Search

Using modern HR FWI for reflection data up to 15Hz and above, in a shallow land environment with limited offset, is a challenge. In particular, without an accurate shallow initial model, HR FWI will have a limited reliability to initiate its process of frequency increase causing convergence difficulties. The need for an accurate enough initial velocity and anisotropy model to feed this process is key to obtain an initial good match between FWI modelled and observed energy. This, not only for diving but also reflection energy. This publication, through the Tilenga processing case study, describes innovative developments and methods to obtain such an initial model, as well as the results with HR FWI.

The main reservoirs in the Balder and Ringhorne areas consist of large Paleocene sand mounds as well as Eocene sands, including thin and steeply dipping injectite bodies. Imaging of the targets in the legacy towed-streamer data is challenging due to sub-optimal signal-to-noise ratio, kinematic distortions caused by the complex overburden, and limited illumination provided by the narrow-azimuth surveys. Recent acquisition of the dense Heimdal Terrace OBN survey provided access to full azimuth, longer offsets, and higher trace density. Through state-of-the-art processing of the up-going wavefield and high-frequency velocity model building in conjunction with better illumination from the OBN data, a significant uplift was achieved in AVO attributes and structural imaging, with less wave-fronting noise that obscured the mapping of targets in the legacy data. In the near surface, the shallow illumination was further expanded by joint primary and multiple imaging, to enhance lateral and vertical resolution and reduce the acquisition footprint to a negligible level.

Permanent reservoir monitoring at Snorre provides high-quality 4D information from PP waves. However, due to the compartmentalized geology, pressure and fluid-saturation effects in the reservoir are difficult to disentangle. We describe 4D processing and imaging of PS mode-converted waves to complement the PP dataset and enable better separation of these effects. The monitoring system at Snorre achieves high PP-wave 4D repeatability using receivers buried in a trench. Unfortunately, the PS-wave data then suffers from a receiver-side trench ghost which can be a significant source of non-repeatability and strong 4D noise when local conditions around the receiver change between acquisition vintages. We designed a careful trench deghosting sequence utilizing a layer-based trench modelling method, cascaded passes of surface-consistent deconvolution, and several rounds of statistical matching. The flow involves separately processing X and Y components until sufficient 4D consistency is obtained for combination into the radial component for imaging. Simplification of the data through deconvolution of X and Y by the downgoing wavefield significantly improved the accuracy of trench deghosting overall. Although PS images have lower frequency than the accompanying PP data, combining PP and PS 4D responses allows better discrimination between pressure and fluid saturation effects than with PP data alone.

The Laconia program marks a major advancement in the design and execution of long-offset, low-frequency (LOLF) ocean-bottom node (OBN) seismic acquisition programs for subsalt imaging in the U.S. Gulf. With more than 18,000 km²of source coverage and more than 8000 node locations, this multiphase program builds on recent breakthroughs in full-waveform inversion (FWI) methodologies, sparse OBN geometry, and new source technology.

This paper explores the potential of leveraging legacy streamer data from the Utsira High in the North Sea to derive valuable exploration insights, despite its limitations. The study focuses on overcoming key geophysical challenges posed by shallow geological features like sand injectites and the highly reflective chalk package overlying the reservoir. These features, characterized by significant elastic contrasts, generate converted waves and complicate velocity model building, especially given the limited 6 km offset of the dataset. Elastic full-waveform inversion (E-FWI) is shown to effectively address these challenges by fully utilizing the available offset range and capturing kinematic information that acoustic FWI cannot. By integrating locally inverted reservoir properties to account for elastic heterogeneities at the injectite level, the E-FWI velocity model is further refined. This improvement is enabled by advanced workflows in deterministic elastic inversion and machine learning-based injectite detection. Extending E-FWI to 45 Hz reveals subtle velocity slowdowns at the target reservoir level beneath the chalk, enabling accurate delineation of reservoir intervals and faults while providing new insights into reservoir characterization.

Imaging deep targets underneath highly heterogeneous overburden using data from non-ideal acquisition is always challenging. In Block SC38, offshore Philippines, the seismic energy propagation is disrupted by a complex volcanic system in the overburden, leading to washed-out imaging beneath the volcano. The absence of diving waves further complicates velocity modeling, and heterogeneous velocity variations challenge depth imaging. An accurate velocity model and advanced migration approach are both crucial for deep target imaging in this area. This paper demonstrates how the combination of Dynamic-Resolution Time-lag Full-waveform Inversion (DR-TLFWI) and Least-squares PSDM (LS-PSDM) addresses these challenges. DR-TLFWI optimizes contributions from low- and high-wavenumber components, improving velocity model accuracy for deep targets beyond the reach of diving waves. With the improved velocity, Least-squares PSDM (LS-PSDM) further compensates for illumination loss and enhances amplitude fidelity in complex subsurface conditions. The results demonstrate improved event focusing and gather flatness in the target reservoir, highlighting the effectiveness of DR-TLFWI with LS-PSDM for sub-volcanic imaging.

Many fields in the Brazilian pre-salt have been historically imaged with narrow-azimuth towed-streamer (NATS) data and only recently have been surveyed with ocean bottom nodes (OBN). Some cases of time-lapse 4D seismic monitoring are limited to hybrid pairs of data, where a NATS baseline acquisition was performed before production start and an OBN data set was acquired later for monitoring. This leads to low repeatability, imposing many challenges for extracting reliable 4D signal, especially in the context of the subtle changes expected from pre-salt carbonate reservoirs. Considering this, we developed a novel joint 4D full-waveform inversion (FWI) formulation to attenuate the adverse effects of low repeatability due to diverse acquisition geometries. Results are demonstrated using field data from the Santos Basin, revealing potentially genuine 4D anomalies.

New long-offset, low-frequency (LOLF) OBN acquired using the innovative TPS low-frequency source provides transformational imaging of the subsalt geology in the complex and inboard, minibasin domain of the central Gulf of Mexico (GOM). The role of low-velocity (LV) shale bodies in the deformation, which fundamentally controls exploration in this domain, has, until now, been under-recognized. The new data, even in the early-out test, is already providing the basis for significantly improved understanding on deepwater Paleogene. Further processing, with the latest elastic FWI, is likely to provide data that will drive renewed exploration activity in this previously challenging part of a prolific basin

Inspired by the successes of sparse ocean bottom nodes (OBN) programs in recent years, especially the industry's first long-offset, low-frequency (LOLF) sparse OBN survey using the Tuned Pulse Source (TPS) as the seismic source (Merritt et al., 2024), Viridien acquired Laconia OBN, its vast multi-client survey of nearly 800 OCS blocks of source coverage to provide significant uplifts to the full-waveform inversion (FWI) velocity update and imaging of underlying staggered-source full-azimuth data. Despite several weather events impacting this 9-month long 5-vessel operation, the project was executed successfully. Starlink was used to transfer hydrophone data down-sampled at 16 ms upon node retrieval offshore, enabling early validation of the data quality through FWI tests and providing interpreters a 3-month lead time to analyze geological features on full-area 5 Hz OBN acoustic FWI (AFWI) before the Laconia Phase I Fast Track volume release.