SEDIMENTARY FACIES ANALYSIS UPON SEISMIC ATTRIBUTES BY K-MEANS CLUSTERING ALGORITHM IN LOW-EXPLORATION AREAS: A CASE STUDY OF WENCHANG FORMATION IN BAIYUN SAG

The Wenchang Formation is a potential hydrocarbon source sequence in the Baiyun Sag, which has been penetrated by drilling one time only up to date and little geological and seismic data are available. In order to tackle these difficulties, the clustering algorithm with seismic attributes based on the method of K-means is recommended in this paper. On the premise of reliable interpretation, seismic attributes should be calculated with different methods for different units. As the results of attributes extraction, optimization and clustering algorithm, a distribution map of seismic attributes clustering would be prepared. By facies analysis of well data, using the seismic profiles and sedimentary models from adjacent areas as references, a comprehensive analysis of Wenchang Formation in the Baiyun Sag could be conducted. We pick RMS, Bandwidth Rating (Debias), and Energy half time for the K-means clustering algorithm and suggest that the research area can be divided into three units: Blue unit, Green unit and Red unit, which represent three kinds of seismic facies: smooth parallel or sub-parallel reflection with middle-high frequency and middle-high amplitude, discontinued sub-parallel reflection with weak-middle amplitude and chaotic high-amplitude reflection with wedge shape. Consequently, it is deduced that there were lakes and fan-deltas in the Eocene Baiyun Sag. In the earlier Wenchang stage, there occurred fan-deltas around the shallow Baiyun lake, as the lake-level rising in the later Wenchang stage, deep-lake area expanded up to 600km². The study gives a powerful contribution to the resource potential evaluation of the Baiyun Sag, which proves that this method is a reasonable and reliable one and useful for similar occasions.