Time-lapse Data Enhancement and Regularization with Common-offset CRS Stack

Abstract

Data quality is extremely important for successful time-lapse experiments. Time-lapse seismic requires estimation of 4D changes that are often smaller than the noise level. Hence, data processing and noise suppression are the key steps for time-lapse analysis. We propose a method for noise suppression and regularization of prestack data. The method is based on the local stack of spatially coherent events along the traveltime surfaces defined by the common-offset common-reflection-surface traveltime operator. Since the data are stacked locally, we don't harm amplitudes and phases of the signal. The coefficients in the traveltime approximation have a definite physical meaning which allows us to enhance particular types of waves. By the example of cross-well dataset we demonstrate, that the proposed method efficiently suppresses random noise, enhances the desired signals and increases the repeatability of the data. The overall benefit is a more reliable estimation of time-lapse changes, providing a more reliable information for enhanced oil recovery or other applications. The proposed stacking technique is not limited to cross-well observation geometries and can be extended to 2D/3D OBN and VSP datasets.