Using Adaptive Differential Evolution algorithm to improve parameter estimation in seismic processing

Since the end of the 1990s, methods of imaging and inversion have been receiving systematic attention, through multiparametric traveltimes, such as the Common-Reflection-Surface (CRS) method, in its two versions zero offset (ZO) and finite offset (FO), and the Offset Continuation Trajectory (OCT). Despite its superior quality to traditional methods, OCT and CRS face the challenges of additional computation costs, which stem from the required multiparameter estimations. The problem of estimating the slope, curvature, and velocity parameters reliably and efficiently has been drawing focus in the seismic literature. Mathematically, approaches to solve that problem rely on global optimization techniques. The main challenges are robustness (small relative sensitivity to given initial values) and convergence speed. The Differential Evolution (DE) has shown promising results. That method has a welcome property of robustness, however also the drawback of undesired convergence speed. In this paper, we propose overcoming this problem upon applying the Adaptive Differential Evolution known as JADE. Qualitative results from synthetic and real datasets show, for similar execution times, the fast convergence of JADE when compared with that of DE. Therefore, JADE presents itself as a great alternative to DE, showing even more promising results regarding estimating the parameters of OCT and CRS.