Cloud resolving model outputs are often used to build databases for satellite microwave remote sensing of precipitating clouds. A known problem of this approach is that cloud resolving models tend to systematically produce excessive amount of high density frozen hydrometeors, causing the cloud/radiation model database to have stronger scattering signatures at high microwave frequencies than those observed by satellite or airborne sensors. Consequently, it lowers the performance of the cloud and precipitation retrieval algorithms that utilize the model database. Since multi-frequency satellite observations contain information on hydrometeors' properties, measured brightness temperatures can give guidance as to how the modeled cloud database may be modified to better mimic natural clouds. Following this philosophy, in this study, we propose a method to adapt the modeled database toward observations. The newly constructed database results in a better match to the characteristics of the satellite observed brightness temperatures.