基于时间序列MODIS的农作物类型空间制图方法

发布时间：2018-09-03 20:10

【摘要】：为快速获取大范围种植结构复杂区域的作物种植面积,以MODIS数据为数据源,选择归一化植被指数(Normalized difference vegetation index,NDVI)、增强植被指数(Enhanced vegetation index,EVI)、宽动态植被指数(Wide dynamic range vegetation index,WDRVI)、地表水分指数(Land surface water index,LSWI)、归一化雪被指数(Normalized difference snow index,NDSI)5种特征,结合同步的实地调查样本点,采用支持向量机算法(Support vector machines,SVM)提取黑龙江省主要农作物的种植面积。研究表明,在待选特征中NDVI、EVI与LSWI指数组合取得了最高的分类精度,总体分类精度为74.18%,Kappa系数为0.60;支持向量机算法与最大似然算法、随机森林算法相比,分类精度更优。该方法为在大区域中提取农作物种植面积提供了参考价值。
[Abstract]:In order to quickly obtain crop planting area in a large area with complex planting structure, MODIS data is used as the data source. Five characteristics of normalized vegetation index (Normalized difference vegetation index,NDVI), enhanced vegetation index (Enhanced vegetation index,EVI), wide dynamic vegetation index (Wide dynamic range vegetation index,WDRVI), surface water index (Land surface water index,LSWI) and normalized snow cover index (Normalized difference snow index,NDSI) were selected. Support vector machine (Support vector machines,SVM) algorithm is used to extract the planting area of main crops in Heilongjiang province. The research shows that the combination of NDVI,EVI and LSWI index achieves the highest classification accuracy in the selected features, the overall classification accuracy is 74.18Kappa coefficient 0.60, and the support vector machine algorithm is better than the maximum likelihood algorithm and stochastic forest algorithm. This method provides a reference value for the extraction of crop planting area in large area.
【作者单位】：中国农业大学信息与电气工程学院;黑龙江省农垦科学院科技情报研究所;
【基金】：国家自然科学基金项目(41671418、41471342、41371326) 国家高技术研究发展计划(863计划)项目(2013AA10230103)
【分类号】：S127;TP79

【相似文献】