Wildfire Early Detection System
Computer Science assignment, 2026
A real-time wildfire risk detector. A Raspberry Pi Pico reads temperature and soil moisture, a historical weather pipeline feeds a fire spread model, and a HIGH reading triggers an SMS and a phone call through Twilio. Once it’s triggered, nobody has to touch it.
The model is based on the Rothermel surface fire spread model, and I calibrated the risk thresholds against the 2020 LNU Lightning Complex fires.
- MicroPython
- Python
- NumPy
- pyserial
- Open-Meteo API
- Twilio
- Shell
Raspberry Pi PicoDHT11, soil sensor and potentiometer send CSV over USB serial
MacBookFetches weather history, runs the spread model, classifies each reading
TwilioTexts and calls you when risk is HIGH
Lowunder 5 km/h
Medium5 to 9 km/h
Highover 9 km/h
ROS = 0.25 × wind_speed area = ROS × 1.852 × (1 − soil_moisture / 100) × temp_factor temp_factor = max(0.1, 1 + (temperature − 20) / 30)