This study introduces a new snowfall detection algorithm over land using AMSU measurements. The algorithm utilizes a combination of channels in the microwave window, water vapor and oxygen absorption regions. The new algorithm is combined with the current NOAA operational AMSU rain rate product for falling snow and rain areal extent retrievals over the U.S. These retrievals are compared with NEXRAD radar precipitation and local meteorological observations. Preliminary results demonstrate that the algorithm is robust in capturing winter precipitation with minimum classification errors.