基于决策树和MODIS植被指数时间序列的中亚土地覆盖分类

利用MODIS植被指数时间序列对中亚土地覆盖类型分类进行了研究。MODIS数据时间分辨率高，时间序列数据可以表征植被生理活动的动态变化。从时间序列数据中提取植被物候信息，可以实现对不同土地覆盖类型的定量描述。MODIS数据质量信息波段，记录了研究区遥感数据质量，提供植被指数可用性、气溶胶处理、云、冰雪可能性、合成方法等信息，为植被指数时间序列噪声的去除提供一种新的方法。决策树分类结构清晰、不基于正态统计分布假设、效率高、分类精度高。分类结果与统计数据比较，两者一致性较好，精度验证总体精度95.76％，kappa系数0.9516。

Land Cover Classification of Middle Asia based on Decision Tree and MODIS Vegetation Index Time Serials

Land cover classification is studied byusing MODIS vegetation index time serials .It can characterize the dynamics of vegetation physiological change . Phenology information can be extracted from the time series tag, which records the quality and describe different land cover of remote sensing data,provides types quatitatively.The quality information on usefullness of vegetation index data,processing of aerosol,possibility of cloud and ice ,composite method It provides a new method of processing the noise including the vegetation index time serials.The classification method of Decision tree has a clear structure,not based on assumption of normal distributing with high efficiency and classification precision.The result is compared with statistic data with high consistency. Accuracy estimate is 95.76%,and kappa coefficient is 0.9516.