Retrieval of Leaf Area Index of Winter Wheat at Different Growth Stages Using Continuous Wavelet Analysis

Leaf area index (LAI) is one of the most basic parameters to characterize the vegetation canopy structure, and is widely used in monitoring crop growth, yield estimation and other fields. Therefore, accurate estimation of LAI has great significance for agricultural precision fertilization and protecting agricultural ecological environment. However, few studies have attempted to estimate LAI of winter wheat using the continuous wavelet analysis (CWA), particularly at different growth stages. This paper aims at studying the spectral estimation of LAI by applying CWA into canopy spectra of 190 samples observed at Guanzhong Plain in China. Two partial least square regression (PLSR) models using six wavelet features and the optimal spectral indices were constructed and compared respectively. Results indicated that the model using wavelet features combination had a considerable improvement than the spectral indices combination for the whole validation dataset. When the validation dataset was separated according to the growth stage, the predictive performance of the wavelet features combination performed well at both growth stages, while the spectral indices combination had not achieved the same effect. The results showed that CWA approach could derive more robust wavelet features to growth stage variation, and wavelet features were more effective than the spectral indices for predicting LAI of winter wheat at different growth stages.