Abstract: Abstract：Carbonatefracture reservoirpredictionisthe keypoint ofpetroleum exploration and production. At present，seismic
reflection dip，coherence and maximum curvature are the common attributes to predict fracture reservoir. Seismic reflection
dip and coherence attributes are used to detect faults and large fractures. Curvature attribute is usually used to describe the
fractures. While single seismic attribute usually has multiple solutions，seismic facies technique for fracture classification
integrates multiattributes by Clustering Analysis，which extracts the common features of the faults and fractures of the seismic
data. The faults and fractures are detected and classified by seismic facies classification. In order to verify the reliability of
the prediction，the results is further confirmed by drilling，MFI and dynamic data. This method is capable of detecting 80%
fractures of a Middle East oilfield in the target formations，which improved the success of the reservoir prediction greatly.

Key words: Key words：fracture classsfication, seismic facies, Carbonate, attributes integration, reservoir prediction