Chlorophyll Inversion of Citrus Leaves in PROSPECT Model with Gaussian White Noise
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:① The PRO-SPA model is the optimal inversion model and its determination coefficient R2 is increased by 0.07 compared with the original spectrum;② In 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 R2 of 0.690 3 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.
chlorophyllPRO-SPA modelglobal sensitivity analysisGaussian white noise
国家科技期刊平台
NSTL
万方数据
