stores may bring in different clientele or still make a profit if the area
is wealthy enough and people find all stores appealing.
Suitable Parcels based on Industry Standards with Emphasis on Low Income and Bus
Routes shows one parcel that is in the same area as in the map assumed to follow grocery store
industry standards. This is a promising discovery since two parcel locations next to each other
are suitable according to the grocery industry and the needs of the underserved. The most
promising parcel according to the lower income bracket and near bus routes is actually located in
an area that is of a high black population, lower income, and without a grocery store for about a
one mile radius. It is identified as the 3rd
suitable parcel insert located in the southwest part of
the city. This would be the most ideal parcel to build on should the issue of underserved areas
needs being met ever become a priority to the grocery store industry.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
20
CONCLUSION
Finding suitable parcels in underserved areas of Rochester, NY, was achieved, but the
study could benefit from further analysis. This kind of analysis can address the problem of
underserved areas in proximity to food retailers more thoroughly if the research question is
expanded beyond simply finding a suitable location. Addressing the grocery gap can be
achieved through accessibility studies as well as focusing on other sources of food suppliers such
as farmer’s markets and local area food banks. This preliminary study doesn’t dig deep enough
into the socioeconomic factors and repercussions of limited access to healthy food sources.
Further analysis into different types of demographic data such as reliance on private
vehicles versus public transit and earned income through work versus public assistance could
shed new light into the depths of this trend as having deeper meaning. Public policy might be
persuaded more if there were enough data supporting the definition of underserved and the
admission of grocery store retailers failing to acknowledge responsibility to provide all income
levels with healthy food choices. On the other hand, reasons why stores don’t locate in areas of
lower income could be supported by crime statistics, inability to capture enough annual revenue
based on the income capture, and/or conditions of the business and social climate.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
21
REFERENCES AND DATA SOURCES
Articles
Abrose, David M. Retail Grocery Pricing: Inner City, Suburban, and Rural Comparisons. (Jan.,
1979). The Journal of Business, vol. 52, No.1 pp. 95-102.
Free Shuttles Can Close the Grocery Gap. 2003-04-15. University of California: UC Newsroom,
http://www.universityofcalifornia.edu/news/article/5318.
Nienow, Sara. (2003, June 6). Grocery Stores along High House Road – MEMO to Town
Council. Cary, North Carolina.
Pothukuchi, Kameshwari. Attracting Supermarkets to Inner-City Neighborhoods: Economic
Development Outside the Box. Economic Development Quarterly, 2005; 19; 232.
http://edq.sagepub.com/cgi/content/abstract/19/3/232.
Supermarket Access in Low-Income Communities. Prevention Institute:
www.preventioninstitute.org.
Winerup, Michael. (1987, January 20). An Inner City Asks For a Supermarket. New York
Times, New York and Region, Column One: Our Towns.
Data Sources
A) Census Block Groups
•Format: Polygon Shapefile
•Includes: Total population, racial population, and household income.
•Coordinate System: GCS_WGS_1984; D_WGS_1984; Greenwich,
Degree
•Details: This layer originally came from the 2000 Census/ TIGER files, joined with the 2000
SF3 survey information.
•Sources: http://www.census.gov/main/www/cen2000.html,
http://www.esri.com/data/download/
census2000_tigerline/index.html
B) Monroe County Geographical Data
•Format: Line and Polygon Shapefiles
•Includes: Parcel data, street_centerlines, town_villages
•Coordinate System: NY State Plane (NAD83, survey feet) FIPS 3103
•Details: These layers provide county boundary, city boundary, street centerlines, and parcel
data.
•Sources: Monroe County Department of Environmental Services: CD format
C) Bus Routes
•Format: Line Shapefiles
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
22
•Includes: RTS bus routes
•Coordinate System: NY State Plane (NAD83, survey feet) FIPS 3103
•Details: This layer shows accessibility to/from existing grocery stores in relation to bus transit.
•Sources: Genesee Transportation Council: emailed shape files based on Agreement of
Conditions.
D) Locations of Grocery Stores
•Format: Point Shapefiles
•Includes: georeferenced points of store locations
•Coordinate System: NY State Plane (NAD83, survey feet) FIPS 3103
•Details: This layer shows the current locations of grocery stores.
•Sources: grocery store websites:
Wegmans Food Markets website for store locations: http://www.wegmans.com
TOPS Markets website for store locations: http://www.topsmarkets.com
PriceRite Supermarkets for store locations: http://www.priceritestores.com
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
23
APPENDIX
A) Geocoding grocery store addresses
Existing locations for major retail grocery stores have been inputted into a table.
Open this table in Arc Catalog and drag it into the Arc Map blank document.
Right click on Major Grocery Stores and click geocode.
Import the Street USA (Street Map) address locator.
Use default settings and make sure that zip code is selected in the dialog box. Click
perform address match.
Check the dialog box to determine how many addresses were matched automatically with
80% accuracy and click the Match Interactively to match those addresses that need to be
updated for matching.
Examine each address and correct any spelling errors or typos and hit enter, match all
addresses within 75-80% accuracy.
Rename the new layer and change symbology to correlate with land use.
B) Displaying census data in Rochester
Obtain demographic information such as Summary Files 3 data within Monroe County
from 2000 Census Bureau. Select all block groups within the county. Download SF3
census tables for: Population Density, Racial distribution of population, and Household
income distribution
Download these files and export them to excel
Download TIGER data at census block level to get shapefiles
Download Monroe County street_centerline and town_village shapefiles
Find the two fields with the exact same information in the TIGER data and the census
data. The STFID field will be the join field in the census data to join to the GEO_ID_2
field in the TIGER data.
Use Arc Toolbox to first define the census tract shapefile and Monroe County GIS
shapefile.
Use Arc Toolbox to reproject the shapefiles as NY State Plane (NAD83, survey feet)
FIPS 3103
Begin a new ArcMap project.
Add the census tracts, town_village shapefiles and street_centerline shapefiles to your
map.
Extract the City of Rochester polygon from the town_villages shapefile and create it as a
new layer. This will be the boundary for the study area.
Clip the street layers to the City of Rochester shapefile.
Add the table of SF3 demographic data which was downloaded earlier to my ArcMap
project.
Join this table to the attribute table for the projected block group boundary using STFID
as the common Field
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
24
C) Suitability Analysis
Setting up spatial analysis
add layers to ArcMap: stores, streets (major), bus (routes), vac_comm (vacant
commercial properties over 2.75 acres), city_limits (roch city boundary), and income
(median income census data)
Activate Spatial Analysis
select Spatial Analysis→Options
on the extent tab, set the analysis extent to same as layer “city_limits”.This becomes the
extent for the outputs for every analysis.
Click OK
Raster Analysis
1) Finding straight line distance: Major Streets. The closer the better.
a. In the Spatial Analyst toolbar, navigate to Spatial Analyst Distance Straight
Line to create a raster based surface of distances based on a straight line.
b. Select streets as the layer in the Distance to: field.
c. In the Output raster: field, I saved the output in my Suitability_Analysis
suitability_analyst_store folder and named it streets_dist.
d. Click OK.
2) Reclassify Distance to Major Streets:
a. On the Spatial Analyst toolbar, navigate to Spatial Analyst Reclassify.
b. Specify streets_dist in the Input raster: field.
c. Click on Classify.
d. In the Classification window, make sure that Method: is set to Equal Interval and
Classes: is set to 10. It is important that the classification is set to equal interval
because this is a type of classification that can be kept the same between all of my
different layers. I want them the same so that I can compare the categories.
e. Click OK.
f. In the Reclassify window, I need to invert the New Values to reflect my
preference of the vicinity to major streets. Right now the closest values to the
highway are classified as 1, and the farthest away as 10. I want to reverse these
numbers, by typing in values from 10 to 1 in the New values field, so that those
parcels that are closest to the major highways will get the highest score.
g. Delete the last row that includes no data. To do so, right click on the last row and
select Remove Entries.
h. In the Output raster: field, I saved the output in my suitability_analyst_store
folder and named it st_dist_re. (Raster file names cannot be more than 13
characters or contain any spaces.)
i. Click OK.
j. Remove streets_dist from my table of contents.
k. I changed the color scheme to a gradual rank to visualize the classifications better.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
25
3) Finding straight line distance: Bus Routes. The closer the better.
a. In the Spatial Analyst toolbar, navigate to Spatial Analyst Distance Straight
Line to create a raster based surface of distances based on a straight line.
b. Select streets as the layer in the Distance to: field.
c. In the Output raster: field, I saved the output in my Suitability_Analysis
suitability_analyst_store folder and named it bus_dist.
d. Click OK.
4) Reclassify Distance to Bus Routes:
a. On the Spatial Analyst toolbar, navigate to Spatial Analyst Reclassify.
b. Specify bus_dist in the Input raster: field.
c. Click on Classify.
d. In the Classification window, make sure that Method: is set to Equal Interval and
Classes: is set to 10. It is important that the classification is set to equal interval
because this is a type of classification that can be kept the same between all of my
different layers. I want them the same so that I can compare the categories.
e. Click OK.
f. In the Reclassify window, I need to invert the New Values to reflect my
preference of the vicinity to major streets. Right now the closest values to the
highway are classified as 1, and the farthest away as 10. I want to reverse these
numbers, by typing in values from 10 to 1 in the New values field, so that those
parcels that are closest to the major highways will get the highest score.
g. Delete the last row that includes no data. To do so, right click on the last row and
select Remove Entries.
h. In the Output raster: field, I saved the output in my suitability_analyst_store
folder and named it bus_dist_re. (Raster file names cannot be more than 13
characters or contain any spaces.)
i. Click OK.
j. Remove bus_dist from my table of contents.
k. I changed the color scheme to a gradual rank to visualize the classifications better.
5) Finding straight line distance: Stores. The further the better.
e. In the Spatial Analyst toolbar, navigate to Spatial Analyst Distance Straight
Line to create a raster based surface of distances based on a straight line.
f. Select streets as the layer in the Distance to: field.
g. In the Output raster: field, I saved the output in my Suitability_Analysis
suitability_analyst_store folder and named it store_dist.
h. Click OK.
6) Reclassify Distance to Stores:
l. On the Spatial Analyst toolbar, navigate to Spatial Analyst Reclassify.
m. Specify store_dist in the Input raster: field.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
26
n. Click on Classify.
o. In the Classification window, make sure that Method: is set to Equal Interval and
Classes: is set to 10. It is important that the classification is set to equal interval
because this is a type of classification that can be kept the same between all of my
different layers. I want them the same so that I can compare the categories.
p. Click OK.
q. I do not need to renumber the current classifications since I feel the further away
the better, and the numbering scheme has already arranged them that way.
r. Delete the last row that includes no data. To do so, right click on the last row and
select Remove Entries.
s. In the Output raster: field, I saved the output in my suitability_analyst_store
folder and named it store_re. (Raster file names cannot be more than 13
characters or contain any spaces.)
t. Click OK.
u. Remove store_dist from my table of contents.
v. I changed the color scheme to a gradual rank to visualize the classifications better.
7) Classifying Income: Demographic data
a. Need to create a field that normalizes median income with census data.
b. Spatial Analyst → Convert → Conversion Tools → Features to Rasters
c. Input feature: income. Field: P053001 (median income field). Output Raster:
Income_Rast.
d. Reclassify: Input raster: Income_Rast.
e. Class: Classification set to Equal Interval, classes set at 10.
f. Click OK
Weighting and Combining Datasets:
1) On the Spatial Analyst toolbar, navigate to Spatial Analyst Raster Calculator
a. Type in an equation that will multiply each raster by the percentage weight I have
given it, then add them all together.
b. Enter in the following equation chosen given the weights I’ve chosen for the analysis
- for the grocery store perspective: income = 45%, streets = 35%, stores = 10%, bus
routes = 10%
- for the lower income and public transit perspective: income = 45%, streets = 15%,
stores = 10%, bus routes = 30%
c. Click Evaluate.
d. A new layer called Calculation will be added to your table of contents. This is
currently a temporary layer, but can be made permanent by right clicking on
Calculation and navigating to Data Make Permanent.
e. In the Make Calculation Permanent window, save the calculation name it weights
f. Click Save.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
27
g. The calculation is saved to my data folder, but the name in the table of contents will
remain the same. Change the name of the Calculation layer to weights.
h. Remove streets_dist_re, stores_dist_re, income_dist_re, and bus_dist_re from my
table of contents
Reclassifying Weights
1. Now, reclassify weights so that it follows the 1-10 ranking system I have been using with
all of my rasters.
a. On the Spatial Analyst toolbar, navigate to Spatial Analyst Reclassify
b. Specify weights in the Input raster window.
c. Click on Classify.
d. In the Classification window, make sure that Method: is set to Equal Interval and
Classes: is set to 10.
e. Click OK.
f. Leave the values in the reclassify window as they are.
g. Delete the last row that includes no data. To do so, right click on the last row and
select Remove Entries.
h. In the Output raster: field, save the output in your data folder and name it
weights_re. (Raster file names cannot be more than 13 characters or contain any
spaces.)
i. Click OK.
j. Remove weights from your table of contents.
k. Note that the value of 10 for weights_re still reflects the most desirable locations.
Combining the Raster and Vector Layers
1. Convert weights_re into a shapefile
a. On the Spatial Analyst toolbar, navigate to Spatial Analyst Convert Raster to
Features
b. In the Raster to Features window, choose inputs that match those shown in the image
below:
c. Save my Output features in my data folder and name it rank.
d. Click OK.
e. Remove weights_re from your table of contents.
f. Open the attribute table of rank.
g. Note that the GRIDCODE field includes the final ranking of the sites.
2. Intersect rank with suitable_areas
a. Open ArcToolbox .
b. In ArcToolbox, navigate to Analysis Tools Overlay Intersect.
c. In the Input Features field, select rank and suitable_areas.
Suitable Locations for Grocery Stores in Underserved Areas, Rochester, NY
Jaclyn Pryll, CRP 386 – Fall 2008
28
d. In the Output Feature Class, save the output in my data folder and name the file
suitable_parcels_ranked.
e. Click OK.
f. Remove rank and suitable_areas from my table of contents.
3. Change symbology of suitable_parcels_ranked in order to make the ranking more legible.
The ranks are contained in the GRIDCODE field.
D. Displaying access from grocery stores by walking (1/4 mile)
Download geocoded existing grocery store shapefiles
Select ArcToolbox – Proximity - Buffer
Input feature: Grocery Store shapefile layer
Output feature class: new shapefile named stores_walk_buff
Distance: 1320 linear feet
Disolve Type: none
click OK