Fully Automatic Picking of Surface Wave Dispersion Curves through Density-Based Spatial Clustering

Summary Rayleigh surface wave inversion can be used to characterize the near surface, which is a major task in desert environments due to the high complexity of the shallow geology. The inversion results depend on the accuracy of the dispersion curves extracted from the seismic measurements. This extraction is commonly obtained through manual picking which is time consuming, highly subjective and not feasible for modern large seismic surveys. In this work we introduce a novel and fully automatic method built on a density-based spatial clustering algorithm to pick surface wave dispersion curves in the frequency-phase velocity spectrum of the seismic gathers. The method was successfully tested on the SEAM Arid model synthetic dataset. The dispersion curves, extracted automatically using the proposed approach, accurately match the theoretical ones and produce results in very good agreement with the ground truth when inverted for the shear-wave velocity distribution. The presented method is currently under test with field data.