Uncertainties in marine controlled source electromagnetic (CSEM) data consist of two independent parts: measurement noise and position uncertainties. Measurement noise can be readily determined using stacking statistics in the Fourier domain. The uncertainties due to errors in position can be estimated using perturbation analysis given estimates of the uncertainties in transmitter-receiver geometries. However, the various geometric parameters are not independent (e.g. change in antenna dip affects antenna altitude, etc.) so how uncertainties derived from perturbation analysis can be combined to derive error-bars on CSEM data is not obvious. In this study, we use data from the 2009 survey of the Scarborough gas field to demonstrate that (a) a repeat tow may be used to quantify uncertainties from geometry, (b) perturbation analysis also yields a good estimate of data uncertainties as a function of range and frequency so long as the components are added arithmetically rather than in quadrature, and (c) lack of a complex error structure in inversion yields model results which are unrealistic and leads to “over-selling” of the capabilities of CSEM at any particular prospect.
The ability of the marine controlled source electromagnetic method to resolve anisotropy in the sediment conductivity is not very well understood. In this study, we address the resolvability of anisotropy using a Bayesian approach. Two markedly different methods, slice sampling and reversible jump Markov Chain Monte Carlo have been used for the Bayesian inversion of a synthetic model of a resistive oil reservoir trapped beneath the seabed. We use this to identify which components of data can provide the strongest constraints on anisotropy in the overburden, reservoir and underlying sediments.
The effect of single-phase fluid saturation on the seismic bulk modulus of a rock is well understood; however, the behavior becomes more complex when multiple fluids are present. Several fluid mixing theories have been developed (e.g., Voigt, Reuss, and Hill) and each is valid in certain situations; however, in some scenarios it is unclear which theory to select, or indeed whether any are accurate. The critical wave propagation behavior depends on the manner that fluids are spatially distributed within the rock, compared to a seismic wavelength. We apply elastic finite-difference modeling to different rock-fluid distribution scenarios and replicate behavior described by various theoretical, empirical and lab data results. Significantly, our results compare well with observations from lab experiments, yet do not rely on poroelastic or squirt-flow models whose parameters are difficult to estimate in real reservoir settings. Our elastic scattering approach is less computationally expensive than poroelastic modeling and can be more easily applied to actual reservoir rock and fluid distributions. Our results provide us with a powerful new tool to analyze and predict the effects of multiple fluids and ‘patchy’ saturation on elastic moduli and seismic velocities. They also challenge assumptions about the controlling factors on saturated bulk moduli, suggesting they are more strongly affected by the spatial fluid distribution properties and wave scattering, than by pore-scale fluid flow effects.