Accurate and non-destructive estimation on the growth status of winter wheat is of significance.The consecutive two-years experiments of nitrogen application in 2011—2012 and 2012—2013 were performed to obtain the canopy spectra and the six growth status indicators of winter wheat(Leaf area index,LAI;Above ground dry biomass,AGDB;Above ground fresh bio-mass,AGFB;Plant water content,PWC;Chlorophyll density,CH.D;Accumulated nitrogen content,ANC).The principle component analysis(PCA)was implemented to construct the comprehensive growth indicator(CGI),which could potentially re-present the growth status of winter wheat.Furthermore,the method of partial least square(PLSR)was applied on constructing the hyperspectral prediction models of all growth indicators and validating the accuracy of CGI.The results showed that the con-structed CGI significantly correlated with all the growth status indicators of winter wheat,excepting for the PWC.It indicated that the CGI could represent most of the information for the six indicators and the CGI also could be used to stand for the grow th status of winter wheat.Moreover,the model performance of CGI and other six indicators were further compared,and it showed that the PLSR model of CGI performed best than other six indicators with R2=0.802,RMSE=1.268,RPD=2.015.The CGI model was validated and proved to be more accurate and robust(R2= 0.672,RMSE= 1.732 and RPD= 1.489).The study showed that the CGI constructed with the PCA method could represent the growth status of winter wheat and the CGI model based on the PLSR method could be used to estimate the growth status of winter wheat.It also indicated that the multivariate statistical analysis had great potential to be applied in the field of crops by using the hyperspectral technology.%利用高光谱分析技术实现冬小麦长势的准确、无损监测具有重要的实践意义.基于连续两年的氮素运筹试验,通过获取叶面积指数(LAI)、地上干生物量(AGDB)、地上鲜生物量(AGFB)、植株含水量(PWC)、叶绿素密度(CH.D)和氮素积累量(ANC)六个冬小麦长势指标及冬小麦冠层高光谱,引入主成分分析法(PCA)构建可表征冬小麦长势的综合长势指标(CGI),并结合偏最小二乘回归法(PLSR)构建CGI的高光谱估测模型.结果表明,除植株含水量外,其他长势指标与所构建的CGI都达到极显著水平,表明利用CGI可以表征各长势指标信息.对比CGI和其他各长势指标的PLSR模型表现可知,CGI光谱监测模型表现最优(R2=0.802,RMSE=1.268,RPD=2.015),也具有较高的预测精度和稳健度(R2=0.672,RMSE=1.732,RPD=1.489).表明基于PCA方法所构建的CGI可以表征冬小麦长势,利用PLSR方法可以实现对冬小麦长势的准确监测,且监测效果要优于单一的冬小麦长势指标.
展开▼
