Both previous theoretical discussion and case studies have shown that seismic reflections do not necessarily follow the geologic time lines. And in this study, we design a reproducible experiment to select seismic attributes that would favor the prediction of the time-transgression for a target seismic event.

Our Tertiary example in South Texas manifested an increasing tendency of time-transgressive reflections, generated by chaotic "labyrinth" velocity model with the most lateral lithofacies variation, compared with smooth "layered-cake" velocity model with the least lateral lithofacies variation. Conforming to this tendency, three-dimensional seismic volumes with known intensiveness and magnitude of time transgression could be prepared accordingly.

Afterward, a quantitative criterion to select time-transgression predictive seismic attributes was proposed. Time transgression for a given event is quantified by "peak-picked error", and the distributions of candidate seismic attributes are estimated within a defined window, centering that event. We inferred that the correlation between a candidate seismic attribute and the "peak-picked error" could serve as one of the quantitative criteria.