Fuel moisture prediction in homogeneous fuels using GIS and neural networks

The goal of this project was to use GIS data layers and a neural network to predict values of fuel (vegetation) moisture across a surface based on other known environmental parameters. The location of this study was the Buenos Aires National Wildlife Refuge in south-central Arizona. The refuge is managing masked bobwhite quail habitat and has an extensive prescribed fire program. The expected relationship between measured environmental variables and fuel moisture values at the sample points in the refuge was not clearly modeled by the combined use of GIS data and neural networks. However, a higher degree of variability in the geographic data would inevitably improve the performance of the neural networks, leading to more accurate predictions.