Land Surface Temperature And Atmospheric Phofile Parameters Retrieval Based On Perturbation Form Of Atmospheric Radiative Transfer Equation

Land surface temperature is a key parameter of surface physical process and plays an important significance to the energy balance in the world, which affects the energy exchange between land surface and atmosphere. It has widely been applied to meteorology, geology, hydrology and zoology, especially the study on water content of soil, monitoring of forest fire.In this process of resolving radiative transfer equation, it is very important to establish the perturbation form of radiative transfer equation. To a great extent, the precision physical algorithm inversion is dependent on the known atmospheric information and urface information. In this algorithm, the initial atmosphere temperature, the water vapor, the surface temperature and two kinds of surface emissivities, are best synchone to the overpass time of satellites. Regarding the random, a series of equations should be built. But as a result of the wave band limit, we can only obtain 16 equations, including 123 unknowns: 40 atmosphere temperature profiles, 40 moisture profiles, 40 ozone profiles, the surface temperature and two surfaces emissivities. In order to reduce the uncertainty and improve the calculation efficiency, an empirical orthogonal function analysis method is employed to reduce the dimension of the original matrix, with the supplementary related new variables to stand for the original variable information. After the empirical orthogonal function analysis, the unknown quantity reduces to 14 (5 atmosphere temperature eigenvector, 3 water vapor eigenvector/feature, 3 ozone eigenvector, 1 surface temperature perturbation quantity and two surface emissivity perturbation quantities ).By statistical regression analysis, two step physical algorithms are achieved to complete the resolving process. First, the Tikhonov Regularization is used to renew initial value for the balance of the residuals radiation value and initial value. Then the non-linear Newton iterative algorithm is applied to obtain the new solution. The nonlinear Newtonian iteration algorithm is further employed to the Tikhonov regularization solution. The regularization parameter plays an important role in the retrieved solution. It balances the residual norm and the side constraint norm.In order to confirm the feasibility of the method, two-steps physical algorithm is investigated in this study. Terra-MODIS data at 11:05 on 30 October 2005, aqua-MODIS data at 13:20 on 31 October 2005 of Poyang Lake area are selected to retrieve surface parameter and the atmosphere parameters in the morning and afternoon time. Also, terra-MODIS data at 11:05 on 4 May 2004, aqua-MODIS data at 02:20 on 5 May 2004 of Luancheng area are selected to retrieve surface parameter and the atmosphere parameters in the morning and afternoon time. In order to validate the retrieval results, the NASA land surface temperature and emissivity products and in situ data are achieved to validate our retrieval results. And the precision and the sensitivities are further analyzed.