The objectives of this study were 1) to modify the site index prediction model for sugi (Cryptomeria japonica) planted forests in Miyazaki Prefecture developed in a previous study and 2) to investigate the effects of quality of DEM and the scale of digital terrain analysis on the performances of site index models. The study site was the Tano Forest Science Station, University of Miyazaki. We acquired 18 data sets of site index estimated using stem analysis and global position of each sample site where a sample tree was felled. Three topographic factors,a solar radiation index, a hydrological upslope contributing area index, and a vertical topographic exposure index, were derived from DEMs generated from three data sources (10-m and 50-m interval point data with 3D coordinates and digitized contour map) and at resolutions of 10, 12.5, 25, and 50 m. Several search ranges (100, 250, 500, and 1000 m) were tested. Correlation analysis between site index and topographic factors as well as regression analysis to develop a site index prediction model using topographic factors as explanatory variables revealed that the hydrological upslope contributing area index requires DEMs generated from more informative DEM data sources (10-m interval point data and digitized contour map) and at fine resolution (10 m or 12.5 m); however, these DEMs were unsuitable for solar radiation index. DEMs generated from a less informative DEM data source (50 m interval 3D point data) and at a coarse resolution of 50 m were suitable for the solar radiation index. The effects of search ranges on topographic factors were clear for vertical topographic exposure index but not for the others. The best model developed in this study accepted the solar radiation index derived from the digital contour map based 10-m resolution DEM, the hydrological upslope contributing area index derived from the digital contour map based 12.5-m resolution DEM, and the vertical topographic exposure index derived from the 50-m interval 3D point data based 50-m resolution DEM as the explanatory variables.
View full abstract