Fourier Transform Analysis of Very High Resolution Remote Sensed Imagery Shows Potential for Estimating Stand Density of Regular-spaced Planted Forests

摘要

An increase in the area of unthinned planted forests is a major issue in Japan, and incomplete record-keeping means that a method for detecting unthinned stands is required. An important indicator of unthinned stands is stand density. In this study, we compared two stand density estimation methods, the discrete Fourier transform method and the local maximum (LM) filtering method, using simulated very high resolution satellite imagery. Even-aged Japanese cedar (Cryptomeria japonica) stands were modeled. Trees were positioned as though planted in a lattice pattern typical of these forests in Japan. The discrete Fourier transform method was more accurate than LM filtering and was stable over a range of stand densities, whereas the accuracy of LM filtering was sensitive to stand density and calculation conditions. Hence, the discrete Fourier transform was superior to LM filtering to estimate even-aged coniferous stand density.