文献类型：期刊文献

中文题名：基于缨帽变换的景洪市时间序列Landsat影像森林扰动自动识别方法研究

英文题名：Forest Disturbance Automatic Identification Method Based on Time Series Landsat Image of Tasseled Cap Transformation

作者：张连华[1,2] 庞勇[2] 岳彩荣[1] 李增元[2]

第一作者：张连华

机构：[1]西南林业大学;[2]中国林业科学研究院资源信息研究所

年份：2013

卷号：38

期号：2

起止页码：6-12

中文期刊名：林业调查规划

外文期刊名：Forest Inventory and Planning

收录：CSTPCD

基金：国家863课题"全球森林生物量和碳储量遥感估测关键技术(编号:2012AA12A306)";亚太森林恢复与可持续管理网络项目"Forest Cover and Aboveground Biomass Mapping in the Greater Mekong Subregion and Malaysia(编号:2011PA004)"资助

语种：中文

中文关键词：时间序列;Landsat影像;扰动指数;缨帽变换;森林扰动;景洪市

外文关键词：time series; Landsat image; disturbance index; tasseled cap transformation; forest disturbance ; Jinghong

分类号：S757;P283.8

摘要：森林扰动是森林生态系统演替的主要过程,对区域生态平衡与稳定起着重要作用。Landsat影像不仅具有较高的分辨率(30 m),而且具有海量的免费数据获取源,适合于做时间序列的森林干扰信息提取。作者以云南省景洪市为例,根据影像的光谱特征,首先自动获取纯净森林训练样本,然后结合缨帽变换过程中得到的土壤亮度、植被绿度以及湿度等信息,建立归一化的扰动指数图像,最后结合时间序列分析提取森林扰动信息,并对精度做了验证与评价。结果表明,该方法在不受季相影响的情况下能准确地检测出森林扰动信息,具有较高的精度。对如何减弱季相与部分农作物的影响还有待于进一步研究。
Forest disturbance is the main course of forest ecological system development, which played an important role of the regional ecological balance and stability. Landsat image not only has the high resolution of 30m, but also has mass of free data acquisition source, which is suitable for time series of forest disturbance information extraction. According to the image spectral features, and taking Jinghong city of Yunnan province as example, pure forest training samples was automatic obtained at first, then combining the information of soil brightness, vegetation greenness and humidity got from tasseled cap transformation process, the normalized disturbance index image was established, finally forest disturbance information was obtained combining with time series analysis, and the accurate was evaluated and verified. The results showed that the algorithm of training data automation could accurately detect forest disturbance information without seasonal effect and have high precision, but how to reduce the impact of seasonal and some crops still needs further studies.