论文部分内容阅读
利用辽宁锦州地区2013年生长季不同土壤水分控制条件下的春玉米冠层高光谱数据,及对应的植株叶面积指数(leaf area index,LAI)数据,分析在不同发育期内不同生长状况下的春玉米冠层高光谱特征及其与植株叶面积指数的关系。采集并计算共313组有效样本,包括350~2 500nm波段范围光谱的反射率、反射率倒数的对数、反射率一阶导数及LAI,应用多元逐步线性回归法和偏最小二乘回归法,对剔除了受大气水分影响较为严重光谱波段的其他波段数据进行降维,构建叶面积指数的全波段冠层高光谱数据模型,并进行精度检验与比较。结果表明,春玉米LAI与光谱反射率在可见光波段(350~680nm)、红外波段(1 430~1 800和1 950~2 450nm)均呈显著的负相关;反射率倒数的对数在对应区间为显著的正相关;反射率一阶导数则在可见光和近红外波段(350~1 350nm)存在较显著相关波段。三种全波段冠层高光谱数据在春玉米LAI的线性回归中,偏最小二乘法在以冠层反射率为自变量的模型构建中,比多元逐步线性回归拟合度好,其总均方根误差为0.480 7;以冠层光谱反射率的倒数的对数及一阶导数为自变量,应用逐步线性回归法建模,拟合度较好,其总均方根误差分别为0.333 5和0.348 8;三种光谱数据的春玉米LAI两种回归算法中,以冠层反射率倒数的对数为自变量,应用逐步线性回归方法建模的拟合度最佳。
Based on hyperspectral data of spring maize and soil leaf area index (LAI) data under different soil moisture control conditions in growing season in Jinzhou, Liaoning Province in 2013, the effects of different growth stages Hyperspectral Characteristics of Spring Maize Canopy and Their Correlations with Plant Leaf Area Index. A total of 313 valid samples were collected and calculated, including spectral reflectance, logarithm of reflectivity, first derivative of reflectivity and LAI in the range of 350-250nm, multivariate stepwise linear regression and partial least-squares regression, The other band data excluding the more severe spectral bands affected by atmospheric moisture were dimensionally reduced to construct a full-canopy hyperspectral hyperspectral data model with leaf area index, and the accuracy was tested and compared. The results showed that the LAI of spring maize showed a significant negative correlation with spectral reflectance in the visible (350-680 nm) and infrared (-1 430-1 800 and 1 950-2 450 nm) bands. The logarithm of the reciprocal of reflectance was in the corresponding range And the first derivative of reflectivity has a significant correlation band in the visible and near infrared (350 ~ 1 350 nm) bands. In the linear regression of LAI of spring maize, PLS data showed that the PLS was better than multivariate stepwise linear regression in constructing model with canopy reflectance as an independent variable. The total mean square The root-mean-square error was 0.480 7. The logarithm of the reciprocal of the canopy spectral reflectance and the first derivative were independent variables and were modeled by the stepwise linear regression method. The fitting error was good, with a total root mean square error of 0.333 5 and 0.348 8. For the two kinds of LAI regression algorithms of spring maize with three kinds of spectral data, the logarithm of reciprocal canopy reflectance was used as the independent variable, and the best fit was obtained by using the stepwise linear regression method.