Project produced by Kelsey Voit and Evan Judge.
INTRODUCTION
Several neighborhoods in Louisville, Kentucky are lacking adequate grocery stores and subsequently have food access issues. The Louisville Food Cooperative is considering building a grocery store in either the Portland, Russell, Shawnee, Parkland, Old Louisville, Smoketown, or Shelby Park neighborhood to address these food access issues. This project seeks to determine an ideal location to build a grocery store by conducting site suitability and network analyses.
INTRODUCTION
Several neighborhoods in Louisville, Kentucky are lacking adequate grocery stores and subsequently have food access issues. The Louisville Food Cooperative is considering building a grocery store in either the Portland, Russell, Shawnee, Parkland, Old Louisville, Smoketown, or Shelby Park neighborhood to address these food access issues. This project seeks to determine an ideal location to build a grocery store by conducting site suitability and network analyses.
METHODS
We conducted this comprehensive analysis in three parts: site suitability, maximum market share, and service areas. We performed a binary site suitability analysis in ArcMap and determined parameters in tandem with Louisville Food Cooperative (LFC) leadership. We extracted the target neighborhoods from the Louisville/Jefferson County Information Consortium (LOJIC) urban neighborhoods layer and clipped Property Value Administrator (PVA) parcel data to that extent. We then extracted both vacant lots with commercial zoning from the parcel layer as well as sites that the LFC is already considering. To differentiate between empty lots and lots that may still have useable buildings we used Open Street Map building data. We then merged both layers and extracted all vacant lots containing buildings to a new layer. In order to ground truth our data, we utilized Google Maps street view (images last updated in September 2015, December 2015 and December 2016), to locate the candidate sites and remove sites without buildings from the data layer.
Using street centerline information we created a network of roads using distance as an impedance value. Using an ERDI online data set of US groceries sorted by NAICS codes, we determined competitor groceries within a five mile radius of the target grocery stores. We assigned weighted valued to competitors: a 2 for supermarkets and a 1 for convenience and specialty stores. Demand points were created using all non-vacant residential parcels of land within target area. Next, we conducted a maximum market share analysis on our car network to determine where a grocery store could be placed to capture the most unmet demand for food access, using a maximum impedance of four miles as an estimated maximum distance that residents would be willing to drive to the grocery. This process was repeated three times eliminating the former choice each time to determine the three top candidates for those with access to cars. To create a multimodal network, we used the LOJIC street centerline data once again, as well as the GTFS to Network Dataset tools created by Melinda Morang, and GTFS data obtained from TransitFeeds.com. The impedance values between the two networks differ. The car network uses length in feet as an impedance value while the bus-pedestrian network uses travel time in minutes. To create the multimodal network, we transformed the street centerline into pedestrian values by using the length a pedestrian can walk in one minute as a conversion factor, since the impedance value for the bus lines was already formated in minutes. We used the GTFS tool to create the multimodal network; this tool creates new intersections to add connectivity between the two modes of transit. We set the impedance to two minutes for transferring from the street network to the bus-pedestrian network and 30 seconds for going from the bus-pedestrian to the street network to account for time spent getting on and off the bus. Using this multimodal network, we ran a maximize market share analysis using the competitor sites to determine the top three results and see how bus-pedestrian access would differ from car access, using an impedance of 30 minutes as an estimated maximum time people would be willing to spend traveling to the grocery for bus-pedestrian network. For each of the candidate locations, we then ran a service area analysis to determine accessibility of each store within a 7.5, 15, and 30 minute trip to visualize which residents could travel to each location using only busing or walking as a means of transport.
We conducted this comprehensive analysis in three parts: site suitability, maximum market share, and service areas. We performed a binary site suitability analysis in ArcMap and determined parameters in tandem with Louisville Food Cooperative (LFC) leadership. We extracted the target neighborhoods from the Louisville/Jefferson County Information Consortium (LOJIC) urban neighborhoods layer and clipped Property Value Administrator (PVA) parcel data to that extent. We then extracted both vacant lots with commercial zoning from the parcel layer as well as sites that the LFC is already considering. To differentiate between empty lots and lots that may still have useable buildings we used Open Street Map building data. We then merged both layers and extracted all vacant lots containing buildings to a new layer. In order to ground truth our data, we utilized Google Maps street view (images last updated in September 2015, December 2015 and December 2016), to locate the candidate sites and remove sites without buildings from the data layer.
Using street centerline information we created a network of roads using distance as an impedance value. Using an ERDI online data set of US groceries sorted by NAICS codes, we determined competitor groceries within a five mile radius of the target grocery stores. We assigned weighted valued to competitors: a 2 for supermarkets and a 1 for convenience and specialty stores. Demand points were created using all non-vacant residential parcels of land within target area. Next, we conducted a maximum market share analysis on our car network to determine where a grocery store could be placed to capture the most unmet demand for food access, using a maximum impedance of four miles as an estimated maximum distance that residents would be willing to drive to the grocery. This process was repeated three times eliminating the former choice each time to determine the three top candidates for those with access to cars. To create a multimodal network, we used the LOJIC street centerline data once again, as well as the GTFS to Network Dataset tools created by Melinda Morang, and GTFS data obtained from TransitFeeds.com. The impedance values between the two networks differ. The car network uses length in feet as an impedance value while the bus-pedestrian network uses travel time in minutes. To create the multimodal network, we transformed the street centerline into pedestrian values by using the length a pedestrian can walk in one minute as a conversion factor, since the impedance value for the bus lines was already formated in minutes. We used the GTFS tool to create the multimodal network; this tool creates new intersections to add connectivity between the two modes of transit. We set the impedance to two minutes for transferring from the street network to the bus-pedestrian network and 30 seconds for going from the bus-pedestrian to the street network to account for time spent getting on and off the bus. Using this multimodal network, we ran a maximize market share analysis using the competitor sites to determine the top three results and see how bus-pedestrian access would differ from car access, using an impedance of 30 minutes as an estimated maximum time people would be willing to spend traveling to the grocery for bus-pedestrian network. For each of the candidate locations, we then ran a service area analysis to determine accessibility of each store within a 7.5, 15, and 30 minute trip to visualize which residents could travel to each location using only busing or walking as a means of transport.
RESULTS: Site Suitability
Upon conducting the binary site suitability analysis depicted in Figures 1 through 19, we found eight suitable sites: 622 E Ormsby Ave., 1700 Owen St., 3800 River Park Dr., 924 S. 2nd St., 431 E. Liberty St., 2809 W. Broadway, 1507 W. Market St., and 3005 Greenwood Ave.
Upon conducting the binary site suitability analysis depicted in Figures 1 through 19, we found eight suitable sites: 622 E Ormsby Ave., 1700 Owen St., 3800 River Park Dr., 924 S. 2nd St., 431 E. Liberty St., 2809 W. Broadway, 1507 W. Market St., and 3005 Greenwood Ave.
|
|
|
RESULTS: Maximum Market Share Network Analysis
The top three sites that reach the maximum market share on the automobile network are 3800 River Park Dr, 2809 W. Broadway, and 3005 Greenwood Ave., respectively. For the bus-pedestrian network, the top three sites that reach the maximum market share are 3800 River Park Dr., 3005 Greenwood Ave., and 622 E Ormsby Ave., respectively. The top site for each network is displayed in Figures 24 and 25 below.
The top three sites that reach the maximum market share on the automobile network are 3800 River Park Dr, 2809 W. Broadway, and 3005 Greenwood Ave., respectively. For the bus-pedestrian network, the top three sites that reach the maximum market share are 3800 River Park Dr., 3005 Greenwood Ave., and 622 E Ormsby Ave., respectively. The top site for each network is displayed in Figures 24 and 25 below.
|
|
|
|
|
RESULTS: Service Areas Network Analysis
Figures 26 through 33 below display the service areas reached for each of the candidate sites on the bus-pedestrian network in 7.5, 15, and 30 minute buffers.
Figures 26 through 33 below display the service areas reached for each of the candidate sites on the bus-pedestrian network in 7.5, 15, and 30 minute buffers.
CONCLUSION
These results indicate that eight sites meet the parameters set for an ideal location for the Louisville Food Cooperative grocery store. 3800 River Park Dr. reaches the maximum number of residents in the target neighborhoods based on both the car and bus-pedestrian networks while considering competitor stores. The service area analyses explore accessibility for each candidate site. This analysis can be used to supplement research being done by the Louisville Food Cooperative to find an ideal grocery store location.
Limitations of this study include currency of data and incongruence in dataset publishing dates and square footage estimated from roof values instead of true building size. The network could be improved by including travel restrictions such as highways, train tracks, handicap accessibility, bike routes, etc. and further refining demand points. Additional factors will need to be considered in order to solidify an ideal location, including the condition of the vacant buildings, handicap accessibility, cost, whether additional grocery stores close or open in these areas, etc. Furthermore, our study limited candidate sites to vacant commercial lots, but the LFC may choose to consider constructing a new building on an empty lot or leasing an existing building. Additionally, they may choose to consider more buildings that are located outside of but still accessible to the target market areas.
These results indicate that eight sites meet the parameters set for an ideal location for the Louisville Food Cooperative grocery store. 3800 River Park Dr. reaches the maximum number of residents in the target neighborhoods based on both the car and bus-pedestrian networks while considering competitor stores. The service area analyses explore accessibility for each candidate site. This analysis can be used to supplement research being done by the Louisville Food Cooperative to find an ideal grocery store location.
Limitations of this study include currency of data and incongruence in dataset publishing dates and square footage estimated from roof values instead of true building size. The network could be improved by including travel restrictions such as highways, train tracks, handicap accessibility, bike routes, etc. and further refining demand points. Additional factors will need to be considered in order to solidify an ideal location, including the condition of the vacant buildings, handicap accessibility, cost, whether additional grocery stores close or open in these areas, etc. Furthermore, our study limited candidate sites to vacant commercial lots, but the LFC may choose to consider constructing a new building on an empty lot or leasing an existing building. Additionally, they may choose to consider more buildings that are located outside of but still accessible to the target market areas.