Time-lapse seismics with randomized sampling

In time-lapse or 4D seismics, repeatability of the acquisition is a very crucial step, as we do not want spurious events that are not there. In this paper, we propose an approach which avoids any requirement to repeat the surveys, by using randomized sampling technique which allows us to be more efficient in the acquisition. Our method applies to sampling data using ocean bottom nodes (OBN) as receivers. We test the efficacy of our proposed randomized acquisition geometry for time-lapse survey on two different models. In the first example, model properties does not change with time, while in the second example, model exhibit a time-lapse effect which may be caused by the migration of fluid within the reservoir. We perform two types of randomized sampling - uniform randomized sampling and jittered sampling to visualize the effects of non-repeatability in time-lapse survey. We observe that jittered randomized sampling is a more efficient method compared to randomized sampling, due to it's requirement to control the maximum spacing between the receivers. The results are presented, in the image space, as a least-squares migration of the model perturbation and they are shown for a subset of a synthetic model - the Marmousi model