Horizon picking for multidimensional data: an integrated approach

3-D seismic surveys generate 5-D data volume. In order to estimate the horizons for interpretation and further processing, the traveltime picking needs to be performed on n-D subsets of this 5-D data volume (n≤5). Horizon picking (HP) is complicated by the irregular sampling, faults, discontinuities, and low signal-to-noise ratio areas. The automatic HP techniques described here are aimed to support the interpreters in the estimation of the events by preserving their depth continuity. The HP is carried out directly on the full n-D dataset and not just iterated on 2-D sliced subsets, this avoids misties among different 2-D slices. The additional advantages are that the proposed method can perform the HP of multiple and irregularly sampled horizons, in addition it can handle discontinuous events by keeping the association with the target in depth. The interpreter is only asked to initialize the HP by providing some seed picks on the target horizon(s), then the algorithm lets the estimate of the horizons grow along all the dimensions simultaneously.