A new automatic tracking system for seismic reflected events based on Kalman filter

Seismic events play an important role in the seismic data interpretation. A new automatic tracking system for seismic reflected events based on Kalman filter is proposed. The whole system includes five main components. Firstly for every seismic trace, the so-called feature points are extracted by using deconvolution and threshold detection. Secondly, use M/N method to start a new possible event. Thirdly, use the probabilistic data association to judge which case every new feature point belongs to. Then use a new proposed Kalman filter model to keep tracking the events. This Kalman model is based on the hyperbolic time-distance equation of the seismic reflected wave, and involves several important physical quantities such as the arrival time of the seismic wavelet, the offset and the average velocity of the seismic wave. Finally we use a set of marking system to manage all of the seismic events such as confirming a possible event, cancelling a false or redundant one, maintaining a normal one and ending a finished one. From the synthetic experimental result, the proposed method can track the seismic events well.