A NEW OPTIMIZATION ALGORITHM IMPROVING REGULARIZED MIGRATION PROCESS IN SEISMIC IMAGING

Reflection seismic survey is the most effective method of imaging the deep subsurface. However, the seismic data obtained from reflection surveys requires careful and tedious steps of processing to generate an image that is closer to the actual configuration of the reflectors. This process gets more difficult as the geological structure deviates from the horizontally layered sequence with regular thick layer boundaries. Regularized migration methods (in the space of time or along the depth) help us find the correct position of dipping geological structures in seismic images. As these methods have the shortcomings due to the ill-posed nature of the inversion process, we need to use hybrid algorithms to achieve optimal solutions both in quality and/or processing time.

In this research, we aim to process the seismic dataset gathered by the University of South Carolina in order to image the Magruder fault near Allendale SC. Specifically, we use a new hybrid algorithm based on a meta-heuristic method called “Dolphin Echolocation Algorithm” to replace the existing optimization algorithms like Genetic Algorithms or Memoryless Quasi Newton-Simulated Annealing Method to migrate the reflections. This algorithm is tested for its efficiency in reducing the processing time and improving the accuracy of seismic images, by better relating the reflection events to the actual structure through its unique iterative approach.