融合高斯白噪声的PROSPECT模型的柑橘叶片叶绿素反演  

Chlorophyll Inversion of Citrus Leaves in PROSPECT Model with Gaussian White Noise

作　　者：窦世卿 张晨红 靖娟利 张文杰 侯以昌 严继池 DOU Shiqing;ZHANG Chenhong;JING Juanli;ZHANG Wenjie;HOU Yichang;YAN Jichi(College of Geomatics and Geoinformation,Guilin University of Technology,Guilin 541006,China;Ecological Spatiotemporal Big Data Perception Service Laboratory of Guangxi Zhuang Autonomous Region,Guilin 541006,China;College of Mechanical and Control Engineering,Guilin University of Technology,Guilin 541006,China)

机构地区：[1]桂林理工大学测绘地理信息学院,广西桂林541006 [2]广西壮族自治区生态时空大数据感知服务重点实验室,广西桂林541006 [3]桂林理工大学机械与控制工程学院,广西桂林541006

出　　处：《无线电工程》2024年第7期1702-1711,共10页Radio Engineering

基　　金：国家自然科学基金(42161028);广西八桂学者专项项目(DT2100001072);桂林市科技局项目(2020010701,20210226-2)。

摘　　要：叶绿素是反映植被光合能力的重要指标。为了建立具有普适性的柑橘叶片叶绿素的反演模型,利用PROSPECT模型建立模拟光谱后采用3种特征选取的方法建模,分别为基于敏感性波段(Sensitivity Band,SA)的特征选取,记为PRO-SA模型;采用竞争自适应加权算法(Competitive Adaptive Weighting Algorithm,CARS)的特征选取,记为PRO-CARS模型;采用连续投影算法(Successive Projection Algorithm,SPA)的特征选取,记为PRO-SPA模型。结果表明,①PRO-SPA模型为最优反演模型,其决定系数R^(2)相较原始光谱提升了0.07;②在最优PRO-SPA模型中重采样并对模拟光谱进行多元散射矫正(Multiple Scattering Correction,MSC)后加入高斯白噪声,加入信噪比为80 dB的高斯白噪声的模拟光谱在PRO-SPA模型中表现出最优的反演精度,其决定系数R 2为0.6903,均方根误差(Root Mean Square Error,RMSE)为12.98,相比较原始光谱反演提升了0.15,表明了该方法对于柑橘叶片的光谱模拟以及叶绿素反演的可行性。Chlorophyll is an important indicator of the photosynthetic capacity of vegetation.In order to establish a universal inversion model of citrus leaf chlorophyll,the PROSPECT model is used to establish simulation spectrum and three feature selection methods is used to model.The three feature selection methods are the feature selection based on Sensitivity Band(SA),denoted as PRO-SA model,the feature selection using the Competitive Adaptive Weighting Algorithm(CARS),denoted as PRO-CARS model,and the feature selection using Successive Projection Algorithm(SPA),denoted as PRO-SPA model.The results show that:①The PRO-SPA model is the optimal inversion model and its determination coefficient R^(2)is increased by 0.07 compared with the original spectrum;②In the optimal PRO-SPA model,Gaussian white noise is added after resampling and Multiple Scattering Correction(MSC)of the analog spectrum,the simulated spectrum with 80 dB white Gaussian noise ratio has showed the best inversion accuracy in the PRO-SPA model,with the determination coefficient R 2 of 0.6903 and Root Mean Square Error(RMSE)of 12.98.Compared with the original spectral inversion,the increase of 0.15 indicates the feasibility of this method for spectral simulation and chlorophyll inversion of citrus leaves.

关 键 词：叶绿素 PRO-SPA模型 全局敏感性分析 高斯白噪声 

分 类 号：S127[农业科学—农业基础科学]