基于Sentinel-2影像的冬小麦种植面积早期提取研究  

Early extraction of winter wheat planting area based on Sentinel-2 images

作　　者：牛鲁燕[1] 封文杰[1] 侯学会 Niu Luyan;Feng Wenjie;Hou Xuehui(Institute of Agricultural Information and Economics,Shandong Academy of Agriculture Sciences,Jinan 250100,Shandong,China)

机构地区：[1]山东省农业科学院农业信息与经济研究所,济南250100

出　　处：《中国农业信息》2024年第1期1-11,共11页China Agricultural Informatics

基　　金：国家重点研发计划项目“农情高频立体与集约共享式精细服务应用示范”(2021YFB3901303)。

摘　　要：【目的】文章旨在探讨基于高分辨率多光谱遥感影像进行冬小麦种植面积早期快速提取、冬小麦空间分布情况快速制图与精度验证的方法,为山东省冬小麦高产、优质种植和农艺肥水的处方决策提供全局性信息。【方法】(1)对Sentinel-2遥感影像数据进行预处理,然后采用历史种植分布数据自动提取与人工选取相结合方式构建冬小麦识别样本库,将样本分为小麦、林地、水体、建筑和道路及其他作物五大类;(2)采用随机森林算法计算机自动分类与影像人工解译相结合的方式,提取研究区冬小麦种植面积,绘制冬小麦种植空间分布图,并进行精度验证。【结果】(1)解译得到研究区冬小麦种植面积为54.41万hm2,冬小麦种植面积的总体分布精度为97.05%,kappa系数为0.94,解译效果良好;(2)该文提出的方法可实现冬小麦种植面积高精度提取以及快速制图。【结论】早期精准掌握冬小麦种植面积及空间分布信息,能够为地方政府和农业部门指导农事活动提供科学依据。[Objective]This study aims to explore a method for early and rapid extraction of winter wheat planting area,rapid mapping and accuracy verification of the winter wheat spatial distribution based on high-resolution multispectral remote sensing images,providing information for high yield and good quality planting and fertilizer and water management of winter wheat in Shandong Province.[Method](1)Sentinel-2 remote sensing images were preprocessed firstly,and then the winter wheat identification sample database was constructed by combining the automatic extraction of historical planting distribution data and manual selection.The samples were divided into five categories including wheat,woodland,water body,buildings,roads and other crops.(2)The winter wheat planting area was extracted by random forest machine learning classification combined with manual interpretation of constitutional images based on the preprocessed Sentinel-2 remote sensing images,so as to realize rapid extraction of winter wheat planting area at an early time and verify the accuracy.[Result](1)The results showed that the winter wheat planting area in the study area was extracted as 544,100 hm2 with the overall distribution accuracy of 97.05%and the kappa coefficient of 0.9410.The extraction effect of the established method was good.(2)The method proposed in this study could achieve high-precision extraction and rapid mapping of winter wheat planting area.[Conclusion]Accurately grasping the planting area and spatial distribution information of winter wheat in the early stage can provide a scientific basis for local governments and agricultural departments to arrange and guide agricultural activities.

关 键 词：Sentinel-2 冬小麦 种植面积 快速提取 

分 类 号：S512.11[农业科学—作物学] S127