Lawrence Berkeley National Laboratory

Methods for geophysical model assessment, in particuale the computation of model parameter resolution, indicate the value and the limitations of time-lapse data in estimating reservoir flow properties. A trajectory-based method for computing sensitivities provides an effective means to compute model parameter resolutions. We examine the common situation in which water encroaches into a resrvoir from below, as due to the upward movement of an oil-water contact. Using straight-forward techniques we find that, by inclusing reflections off the top and bottom of a reservoir tens of meters thick, we can infer reservoir permeability based upon time-lapse data. We find that, for the case of water influx from below, using multiple time-lapse 'snapshots' does not necessarily improve the resolution of reservoir permeability. An application to time-lapse data from the Norne field illustrates that we can resolve the permeability near a producing well using reflections from three interfaces associated with the reservoir.