Retrieval Of Leaf Area Index And Foliage Clumping Index From Remotely Sensed Data

Leaf area index (LAI) is one of the most important structural parameters controlling the vegetation biophysical processes, and is therefore needed in most land process modeling. As an additional biophysical parameter of comparable importance to LAI, the clumping index quantifies the deviation of the leaf spatial distribution from the random case. It is useful for converting the effective LAI to the true LAI and for separating the canopy into sunlit and shaded leaves in photosynthesis and evapotranspiration modeling. However, due to the lack of high resolution multi-angular remote sensing data, there are only few clumping index maps available presently. This thesis is focused on a new inversion algorithm of the 500 m resolution clumping index and true LAI from MODIS (Moderate Resolution Imaging Spectroradiometer) data. The 500 m resolution clumping index and true LAI over China are first presented in this study, which can be supplied to users for various land surface studies.The 500 m resolution MODIS BRDF (Bidirectional Reflectance Distribution Function) model parameters product (MCD43A1) makes use of multi-date observation data and a semi-empirical kernel-driven BRDF model (Ross-Li model) to determine a global set of BRDF parameters describing the anisotropic behavior of surface scattering, which provides a new way to retrieve the clumping index of vegetated surfaces over large areas. The ability of the Ross-Li model to simulate the bidirectional reflectance is assessed by comparing its simulated reflectance based on the MCD43A1 data with observed POLDER-3 (Polarization and Directionality of Earth Reflectance) BRDF data. It is found that the Ross-Li model simulates well the bidirectional reflectance of ground objects at most observation angles, but significantly underestimates the hotspot reflectance. A modified Ross-Li model is developed to simulate the reflectance at hotspot and darkspot, which is used to calculate the Normalized Difference between Hotspot and Darkspot (NDHD). With the relationship between clumping index and NDHD simulated by the 4-Scale geometrical model, the clumping index over China's landmass at 500 m resolution is retrieved every 8 days during the period from 2003 to 2008. The effect of topography on the retrieved clumping index is corrected using a topographic compensation function calculated from the digital elevation model (DEM) at 90 m resolution. The topographically corrected clumping index values correlate well with field measurements in 5 areas over China (R2=0.61, RMSE=0.08), indicating the feasibility of the algorithm developed in this study for retrieving the clumping index from the MCD43A1 product.Based on the new clumping index product, the 8-day 500 m resolution MODIS-based LAI (MOD_NJU LAI) maps over China's landmass from 2003 to 2008 are generated. The new LAI product has been validated using the field measurements in 2 forest areas and 2 grassland areas in China through an upscaling process using 30 m resolution Landsat5-TM images. Compared with MOD15A2 LAI, MOD_NJU LAI shows much stronger correlation with the reference LAI in these validation areas. The mean overall evaluation accuracy for the 2 forests areas can reach 73.5%, suggesting satisfactory performance of the algorithm in retrieving forest LAI. However, for the 2 grassland areas, this algorithm significantly underestimates grassland LAI.In this study, we propose a moisture adjusted vegetation index (MAVI) by replacing the soil adjusted factor L in soil adjusted vegetation index (SAVI) with the SWIR reflectance to constrain the background effects on LAI retrieval. The superior performance of MAVI over other Vis in retrieving LAI is investigated using the modeled data using the 4-Scale model and field measurements in two forest and two grassland areas in China. The new vegetation index defined as the ratios of bands is able to reduce a large proportion of noise caused by topography variations. This advantage is demonstrated in LAI mapping in a mountainous forest area. The new MAVI is used to retrieve the reference LAI in our LAI validation, and is expected to minimize the topographic effects on the LAI validation.