Brownfield Opportunities
Case study · Infrastructure
A Reindustrialization & Energy Exploration
- What
- ~46,800 federally-tracked contaminated sites, scored for redevelopment and new power.
- Stack
- Pure-Python spatial pipeline over federal ArcGIS services; Leaflet + Canvas map.
- Status
- LIVE
- Dates
- Published May 2026 · Updated July 2026
- Links
- Live site · Source code
Background
Where are the contaminated and underused sites that could be redeveloped, and which of them already sit next to the power, water, and transport you'd need to actually build?
This dashboard maps ~46,800 federally-tracked contaminated sites and scores each one on how close it sits to the transmission, generation, water, and transport a build actually needs. The same proximity that suits energy redevelopment also suits AI and data-center load growth. One site can carry the build-out, bring new generation online, and send jobs and cleanup to a community that lived with the contamination for decades. EPA's RE-Powering program already flags hundreds as redevelopment-ready.
A Look Inside
Each view shown on mobile and desktop — tap any image to open the live site.
How It Works
I vibe-coded the data pipeline and the dashboard, leaning on official federal sources so the map is defensible. The whole thing is static and refreshes from one command:
- Data sources: ~46,800 federally-tracked sites across EPA Superfund (NPL), EPA ACRES brownfields, USACE Formerly Used Defense Sites, and DOD BRAC, pulled from their official ArcGIS/REST services.
- Enrichment: a pure-Python spatial grid computes mile-precision distance from every site to the nearest transmission line (with kV), substation, power plant (with MW and fuel), road, rail, and gas pipeline.
- Scoring: FEMA flood zones, EPA ECHO enforcement history, and EPA's RE-Powering signals, which flag data-center-reuse candidates — ≥50 acres with power and water on site.
- Narratives: Claude Haiku turns each site's status, acreage, infrastructure, and enforcement into a short buyer-oriented summary, cached by content hash so re-runs only re-bill on real changes.
- UX: a Leaflet + Canvas map and a sortable table with filters, shareable URL state, CSV export, and a mobile bottom-sheet detail panel, all served as static files on GitHub Pages.