GF-1卫星影像的空间分辨率对水稻识别精度的影响  

Effect on rice identification accuracy based on different spatial resolution images of GF-1

作　　者：邱琳[1] 卢必慧[1] 孙玲[1] 单捷[1] 毛良君 QIU Lin;LU Bi-hui;SUN Ling;SHAN Jie;MAO Liang-jun(Institute of Agricultural Information, Jiangsu Academy of Agricultural Sciences, Nanjing 210014, China)

机构地区：[1]江苏省农业科学院农业信息研究所,江苏南京210014

出　　处：《江苏农业学报》2019年第1期70-75,共6页Jiangsu Journal of Agricultural Sciences

基　　金：国家科技重大专项课题(09-Y30B03-9001-13/15-006);江苏省农业科技自主创新基金项目[CX(17)3020];江苏省农业科学院基金项目(6111651;6111650)

摘　　要：基于GF-1卫星影像数据,选择建湖县3块5 km×5 km水稻种植区为试验区,分析了水稻识别精度与GF-1卫星3种分辨率之间的关系。结果表明,随着空间分辨率由2. 0 m变化到16. 0 m,混合像元内水稻所占比例从0. 92下降到0. 73,水稻图斑平均面积从1. 66 hm2上升到13. 28 hm2,水稻识别的总体精度由92. 41%降低到89. 20%;水稻识别精度与高、中、低3个农田景观破碎度密切相关,破碎度增加导致水稻识别精度下降;同等分辨率下,破碎度越高水稻识别精度越低。In this study,three test areas of 5 km×5 km were chosen in rice planting area of Jianhu county for studying the relationship between accuracy of rice identification and GF-1 images with different spatial resolutions.The results showed that:with the decreasing of spatial resolution of GF-1 images from 2.0 m to 16.0 m,rice component proportion of mixed pixel fell off from 0.92 to 0.73,the average area of all rice polygons derived from GF-1 images increased from 1.66 hm^2 to 13.28 hm^2,and the accuracy of rice identification decreased from 92.41%to 89.20%.The accuracy of rice identification with GF-1 images is closely related to the landscape fragmentation of farmland,and the same trends occur at all levels of spatial resolution images,that the increasing fragmentation leads to the decreasing accuracy of rice identification.

关 键 词：水稻 GF-1卫星 空间分辩率 破碎度 面积监测 

分 类 号：S127[农业科学—农业基础科学]