基于高分一号遥感影像湿地变化检测

Wetland change detection based on GF-1 remote sensing image

湿地是地球生态环境的一个重要组成部分,是自然界中最富生物多样性的生态景观之一,是人类赖以生存的自然资源宝库和生存环境。本研究以高分一号卫星数据为数据源,以东洞庭湖西北部区域为研究区域,通过改进植被指数检测法,经缨帽变换处理后,选取与湿地相关密切的土壤亮度植被指数SBI、"绿度"植被指数GVI、"黄度"指数YVI,以及归一化差异植被指数和归一化差异水体指数。通光谱分析,找出提取各变化类型的阈值,并利用决策树模型进行变化信息提取,以获取湿地变化与未变化的信息。结果表明:通过改进植被指数检测法,检测精度较高,总体精度为88.8174%,Kappa系数为0.8509,制图精度高、错分精度低,得到了较好的检测效果,能够准确掌握湿地的变化与未变化情况,为湿地资源的保护与管理以及退化湿地的生态恢复等提供科学重要的决策依据。

Wetland is an important part of the earth＇s ecological environment. It is one of the most abundant biodiversity in nature, and it is a treasure house of natural resources and living environment for human. This study considers GF-1 remote sensing image as the data source, Northwest of East Dongting Lake as the study area. Through the improvement of the inspection method of the vegetation index, after tasseled cap transform, some indexes associated wetlands close are selected： soil brightness vegetation index （SBI）, ＂green＂ vegetation index （GVI）, ＂yellow＂ vegetation index （YVI）, and normalized difference vegetation index and normalized difference water index. By spectral analysis, the threshold value of each change is found, and the decision tree model is used to extract the change information in order to obtain the information of wetland change and unhange. The results show that by improving the inspection method of the vegetation index, detection accuracy is high, overall accuracy for 88.8174%, kappa coefficient for 0.8509, high drawing accuracy, fault classification accuracy for low with better detection effect, and it can accurately grasp the wetland change and no change, provide scientific basis on decision-making for the protection and management of wetland resources and the ecological restoration of degraded wetlands.