A Critical Comparison of Three Methods for Time-Lapse Time-Shift Calculation

Time-shift, one of the most popular time-lapse seismic attributes, has been widely used in dynamic reservoir characterization by linking it with pressure and geomechanical changes. Therefore, it is important to select appropriate calculation methods according to different time-lapse seismic data quality and time-shift magnitude. To date, there have been various published works comparing different time-shift calculation methods and discussing their advantages and disadvantages. However, most of these comparisons are based only on synthetic tests or single field applications. As the quality of time-lapse seismic data and time-shift magnitude can vary in different fields, one method may not work consistently well for each case. In this paper, a critical comparison of three different time-shift calculation techniques (Hale’s fast cross-correlation, Rickett’s non-linear inversion, and Whitcombe’s correlated leakage method) is provided. The three methods are applied to a set of synthetic data sets that are designed to account for various seismic noise and time-shift magnitudes. They are also applied to four real time-lapse seismic data sets from three North Sea fields. The calculated time-shift results are compared with the input (in synthetic tests) or the real observations from information such as seabed subsidence and compaction (in field applications). Both qualitative and quantitative comparisons are performed. At the end, each of the time-shift methods is evaluated based on different aspects, and the most appropriate method is suggested for each data scenario. All three time-shift methods are found to successfully measure time-shifts. However, Rickett’s non-linear inversion is the most outstanding method, as it gives smooth time-shifts with relatively good accuracy, and the derived time strains are more stable and interpretable.