HORIZON PICKING ALGORITHM USING GLOBAL OPTIMIZATION AND COHERENCE MEASUREMENT IN POST-STACKED SEISMIC DATA

Abstract

Accurate horizon recognition within post-stack seismic sections stands as a pivotal stride in seismic interpretation. Horizon-picking techniques wield importance in diverse seismic structural analyses and inversion methodologies. Despite encountering challenges like computational demands and time constraints, the past years have witnessed the evolution of numerous 2-D and 3-D methods. With the progress of modern computing, more resilient and efficient techniques, harnessing artificial intelligence, have come to the fore. In this context, we propose an innovative, cost-effective algorithm grounded in a global optimization framework. This algorithm amalgamates the Very Fast Simulated Annealing method with a set of stability parameters and coherence measurement of neighboring seismic traces, employed as the objective function. The search is executed on continuous and sequential clusters of seismic traces, with a focus on maximizing coherence amidst the presented events. We evaluate the algorithm's efficacy by applying it to two distinct input data sets—enveloped and non-enveloped seismic traces. The initial application pertains to the time-migrated Marmousi dataset, a synthetic marine dataset originating from a complex geological sedimentary basin situated in Angola, Africa. Subsequently, in the context of land seismic data, we apply the algorithm to depth-migrated field data extracted from a past survey conducted in the Tacutu basin in northern Brazil. Outcomes arising from the application to 2-D synthetic and field data sets underscore the method's viability as a compelling alternative for seismic horizon picking within the time or depth domain.#xD;