Abstract
Stands of Cryptomeria japonica and Chamaecyparis obtusa have different functions: in particular, diversity of understory vegetation and prevention of soil erosion, but their distribution is not always delineated and distinguished on forest maps. This study investigated the relationship between parameters in the image segmentation procedure for high-resolution satellite data and generated object size in object-oriented classification, and distinguished objects of C. japonica and C. obtusa by means of discriminant analysis. Thirty-six sample plots were established in C. japonica and C. obtusa stands in the national forests of the Koisegawa watershed. QuickBird panchromatic and multispectral data was used for this study. Segmentation, which is the first step in object-oriented classification, was applied to the image data of the sample plots, and species were assigned to corresponding objects generated from the image segmentation. For these objects, the average and standard deviation of digital number for the four multispectral bands and panchromatic band and the Normalized Difference Vegetation Index (NDVI) were calculated within them. Discriminant analysis to distinguish C. japonica and C. obtusa was conducted using these twelve variables of average and standard deviation as independent variables. The correct distinction was made for 100% of C. japonica objects and 95.5% of C. obtusa objects. The results clarified that it was possible to distinguish C. japonica and C. obtusa patches using high-resolution satellite data.
