基于多时相遥感数据提取旱作苜蓿人工草地空间分布信息

Extracting Spatial Distribution of Rainfed Artificial Alfalfa Grassland Based on Multi-Temporal Remote Sensing Data

以我国北方农牧交错带典型地区内蒙古林西县为研究区,利用国产卫星高分一号的多时相遥感图像,构建归一化差植被指数(Normalized Difference Vegetation Index,NDVI)时间序列数据集,反映各类地物的物候特征,同时集成坡度等地形数据,在增强目标地物光谱特征的基础上,采用阈值法逐一去除干扰地物,提取旱作苜蓿人工草地的空间分布。结果表明:森林的NDVI在2017年4月明显高于苜蓿和其他地物,可运用阈值法将森林剔除;4月29日至5月28日期间,苜蓿NDVI的增加高于玉米、小麦等农田与草地,利用变化率加和法与阈值法相结合的途径剔除农田与草地;灌丛NDVI在5月12日至5月28日期间的增幅明显高于苜蓿,并且天然灌丛植被主要分布于山地的陡坡,因此结合斜率法、阈值法以及坡度阈值法将灌丛剔除。

The objective of this study was to realize the automatic extraction of spatial distribution information of the dry-cultivated artificial alfalfa grassland.We utilized multi-temporal remote sensing image of the rainfed artificial alfalfa grassland in Linxi County,Inner Mongolia,in 2017,which is a typical area in the agro-pastoral ecotone of northern China,taking by China-made satellite GF-1.We constructed the dynamic changes of Normalized Difference Vegetation Index(NDVI)time series data,in order to reflect the phenological characteristics of various types of features,and integrate the data of slope and other terrain.On the fundamental of enhancing the spectral characteristics of the target objects,We utilized the threshold method to remove the noise one by one.The results showed:the NDVI value of the forest was significantly higher than that of alfalfa and other features in April,and the threshold method was used to remove the forest part;from April 29 to May 28,the increase of NDVI value of alfalfa was higher than that of the grassland and farmlands such as corn and wheat,which were then removed by using the summation of the changes rate;from May 12 to May 28,the slope of shrub NDVI value changes was significantly higher than that of alfalfa,and the natural shrub vegetation was mainly distributed on the steep slopes of the mountainous region,therefore,the shrub was removed by combining slope method,threshold method and slope threshold method.