Data‐Driven Velocity Model Evaluation Using K‐Means Clustering

Abstract We develop a data‐driven clustering method to evaluate velocity model using surface wave dispersion. This is done by first computing theoretical dispersion curves for 1‐D profiles of all the grid locations and then splitting resulting into certain number groups via K‐means clustering. The observed are also clustered following same procedure assessed comparing spatial patterns obtained synthetic data sets. applied two community models in southern California, CVM‐S4.26 CVM‐H15.1, phase maps derived 3–16 s Rayleigh waves. found good correlation distribution clusters between result that data, suggesting fits better than CVM‐H15.1 terms features extracted from analysis.