AI-powered climate equity simulation platform — place green infrastructure without displacing vulnerable communities.

GreenWatch helps policymakers model the impact of green infrastructure investments (parks, greenways, transit, tree planting) on both environmental quality and displacement risk across 85,000+ US census tracts. It makes the tradeoffs of green gentrification visible before decisions are made.

Built for HooHacks 2026.

• Interactive Map — Visualize every US census tract colored by Displacement Risk Score (DRS) or Environmental Benefit Score (EBS)
• Intervention Placement — Place 6 types of green infrastructure with configurable scale and see their impact radius
• What-If Simulation — Predict how interventions affect surrounding tracts using a spatial distance decay model
• Policy Mitigations — Model rent stabilization, community land trusts, and affordable housing mandates to counteract displacement
• AI Optimization — Get recommended placement locations that maximize environmental benefit while minimizing displacement risk

greenwatch/
├── frontend/          # Next.js app (map UI, simulation controls, results)
├── scoring/           # FastAPI scoring & simulation service
├── pipeline/          # ETL data ingestion (Census, EPA, CDC, FEMA)
├── db/                # PostgreSQL/PostGIS schema & migrations
└── docker-compose.yml # Full-stack local orchestration

Data flow:

1. pipeline/ ingests federal datasets → PostgreSQL
2. scoring/ computes DRS & EBS for all tracts, runs simulations
3. frontend/ proxies simulation requests to scoring service, renders results on map

• 40% Vulnerability: renter %, rent burden, poverty, race, education, eviction rate, SVI
• 35% Market Pressure: 5-year % change in rent, home value, household income
• 25% Green Proximity: proximity to existing green infrastructure

• 30% Air Quality: PM2.5, ozone, diesel particulate
• 30% Green Infrastructure: tree canopy, park access, impervious surface
• 25% Climate Resilience: flood, heat, hurricane risk (inverted)
• 15% Health: asthma & mental health prevalence (inverted)

All indicators are percentile-ranked before weighting to normalize across disparate units.

• Docker & Docker Compose
• Node.js 20+
• Python 3.11+
• Census API key (get one free)

git clone https://github.com/JAGAN666/GreenWatch.git
cd GreenWatch
cp .env.example .env
# Fill in your API keys in .env

docker-compose up --build

This starts:

• PostgreSQL + PostGIS on port 5432
• Redis on port 6380
• FastAPI scoring service on port 8000
• Next.js frontend on port 3000

cd pipeline
pip install -r requirements.txt
python run_pipeline.py

This ingests all federal datasets and computes initial DRS/EBS scores for all US census tracts (~85,000).

Visit http://localhost:3000

All scoring endpoints are available at http://localhost:8000.

Interactive API docs available at http://localhost:8000/docs.

# Database
DATABASE_URL=postgresql://greenwatch:greenwatch@localhost:5432/greenwatch

# Redis
REDIS_URL=redis://localhost:6379

# Scoring service (used by Next.js API routes)
SCORING_SERVICE_URL=http://localhost:8000

# Census API key
CENSUS_API_KEY=your_census_api_key_here

• Frontend: Deployed on Vercel — connect your repo and it deploys automatically
• Map Tiles: PMTiles file hosted on Cloudflare R2 — upload via wrangler r2 object put
• Scoring Service: Containerized FastAPI — deploy to any Docker-compatible host (Railway, Fly.io, GCP Cloud Run)

MIT