Overcoming Deep-Water Velocity Model Challenges with FWI

Fig 1. Map highlighting the offshore area of the Mahanadi Basin (Kumar et.al. 2016)

Geological Landscape

The Mahanadi Basin is a significant rifted basin formed by the breakup of Gondwanaland during the Late Jurassic to Early Cretaceous period. The offshore basin area covers approximately 80,000 sq. km, and extends into deep waters of the Bay of Bengal (Dangwal et.al. 2008). The basin is bounded onshore by Pre-Cambrian Indian Crystalline Shield rocks to the northwest and is structurally influenced by major fault systems trending ENE-WSW, NNW-SSE, and NNE-SSW, giving rise to horst and graben features (DGH 2025).

North of the Eocene Hinge Zone, the continental basement consists mainly of Precambrian Eastern Ghat granulites and gneisses, while the south features oceanic-type crust likely of Early Cretaceous volcanic origin (e.g., Rajmahal Trap basalts). This basement geometry strongly influences sediment patterns and seismic velocity contrasts.

Key geologic developments:

• Late Albian: Marine transgression reversed paleoslope, initiating shelf-to-slope sedimentation.
• Miocene: Himalayan uplift drove massive sediment influx from the Ganga and Brahmaputra, forming shelf-edge deltas, channel-levee complexes, and basin-floor fans, important reservoir systems.

Over 8 km of Upper Cretaceous to Recent sediments accumulated offshore. Notable lithologic markers include:

• Rajmahal Trap flows separating Lower and Upper Cretaceous units.
• Middle Eocene carbonates as seismic markers near the Eocene hinge zone.
• Extensive Mio-Pliocene channel-levee complexes as key reservoir facies.

Petroleum Systems

Within this setting, multiple petroleum systems are postulated:

• Late Cretaceous–Paleogene System: Originates from syn-rift graben lows where marine conditions facilitated preservation of organic matter in restrictive/anaerobic environments. Source rocks matured predominantly during the Early to Late Miocene (ca. 23–9 Ma), reaching oil generation windows. Reservoirs include syn-rift clastics and carbonate buildups along the Eocene Hinge.
• Paleogene–Neogene System: Characterized by mature Paleogene sediments in the northeast deepwater basin. Reservoir facies are mainly deepwater channels and slope fan deposits capable of forming stratigraphic traps. Vertical and lateral migration through faults and turbidite channels enhance hydrocarbon prospectivity.
• Neogene–Neogene System: Involves younger Mio–Pliocene bathyal turbidite fans and channel-levee complexes with biogenic gas potential in younger channel-fill sediments.

Recent onshore and shallow offshore drilling has confirmed significant hydrocarbon shows, including gas with heavier hydrocarbons in Miocene sandstones and Eocene carbonates, underscoring active petroleum systems.

Key Basin Elements

Key basin elements supporting hydrocarbon accumulation include:

• Source rocks with high TOC values up to 9.5% in onshore Early Cretaceous intervals and mature Paleogene source facies offshore.
• Reservoirs mainly composed of coarse-grained sandstones and fractured volcanic flows within channel-levee complexes and syn-rift clastic sequences.
• Extensive regional cap rocks formed by thick Miocene claystone formations and marine shales.
• Diverse trap types, such as fault-related structural traps in syn-rift sediments, carbonate buildups adjacent to the Eocene Hinge, wedgeouts and pinchouts notably against the 85°E Ridge, and stratigraphic traps within submarine canyon and turbidite channel fill systems.

This integrated geotectonic and stratigraphic evolution framework, combined with multiple working petroleum systems, establishes the Mahanadi Basin as a large, geologically complex, and under-explored hydrocarbon province with significant exploration potential.

S-Cube's FWI Approach

Traditional tomography and early FWI attempts have struggled to create a subsurface model here:

• Narrow-azimuth streamer data limited raypath coverage.
• Absence of ultra-low frequencies made deep velocity updates unreliable.
• Structural sag artifacts mispositioned reflectors, creating drilling uncertainty.
• Channel features and compacted clastics looked similar on seismic amplitudes, risking prospect misinterpretation. amplitudes, risking prospect misinterpretation.

To overcome these limitations, ONGC, S-Cube, and Imperial College London deployed a multi-stage Full-Waveform Inversion strategy built on S-Cube's Adaptive Waveform Inversion (AWI) and Reflection Waveform Inversion (RWI), followed by Least-Square Full Waveform Inversion (LSFWI).

This approach was specifically designed to address the data and geological challenges of the Mahanadi Basin:

AWI (Adaptive Waveform Inversion)

AWI focuses on matching the overall shape and timing of seismic waveforms, which helps to avoid the common problem of cycle-skipping in inversion. By comparing seismic traces using convolutional filters instead of direct amplitude differences, AWI effectively builds a reliable shallow velocity model based on diving waves. This approach ensures a more accurate and stable starting point for deeper imaging.

RWI (Reflection Waveform Inversion)

RWI uses information from reflected seismic waves and their multiples to update the larger-scale, deeper velocity structure. Even when very low-frequency signals are missing, RWI can extract meaningful velocity updates by carefully analyzing subtle differences in reflection timing. This method is especially valuable where diving waves do not penetrate deeply enough, such as in streamer data, helping to refine the velocity model in those challenging zones.

LS-FWI (Least Squares Full Waveform Inversion)

Once AWI and RWI have established a solid velocity foundation, LS-FWI is applied to add high-resolution detail. This stage uses full wave-equation modelling sensitive to amplitude variations to capture fine-scale velocity heterogeneities. In this case, LS-FWI helped to reveal subtle variations within and around the Channel-Levee Complexes (CLCs), improving the geological accuracy of the final velocity model.

Fig 2. The Workflow

This staged approach is computationally efficient, progressively building a reliable macro-model before resolving fine detail — reducing both cost and risk of failure.

What the Results Yielded

The inversion produced a geologically consistent, high-resolution velocity model that revealed details absent in previous efforts:

Gas-Charged Channels

Fig 3. Inverted model of the 1750m depth slice clearly highlighting low-velocity anomalies in the Mio–Pliocene CLCs, consistent with proven gas-charged channels.

Compacted Clastics Differentiated

Fig 4. Inverted model of an arbitrary depth slice revealing High-velocity Oligocene–Miocene clastics.

These localized high-velocity anomalies are interpreted as compacted clastic units, supported by their strong RMS amplitude response. Though they resemble channel features in amplitude and frequency content, their distinct velocity signature allows effective separation.

Structural Accuracy Improved

Fig 5. Initial velocity model (left) and inverted velocity model (right). The inversion successfully flattens the structural sags by incorporating realistic low-velocity anomalies, leading to improved structural positioning and imaging accuracy.

Validated with Wells

Fig 6. Strong agreement between inverted velocities and well log trends

Compared to earlier models, the new velocity field delivers sharper imaging, better stratigraphic boundary definition, and higher confidence in structural interpretation.

Significance of the Results

The high-resolution FWI velocity model is critical in the geologically complex Mahanadi Basin, revealing heterogeneities in Mio–Pliocene channel-levee complexes (CLCs) and turbidite systems that directly inform reservoir quality and distribution. Beyond structural traps like fault-bounded horsts and Late Cretaceous–Paleogene features, the model captures subtle lateral and vertical variations within the CLCs, aiding the evaluation of stratigraphic prospectivity. Improved depth positioning resolves ambiguities caused by lateral velocity variations and structural sags, while supporting integration with basin petroleum system models to characterize source rocks, reservoir facies, and seals.

In sum, the FWI results:

• Corroborate geological and petroleum system models indicating multiple active hydrocarbon systems.
• Enable higher confidence in mapping reservoirs within complex stratigraphic and structural frameworks.
• Provide a velocity framework that underpins safe and effective exploration and development planning.

S-Cube's multi-stage FWI workflow has sharpened exploration imaging in the Mahanadi and offers a scalable approach for other deepwater basins facing similar velocity-model challenges.

References

Questions and Answers