基于多时相Landsat8 OLI影像的作物种植结构提取

针对基于多时相遥感影像、多种特征量提取多种作物种植结构在我国研究较少的现状,利用多时相Landsat8OLI影像数据,根据温宿县不同作物的农事历,通过分析主要地物的光谱特征和归一化植被指数的时间变化信息,构建不同作物种植结构提取的决策树模型,实现了对温宿县多种作物种植结构信息的提取。结果表明:1水稻的最佳识别依据是5月20日影像的近红外波段和7月23日影像的NDVI值;棉花和春玉米的最佳识别依据是5月20日~9月9日影像的NDVI变化值;冬小麦—夏玉米和林果的最佳识别依据是5月20日~7月23日影像的NDVI变化值;2与单时相监督分类相比,多时相决策树法对多种作物种植结构的提取效果更理想,总体精度提高了7.90%,Kappa系数提高了0.10;3Landsat8OLI影像数据分辨率高、成本低、获取方便,是农作物遥感的良好数据源。

Extraction on Crops Planting Structure based on Multi-temporal Landsat8 OLI Images

Crops planting structure and spatiotemporal change information have become the focus of the current agricultural remote sensing application.However,using the spectral characteristics and vegetation index of multi\|temporal remote sensing data to extract the crops planting structure was still few in our country.The paper took Wensu county as the study area,used Landsat8 OLI satellite data,analyzed the difference of spectral characteristics and vegetation index of the main land cover types according to the crop phenological calendar,and modeled a decision tree to identification crops,then used the model to extract crops planting structure precisely on Wensu county.The results suggested that:① The optimal choice to identify paddy rice is near infrared band on May 20 and NDVI value on July 23;NDVI change value from May 20 to September 9,which is most helpful for the identification of cotton and corn;NDVI change value from May 20 to July 23,which is the best choice for wheat and forest fruit.② the decision tree classification based on multi\|temporal Landsat8 OLI images outperformed the single\|temporal supervised classification results with the overall accuracy increased 7.90% and Kappa coefficient increased 0.10.③Landsat8 OLI satellit data has high spatial resolution and convenient access,so it is a potential data source for crops information identification,monitoring,extracting in the future.