摘要
叶面积指数(Leaf area index,LAI)与植物的光合能力密切相关,是评价作物长势和预测产量的重要农学参数,利用高光谱遥感能够实现农作物LAI快速无损监测。为了建立不同播期条件下冬小麦LAI反演的最佳高光谱监测模型,提高冬小麦LAI估算模型精度,将地面实测冬小麦LAI数据和冠层高光谱数据相结合,对4个播期及4个播期组合模拟的混合播期数据进行分析,选取8种植被指数,通过相关分析、回归分析等统计方法,构建不同播期冬小麦叶面积指数监测模型。结果表明,在4个播期处理和由一个所有播期组合下(即混合播期)建立的LAI光谱监测模型中,播期1和播期4分别以EVI2和mNDVI拟合效果较好,播期2、播期3及混合播期均与NDGI拟合效果最好。不同播期及混合播期的拟合方程决定系数(R^2)分别为0.803,0.823,0.907,0.819和0.798;通过试验田实测LAI与反演LAI数据进行拟合模型验证,均方根误差分别为0.81,0.78,0.63,0.82,0.91。通过分析可知,不同播期的分期监测模型比混合播期统一监测模型的拟合效果更好,精度更高。因此,播期1、播期2、播期3、播期4分别选用植被指数EVI2、NDGI、NDGI、mNDVI建立冬小麦LAI反演模型。该结果可为实现不同播期下冬小麦长势精确监测提供理论依据和技术支撑。
Leaf area index(LAI)is closely related to the photosynthetic ability of plants,and its measurement helps to evaluate crop growth status and forecast yield.Hyperspectral remote sensing can be used to acquire crop LAI in real time.This research aimed to establish the best hyperspectral monitoring model for winter wheat LAI under different sowing dates and to improve the forecast precision of the LAI estimation model.The experiments combined ground measurements of winter wheat LAI data with canopy hyperspectral data from four sowing dates.Eight kinds of vegetation indices were comparatively analyzed,then LAI monitoring models for different winter wheat sowing dates were constructed using correlation and regression analyses.The results showed that in comparison to LAI,spectrum monitoring models established for four different sowing dates and from all sowing dates together,the first and fourth sowing dates were better fitted using EVI2 and mNDVI,respectively.The second and third sowing dates and all sowing dates together were best fitted using NDGI.The determination coefficients(R2)for the first,second,third,fourth and all sowing dates together were 0.803,0.823,0.907,0.819,and 0.798,respectively.The model was validated using experimentally collected LAI data and inversion LAI data.The root mean square errors for the fits of the first,second,third,fourth and all sowing dates together were 0.81,0.78,0.63,0.82,and 0.91,respectively.Our results show that monitoring models from different plant stages with different sowing dates were better than a unified monitoring model with a mix of sowing dates,and the precision was higher.Therefore,the vegetation indices EVI2,NDGI,NDGI,and mNDVI were selected separately to establish the LAI monitoring models for the first,second,third and fourth sowing dates.This result provides technical support for growth monitoring of winter wheat at different sowing dates for farmers.
作者
范剑
尤慧
刘凯文
高华东
Fan Jian;You Hui;Liu Kaiwen;Gao Huadong(Jingzhou Meteorological Bureau,Jingzhou 434020;Collaborative Innovation Center of Remote Sensing Technology in Ecological and Meteorological Monitoring in the Jianghan Plain,Jingzhou 434025;Jingzhou Agrometeorological Trial Station,Jingzhou 434025)
出处
《气象科技进展》
2018年第5期72-77,共6页
Advances in Meteorological Science and Technology
基金
荆州市气象局科技课题项目(JZ201701)
湖北省气象局科技发展基金项目(2015Q08)
关键词
冬小麦
播期
叶面积指数
高光谱遥感
估算模型
winter wheat
sowing date
leaf area index
hyperspectral remote sensing
estimation model