Trace Warping vs. Impedance Warping in 4D Seismic Inversion

During hydrocarbon production variations in fluid saturation and pressure cause density and wave velocity changes, resulting in reflectivity amplitude and travel-time changes (Calvert, 2005). Where multiple surveys have been acquired over a producing reservoir these changes manifest themselves as differences in the seismic response between surveys. 4D or time-lapse seismic inversion involves finding a model which explains these differences. Existing methods of 4D inversion correct for the time shift by warping the seismic traces in time and solving for small changes in impedance. Here we compare this approach with that of warping the impedance traces and solving for changes in impedance. We make these comparisons on synthetic and real data to highlight the importance of adhering to the physical forward model (1D convolution) and demonstrate that improved results are achieved by doing so. We show that an inversion which warps the trace produces artefacts in the estimated model, most prominent when the 3D and 4D signals are strong. An inversion which warps the impedance does not suffer the same problems.