2004 Volume 20 Issue 1 Pages 86-96
Unmanaged bamboo forest worsens the landscape of relative regions and river basin ecosystems. Besides this, its consistent expanding could erode artificially planted forest resources and agricultural fields, and may also induce some serious environmental problems. In this study, the distribution of bamboo in the surrounding areas of Gifu-city, with a total area of 14,280 ha, was extracted using the decision tree classifier method and the maximum likelihood classifier method, respectively. For this purpose, high-resolution satellite IKONOS data and mediate-resolution satellite Landsat ETM+ data were utilized, along with fusion images generated using IHS conversion based on both satellite data. Using field-surveyed results, the accuracy of the classification was assessed. In five data sets of Landsat ETM+ acquired in different seasons, the ETM+ data acquired in March was found to be the best bamboo classification (with 61.9% of precision). By applying generated fusion images, the classification precision was improved as much as 84.9%, i.e., a level higher than individual use of either IKONOS or Landsat ETM+ data. Moreover, based on fusion images, the bamboo distribution area was estimated to be 89.6 hectares in the study area.