Wildfire Forecasting from Sparse Observational Measurements

Louisa Ebby

North Carolina State University

"Wildfire Forecasting from Sparse Observational Measurements"

As wildfires increase in frequency and intensity due to climate change, so does the need to create better forecasts. The wildfire perimeter is seldom fully observable, but the geospatial locations of first responder and 911 civilian cellphone calls provide a sparse representation of the wildfire in real time. We use these calls to estimate the complete fire perimeter at specific times. We present a state estimation method that yields smaller reconstruction errors than existing methods. Using the reconstruction as an initial condition, we run a cellular automaton to predict the future state of the fire. As a case study, we use calls from Maui during the devastating wildfires in August 2023 to predict the final fire perimeter.

---