欢迎访问《麦类作物学报》编辑部网站

作者	单位
芮婷婷，徐云飞，程琦，杨斌，冯志军，周涛，张世文	(1.安徽理工大学空间信息与测绘工程学院，安徽淮南 232001 2.安徽理工大学矿山采动灾害空天地协同监测与预警安徽普通高校重点实验室，安徽淮南 232001 3.安徽理工大学矿区环境与灾害协同监测煤炭行业工程研究中心，安徽淮南 232001；4.安徽理工大学地球与环境学院,安徽淮南 232001)

摘要点击次数: 399

全文下载次数: 312

中文摘要:

为了快速监测小麦叶片水分含量，以敏感波段组和植被指数组2种变量分别作为输入变量，以地面同步观测的冬小麦叶片含水量作为输出变量，分别采用偏最小二乘(partial least squares,PLS)、极限学习机(extreme learning machine,ELM)和粒子群算法(particle swarm optimization,PSO)优化极限学习机，建立冬小麦叶片含水量预测模型，并对其反演效果进行比较。结果表明，光谱反射率和植被指数与叶片含水量之间存在较为密切的相关性，依此确定的敏感光谱波段为红光、蓝光和近红外波段，敏感植被指数为绿度指数、过红指数、归一化绿红差值指数、三角形植被指数和过绿指数。从2种变量的建模效果看，基于植被指数组构建的模型的精度和稳定性均优于敏感波段组，其中基于植被指数组的PSO-ELM模型在6个叶片水分含量反演模型中表现最佳，其R²和RMSE分别为0.98和0.26%。利用最优模型反演得到研究区冬小麦叶片含水量的分布范围为45%~75%，平均为64.57%，反演结果与地面实测较相符，说明基于无人机光谱数据通过建立以植被指数为变量的PSO-ELM模型可实现对冬小麦叶片水分含量的精准预测。

英文摘要:

Rapid monitoring of crop leaf water content can provide data support for drought monitoring and farmland irrigation. In this study,two variables,sensitive band group and vegetation index group,were used as the input variables of the model,and the leaf water content of winter wheat observed synchronously on the ground was used as the output variable of the model. Partial least squares(PLS),extreme learning machine(ELM) and particle swarm optimization(PSO) were used. Three models of limit learning machine were optimized to establish the prediction model of winter wheat leaf water content. The results showed that there was a close correlation between spectral reflectance,vegetation index and leaf water content,and the sensitive bands were red light band,blue light band and near infrared band. The selected vegetation indices were greenness index,red index,normalized green red difference index,triangular vegetation index and green index. The modeling effects of two variables of sensitive band group and vegetation index group were compared.The accuracy and stability of the model based on vegetation index group were better than that of sensitive band group. Among the six leaf moisture content inversion models,PSO-ELM model based on vegetation index group was the best retrieval model,and its R² and RMSE were 0.98 and 0.26%,respectively. The spatial distribution of winter wheat leaf water content in the study area was retrieved using the optimal model,and the distribution range ranked from 45% to 75%,with an average of 64.57%. The retrieval results were in good agreement with the ground measurement,which showed that the accurate prediction of winter wheat leaf water content could be realized by establishing PSO-ELM with vegetation index as the variable based on UAV spectral data.

查看全文查看/发表评论下载PDF阅读器

关闭