Integration of digital satellite data and forest inventory data for forest cover mapping in Korea

Abstract

Forest cover maps are a major product of the National Forest Inventory (NFI) system in Korea. The main objective of this study is to evaluate the potential of digital satellite imagery in combination with field plot data from the NFI to support forest cover classification. Field plot data from the NFI and Landsat TM for a test area were used to generate a forest cover map through pixel‐wise classifiers. For classification, two pixel‐wise classifiers, the Nearest Neighbor (NNC) and the Maximum likelihood (MLC) were applied and their results were compared with a classification from field plot data per sub‐plot (n = 970). The NNC yielded higher accuracy than the MLC. The estimated kappa for NNC was about twice as high as for MLC. The NNC classified image was also assessed using existing digital forest map derived from aerial photo interpretation as a reference. The accuracy, however, was modest (κ = 0.28). The goodness‐of‐fit test indicates that the digital forest map and the MLC classified image differ significantly from the result of field plots, while a statistically significant difference between field data and the NNC classified image was not found.