Atmospheric retrievals using regularization methods

A radiometer measures microwave emissions from the atmosphere and surface. The radiance measured by the radiometer is proportional to the brightness temperature. This brightness temperature can be used to estimate atmospheric parameters such as temperature and water vapor content. Depending on the range in the electromagnetic spectrum being measured by the radiometer and the atmospheric quantities to be estimated, the retrieval or inverse problem of determining atmospheric parameters from brightness temperature might be linear or nonlinear. A difficulty associated with the retrieval problem is that they are under constrained because there are only few brightness temperatures being measured by the sensor while the quantities of interest are estimated at a higher vertical resolution (more unknowns). To further constrain the problem, prior information is incorporated into the problem to convert the ill-posed problem into a well-posed one. The author presents the results of work in applying different regularization techniques to the discrete problem associated with temperature retrieves using brightness temperatures from the SSM/T-1 sensor. Simulation results are presented which show the potential of these techniques to improve the performance of retrieval. In particular, no statistical assumptions are needed and the algorithms were capable of correctly estimating the temperature profile corner at the tropopause independent of the initial guess