Research On One Dimensional Variational Inversion Methed For Ground-based Microwave Radiometer

Atmospheric profile is a very important parameter for human's research on weather prediction. In the past time, atmospheric profile was usually achieved by radiosonde which was launched twice a day. Due to the reason that the radiosonde cost a lot, can't be reused, and dosen't ascend vertically, human gradually used remote sensing instrument instead of radiosonde. And ground-based microwave radiometer system as an important kind of atmospheric remote sensing instrument, is widely used to observe atmosphere.Ground-based microwave radiometer system observes the atmospheric downward radiation brightness temperature, and inputs the brightness temperature into the inversion model, and then achieves the inversion results of atmospheric profile. This article focuses on the one-dimensional variational inversion algorithm of the microwave radiometer, and validates the performance of this algorithm.Based on the atmospheric radiation transmission theory, this article firstly introduces the atmosphere attenuation model. And the article simulates the main absorption coefficient of atmospheric radiation in microwave band, this simulation shows how to select radiometer frequency band. Then the article introduces the atmospheric downward transmission equation, and its physical meaning which lead to the atmospheric stratification and discretization of radiative transfer forward model. And then the article turns those integral equations to multivariate equations, which equals the inversion problem to the solution of equations.Then this article introduces several kinds of inversion algorithm for ground-based microwave radiometer. At first, article mainly introduces linear regression and neural network, and then secondly includes regularization method, the smoothing method, bayesian maximum probability method and iterative method. the mathematical formula of each inversion algorithm is also given.This article emphasizes the theory of one-dimensional variational inversion algorithm and the analysis of experiment results. The article simply induces the data assimilation method. Then article elaborates the mathematics principle of one dimensional variational algorithm, the background and observation errors, the jacobian matrix and the conjugate gradient method. At last the article shows the simulation result analysis and the comparation of inversion result.