Vashan Wright , Matt Hornbach

How effective are rock physics models for relating seismic velocities to the physical properties of sediments, fluids, and cement within the critical zone, and what factors most substantially influence the models’ accuracies? We answer these questions by testing and analyzing the accuracies of seven rock physics models (Hertz-Mindlin, Walton, Jenkins, Digby, stiff sand, soft sand, and contact cement) for estimating seismic velocities of vadose zone sands at Port Royal Beach in Jamaica. These sands are clean, well-rounded, and highly-spherical, which are ideal for rock physics model testing. Measured velocities and model input parameters (e.g., porosity, density, grain size, and fluid saturation percentage) derive from seismic refraction surveys and sidewall sediment cores, respectively. We find that, in their current forms, all seven rock physics models overpredict seismic velocities for sands deposited within the last forty-three years. Misfits between measured and predicted velocities reduce with time since deposition, with all but one (Digby) cementless models accurately predicting the seismic velocities for sands older than ninety-five years. Jenkins, followed by Walton, Hertz-Mindlin, and soft sand models are generally most accurate (i.e., have the lowest misfits), possibly because high porosity sands are more susceptible to tangential slip during seismic wave propagation. We conclude that the models will most substantially improve when the effects of the existence and locations of strong versus weak force-chain links are included in their respective equations.