基于地震纹理属性聚类分析的裂缝分布预测

裂缝在地震数据中的特征可视为一种纹理,因此可以使用纹理属性表征裂缝,基于纹理属性聚类分析方法对裂缝发育带进行预测。计算地震数据体不同灰度级和不同三维数据基元大小的体共生矩阵,以此提取多组纹理属性,并进行对比和优选,最后使用模糊C均值聚类法对优选的纹理属性进行分析,根据聚类结果对裂缝的分布范围进行预测。方法减少了人工预测裂缝的主观性,而且从多个方面表征裂缝来提高预测结果的可信度。对野外数据进行分析,其结果表明分析获得的裂缝预测范围与勘探结果具有较高的吻合度。

Prediction of Fracture Distribution Based on Clustering Analysis of Seismic Texture Attributes

Fracture can be described as texture feature of seismic data. Therefore, texture attributes of seismic data can be used to identify fracture. Based on the amplitude of the seismic data volume, multi-group texture attributes be extracted with different gray levels and different size of three-dimensional data element. The optimum texture attribute is obtained by comparing the features of texture attributes. The fuzzy C-means clustering method is used to analyze the texture attribute, and the clustering result can be used to predict the fracture distribution. The proposed method is more objective and credible compared with manual intervention. The result of practical application shows that the results of fracture prediction is consistent with the exploration results.