地球信息科学学报 ›› 2020, Vol. 22 ›› Issue (4): 731-742.doi: 10.12082/dqxxkx.2020.190726

• • 上一篇    下一篇

时空协同的精准农业遥感研究

吴志峰1, 骆剑承2,3,*(), 孙营伟2,3, 吴田军4, 曹峥1, 刘巍2,3, 杨颖频2,3, 王玲玉5   

  1. 1. 广州大学地理科学学院,广州 510006
    2. 中国科学院空天信息创新研究院,北京 100101
    3. 中国科学院大学,北京 100049
    4. 长安大学地质工程与测绘学院,西安 710064
    5. 贵州师范大学喀斯特研究院,贵阳 550001
  • 收稿日期:2019-11-29 修回日期:2019-12-26 出版日期:2020-04-25 发布日期:2020-06-10
  • 通讯作者: 骆剑承 E-mail:luojc@radi.ac.cn
  • 作者简介:吴志峰(1969— ),男,湖南湘潭人,博士,教授,博士生导师,研究方向为城市遥感与陆地生态遥感、复杂地表格局—过程—效应。E-mail:zfwu@gzhu.edu.cn
  • 基金资助:
    国家自然科学基金项目(41671430);国家自然科学基金项目(41631179);国家自然科学基金青年科学基金项目(41701472);NSFC-广东联合基金重点项目(U1901219)

Research on Precision Agricultural based on the Spatial-temporal Remote Sensing Collaboration

WU Zhifeng1, LUO Jiancheng2,3,*(), SUN Yingwei2,3, WU Tianjun4, CAO Zheng1, LIU Wei2,3, YANG Yingpin2,3, WANG Lingyu5   

  1. 1. School of Geographical Sciences, Guangzhou University, Guangzhou 510006, China
    2. Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, China
    3. University of Chinese Academy of Sciences, Beijing 100049, China
    4. School of Geology Engineering and Geomatics, Chang'an University, Xi'an 710064, China
    5. Institute of Karst Science, Guizhou Normal University, Guiyang 550001, China
  • Received:2019-11-29 Revised:2019-12-26 Online:2020-04-25 Published:2020-06-10
  • Contact: LUO Jiancheng E-mail:luojc@radi.ac.cn
  • Supported by:
    National Natural Science Foundation of China(41671430);National Natural Science Foundation of China(41631179);National Natural Science Foundation of China Youth Science Foundation Project(41701472);NSFC-Guangdong Joint Foundation Key Project(U1901219)

摘要:

高分辨率遥感对地观测为我们从空间与时间2个维度客观反演地表格局—过程提供了有效的技术支撑。本文遵循时空协同的研究思路,基于高分辨率遥感影像,开展了农业遥感领域2个典型的问题研究:① 提出了一种基于影像视觉特征的耕地分区分层提取方法,该方法在利用DEM数据进行分区的基础上,根据不同区域内耕地所呈现的几何特征和纹理特征差异,分别设计了不同的耕地提取模型;② 构建了一种地块尺度的作物生长参数反演方法,方法以地块为基本单元,在空间、时间及属性组合约束下进行作物理化参数反演。本研究以贵州省安顺市西秀区和广西扶绥县耕地提取进行了耕地地块提取示范,以扶绥县进行了基于耕地地块和中空间分辨率时间序列遥感数据的甘蔗叶面积指数反演。其中,对于安顺市西秀区的耕地地块提取结果而言,形态精度(IoU)大于0.7的地块超过60%,规则耕地、梯田以及林草地等的类型精度均超过了80%;对于扶绥县甘蔗叶面积指数反演的结果而言,其结果可以较为精确地反映出基地甘蔗与非基地甘蔗的差异,基地甘蔗在品质上要优于非基地甘蔗。西南山地区的耕地形态提取/类型判别和地块甘蔗叶面积指数应用验证均证明了方法的可行性。结果表明,协同使用多源高分辨率数据是实现精准农业遥感研究的有效途径。

关键词: 精准农业, 高分遥感, 遥感图谱认知, 机器学习, 分区分层, 时空协同, 地块, 参数反演

Abstract:

High-resolution remote-sensing earth observation provide us with effective technical support for objectively inverting the surface patterns-process from the dimensions of space and time. This paper follows the research idea of space-temporal collaboration, and based on the high-resolution remote sensing images, we explored two typical problems in the agricultural remote sensing field: (1) proposed a division control and stratified extraction method for geo-parcel based on visual characteristics of images. Based on the division of DEM, we have designed different geo-parcel extraction models based on the differences in geometric and texture features in the division regions; (2) proposed a method for crop growth parameters inversing at the geo-parcel scale. Geo-parcel is the basic unit to perform physical parameter inversion under the constraints of space-time-attribute combination. The study taking geo-parcels extraction in in Xixiu District of Anshun City in Guizhou Province and Fusui County in Guangxi as examples for the division control and stratified extraction method, and taking inversion of sugarcane leaf area index in Fusui County of Guangxi Province as examples for the method of crop growth parameters inversing at the geo-parcel scale. For the extraction of cultivated land in Xixiu District, The number of geo-parcels with morphological accuracy (IoU) greater than 0.7 accounts for more than 60%, and the accuracy of the types of regular geo-parcels, terraces, forests and grasslands exceeded 80%; also, for the inversion results of sugarcane leaf area index in Fusui County, the results can accurately reflect the difference between base sugarcane and non-base sugarcane, and the base sugarcane is superior in quality to non-base sugarcane. It shows that Spatial-temporal collaboration use of multi-source high-resolution data is an effective way to achieve accurate agricultural remote sensing research.

Key words: precision agriculture, high-resolution remote sensing, spatial-spectral cognition of remote sensing, machine learning, division and stratify, spatial-temporal collaboration, geo-parcel, parameter inversion