Inversion and validation of leaf area index based on the spectral&knowledge database using MODIS data

LAI (Leaf area index) is an important parameter for describing vegetation canopy structure in the terrestrial ecosystem, and is closely related to plant evaportranspiration, sunlight intercept, photosynthesis and net primary productivity. It is feasible to retrieve LAI of large area from remote sensing data with physical models; however, it is quite difficult to get accurate LAI without enough prior knowledge due to the underdetermined parameters in the physical inversion models. A spectrum database system of typical objects in China (SpecLib) has been set up recently, which may provide a priori knowledge of typical land cover for LAI inversion. MODIS data is used to retrieve LAI after atmosphere correction, geometrical correction and cloud identification. The SAIL (Scattering by Arbitrarily Inclined Layers) model is applied for the inversion of LAI for MODIS data. The vegetation coverage of the mixed pixels of the MODIS data are calculated based on the TM data sets. The LAIs of pure pixels (computed from the retrieved LAIs and vegetation coverage) are compared with the field measurement data in Luancheng, Heibei Province, China. Meanwhile, the LAIs of pure pixels are also compared with the MODIS LAI data products. The inversion results show that the SpecLib effectively improved the accuracy of leaf area index inversion.