The importance of soil state in initializating atmospheric models is studied in a wide number of studies for climatic, meteorological and hydrological applications and impacts of soil moisture fields on atmospheric behavior were analyzed as one of the dominant slow component of Earth climate system in controlling continental precipitation especially in midlatitude summers. In this study precipitation and temperature sensitivity impacts were analyzed for a deep convective event on August 2002 in Italy. The last parallel version (5.04) of the Regional Atmospheric Modeling System, RAMS, is used to perform the convective event simulation. A modified version of RAMS, called RAPI, is adopted to produce initial soil state. Such initial state is made running RAPI, which is made essentially by the soil-vegetation-atmosphere RAMS scheme (called LEAF-3), using only two information as inputs for the specific period: atmospheric fields, coming from a previous run, and satellite observed precipitation. Using such two inputs RAPI produces an evolution of the soil state, for a the defined period. The RAPI soil output fields provide the soil input fields for the RAMS simulation. A quantitative description of atmospheric behavior divergences coming from an ensemble of soil moisture initialization is provided in order to study aspects of possible forecast skill improvement on a short time range. The proposed approach, due to his extreme flexibility, is designed non only for a hindcast utilization, but for an operational framework too.