New spectral model for estimating leaf area index based on gene expression programming

Abstract

The accurate evaluation of leaf area index (LAI) in paddy rice is of great importance in predicting the production and assessing food security. Traditional LAI estimation based on the PROSAIL model requires a large number of parameters. To solve this problem, this paper proposes a new estimation model between LAI and spectral reflectance for paddy rice canopies based on gene expression programming. And, we suggest a new spectral index for maximizing the response to LAI and weakening the sensitivity to chlorophyll concentration in the proposed LAI estimation model. Experimental results show that compared with existing indices, the proposed index (VI-R2) maximized the response to green LAI and was less sensitive to chlorophyll content variations, and in the validation with measured LAI values, VI-R2 shows the excellent fitting performance with coefficient of determination (r2) of 0.954. Meanwhile, based on a comparison of the estimated LAI values among existing indices using hyperspectral image, the results indirectly demonstrate that VI-R2 is usable for estimating LAI at the remote sensing pixel scale.