Utilizing Full Waveform Inversion to Improve Reservoir Imaging and Inversion in East Breaks, Gulf of Mexico

Abstract

Abstract

Ray-based tomography usually provides adequate velocity updates when the velocity model is relatively simple. Yet, the presence of small-scale velocity anomalies complicates ray tracing and poses difficulties for ray-based tomography. Full waveform inversion (FWI) has the potential to resolve high-resolution lateral velocity variations. In this paper, we used high-frequency FWI on an East Breaks data set to derive a high-resolution velocity model that conformed to geology and provided improved reservoir images.

The study area covers approximately 100 km², where we observed a roughly 1-km-thick mass-transport complex (MTC) below the seabed. Its complex dips and internal structures not only impede proper imaging of two deeper reservoir events, but could contain shallow hazards for drilling.

The starting model for FWI was derived from ray-based tomography that honored tilted transverse isotropy. While the starting model provided a reasonable background velocity trend and generated flat gathers in most places, the tomography-derived model failed to provide sufficient resolution for the small-scale velocity anomalies within the MTC. As a result, the events below the velocity anomalies appeared distorted and out of focus.

We performed multiple iterations of FWI from 3 Hz to 18 Hz using both diving wave and reflection energy. As the frequency increased, the resulting velocity model progressively revealed more details. As expected, most of the velocity anomalies within the MTC were slower than the background velocity in the starting model. Identification of slow velocity anomalies can help pinpoint potential shallow hazards for drilling. In addition, slower velocities at the reservoirs, associated with hydrocarbon accumulation, were also identified and closely matched the well logs.

The migrated stack image using the FWI-derived velocity model showed that the two reservoir events were less distorted and more focused than with the tomography-derived velocity model. Furthermore, anomalous amplitude stripes at the reservoir tops resulting from heterogeneity within the MTC were visibly diminished. Non-hyperbolic move-out on migrated gathers was reduced, and the resulting AVO responses at the reservoir levels exhibited a simple linear trend. Finally, prestack p-impedance inversion at both reservoirs appeared more interpretable for quantitative analysis.

AAPG Datapages/Search and Discovery Article #90260 © 2016 AAPG/SEG International Conference & Exhibition, Cancun, Mexico, September 6-9, 2016