| dc.description |
Submarine mud diapirs (MDs) and volcanoes, often linked with gas hydrates and hydrocarbons, are prevalent features in the offshore region of SW Taiwan. Despite their significance, the gravity signatures of these MDs have not been investigated except for a few studies. Therefore, our study aims to address this gap by employing a recently proposed optimization algorithm, Hunger Games Search (HGS), to estimate model parameters of an inclined 2D anticlinal structure representative of an active MD. This approach was implemented to gravity anomaly obtained from SW coast of Taiwan. Before its application, we explored the nonlinear and high-dimensional characteristics of the cost function associated with this inverse problem by generating error landscape maps for model parameter pairs. Subsequently, to mitigate the challenges, a parameter tuning procedure was performed. Using the optimal control parameters derived from this procedure, we increased the efficacy of the HGS algorithm for the problem handled. Furthermore, we employed standard Particle Swarm Optimization (PSO), a widely utilized metaheuristic in geophysical inversion studies, for a comparison. The findings showed that the proposed novel algorithm yielded superior accuracy and reliability compared to PSO. Additionally, the inclined anticlinal structure exhibited better agreement with interpretation obtained from available seismic reflection outcomes, in contrast to approximation based on inclined dike-like structure. Besides, comparative analyses between HGS and two well-established techniques, namely correlation imaging and compact inversion algorithms further validated the effectiveness of HGS. Consequently, we assert that the novel HGS optimizer is a robust and appealing tool for inverse gravity problems. Moreover, assuming a 2D inclined anticlinal source type holds promise for extended investigations into other MDs in SW Taiwan's upper slope region. Such efforts can significantly enhance the understanding of regional mud diapirism and volcanism on a broader scale. |
|