Global 4-D Seismic Inversion And Time-Lapse Fluid Prediction

Independent inversion of base and monitor seismic surveys can yield estimates of elastic properties that are inconsistent with expected production effects. We therefore propose a global time-lapse inversion scheme, involving joint inversion of base and monitor data. All vintages and input angle stacks are combined in a single objective function, which is optimized using simulated annealing to estimate the time-variant distribution of elastic attributes that best matches all available data. The multi-vintage nature of the optimization allows us to incorporate flexible, user-defined rock physics constraints on the evolution of Vp, Vs and density between consecutive surveys. There are no restrictions on the number of input angle stacks or number of monitor surveys. The constrained, global inversion solution can therefore be easily updated as new data become available. We apply the global 4-D inversion with rock physics coupling to data from the Brage Field and compare results with a workflow involving separate inversion of base and monitor data. The global 4-D inversion results are combined with a time-lapse Bayesian fluid classification scheme to map production-induced fluid movement and quantify associated uncertainty.