首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于多源Sentinel数据的县域冬小麦种植面积提取
引用本文:李长春,陈伟男,王宇,马春艳,王艺琳,李亚聪.基于多源Sentinel数据的县域冬小麦种植面积提取[J].农业机械学报,2021,52(12):207-215.
作者姓名:李长春  陈伟男  王宇  马春艳  王艺琳  李亚聪
作者单位:河南理工大学
基金项目:国家自然科学基金项目(41871333)和河南省科技攻关项目(212102110238)
摘    要:冬小麦是我国主要的粮食作物之一,及时准确地获取冬小麦种植面积对农业政策的制定具有重要意义。以河南省扶沟县为研究区域,以多生育期Sentinel-1A和Sentinel-2A/B遥感影像为数据源,构建光谱特征、植被特征和极化特征的多生育期数据集,分析各类地物的特征曲线,采用随机森林算法对单生育期单传感器、单生育期多传感器、多生育期单传感器和多生育期多传感器的遥感影像进行精细分类,实现县域冬小麦制图。结果显示:单生育期的雷达影像无法满足制图要求,拔节期的总体精度最高,仅为62.9%,多生育期雷达影像分类精度达到81.9%,基本满足制图要求;单生育期的光学影像和融合影像在成熟期的精度最高,总体精度分别为93.4%和95.1%,Kappa系数分别为92.4%和94.8%,可以绘制较为精准的冬小麦分布图;多生育期融合影像绘制的扶沟县2019年冬小麦空间分布图,总体精度为96.8%,结果最优。研究结果表明融合的多生育期遥感影像可以为县域冬小麦种植面积的提取提供技术依据。

关 键 词:冬小麦  种植面积  Sentinel数据  生育期  融合影像  随机森林
收稿时间:2021/7/27 0:00:00

Extraction of Winter Wheat Planting Area in County Based on Multi-sensor Sentinel Data
LI Changchun,CHEN Weinan,WANG Yu,MA Chunyan,WANG Yilin,LI Yacong.Extraction of Winter Wheat Planting Area in County Based on Multi-sensor Sentinel Data[J].Transactions of the Chinese Society of Agricultural Machinery,2021,52(12):207-215.
Authors:LI Changchun  CHEN Weinan  WANG Yu  MA Chunyan  WANG Yilin  LI Yacong
Institution:Henan Polytechnic University
Abstract:Winter wheat is one of the main food crops in China. Timely and accurate localization of the planting areas of this crop is highly crucial for making adequate agricultural policies. The feasibility of a remote-sensing system for mapping winter wheat in the Fugou County was explored, combining Sentinel-1A and Sentinel-2A/B remote-sensing images. Firstly, remote-sensing images were collected to reflect the different phenological patterns of winter wheat. In particular, these images were sampled across five typical growth stages, namely, jointing, heading, flowering, milk mature, and mature stages. Then, spectral, vegetation, and polarization features were extracted from the collected images, and the characteristic curves of various ground objects were analyzed. Last, random forest classifiers were trained to accurately classify the remote-sensing images associated with four possible winter wheat models: a single-growth-stage single-sensor model, a single-growth-stage multi-sensor model, a multi-growth stage single-sensor model, and a multi-growth-stage multi-sensor model. The results showed that the single-growth-stage models cannot meet the crop mapping requirements, where the highest attained accuracy reached only 62.9% for the jointing stage. Additionally, these requirements were met by the multi-growth-stage models whose highest classification accuracy reached 81.9%. The optical and fusion images associated with the single-growth-stage models achieved the highest accuracy for the mature stage, with overall accuracies of 93.4% and 95.1%, and Kappa coefficients of 92.4% and 94.8%, respectively. These results could lead to more accurate mapping of the spatial distribution of the winter wheat crop. Also, the spatial distribution map of winter wheat in Fugou County in 2019 drawn by multi-growth-stage model has the overall accuracy of 96.8%, and the result is the best. Thus the proposed multi-growth-stage fusion model can be effectively employed in the localization and mapping of winter wheat planting areas.
Keywords:winter wheat  planting area  Sentinel data  growth stages  fusion image  random forest
点击此处可从《农业机械学报》浏览原始摘要信息
点击此处可从《农业机械学报》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号