A neural network refinement of seismic data processing

Seismic reflection data processing that is widely applied to oil exploration uses data acquired with low frequency ranges that are in the order of tens to hundreds hertz. This range of frequencies allow very deep penetration and low resolution data acquisition. Engineering and environmental applications require high resolution shallow subsurface seismic reflection data acquired using frequencies that range on the order of thousands hertz. Processing of high resolution shallow subsurface seismic reflection data has not been addressed in detail in the seismic exploration literature. This research presents a technique including Artificial Neural Networks to process high resolution shallow subsurface seismic reflection data.; This technique is applied to locate oyster reefs and paleochannels in a seismic reflection survey performed by The Mississippi Mineral Resources Institute near Cat Island, Mississippi. Artificial Neural Networks that allow the selection of positive picks and the enhancement of reflectors in seismic reflection data are developed and applied to seismic reflection data processing.; Seismic sections of the subsurface of the studied area are developed and maps depicting the location of oyster reefs and paleochannels near Cat Island, Mississippi are produced. A stepwise procedure to apply Artificial Neural Networks to the seismic data processing is also presented.