Multi-Scale Fault Recognition Method Based on Generalized W Transform

Faults control the formation and distribution of hydrocarbon reservoirs. Accurate fault characterization effectively reveals the rules of hydrocarbon accumulation and reservoir enrichment. Conventional fault recognition methods, such as coherence and curvature attributes, are based on the assumption of continuity anomalies in seismic data near faults. However, these methods are highly influenced by seismic data quality and often fail to accurately delineate fault distributions, resulting in imprecise fault positioning and unclear contour depiction. This paper proposes a multi-scale fault recognition method based on generalized W transform (GWT). After applying the GWT for frequency decomposition of seismic data, the gradient structure tensor fault attributes of different scales are calculated. Large, medium, and small scale fault features are integrated to obtain multi-scale fault attributes. Subsequently, the optimal surface voting is applied to enhance fault delineation and produce high precision multi-scale fault distributions. This method combines large scale fault development trends, medium scale fault distributions, and small scale fault details to effectively utilize features across scales, thereby improving fault recognition accuracy. Applications show that the GWT-based method provides clear fault positioning and contours, improves lateral and vertical continuity, and significantly enhances the precision of fault characterization.