We describe the planning, processing, and uncertainty analysis for a marine CSEM survey of the Scarborough gas field off the northwest coast of Australia, consisting of 20 transmitter tow lines and 144 deployments positioned along a dense 2D profile and a complex 3D grid. The purpose of this survey was to collect a high-quality data set over a known hydrocarbon prospect and use it to further the development of CSEM as a hydrocarbon mapping tool. Recent improvements in navigation and processing techniques yielded high-quality frequency domain data. Data pseudosections exhibit a significant anomaly that is laterally confined within the known reservoir location. Perturbation analysis of the uncertainties in the transmitter parameters yielded predicted uncertainties in amplitude and phase of just a few percent at close ranges. These uncertainties may, however, be underestimated. We introduce a method for more accurately deriving uncertainties using a line of receivers towed twice in opposite directions. Comparing the residuals for each line yields a Gaussian distribution directly related to the aggregate uncertainty of the transmitter parameters. Constraints on systematic error in the transmitter antenna dip and inline range can be calculated by perturbation analysis. Uncertainties are not equal in amplitude and phase, suggesting that inversion of these data would be better suited in these components rather than in real and imaginary components. One-dimensional inversion showed that the reservoir and a confounding resistive layer above it cannot be separately resolved even when the roughness constraint is modified to allow for jumps in resistivity and prejudices are provided, indicating that this level of detail is beyond the single-site CSEM data. Further, when range-dependent error bars are used, the resolution decreases at a shallower depth than when a fixed-error level is used.
Bayesian inversion of marine CSEM data from the Scarborough gas field using a transdimensional 2-D parametrization