基于Sentinel-2和MODIS数据的宣恩县水稻遥感估产研究  

Study on Rice Yield Estimation by Remote Sensing in Xuanen County Based on Sentinel-2 and MODIS Data

作　　者：吴燕平 洪勇 姜益民 罗冷坤 杜庭晖 龙守勋 WU Yanping;HONG Yong;JIANG Yimin;LUO Lengkun;DU Tinghui;LONG Shouxun(Wuhan Optics Valley Information&Technology Co.,Ltd.,Wuhan 430079,China;State Key Laboratory of Information Engineering in Surveying,Mapping and Remote Sensing,Wuhan University,Wuhan 430079,China;Xuan′en Agriculture and Rural Bureau,Xuan′en 445500,China)

机构地区：[1]武汉光谷信息技术股份有限公司,湖北武汉430079 [2]武汉大学测绘遥感信息工程国家重点实验室,湖北武汉430079 [3]宣恩县农业农村局,湖北宣恩445500

出　　处：《测绘与空间地理信息》2022年第12期33-37,共5页Geomatics & Spatial Information Technology

基　　金：中国工程科技发展战略湖北研究院咨询项目(HB2020B12);2020年度中国(湖北)广播电视媒体融合发展创新中心专项课题研究项目(HBZSXWHCG-2020-9)资助。

摘　　要：基于作物的波谱反射特征,利用公开的多源遥感数据和相关技术能够实现农作物种植面积提取和产量预估,为作物长势监测等农业需求提供科学决策依据。本文首先基于Sentinel-2卫星影像,结合基于人工目视解译的监督分类、基于规则的面向对象分类和基于专家知识的决策树分类3种影像分类方法综合确定县级研究区的水稻种植范围,再选取水稻生育物候期内的多时相多光谱MODIS13Q1影像产品,建立影像提取出的植被指数EVI与水稻年产量之间的多元统计回归模型并应用于年产量预估,估产结果精度均达94%以上,符合实际需求。该模型可用性较强,对县域农作物遥感估产应用具有一定的指导意义。Based on the spectral reflection characteristics of crops,publicly available multi-source remote sensing data and related technologies can be used to achieve crop planting area extraction and yield estimation,then to provide scientific decision-making basis for crop growth monitoring and other agricultural needs.In this paper,the rice planting range in the research county is comprehensively determined based on Sentinel-2 satellite imagery and a combination of three image classification methods:supervised classification based on human visual interpretation,object-oriented classification based on rules,and decision tree classification based on expert knowledge.Then the vegetation index EVI is extracted from the multi-temporal and multi-spectral MODIS13Q1 image products during the phenological period of rice growth.A multiple statistical regression model for rice yield estimation between EVI and the rice annual output is established,and the accuracy of its application result in annual yield estimation has reached over 94%,which meets an actual need.The model is highly usable and of a certain guiding significance for the application of remote sensing crop yield estimation at the county level.

关 键 词：水稻 估产 遥感 植被指数 回归 

分 类 号：P237[天文地球—摄影测量与遥感]