Near-surface Earth Parameter Estimation via Recompressive Filtering

Utilization of dispersive surface waves is a promising technique for velocity model building of the near-surface. To analyse the dispersive properties of surface waves in case of a 3D reflection seismic acquisition, where surface waves are usually regarded as noise, we consider a recompression technique. Recompressive filtering collapses the long wave-train of dispersive waves into an impulse-like signal. This feature of the filtering method proved to be useful in wave type and wave mode separation, which is highlighted in this paper. When recompressive filtering is incorporated into an optimisation procedure the wavelet shape can drive the parameter estimation (dispersion curve and/or model parameters). We have developed a processing sequence, where traces of reflection records containing dispersive arrivals can be used to get information on the near-surface. On synthetic data we show how well the results of the recompression based Simulated Annealing optimisation works even in noisy conditions. The near-surface earth parameters which can be estimated with the help of recompressive filtering, presented in this paper, can support noise removal techniques, help the standard static calculations, intra-array perturbation correction, depth-velocity function estimation near the surface, etc. even in the latest 3D single source and/or single receiver field acquisition methods.