大豆叶片水平叶绿素含量的高光谱反射率反演模型研究

HYPERSPECTRAL REFLECTANCE MODEL TO ESTIMATE CHLOROPHYLL CONTENT IN SOYBEAN LEAVES

利用便携式光谱仪和分光光度计法分别测得大豆生长期内叶片的高光谱数据及相应的叶绿素质量分数w,利用多元统计分析和红边参数反演提取与叶片水平的w(叶绿素)相关性较高的敏感波段和光谱形式,及特征光谱位置,并以此为基础推演得到一个基于神经网络算法的叶绿素含量反演模型,结合了3种方法的优点,具有较高的反演精度.

Good correlation between hyperspectral reflectance of plant leaf and chlorophyll content would make possible to estimate vegetation chlorophyll content by hyperspectral remote sensing.Scholars in the past often generated chlorophyll models on the basis of statistical regression,red edge parameters or neural network theories.Since leaf structure and biochemical compositions vary among different species,these models are not universally applicable.Parameters and methods must be modified to suite specific plants.Soybean was taken as an example,its leaf hyperspectral reflectance and chlorophyll content during vegetative period in soybean field were measured by ASD portable spectrometer and spectrophotometer.Sensitive wave bands,proper hyperspectral forms,red edge parameters were extracted and used to generate a model based on neural network theory.This model took advantages of all existing inversing methods and proved to have high accuracy.