Tracking of Seismic Events Using a Combined Global and Local Optimization Strategy

Some seismic processing algorithms require a smooth and continuous track of a seismic event in a multi-dimensional domain. Often tracking algorithms are based on trace-to-trace correlations and have a local search character. They show a discontinuous behaviour, are sensitive to cycle-skipping and cannot bridge bad data zones. Therefore, it is proposed to design a multi-dimensional tracking algorithm that is based on a global solution that can avoid local minima, produces smooth and laterally continuous tracks and can be optimally guided by the user. In order to keep the algorithm efficient a hybrid optimization strategy is developed, in which a global optimization method - being a Genetic Algorithm - is combined with a local optimization method - the Simplex algorithm. For a 2D section the track is parameterized by spline functions and for a 3D volume the parameterization is extended with a polynomial description. Results on field data show the effectiveness of the tracking procedure.