多遥感光谱指标结合进行大田冬小麦叶片叶绿素含量估测研究  

Study on Chlorophyll Content Estimation of Winter Wheat Leaf Based on Multiple Remote Sensing Spectral Indices

作　　者：韩振强 李卫国 张晓东[1] 李伟[3] 马廷淮[4] 张宏[1,2] 姚永胜 HAN Zhenqiang;LI Weiguo;ZHANG Xiaodong;LI Wei;MA Tinghuai;ZHANG Hong;YAO Yongsheng(College of Agricultural Engineering,Jiangsu University,Zhenjiang,Jiangsu 212013,China;Institute of Agricultural Information,Jiangsu Academy of Agricultural Sciences,Nanjing,Jiangsu 210014,China;Fluid Machinery Engineering Technology Research Center,Jiangsu University,Zhenjiang,Jiangsu 212013,China;Nanjing University of Information of Science and Technology,Nanjing,Jiangsu 210044,China)

机构地区：[1]江苏大学农业工程学院,江苏镇江212013 [2]江苏省农业科学院农业信息研究所,江苏南京210014 [3]江苏大学流体机械工程技术研究中心,江苏镇江212013 [4]南京信息工程大学,江苏南京210044

出　　处：《麦类作物学报》2023年第11期1467-1474,共8页Journal of Triticeae Crops

基　　金：国家重点研发计划项目(政府间重点专项)(2021YFE0104400);江苏省农业科技自主创新资金项目(CX(20)2037)。

摘　　要：为解决大田冬小麦叶片叶绿素含量估测模型精度低、通用性弱的问题,在获取冬小麦拔节期和抽穗期冠层红光波段反射率(BR_(red))和近红外波段反射率(BR_(nir))的基础上,计算归一化差值植被指数(NDVI)、差值植被指数(DVI)、比值植被指数(RVI)、土壤调节植被指数(SAVI)、改进型比值植被指数(MSR)、重归一化植被指数(RDVI)、II型增强植被指数(EVI2)和非线性植被指数(NLI)等8个植被指数。经统计分析,选择与叶片叶绿素含量(SPAD值)相关性较好的5个遥感光谱指标(NDVI、MSR、NLI、BR_(red)和RVI)作为输入变量,建立了冬小麦叶片叶绿素含量的BP神经网络估测模型(WWLCC_(BP)),并对估测模型进行精度验证。结果表明,WWLCC_(BP)估测模型在拔节期估测的决定系数(r^(2))为0.84,均方根误差(RMSE)为5.39,平均相对误差(ARE)为9.87%。抽穗期的估测效果与拔节期较为一致。将WWLCC_(BP)和高分六号影像结合监测了研究区域冬小麦叶片叶绿素含量的空间分布信息,叶片SPAD值在43.2~53.7之间的冬小麦长势正常,种植面积为25483 hm^(2),占冬小麦总播种面积的69.81%。这说明多遥感光谱指标结合建立的神经网络估测模型可以实现对大田冬小麦叶片叶绿素含量的有效估测。In order to solve the problems of low precision and low universality of the model for estimating the chlorophyll content of winter wheat leafin the field,an accurate and efficient method was proposed by combining multiple remote sensing spectral indices and neural networks.Based on the red band reflectance(BR_(red))and near infrared band reflectance(BR_(nir))of winter wheat canopy at jointing and heading stages,the normalized difference vegetation index(NDVI),differential vegetation index(DVI),ratio vegetation index(RVI),soil adjusted vegetation index(SAVI),modified simple ratio vegetation index(MSR),renormalization difference vegetation index(RDVI),enhanced vegetation index of type II(EVI2)and nonlinear vegetation index(NLI)were calculated.After statistical analysis,five remote sensing spectral indicators(NDVI,MSR,NLI,BR_(red),and RVI)well correlated with leaf chlorophyll content were selected as input variables to establish a BP neural network estimation model(WWLCC_(BP))for winter wheat leaf chlorophyll content,and the accuracy of the estimation model was verified.The results showed that the determination coefficient(r^(2)),root mean square error(RMSE),and average relative error(ARE)of WWLCC_(BP)estimation model at jointing stage were 0.84,5.39,and 9.87%,respectively.The estimation effect of heading stage was consistent with that of jointing stage.The spatial distribution information of chlorophyll content in winter wheat leaf in the study area was monitored by combining WWLCC_(BP)and GF-6 image.The winter wheat with leaf SPAD value between 43.2 and 53.7 grew normally,and the planting area was 25483 hm^(2),accounting for 69.81%of the total planting area of winter wheat.The neural network estimation model based on multiple remote sensing spectral indices can effectively estimate the chlorophyll content of winter wheat leaf in the field.

关 键 词：冬小麦 遥感光谱指标 神经网络 叶片叶绿素含量 估测模型 

分 类 号：S512.1[农业科学—作物学]