Value of elastic FWI in derisking clastic reservoirs in presence of noise

Abstract In complex geologic environments, overburden noise often interferes with primary seismic reflection signals, leading to erroneous subsurface property predictions. Conventional inversion methods struggle with this interference, especially when amplitude variation with offset/angle is compromised. The issue lies in the input data, as migrated angle stacks with elastic noise are incompatible with conventional physics assumptions. This paper presents a 3D real-data example from a clastic setting where overburden noise interferes with far- and ultra-far-offset seismic responses. Despite using established broadband seismic processing workflows, the noise resulted in bright far-offset amplitudes, misleadingly suggesting hydrocarbon-bearing sands. Well results confirmed wet sands, and synthetic well ties showed significant mismatches, bringing the predictive capability of conventional seismic data into question. This mismatch was our starting point for this research. We used an anisotropic viscoelastic simulation to understand noise interference with the signal. This study provided important insights into noise-signal coupling at the reservoir. We demonstrate that 3D elastic full-waveform inversion (EFWI) can directly estimate reservoir elastic parameters by fitting the amplitude and phase of both signal and noise in raw shot gathers without the well information. The EFWI parameters, P-impedance and VP/VS, enable derisking fluid prediction at the reservoir with higher probability. Our results show that EFWI provides a more accurate depiction of subsurface elastic properties compared to conventional depth-migrated angle stacks.