A SHALE Exclusive by Ameur Hamdane
Imagine drilling into a deepwater reservoir with the confidence that comes from knowing that every twist of the well path is guided by an ultra-clear image of the subsurface. This is the promise of Full Waveform Inversion (FWI), a seismic imaging technique revolutionizing the oil and gas (O&G) industry. First proposed by visionary geophysicists in the 1980s, FWI was once deemed computationally impractical. Today, thanks to massive advances in computing power and algorithms, it has evolved into a game-changing tool for exploration and production. FWI enables geoscientists to extract unprecedented detail from seismic data – resolving elusive geological features and reducing drilling risk in ways earlier methods could never achieve.
FWI’s impact is being felt from the deepest Gulf of Mexico subsalt plays to the most challenging onshore fields. It is helping companies pinpoint drilling targets under complex salt sheets, correct for harsh near-surface distortions, and even monitor producing reservoirs over time. Crucially, FWI is also translating into business value: better well placement, fewer surprises, and improved production performance.
Differentiating FWI
FWI is an advanced method of building a high-resolution image (or model) of the earth’s subsurface by harnessing the entire seismic wavefield recorded in a survey. Traditional seismic processing workflows – like travel-time tomography or migration – use only portions of the data (for example, the first-arriving waves or reflected echoes) to infer subsurface velocities. FWI, by contrast, tries to use all the information in the seismic traces, including subtle amplitude and phase details, to directly calculate the rock properties that produced those recordings. It does this by iteratively adjusting a computer-simulated earth model so that its synthetic seismic data matches the field-recorded data as closely as possible. The result is like focusing a camera: previously blurry subsurface features snap into a much sharper velocity picture.
The practical value of FWI is a step-change in resolution and accuracy for seismic imaging. Standard velocity modeling might smear out a salt boundary or miss a narrow high-velocity streak; FWI often can capture those in crisp detail. This means geophysicists can better resolve complex geology – such as the intricate shape of a salt dome, or the rugose top of a carbonate reef – that could make or break an oil trap. By improving the velocity model fed into seismic migration, FWI sharpens the final seismic image used by interpreters and drilling teams. This enables FWI to help solve key problems that have long plagued oil and gas explorers: uncertain depths in subsalt plays, fuzzy imaging below gas clouds or basalt layers, and poorly understood near-surface distortions on land. Each of these problems can translate into real business risks – whether it’s a well landing off-target due to velocity errors, or missing a subtle trap altogether. FWI is mitigating those risks by illuminating the subsurface with unprecedented clarity.
Equally important, FWI-derived models often reveal geologic details directly. Unlike a standard velocity model that is just a means to an end (migration), an FWI velocity volume can itself hint at lithology changes or fluid effects. For example, FWI might highlight a low-velocity anomaly in the shallow section that turns out to be a gas pocket or weak zone to avoid when building a platform. In deep reservoirs, subtle velocity slowdowns could indicate high pore pressure zones or remaining oil saturation. This direct information from FWI can guide drilling decisions and hazard assessments beyond just producing a prettier seismic section. In sum, FWI expands the role of seismic from pure imaging into a tool for quantitative subsurface characterization – bringing interpreters closer to the rock properties and conditions in the field.
Evolving from acoustic beginnings to elastic multi-parameter inversion
FWI’s journey from academic curiosity to industry mainstay reflects steady advances in physics and scope. Early implementations in the 1980s and 90s were mostly research prototypes, often two-dimensional and based on simplifying assumptions. The classic starting point was acoustic FWI – treating seismic propagation as if the earth had no shear waves, only compressional waves. This made computation more tractable and worked well for offshore data dominated by P-waves, typically updating only P-wave velocity. By the 2000s, with greater computing power, acoustic FWI proved its value on marine field data, especially long-offset surveys providing low-frequency refracted waves. By the mid-2010s, diving-wave FWI for P-wave velocity had become standard in marine velocity model building, sharpening images in areas like the subsalt Gulf of Mexico and North Sea chalk plays.
But the earth is not acoustic – especially onshore. Real seismic data include shear waves, mode conversions, surface waves, and attenuation. Ignoring these can limit FWI’s success or create artifacts. Elastic FWI arose to incorporate shear (S-wave) velocity and often density, but it is far more data-hungry and computationally intensive. Full elastic-wave modeling and absorption remain rare in routine industry use due to cost, so many projects still rely on acoustic FWI with pre-processed data to suppress elastic arrivals. Yet, momentum toward elastic FWI is growing. Researchers and advanced projects have shown its ability to capture near-surface complexities on land and delineate features such as gas clouds or fracture zones in seabed node data. Elastic FWI provides a richer picture: simultaneous P- and S-wave velocity models and insights into rigidity or fluid content from VP/VS ratios.
Adding parameters introduces trade-offs, as seismic waves can blur velocity and density effects. Multi-parameter FWI addresses this with strategies like hierarchical inversion and geological constraints to stabilize updates. In practice, successes include a North Sea case where inverting for both velocity and density identified a gas-saturated zone that velocity-only inversion would have smeared. These advances are pushing FWI from a one-parameter tool toward a fuller inversion of subsurface physics.
Another leap is reflection FWI, which extends velocity updates to depths unreachable by transmitted waves. Classical FWI relies mainly on diving or head waves, restricting updates to shallow and mid sections. Reflections, however, bounce off deeper interfaces and can inform velocities at reservoir scale. The challenge lies in their sensitivity to deeper structure not yet known – a chicken-and-egg problem. Recent algorithms and workflows solve this by separating smooth velocity updates from reflectivity, isolating travel-time shifts of reflections as tomography-like signals. Reflection FWI has successfully refined deep targets, such as velocity below salt or within reservoir intervals after diving-wave FWI plateaued. This extends FWI’s reach to true reservoir depth, well beyond the traditional limits of acquisition aperture.
FWI Meets AI: Faster, Smarter Inversion
One of the key factors that enabled FWI to flourish in the 2010s was better computing hardware – clusters and cloud systems able to handle the immense number-crunching required. Looking ahead, much attention is now on artificial intelligence (AI) and machine learning (ML) to accelerate and enhance inversion workflows. AI is emerging as a natural ally for FWI in several ways:
- Speeding up computations: Neural networks are being trained as surrogates for parts of the FWI process. For example, a deep neural net can predict a velocity model directly from seismic shot gathers, acting as an “emulator” of FWI. While these networks don’t replace physics-based inversion, they can provide physics-guided starting models much closer to the solution, reducing the number of iterations. Others are using Fourier neural operators, which learn to solve PDEs, to perform forward modeling and inversion at a fraction of the cost of traditional simulators. At Lawrence Livermore Lab, a trained neural operator produced FWI results orders of magnitude faster than iterative inversion, albeit with some fidelity loss. The vision is that after a heavy training phase, such AI tools could enable near real-time FWI updates, with networks inferring velocity changes without simulating every wave propagation.
- Improved convergence and robustness: FWI’s “Achilles heel” is false minima from cycle skipping. Machine learning helps by regularizing or guiding inversion. One method trains a neural net on realistic velocity models and uses it as a constraint, preventing geologically implausible results. Another approach uses deep learning for bandwidth extension, estimating missing low frequencies critical for convergence. In one experiment, a self-supervised model generated low-frequency trends that allowed FWI to start from a better place and avoid cycle skips that band-limited data would have caused.
- Workflow automation: Beyond the core math, FWI involves practical choices – shot selection, denoising, frequency progression, switching from acoustic to elastic. AI can assist by monitoring results and suggesting adjustments, such as switching metrics or injecting randomness if inversion stalls. AI-driven tools also interpret FWI output, classifying anomalies (e.g., shallow gas) to integrate velocity models with geological knowledge more effectively.
AI is transforming FWI by combining physics-based accuracy with machine learning speed, making workflows more automated and reducing reliance on scarce experts. As software integrates AI and cloud services, high-end FWI is becoming accessible beyond major companies, lowering entry barriers and democratizing its use across the industry.
The road ahead: FWI’s future in the energy industry
While FWI has come a long way, its next chapter may promise more meaningful use across a broader canvas of applications. Looking ahead, several developments are poised to shape the next chapter of FWI in the O&G industry, as well as in the wider realm of energy and geoscience.
- Routine Use and Real-Time Updates: Just as 3D seismic itself went from novel to standard practice, FWI is likely to become a routine part of subsurface studies. We can expect even real-time or iterative updating of models while drilling. For example, with the advent of cloud computing and edge devices, one can imagine running a localized FWI on new seismic-while-drilling data or on streaming field recordings, to continually refine the velocity model ahead of the bit. Some pilot projects are already using FWI on time-lapse VSP (vertical seismic profile) data gathered during drilling pauses, to update pore pressure models in near-real time. This proactive use of seismic inversion could significantly improve well safety and targeting. Real-time FWI will require extremely efficient algorithms (where again AI may help) and tight integration of acquisition and processing, but it aligns with the industry’s push for adaptive drilling and reducing non-productive time through better information.
- Beyond Oil and Gas – the Energy Transition: As the podcast title “AI, FWI, and the Future of Subsurface Imaging” (Society of Exploration Geophysicists, Episode 248) suggests, the role of FWI and advanced seismic is expanding beyond traditional oil and gas exploration. Carbon capture and storage (CCS) projects need high-resolution imaging of injection sites to ensure carbon dioxide (CO₂) is contained; FWI can provide the detailed velocity and potentially even monitor CO₂ plume movement with 4D seismic. Geothermal energy exploration benefits from FWI by characterizing fracture zones and fluid pathways in hot reservoirs where conventional methods struggle to provide clarity. Even emerging areas like hydrogen storage in salt caverns require precise imaging of cavern integrity and caprock – again a task where full-wavefield methods excel. In essence, FWI is becoming a go-to technique for any subsurface characterization problem that demands detail and accuracy, reinforcing that reservoir characterization is central not just to oil and gas, but to our clean energy future.
- Multiphysics and Joint Inversions: In the future, we will also likely see FWI synergize with other geophysical data. For instance, joint inversion of seismic and electromagnetic (EM) data could combine FWI’s structural detail with EM’s fluid sensitivity to better discern lithology and saturation. Efforts are underway to do “FWI -style” inversion on different wave phenomena – even on gravity data or controlled-source EM – borrowing the computational advances from seismic FWI. The ultimate goal is a more holistic earth model that honors all data types, improving confidence in interpretations for complex reservoirs. In drilling terms, that means fewer surprises: if FWI (seismic) suggests a low velocity (possible gas), but EM suggests no resistivity anomaly (so likely just low-pressure water), a drilling engineer can plan accordingly. These integrated models will be important as easy reservoirs are gone, and the remaining ones have subtle, mixed signals that no single method can fully resolve.
- Greater Physics: Anisotropy and Attenuation: On the algorithmic front, tomorrow’s FWI will incorporate even more physics. Anisotropy (the directional dependence of seismic velocity) is prevalent in subsurface due to layering and fractures, and incorporating anisotropic parameters into FWI can further sharpen images and align them with well data. In fact, some of the latest FWI case studies include updating anisotropy as part of the inversion, which can correct mis-ties in depth conversion and lead to better well placement confidence. Similarly, accounting for attenuation (energy loss) in FWI – solving for a Q factor or using viscoelastic simulation – could allow the inversion to also highlight absorption anomalies (often associated with gas or fractures). While these add computational load and complexity, steadily improving computing resources and algorithm efficiency make it plausible that in a few years a “full physics” FWI (viscoelastic, anisotropic, multi-parameter) might be run on large 3D surveys as routinely as today’s acoustic FWI.
- User-Friendly FWI and Workforce: The human factor is also important. As FWI becomes more common, there is momentum to make it more user-friendly for geoscientists. Expect software interfaces that let you set up an FWI run with guided parameters, and visualization tools that allow you to easily see how the model is updating iteration by iteration. Uncertainty estimation is another area of growth – future FWI results might come not just as one model, but with an ensemble of possible models or uncertainty volumes, so that decision-makers can gauge the risk (e.g., “95% chance the top of that high-speed layer is within ±10 feet”). Because training the next generation of geophysicists in these tools is paramount, industry and professional societies are addressing technical education through workshops and new graduate programs focusing on computational geophysics.
FWI, clearly optimizing subsurface exploration
As FWI matures, it is becoming a cornerstone technology for subsurface exploration and monitoring. In merging computational science with field applications, FWI is now delivering accurate models that underpin better decisions. For O&G operators, this reduces uncertainty, optimizes well trajectories, and improves production strategies by illuminating the subsurface with unparalleled clarity.
Beyond hydrocarbons, FWI is vital for carbon storage, geothermal development, and hydrogen storage, exemplifying how seismic advances serve both oil and gas and the energy transition.
Thus, by harnessing the full wavefield, FWI provides clarity where earlier methods left uncertainty. With ongoing advances in algorithms, computing, and AI, the future points to routine, real-time, and multi-physics inversion. For the energy industry, FWI represents not only better images, but better decisions, safer drilling, and deeper understanding of the reservoirs that drive our future.
About the author:
Ameur Hamdane is an accomplished Senior III Geophysicist and Project Leader at one of the world’s leading oilfield services companies. With 15 years of experience in all aspects of advanced seismic processing and imaging, including high-resolution imaging for both marine and land datasets in complex environments, from onshore Europe to the deepwater Gulf of Mexico, he is widely recognized for his leadership in global geophysics and for fostering innovation and collaboration across international teams. Ameur is a member of the Society of Exploration Geophysicists (SEG) and the American Society of Civil Engineers (ASCE), and an organizing member of the Geophysical Society of Houston (GSH). He has published numerous technical case studies and guidelines, and contributes to the advancement of geophysical research and standards as a peer and journal reviewer for SEG and ASCE.
Stay In The Know with Shale
While the world transitions, you can count on Shale Magazine to bring me the latest intel and insight. Our reporters uncover the sources and stories you need to know in the worlds of finance, sustainability, and investment.
Subscribe to Shale Magazine to stay informed about the happenings that impact your world. Or listen to our critically acclaimed podcast, Energy Mixx Radio Show, where we interview some of the most interesting people, thought leaders, and influencers in the wide world of energy.