Inversion of Winter Wheat Leaf Area Index Based on Canopy Reflectance Model and HJ CCD Image

Leaf area index is an important vegetation canopy structure parameter for vegetation monitoring, climate change, ecological process and data assimilation system. The method of LAI inversion is one of the focuses of quantitative remote sensing research. In this study, the global optimal algorithm SCE-UA (Shuffled Complex Evolution method developed at the University of Arizona) was integrated into canopy reflectance model ACRM (A Two-Layer Canopy Reflectance Model) to improve the process of inversion LAI. A global sensitivity analysis method EFAST (Extended Fourier Amplitude Sensitivity Test) was used firstly to analyze sensitivity of parameters in ACRM. The most sensitive parameters were optimized iteratively until difference between simulated and measured canopy reflectance was minimized in order to obtain optimal LAI. Accuracy verification and feasibility analysis were carried out with simultaneous observation data on field sites and regional scale. The results show that LAI, chlorophyll content Cab, leaf structure parameter N, Markov (clumping) parameter Sz, relative leaf size SL and soil reflectance parameter rsl1 are optimized variables which are most sensitive for the four bands of HJ CCD image. The acceptable and desired inversion of winter wheat LAI was performed successfully. On field sites, coefficient of determination R-2 is 0.88, Normalized Root Mean Square Error (NRMSE) is 9.95%, and the Relative Error (RE) is 8.45%. On regional scales, R-2 for LAI is 0.66, NRMSE is 26.13%, and RE is 20.62%. The proposed inversion approach of vegetation canopy LAI based on ACRM is feasible and effective, and will be most potential method of application.