期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2010
卷号：XXXVIII - Part 8
页码：690-695
出版社：Copernicus Publications
摘要：Global percent tree cover is an important parameter to understand global environment. Some attempts to produce global percent tree cover maps have been made so far. But the accuracy of these maps is not so high. In this study, percent tree cover of some areas in Eurasia was estimated using a supervised regression tree algorithm from MODIS data in 2003 as a preliminary research. Simulated training data were created from a lot of ground truth data consisted of various land cover types to improve the accuracy of the estimate. The ground truth data were collected from QuickBird images and Google Earth images. In South Asia and a part of Indonesia, the percent tree cover in 2008 was also estimated and compared with the result in 2003 to investigate the stability of the estimation result and the possibility of change detection. In areas where training data were collected, the accuracy of the estimate improved. This means the necessity of constructing regression tree models area by area to increase the accuracy
关键词：Forestry; Land Cover; Mapping; Model; Estimation; High resolution; Comparison
