Upwind

An AQI map for the US that explains why the air is bad instead of just displaying a number. Upwind breaks down local air quality spikes, pulling satellite feeds, sensor data, and current events to rank the most likely causes.

Class	Likely Triggers
Wildfire Smoke	PM2.5 + satellite fire hotspots + AOD plumes
Ozone Episode	High ground O3 + high heat (≥75°F)
Windblown Dust	High coarse PM10 + high winds
Thermal Inversion	Fine PM2.5 + cold, calm weather + shallow mixing layer
Urban Pollution	Local PM2.5 + clean overhead AOD + no fire hotspots

Note: Upwind provides evidence and heuristics based on public satellite and monitor feeds. It is not physical air sampling or lab testing. The application's conclusions are ranked observational hypotheses, not verdicts.

---

Stack

• Backend: Python 3.11, FastAPI, SQLite
• Frontend: React 18, Vite, MapLibre GL / Leaflet
• Data Sources: AirNow API, Open-Meteo, NASA FIRMS Thermal Hotspots

---

Environment Variables

Copy .env.example to .env before running:

Variable	Type	Description
AIRNOW_KEY	Optional	Primary US monitor data feed
FIRMS_MAP_KEY	Optional	NASA satellite thermal hotspot detection
DEEPSEEK_API_KEY	Optional	My briefing LLM of choice
GROQ_API_KEY	Optional	LLM Judge
RATE_LIMIT_AQI_PER_MIN	Optional	Max /api/aqi requests/min per IP

---

Local Dev

1. Backend API

python3 -m venv venv
source venv/bin/activate
pip install -r backend/requirements.txt
uvicorn backend.main:app --reload --port 8000

2. Frontend

cd frontend
npm install
npm run dev

Open http://localhost:5173