Portland State University - Capstone 2025-26

Wildfire Forecasts For Everyone.

Physics-driven WRF‑SFIRE simulations and FourCastNet AI - so communities can see how fires spread before they do.

WRF-SFIRE FourCastNet AI Portland State University Real-Time Forecasting
Palisades fire spread simulation - animated WRF-SFIRE output
WRF-SFIRE Output Palisades Fire · 2025

Recent Fire Simulations

High-resolution wildfire spread simulations from our coupled WRF‑SFIRE model.

Palisades fire simulation - animated fire spread
WRF-SFIRE

Palisades Fire

Physics-based two-way fire-atmosphere coupling. Terrain, fuel, and live wind dynamics captured in real time.

Caldor fire simulation - animated fire spread
Model Output

Caldor Fire

Validated against observed fire perimeter progression and satellite imagery data.


Powered By

Physics meets AI - two complementary systems, one unified forecasting platform.

Dynamical Model

WRF‑SFIRE

A physics-based coupled model simulating the two-way interaction between wildfires and the atmosphere. Captures fire behavior driven by terrain, fuel, and live wind dynamics in real time.

Learn how it works →
AI / Deep Learning

NVIDIA FourCastNet

State-of-the-art deep learning for fast, accurate atmospheric forecasting. We fine-tune FourCastNet on WRF‑SFIRE outputs to accelerate wildfire spread prediction at scale.

Learn how it works →

7 Team Members
2 Model Tracks
PSU Portland State
2025 Capstone Year

Making wildfire prediction accessible to everyone.

Wildfires are a growing threat - causing billions in damage, displacing thousands, and degrading air quality across entire regions. Our PSU Capstone team is building a platform that makes state-of-the-art fire forecasting accessible, interpretable, and actionable for the public.