Model uncertainty analysis for de-risking seismic image accuracy

Abstract

With exploration moving towards areas of increasing geological complexity, reservoir evaluation is often based on the interpretation of a single seismic image. Recovering a suitable velocity model employed in pre-stack depth migration plays a crucial role in the creation of this image on which economic evaluations are drawn. Typical depth imaging projects provide final velocity model attributes and their associated seismic image. The amount of uncertainty associated with this image is poorly understood; the only quantitative measures of reliability are provided through analysis of volumetric residual move-out and by comparison with available auxiliary data. We aim to address this situation by using the same tomography method we use to derive the model parameters. The workflow allows us to establish the resolution of the tomography, in establishing the model parameters. It also enables us to determine the recoverable degree models may be perturbed prior to the tomography. Using the workflow in conjunction with these criteria we generate a population of solution models which equally conform to the observed data. We then analyse the variance of this model population to derive confidence attributes to assign to both the target model and its associated seismic image.