Roughness Retrieval And Topographic Correction Based On Bare Surface Radiative Transfer Model

This dissertation was designed to study the directional reflectance properties of the bare surface consisted with the densely packed particles media. The physical parameters (sub-pixel roughness, the average single-scattering albedo, the asymmetric parameters of phase function) are retrieved using the bidirectional reflectance model and multi-angle remote sensing data (CHRIS / PROBA). Furthermore, on mountain area, the hyperspectral data processing, includes the topographically radiometric correction and reflectance retrieval, are studied based on atmospheric radiative transfer mode and the DEM elevation data. The work mainly includes three aspects.Analysis of the bare surface bidirectional reflectance model and environmental characteristics (particles size distribution, porosity, the opposition effect ) has been carried out to investigate the model characteristic. Four bidirectional models have been compared to each other: the Hapke model, the improved Hapke model, the DISORT model and the Mishchenko model. The result of analysis shows that the improved Hapke model is accurate and simple for inversion of the surface physical parameters. The improved Hapke model was fit to the laboratory reflection infrared spectra from 5 to 25 micron wavelength that include the United States Geological Survey spectral library of minerals (kaolinite, serpentine, montmorillonite, saponite) and different size powder of quartz powder, and the model spectra was very similar to these laboratory spectra.Roughness parameters (root-mean-square elevation, root-mean-square slope, the mean slope) and the shadowing model of roughness (Smith shadowing function, Beckmann shadowing function, Hapke shadowing function) have been analyzed for characteristics. The parameters sensitivity of the bidirectional model have been studied with the imaging geometry of CHRIS/PROBA data, which optimizes initial value of the parameters inversion. Moreover, with a prior parameters sensitivity information, a new method called 'the initial value optimization inversion'has been presented to improve inversion results. Comparison between the model calculation of different parameters and inversion results shows that the accuracy of 'the initial value optimization inversion'is perfect, and the simulation image of the retrieval parameters was very similar to the multi-angle remote sensing data.On mountain area, topographic correction and reflectance retrieval of hyperspectral data are studied, moreover, the solar direct radiation and sky diffuse radiation have been simulated using MODTRAN atmospheric radiative transfer model and DEM data . Analysis of the topographic correction(Statistical Empirical Method SEM, C-correction, Minnaert-correction, Cosine-correction) has been investigated. The establishment of hyperspectral data 'topographically radiometric correction -- reflectance retrieval' module provides a good processing for remote sensing data on mountain area, which is based on MODTRAN model and the DEM data. The modules related to topographically radiometric correction, atmospheric water vapor and aerosol retrieval, look-up table (LUT) creation and interpolation technique etc.