兼容光学雷达影像及地形辅助数据的扎龙湿地遥感分类  

Wetland Mapping in the Zhalong Natural Reserve Using Optical and Radar Remotely Sensed Data and Ancillary Topographical Data

作　　者：刘蕾[1] 臧淑英[1] 那晓东[1] 裴雪原[1] 

机构地区：[1]黑龙江省普通高等学校地理环境遥感监测重点实验室,哈尔滨师范大学,黑龙江哈尔滨150025

出　　处：《地理与地理信息科学》2013年第1期36-40,F0002,共6页Geography and Geo-Information Science

基　　金：国家自然科学基金青年项目(41001243);国家自然科学基金重点项目(41030743)

摘　　要：结合Landsat TM影像、Envisat ASAR的C波段雷达影像和地形辅助数据,采用决策树方法,包括分类回归树(C1assification and Regression Tree,CART)和随机森林(Random Forest,RF)算法,对扎龙湿地进行遥感分类。用实测GPS样本点对分类结果进行精度验证,并与最大似然监督分类方法(Maximum Likelihood Classification,MLC)对比。结果表明,地形辅助数据和雷达后向散射系数对湿地分类精度的提高起重要作用。基于RF算法分类结果的总精度和Kappa系数分别为92.6%和0.901,沼泽湿地的分类精度达到96.3%,较CART算法和MLC监督分类方法有明显提高。该研究提供了一种快速、高效的内陆淡水沼泽湿地遥感分类技术。This study develops an expedient digital mapping technique using passive optical remotely sensed imagery of the Landsat Thematic Mapper（TM）, Envisat ASAR active radar （;-band imagery, and topographical indices. All data inputs were re- sampled to a common 30 m resolution grid. An ensemble classifier based on trees （Random Forest for example） procedure was employed to produce a final map of pe^grid cell wetland probability map. A detailed accuracy assessment showed that ancillary topographical data and radar backscattering coefficients effected prominently in contributing to the classification tree procedure, reinforcing the idea that multi source data are useful in the characterization of wetland land cover. Random Forest （RF） pro duced the highest overall accuracy （92. 6%） and the Kappa coefficient is 0. 901, with marsh class accuracy 96. 3%. It outper- forms standard approaches like a single classification and regression tree （CART） and a conventional maximum likelihood clas- sification （MLC）. The method employed freely available data and a fully automated process. This study provides a fast, efficient remote sensing classification method of freshwater marsh wetland and has an important meaning in wetland information extrac tion,management and protection.

关 键 词：扎龙湿地 随机森林 ENVISAT ASAR地形辅助数据 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]