Integration of Density-Based Spatial Clustering with Noise and Continuous Wavelet Transform for Feature Extraction from Seismic Data

Seismic reflections are crucial for obtaining information about subsurface structures and lithologies for oil and gas exploration. Several techniques have recently been introduced which improve the visualization of subsurface structures, lithologies, and facies. This article proposes a novel method of seismic reflection identification through the integration of continuous wavelet transform (CWT) and density-based spatial clustering of applications with noise (DBSCAN). Here, a three-layer geological model is adopted. Initially, 2D seismic reflection data with 5%, 8%, and 10% Gaussian noise are generated. Later, the DBSCAN algorithm is applied to 2D noise seismic data, and clusters are generated at their respective times for each reflector. Next, to confirm and validate the results of DBSCAN, CWT is executed on the cluster data set. Based on our results of CWT, the true representation of seismic data with minimum noise in the time domain is achieved. The successful integration of DBSCAN and CWT is achieved in terms of identification of true seismic reflections as localized anomalous zones at 0.8 s, 1 s, and 1.07 s, which exactly match the geological model of this study.