Seismic horizon-tracking strategy integrating local and global constraints

ABSTRACT Seismic horizon tracking is fundamental for reconstructing subsurface geologic history, yet traditional methods often struggle with complex geologic features, such as faults and unconformities. To address this challenge, horizon tracking was reformulated as a 3D scatter-point classification problem, balancing local and global tracking. A density-based spatial clustering of applications with noise (DBSCAN) was applied to group spatially continuous stratigraphic interface points extracted from seismic phase attributes, ensuring accurate local horizon tracking and generating multiple horizon patches. These patches were then treated as states within a Markov chain, where their connectivity was optimized using an accessibility matrix, enhancing global horizon-tracking accuracy. This framework enabled the delineation of multiple stratigraphic interfaces while preserving fine-scale structural details in regions with fault-induced stratigraphic offsets and unconformity-induced stratigraphic terminations. The integration of DBSCAN clustering and Markov modeling with seismic phase attributes allowed for an adaptive and scalable approach to horizon tracking. Benchmark tests on seismic data sets demonstrated the method’s effectiveness in capturing geologic complexities and refining subsurface structural interpretation. By improving the continuity of stratigraphic surfaces and accommodating structural discontinuities, this approach provided insights into depositional environments, tectonic activities, and erosional events. The proposed methodology enhanced the accuracy and reliability of horizon tracking, making it a valuable tool for reservoir characterization, seismic interpretation, and subsurface exploration.