To be able to study information content of satellite observations with high spatial and temporal resolutions and where the emphasis is on the analysis of water cycle including clouds and soil moisture we developed a mesoscale data assimilation numerical model using four dimensional variational (4DVAR) data assimilation approach applied to the Regional Atmospheric Modeling System (RAMS). We called this data assimilation model RAMDAS (Regional Atmospheric Modeling and Data Assimilation System). An overview of RAMDAS in presented in the poster session at this conference (Vukicevic et al). The 4DVAR technique applied in RAMDAS is similar to the 4DVAR used with the Eta model at NCEP: a) data assimilation problem is solved optimally with the forecast, background and observational errors included, b) Zupanski's preconditioning is applied which allows for small number of iterations and c) the digital filter is used for control of high frequency modes. The RAMS and associated adjoint model in RAMDAS include state of the art explicit microphysics and interactive land surface model (LEAF-2). This property allows for assimilation of satellite radiance observations under both clear and cloudy conditions. To date we developed observational operators for use of the GOES imager measurements. The observational operators are presented in companion paper at the same conference (Greenwald and Vukicevic). This paper discusses RAMDAS properties relevant to the cloudy radiance assimilation and the results of data assimilation experiments using GOES channel 1 measurements for a case of stratiform cloud.