面向对象的黑河下游河岸林植被覆盖信息分类!

地表植被覆盖是描述区域生态系统的基础数据,也是全球及区域陆面过程、生态与水文众多模型中所需的重要地表参数。对于黑河下游额济纳绿洲,以Landsat 30m分辨率为主的遥感影像难以真实提取下游绿洲河岸林植被覆盖信息,而高分辨率影像目标地物轮廓清晰、空间细节信息丰富,有利于干旱背景下景观破碎、异质性强的植被覆盖信息分类。基于黑河下游额济纳绿洲QuickBird影像,通过面向对象的分类方法提取耕地、胡杨、柽柳、草地和裸地等主要植被覆盖类型,分类总体精度和Kappa系数分别为84.71%和0.7986。结果表明:利用面向对象分类方法对高分辨率影像进行植被覆盖信息分类,分类结果较好,能够满足精度要求。

Riparian Forest Vegetation Coverage Information Classification based on Object\|oriented Method in Heihe River

Vegetation coverage is not only the basic data to describe the regional ecosystem,but the important surface parameter in many global and regional land surfaces process,ecological and hydrological models.For Ejina Oasis,the desert riparian forest vegetation coverage can not be accurately described by remote sensing images mainly based on LandSat 30 m resolution.However,the characteristics of the high resolution imagery are the clear outlines of target objects,the abundant spatial detail information and so on,which contribute to classify the fragmental and strongly heterogeneous vegetation coverage information based on arid background.This study mainly extracts primary vegetation cover types,including farmland,Populus euphratica,Tamarix chinensis,grassland,barren\|land and so on,from QuickBird imagery in Ejina Oasis through using object\|oriented classification method.The overall accuracy and Kappa coefficient of object\|oriented classification result are 84.71% and 0.7986 respectively.This study shows,the classification results are better and able to meet the accuracy requirements through the use of object\|oriented classification method to discriminate vegetation cover information from the high resolution images.
