Here we aim to develop the technique to define the life stage of deep convection from the split window (11 and 12 mm )of GOES snap shot image. In a simple deep convection, we realize that the life cycle starts from the cumulus type convective cell, cumulonimbus type cloud at mature stage, then anvil cloud dominance at decaying stage. The split window is effective in classifying the cirrus cloud that consists of ice crystals. We can track the change of dominant cloud type in the deep convection with the use of split window cloud type classification technique. Therefore, we are going to develop the scheme to define the life stage of deep convection with the ratio of cirrus cloud coverage within the deep convection. The life stage of deep convection is important in terms of rainfall. We found that convective rain observed by the PR/TRMM is dominant at the developing stage of deep convection by the split window technique and the stratiform rain is dominant at the decaying stage. Further, current microwave rainfall estimation tends to underestimate the rainfall amount at the developing stage and to overestimate the rainfall amount at the decaying stage. The information of life stage of deep convection is important for the improvement of a passive microwave rainfall estimation algorithm.