Monitoring Leaf Area Index after Heading Stage Using Hyperspectral Remote Sensing Data in Rice.

Due to the influence of the rice panicles on the spectral reflectance of rice canopy, the sensitivity of the traditional vegetation indices to rice leaf area index (LAI) was greatly reduced after the heading stage. The main goal of this work was to improve the estimate accuracy of leaf area index by eliminating the effects of panicles based on field measurements of rice under different treatments in late stages of rice growth. The results demonstrated that the removal of rice panicles had evident effects on the canopy spectral reflectance and there were differences in the types and degrees of these effects at different wavelengths, especially in 550—690 nm and 750—1000 nm. In most case, the vegetation indices (i.e., DI, SAVI, GNDVI, TVI, RDVI, MTVI2) were more sensitive to LAI by using the canopy spectral reflectance with panicles being removed. Moreover, to eliminate the effect of rice panicles on the canopy reflectance, we constructed a simple and effective linear correction model to modify the entire canopy reflectance by using the dry weight ratio of panicles and leaves. The results showed that, by using the modified canopy reflectance, the determination coefficients (R²) of vegetation indices (DI, SAVI, GNDVI, TVI, RDVI, MTVI2) and leaf area index increased by 20.46%, 16.08%, 11.85%, 28.20%, 13.85% and 8.82%, respectively. The above findings indicated that the modified canopy spectral reflectance could eliminate the influence of rice panicles and it could be used to improve the estimate accuracy of leaf area index in the late stages of rice growth.