基于主成分分析的叶面积指数尺度效应

为描述多空间尺度观测数据在表达同一区域农作物叶面积指数(LAI)分布特征时存在的差异,该文提出了一种基于主成分分析(PCA)的LAI尺度效应分析方法。该方法充分考虑了多尺度数据的相关性与差异性,从统计分析角度出发,采用PCA进行数据挖掘和信息重组,引入动态多元线性回归模式基于主成分信息(PCs)反演估算LAI,进而定量描述尺度效应。选取大麦和玉米为试验对象,先以地面最细空间尺度观测数据为基准,通过尺度上推构建一系列不同空间尺度数据;再依据上述尺度效应分析方法进行有效信息提取和LAI估算,并纳入有效主成分个数(NEPCs)、决定系数(R2)和平均相对精度(MRA)等参数定量描述尺度效应。理论分析和数值实践证实了该方法在农作物LAI尺度效应定量分析中的可行性和有效性。

Analyzing scale effects of crop LAI based on PCA method

In order to quantitatively analyze the scale effects of leaf area index (LAI) of crop canopies, an analyzing method based on principal component analysis (PCA) theory was proposed in this paper. In this method, PCA theory was introduced for data mining and reorganization, which fully considered the correlation and variability of multi-resolution data. Dynamical multiple linear regression theory was selected for LAI estimation by taking principal components (PCs) as independent variables. Barley and corn were chosen as experimental objects. Firstly, observed data at different spatial scales were constructed by polymerization method based on the small scale observed data. Secondly, the scale effects analyzing method proposed in this work was used for data processing and LAI estimation. Finally, the number of effective principle components (NEPCs), coefficient of determination (R2), and mean relative accuracy (MRA) were selected as testing indicators to analyze the above results to quantitatively describe scale effects of crop LAI. The theory analyses and numerical practices verified the feasibility and validity of this proposed method in analyzing scale effects of crop LAI.