Uncertainty estimates on prestack inversion

Deterministic prestack inversion results represent a minimum of a cost function in some parameter space, and in most cases, the uncertainty or variability of the inversion at this minimum is not addressed. In this study, we calculate spatially varying Cramer-Rao lower bound (CRLB) uncertainty estimates on a set of inversion results from a North Sea injectite field using a spatially varying signal to noise model. The CRLB uncertainty estimates by themselves provide a platform for interpretation of the

inversion results, but using the CRLB estimates and a defined direction of interest in the inversion parameter space, we investigate the P10 and P90 percentile halfspaces along this direction. Finally we perform a lithology classification on the P10, P50 and P90 inversion results and compare the hydrocarbon earing sands lithology probability for the three cases. This provides a range of possibilities for the extent of the hydrocarbon bearing sands within the uncertainty of the inversion results