SEMIVARIANCE ANALYSIS OF FOREST STRUCTURE AND REMOTE SENSING DATA TO DETERMINE AN OPTIMAL SAMPLE PLOT SIZE

摘要

In a preliminary study to determine the optimal sample plot size for a mixed forest region, semivariograms have yielded useful statistics from analysis of forest structural data and remotely sensed imagery. Omni-directional semivariograms were extracted from forest structure attributes; DBH, tree height and average crown diameter, to derive second moment statistics describing the spatial dependence of each data set. The variables were then related using the relative nugget effect (ratio of nugget: sill) to determine the most reliable range value to use for the determination of an optimal ground sample resolution. The same analysis was performed on a series of degraded red and near infrared images covering the same study site. The results show that each data set usually characterizes spatial dependence between 8 and 10m.