A model-based approach to estimating forest area

A logistic regression model based on forest inventory plot data and transformations of Landsat Thematic Mapper satellite imagery was used to predict the probability of forest for 15 study areas in Indiana, USA, and 15 in Minnesota, USA. Within each study area, model-based estimates of forest area were obtained for circular areas with radii of 5 km, 10 km, and 15 km and were compared to design-based estimates based on inventory plot data. Precision estimates for the circular areas were also obtained using variance formulae developed for this application that incorporated spatial correlation among model predictions for individual pixels. The model-based estimates were generally comparable to the design-based estimates. The advantages of the model-based approach are that maps and small areas estimates may be obtained and the necessity of releasing exact plot locations for user-specific applications is alleviated.