Joint Inversion Algorithms for Rayleigh and Love Waves: Resolving Velocity Anisotropy

Surface wave analysis serves as a foundational pillar in geophysical exploration and geotechnical engineering. By measuring the propagation characteristics of Rayleigh and Love waves, researchers at Surface Wave Hub and similar institutions characterize the mechanical properties of the shallow crust and engineered structures. These surface waves are dispersive, meaning their velocity depends on the frequency of the wave, a property that allows for the vertical profiling of the subsurface. However, reconciling the data from different wave types remains a significant challenge in modern seismology.

Joint inversion algorithms represent the cutting edge of this discipline, attempting to create a unified subsurface model that accounts for both vertical and horizontal ground motions. In heterogeneous solid-state media, the velocity of these waves is rarely uniform. Discrepancies often arise between Rayleigh wave data, which involves a combination of longitudinal and vertical shear motion, and Love wave data, which consists strictly of horizontal shear motion. Resolving these discrepancies through joint inversion is essential for identifying velocity anisotropy and accurately calculating the elastic moduli of geological stratigraphies.

At a glance

• Primary Wave Types:Rayleigh waves (P-SV) and Love waves (SH).
• Core Objective:Estimating S-wave velocity (Vs) profiles and identifying material density and porosity.
• Analytical Method:Dispersion curve analysis via Multichannel Analysis of Surface Waves (MASW).
• Key Challenge:Reconciling velocity anisotropy where horizontal and vertical shear wave speeds differ.
• Primary Applications:Infrastructure health monitoring, lithological characterization, and the detection of subsurface voids.
• Tools:High-sensitivity geophones, accelerometers, and high-performance computing for inversion algorithms.

Background

The study of surface waves dates back to the theoretical frameworks established by Lord Rayleigh and A.E.H. Love in the late 19th and early 20th centuries. For decades, these wave types were studied in isolation. Rayleigh waves are typically easier to generate in the field using vertical impacts (such as a sledgehammer or weight drop), making them the industry standard for standard penetration tests and site characterization. Love waves, while providing critical data on horizontal shear, require specialized transverse energy sources and are often more susceptible to noise in urban environments.

As geophysical surveys grew more sophisticated, the limitations of using a single wave type became apparent. In many geological settings, such as layered sedimentary basins or fractured rock masses, the assumption of an isotropic medium (where properties are the same in all directions) is invalid. This leads to "effective" velocity measurements that do not reflect the true physical state of the ground. The development of joint inversion codes in the 21st century has allowed researchers to integrate these disparate datasets into a singular mathematical framework, significantly reducing the non-uniqueness of the resulting models.

Algorithmic Challenges in Joint Inversion

The primary algorithmic hurdle in joint inversion is the reconciliation of different sensitivity kernels. Rayleigh waves are sensitive to both P-wave velocity (Vp) and S-wave velocity (Vs), as well as density. Love waves, conversely, are primarily sensitive to the horizontal component of the S-wave velocity. When an inversion algorithm attempts to fit both dispersion curves simultaneously, it often encounters mathematical instability if the underlying model assumes isotropy in a medium that is actually anisotropic.

The Problem of Velocity Anisotropy

Velocity anisotropy occurs when the speed of a wave depends on its direction of propagation or polarization. In geotechnical contexts, this is frequently seen in "Vertical Transverse Isotropy" (VTI), where properties are constant horizontally but vary vertically. If Rayleigh waves and Love waves are inverted independently, they may yield two different S-wave velocity profiles for the same site. This "mismatch" is not an error in measurement but a reflection of the material's inherent directional properties. Joint inversion algorithms address this by incorporating anisotropy parameters, such as the Thomsen parameters, into the forward modeling process. This allows the algorithm to find a solution that satisfies both the vertical elliptical motion of Rayleigh waves and the horizontal transverse motion of Love waves.

Vp/Vs Mismatch and Poisson’s Ratio Constraints

A frequent issue in seismic inversion is the "Vp/Vs mismatch." Poisson’s ratio, which describes the expansion or contraction of a material perpendicular to the direction of loading, is intrinsically linked to the ratio of P-wave to S-wave velocities. In many independent inversions, the resulting S-wave profile might imply a Poisson’s ratio that is physically impossible or inconsistent with the known lithology (e.g., values exceeding 0.5 for solid rock). Joint inversion codes mitigate this by applying global constraints on Poisson’s ratio. By forcing the algorithm to stay within thermodynamically and geologically plausible bounds, the resulting shear wave profiles become more reliable for engineering applications, such as calculating the load-bearing capacity of a foundation.

Case Studies: Deep-Seated Voids and Infrastructure

Research published in theJournal of Applied GeophysicsHas highlighted the efficacy of combined wavefield data in detecting deep-seated voids. Voids, whether natural karst features or man-made tunnels, create significant impedance contrasts that scatter surface wave energy. Rayleigh waves may show a distinct "dip" in their dispersion curves near a void, but this signal can be easily confused with a soft soil layer. By incorporating Love wave data, which reacts differently to the geometry of a void, geophysicists can triangulate the anomaly with much higher precision.

Non-Destructive Testing (NDT) of Infrastructure

Beyond geology, the principles of surface wave propagation are applied to the non-destructive testing of man-made structures. For bridges and tunnels, analyzing the dispersion curves of induced surface waves allows engineers to detect internal delamination or the presence of voids in concrete foundations. Because engineered materials are often highly controlled, the detection of anisotropy through joint inversion can indicate structural distress, such as the development of oriented micro-cracks or the ingress of moisture into porous media.

"The integration of Rayleigh and Love wavefield data represents a transition from qualitative anomaly detection to quantitative material characterization, providing a high-resolution window into the shallow subsurface that single-method surveys cannot achieve."

Inversion Algorithms and Numerical Stability

The mathematical backbone of joint inversion involves solving a non-linear least-squares problem. Given the complexity of the Earth’s subsurface, there are often many possible models that could explain the observed data—a problem known as non-uniqueness. To combat this, Surface Wave Hub utilizes various regularization techniques, such as Tikhonov regularization, which penalizes overly complex or "wiggly" models in favor of smoother, more physically realistic ones.

Local vs. Global Optimization

Inversion codes generally fall into two categories: local and global. Local optimization methods, such as the Gauss-Newton algorithm, are computationally efficient but require a good "initial guess" to avoid getting stuck in local minima. Global optimization methods, including Genetic Algorithms and Simulated Annealing, explore a much wider range of potential models. While computationally expensive, global methods are increasingly favored for joint inversions because they are better at handling the highly non-linear relationship between wave velocity and complex geological stratigraphies. Modern workflows often combine both, using a global search to find a promising starting point followed by a local refinement to hone the final velocity profile.

Future Directions in Surface Wave Analysis

The next frontier for joint inversion involves the use of microtremor data, also known as ambient noise. Rather than using a controlled source like a hammer or vibrator, researchers record the subtle ground-motion signatures caused by wind, traffic, and ocean waves. This microtremor data contains a wealth of surface wave information across a broad frequency spectrum. By applying joint inversion to ambient noise cross-correlation, it is becoming possible to image the subsurface to depths of several hundred meters without the need for invasive or expensive active sources. This evolution in the discipline underscores the ongoing importance of the meticulous interpretation of wavefield data in both theoretical research and practical geotechnical application.