Depth characterization of shallow aquifers with seismic reflection, Part I - The failure of NMO velocity analysis and quantitative error prediction

As seismic reflection data become more prevalent as input for quantitative environmental and engineering studies, there is a growing need to assess and improve the accuracy of reflection processing methodologies. It is common for compressional-wave velocities to increase by a factor of four or more where shallow, unconsolidated sediments change from a dry or partially water-saturated regime to full saturation. While this degree of velocity contrast is rare in conventional seismology, it is a common scenario in shallow environments and leads to significant problems when trying to record and interpret reflections within about the first 30 m below the water table. The problem is compounded in shallow reflection studies where problems primarily associated with surface-related noise limit the range of offsets we can use to record reflected energy. For offset-to-depth ratios typically required to record reflections originating in this zone, the assumptions of NMO velocity analysis are violated, leading to very large errors in depth and layer thickness estimates if the Dix equation is assumed valid. For a broad range of velocity profiles, saturated layer thickness will be overestimated by a minimum of 10% if the boundary of interest is <30 m below the water table. The error increases rapidly as the boundary shallows and can be very large (>100%) if the saturated layer is <10 in thick. This degree of error has a significant and negative impact if quantitative interpretations of aquifer geometry are used in aquifer evaluation such as predictive groundwater flow modeling or total resource estimates.