ИСТИНА

4D seismic is the main tool for monitoring CO2 geosequestration. Traditionally, time-lapse signal is being interpreted qualitatively - spatial location of the main body of CO2 plume. However, storage site operators are legally bound to validate conformance of the injection to dynamic models, which requires estimates of the plume parameters: plume thickness and spatial distribution of CO2 saturation etc. Diffusion of CO2 is driven by buoyancy at relatively short distance from injection wells, so the plumes tend to form heterogeneous pack of thin low-saturated interlayers that may not be resolved accurately by surface seismic (Williams and Chadwick 2012). Response from a thin but contrast layer is a fundamental problem for seismic monitoring, e.g., active detection of hydraulic fractures (Oelke et al. 2013). Widess (1973) developed approximate reflectivity of an isolated thin layer to answer ‘how thin is a thin bed’. Here, we have to do with a similar question – what is a ‘seismic CO2 plume’? Glubokovskikh et al. (2016) examined detectability of a very small CO2 leakage through a computationally expensive simulations. Extending that study, we focus on accuracy of 4D seismic estimates of the plume parameters relevant for dynamic simulations. To this end, we develop stochastic rock physics/seismic modelling workflow to estimate the sensitivity to the plume parameters and inherent accuracy limits of 4D seismic.