By appropriately decimating a comprehensive shallow 3‐D seismic reflection data set recorded across unconsolidated sediments in northern Switzerland, we have investigated the potential and limitations of four different source‐receiver acquisition patterns. For the original survey, more than 12 000 shots and 18 000 receivers deployed on a [Formula: see text] grid resulted in common midpoint (CMP) data with an average fold of ∼40 across a [Formula: see text] area. A principal goal of our investigation was to determine an acquisition strategy capable of producing reliable subsurface images in a more efficient and cost‐effective manner. Field efforts for the four tested acquisition strategies were approximately 50%, 50%, 25%, and 20% of the original effort. All four data subsets were subjected to a common processing sequence. Static corrections, top‐mute functions, and stacking velocities were estimated individually for each subset. Because shallow reflections were difficult to discern on shot and CMP gathers generated with the lowest density acquisition pattern (20% field effort) such that dependable top‐mute functions could not be estimated, data resulting from this acquisition pattern were not processed to completion. Of the three fully processed data subsets, two (50% field effort and 25% field effort) yielded 3‐D migrated images comparable to that derived from the entire data set, whereas the third (50% field effort) resulted in good‐quality images only in the shallow subsurface because of a lack of far‐offset data. On the basis of these results, we concluded that all geological objectives associated with our particular study site, which included mapping complex lithological units and their intervening shallow dipping boundaries, would have been achieved by conducting a 3‐D seismic reflection survey that was 75% less expensive than the original one.
Optimization through Bayesian Classification on the k-Anonymized Data
