2006 Volume 12 Issue 2 Pages 49-58
Five images of Landsat Thematic Mapper (TM) and one image of ASTER data at different years were taken to determine the topographic correction effects in the mountainous area in Tsugawa region, Niigata. An automatic scattergram-controlled regression approach was applied for radiometric normalization correction using sequence images obtained using the Landsat TM. The forest was classified as either a deciduous, low-middle volume sugi or high volume sugi forest based on its characteristics and volume by separability analysis. We examined four topographic correction methods to reduce the slope-aspect effect in each forest class compared. In order to elucidate the effect of the applied correction method, we also analyzed the correlation between the digital number and ground measurement. The results showed that the specific forest class type should be considered as input when applying a topographic correction method in satellite images analysis. The two-stage topographic normalization proposed by CIVCO (1989) and statistic-empirical methods appeared to be the best for statistical analyses but were unsuccessful regarding actual correlation for analysis of sugi stand volume. The results indicated that, the C-correction for visible bands and MINNAERT correction methods for infrared band are adequate methods for improving the correlation accuracy between derived digital number and sugi stand volume in the Tsugawa region.
