Warp and warp again: Empirical evaluation of an automatic method for the estimation of modal time-frequency dispersion in shallow-water

The time-frequency (TF) dispersion of normal-modes is known to be a relevant waveguide feature that characterize shallow-water acoustic propagation. A popular way to estimate modal TF dispersion is a non-linear signal processing method called warping. It has been extensively used for geoacoustic inversion and source-localization. However, a known drawback of warping is that it requires a manual trial-and-error tuning process. In this paper, we investigate a previously published method for automated estimation of modal TF dispersion. Each mode is extracted using optimal warping parameters that minimize an explicit quality factor, and the resulting warped modes are separated using a watershed segmentation algorithm. The contribution here is an empirical validation of the method on airgun data collected on the New England Mud Patch. It is successfully applied to 50 airgun shots, recorded on a set of 11 distant receivers, demonstrating automatic modal TF estimation for more than 500 source-receiver pairs. The estimated modal dispersion data is further used to perform geoacoustic inversion, illustrating the practical performance of the fully automated warping-based processing chain. [Work supported by the Office of Naval Research.]