论文部分内容阅读
土壤盐碱化是制约干旱区植被生长最主要的生态环境地质问题,也是土地资源利用的主要障碍因子之一。利用遥感数据准确自动的提取盐碱化土壤信息对土壤的盐碱化的提取和动态监测具有重要意义。本文以新疆石河子150团为研究区域,使用2010年8月和2005年7月的ETM数据提取归一化植被指数(NDVI)、第一主成分(PC1),分别作为非盐碱化土壤和盐碱化土壤信息提取的主要特征变量,然后用改进归一化差异水体指数(MNDWI)、TM7、Mean作为辅助特征变量,结合研究区的实地考察和典型土样的遥感信息以及与缨帽变化(K-T)对比分析来确定各个特征变量的阈值。利用决策树对研究区遥感图像进行了分类。研究结果显示,基于决策树的干旱区盐碱化土壤信息提取方法是可行的并能达到较高的精度。
Soil salinization is the most important eco-environmental geological problem restricting the growth of vegetation in arid areas, and is also one of the main obstacles to the utilization of land resources. The accurate and automatic extraction of salinized soil information by using remote sensing data is of great significance to the extraction and dynamic monitoring of soil salinization. In this paper, the 150th group of Shihezi in Xinjiang was selected as the study area. The normalized vegetation index (NDVI) and the first principal component (PC1) were extracted using the ETM data of August 2010 and July 2005, respectively, as non-salinized soil and salt The main characteristic variables of soil alkalization were alkalized, and then MNDWI, TM7 and Mean were used as auxiliary characteristic variables, combined with field investigation and typical remote sensing information of the study area, KT) comparative analysis to determine the threshold of each characteristic variable. The decision tree is used to classify remote sensing images in the study area. The results show that the method of soil salinization based on decision tree is feasible and can achieve high precision.