Estimation Of Fuel Moisture Content by Integrating Surface and Satellite Observations Using Machine Learning

Fuel moisture content (FMC) is an important fuel property and an important parameter controlling the rate spread of a wildland fire. Currently the dead FMC is estimated based on relatively sparse observations over Conterminous United States while the live FMC is sampled manually and infrequently. An effective operational wildland fire prediction requires real-time, high-resolution fuel moisture content data set. We have therefore developed a fuel moisture content data set by combining satellite and surface observations as well as National Water Model output using a machine learning model. The new FMC data set is integrated in the Colorado Fire Prediction System (CO-FPS) for operational wildland fire prediction.