摘要：Continuous wavelet analysis (CWA) has been applied to leaf-scale spectral data for quantifying leaf chlorophyll content, but its application to canopy-scale spectral data for estimating the canopy chlorophyll density (CCD) of winter wheat at different growth stages requires further analysis. This study aims to estimate CCD by applying CWA to the canopy spectra of 185 samples from Guanzhong Plain, China. The five most informative wavelet features related to CCD were identified using the CWA method. Meanwhile, 10 commonly used spectral indices were selected to compare with the CWA method. Two partial least square regression (PLSR) models based on wavelet features and spectral indices were developed and compared. Results showed that the PLSR model using wavelet features (R 2 = 0.64, RMSE = 0.43 g/m 2 ) was better than that using spectral indices (R 2 = 0.57, RMSE = 0.48 g/m 2 ) and wavelet features were less sensitive to the growth stage variation than spectral indices. This result suggested that the CWA approach can derive robust wavelet features and was more effective than spectral indices for predicting CCD from canopy-scale spectral data for an agricultural ecosystem.
