NYC Grocery Access Supermarket Site Selection

How this prototype scores sites

Question the tool answers: if the city were to site a full-service grocery store on land it already owns, where would it close the largest food-access gap? Candidate pool: vacant, publicly owned lots (PLUTO land use = vacant, owner type city / other public / public authority), cross-referenced with the City Owned and Leased Property list. Lots in pure manufacturing zoning districts are excluded.

Score = 50% food access gap + 25% need + 15% reach + 10% transit

• Food access gap — distance to the nearest full-service supermarket (80% of the component; capped at 1 mile) plus USDA designation of the surrounding tract as low-income/low-access (20%). “Full-service supermarket” borrows the FRESH program threshold: a licensed retail food store of 6,000+ sq ft.
• Need — tract deprivation (poverty, unemployment, housing cost burden; 60%) plus households without a vehicle (40%), population-weighted and distance-decayed. No car makes supermarket distance bind much harder.
• Reach — residents below 150% of the poverty line within a 10-minute walk of the site.
• Transit — distance to the nearest subway station.

Each measure is a percentile rank against all candidate sites citywide, then combined with the weights above. Adjusting weight sliders re-ranks instantly.

Data sources (all public)

• Retail Food Stores — NYS Dept of Agriculture & Markets license data (store locations, square footage)
• USDA Food Access Research Atlas 2019 — low-income/low-access census tracts
• MapPLUTO & COLP — NYC Dept of City Planning / NYC Open Data (lots, zoning, ownership, FEMA 2015 preliminary flood flag)
• CDC/ATSDR Social Vulnerability Index 2022 — tract demographics incl. vehicle access
• Facilities Database (FacDB) — NYC DCP (food pantry context layer); MTA subway stations — NYC Open Data

Known limits — read before trusting a ranking

• Candidate ≠ available. Public data cannot see encumbrances, commitments, or disposition status. This is not an acquisition list.
• A license is not an open, stocked store. License data measures permission to operate, not store quality, and ~25% of licensed stores lack recorded square footage, so some real supermarkets are missing from the layer.
• The USDA Atlas is a 2019 vintage on 2010 census tracts; roughly 4 in 10 lots could not be matched to an Atlas tract and default to “not designated.” The designation is a sparse booster, not the main signal.
• Store viability (sales volume a location could support) is not modeled. High gap does not guarantee a store can operate profitably there.
• Tract locations are approximated from facility positions; transit uses subway only, understating access in eastern Queens and Staten Island.