Abstract: For the specific issues, this paper, based on all kinds of seismic attributes, using the technology of the forward modeling, analysis the sensitive attributes of various types of sand body development model, select the seismic attributes which can reflects the reservoir development best and optimalize the seismic attributes. Based on this optimization, letting those seismic attributes doing cluster analysis attributes fusion. It overcomes the one-sidedness of the single seismic attribute for reservoir prediction. Because of this, it achieved the fine characterization of channel sand body in the layer of H22 in the area of Gu109 and it is confirmed by seismic data, forward modeling, log data and sedimentary facies and so on. Finally, through the analysis of the sand body of the different kinds of the reservoir, achieving the fine characterization of channel sand body under the controlling of combination of seismic and logging data, it gives the actual geological significance for the different kinds of seismic attributes, identifying, predicting and describing the situation of the reservoir.