In the current AMSU-B rain rate retrieval algorithm, the cloud ice water path (IWP) and ice particle effective diameters (De) are simultaneously derived using the AMSU channels at 89 and 150 GHz. The derived IWP is converted into the surface rainfall rate through an IWP and rainfall rate relationship developed from cloud model results. It is found that the AMSU-B rain rate algorithm tends to overestimate stratiform precipitation and underestimate convective precipitation. To improve the performance of this algorithm, different rain rate algorithms are developed according to rain types. This is done by using both AMSU window and sounding channels. An approach is made to use the AMSU 183 GHz moisture sounding channels to classify different rain types. It is found that the three sounding channels at 183 ±1, ±3, ±7 GHz which are sensitive to the water vapor burden above different levels provide unique signals on the vertical extent of frozen hydrometeors. In particular, brightness temperature at 183±7 GHz is sensitive to the presence of ice particles at lower altitudes, whereas the measurements at 183±1 GHz primarily responds to deep convection where large ice particles are thrust substantially higher into upper atmospheres. This sensitivity is studied here with AMSU 183 GHz measurements. An indicator of the convective strength of cloud systems is calculated based on the information inferred from the AMSU 183 GHz measurements, and is then used to separate convective regime from stratiform. This classification scheme is integrated as part of AMSU precipitation algorithm. The preliminary result shows that the rainfall rate is improved with the rain type information. Furthermore, the improved algorithm enhances the dynamic range of the rain rate.