基于BP神经网络模型的水体叶绿素a浓度高光谱反演
首发时间:2023-02-22
摘要:本研究基于BP神经网络,为探究水库水体最佳光谱预处理方法,以福建省莆田市东圳水库为研究区,获取95组实测数据,将不经过预处理、经过Min-max标准化、SNV、一阶微分和900nm处归零化处理的光谱结果进行BP神经网络运算,对比其模型精度,发现经过900nm处归零化预处理的模型精度更高,其模型精度R2达到0.85,MAE为2.64,MAPE为30.02%。
关键词: 高光谱 叶绿素a浓度 BP神经网络 光谱预处理方法
For information in English, please click here
Hyperspectral inversion of chlorophyll a concentration in water based on BP-neural network modle
Abstract:This study is based on BP-neural network. In order to explore the best spectral preprocessing method for reservoir water body, Dongzhen Reservoir in Putian City, Fujian Province is taken as the research area, and 95 sets of measured data are obtained. The spectral results without preprocessing, Min-max standardization, SNV, first-order differential and 900 nm are processed by BP neural network. Comparing the accuracy of the model, it is found that the model accuracy is higher after the preprocessing at 900 nm.
Keywords: Hyperspectral data Chlorophyll-a concentration BP-neural network Spectral pretreatment
基金:
引用
No.****
动态公开评议
共计0人参与
勘误表
基于BP神经网络模型的水体叶绿素a浓度高光谱反演
评论
全部评论0/1000