A Variational Data Assimilation Technology Of The Cloud And Rain Affected Satellite Microwave Data

A wealth of information which is closely related to the development of the weather systems is contained in cloud and rain in the atmoshphere, improving the initial conditions in these regions is important for enhancing the numerical prediction skills. Microwave remote sensings are able to penetrate through the clouds and precipitation, especially with the rapid development of microwave imagers, which can provide a richer information of cloud and rain. However, the satellite data assimilation currently are still mainly concentrated on the data of clear sky in weather prediction, a lot of satellite observations which are affected by the cloud and rain are discarded. Therefore, we carried out the research on satellite microwave data which is affected by the cloud and rain in this paper.In this paper, we carefully study the basic principles of the 1D+4D-Var two-step method and design a one-dimensional+four-dimensional variational assimilation system process which is used for assimilating the cloud and rain affected Special Sensor Microwave/Imager (SSM/I) data. First, the cloud and rain affected SSM/I observations are used to constrain the one-dimensional variational assimilate (1D-Var) system which retrieves the total column water vapor (TCWV) , and then the TCWV amount which is as a type of pseudo observation is passed to the four-dimensional variational assimilation (4D-Var) system to get the atmospheric analysis fields with other observations. we mainly do the following aspects of work in the 1D+4D-Var system: the analysis of the deviation of SSM/I database; the application of large-scale condensation and convective parameterization schemes as moist physical operator; the application of the radiative transfer model RTTOV10's new features at microwave wavelength; the design of the background fields and error covariance matrics of cloud cover, liquid water and ice water; the design of the TCWV observation operator in the incremental method 4D-Var system. For the bias correction of SSM/I database, the lower three channels 19v, 19h and 22v are better, and they are conformed to the Gauss state. Consequently, only the TCWV information retrieved from the lower three channels would be transferred to the 4D-Var system. The 1D-Var experiments show the retrieve performance of 1D-Var system is good. To test the 1D+4D-Var system, we design a group of comparative experiments, where the control experiment didn't use the SSM/I observations which are affected by the cloud and rain, and joined the observations in experiment RAIN. Results show that the experiment RAIN is significantly better than the control experiment, and TCWV 1D-Var and 4D-Var increment is strong correlative, the correlation of analysis is significantly better than first-guess, which fully shows that the 1D+4D-Var system runs well.