Finite‐offset common reflection surface stack using global optimisation for parameter estimation: a land data example

In recent years, several research works dealing with velocity model independent seismic imaging have been published. These methods are capable of simulating arbitrary offset seismic sections by stacking a set of measured prestack seismic data along paraxial travel‐time surfaces. Hyperbolic common‐reflection‐surface travel‐time approximation is one of the most robust descriptions, which simulates not only zero‐offset but also finite‐offset sections with high accuracy from noisy multi‐coverage seismic data. In order to reconstruct seismic reflection events in common‐offset sections, the common‐reflection‐surface travel‐time approximation depends on five kinematic attributes (or parameters) for each selected point of the common‐offset seismic section. The main challenge of this method is to provide a computationally efficient data‐driven strategy for accurately determining the best set of parameters. Here, we introduce an approach for simultaneously estimating the five parameters from prestack seismic data by a very fast simulated annealing optimisation algorithm. For each sample point of the common‐offset section to be simulated, we determine only one set of common reflection surface attributes corresponding to the global maximum or the event with highest coherency. We applied our method of simultaneous global optimisation on synthetic and real data examples and showed the potential of the proposed strategy to enhance the reflection events in noisy data, even with very low signal‐to‐noise ratio. Finally, we demonstrate the regularisation capability of our method in a land seismic data example with missing traces for near, middle, and far offsets. In order to better appreciate the field data results, we present the time‐migrated sections with and without application of the proposed regularisation strategy.