Thomsen's Parameters

The way seismic waves travel through the Earth holds the key to mapping its hidden structures. For decades, a simplifying assumption was made: that the Earth's subsurface is isotropic, meaning waves travel at the same speed in all directions. However, the reality of geology—with its layered sediments and stress-induced fractures—is one of anisotropy, where properties like wave speed are inherently directional. This directionality can distort our picture of the subsurface, leading to inaccurate maps and misinterpreted data. The primary challenge has been the complexity of describing this phenomenon, which can require up to 21 separate elastic constants, offering little physical intuition.

This article addresses this challenge by exploring the groundbreaking work of Leon Thomsen. We will delve into a simplified yet powerful framework that has become a cornerstone of modern geophysics. In the "Principles and Mechanisms" chapter, you will be introduced to the three intuitive Thomsen's parameters—epsilon (ε), delta (δ), and gamma (γ)—and discover how they govern the strange and fascinating physics of anisotropic wave propagation. Following this, the "Applications and Interdisciplinary Connections" chapter will demonstrate how these parameters are not merely theoretical constructs but are actively used to correct seismic images, deduce rock properties, and guide the quest to unravel the Earth's secrets. To begin, we must first understand the foundational principles of these parameters and the new view of wave physics they unlock.

Imagine a sound wave traveling through a perfectly uniform block of steel. No matter which direction it sets off—up, down, or sideways—it travels at the same speed. The universe, from the wave's point of view, looks the same in every direction. This beautifully simple property is called ​​isotropy​​. For a long time, it was the default assumption for physicists and geophysicists modeling how waves travel through the Earth. It's an elegant picture, and for many purposes, a useful one. But like many simple pictures, it is profoundly incomplete.

The Earth's subsurface is not a uniform block of steel. It's a complex tapestry woven from layers of sediment deposited over millions of years, fractured by tectonic stress, and composed of crystals that are themselves directionally aligned. Think of a piece of wood. It splits easily along the grain but resists splitting across it. Its strength is direction-dependent. This is the essence of ​​anisotropy​​. Seismic waves traveling through the Earth's crust find themselves in a world that is much more like wood than steel. A wave traveling parallel to sedimentary layers might move significantly faster than one trying to cut across them. This directional dependence doesn't just change the speed; it alters the very path that wave energy takes, leading to phenomena that an isotropic model could never predict. To understand the Earth, we must first learn to speak the language of anisotropy.

Describing a fully anisotropic material can be a nightmare. In the most general case, it takes 21 independent elastic constants to fully characterize its behavior. Even for a common and simpler case, called ​​Transverse Isotropy (TI)​​, where there's one special axis of symmetry (like the vertical direction in flat-lying sediments), we still need five constants ($c_{11}, c_{33}, c_{44}, c_{66}, c_{13}$) to describe the physics. While these numbers contain all the information, they offer little physical intuition. What does it mean if $c_{11}$ is 10% larger than $c_{33}$?

This is where the physicist Leon Thomsen provided a moment of brilliant clarity. In 1986, he proposed a new language to describe "weak" anisotropy—the case, common in geology, where the directional differences are present but not extreme. Instead of a handful of cryptic stiffness constants, he gave us three intuitive, dimensionless numbers: $\epsilon$ (epsilon), $\delta$ (delta), and $\gamma$ (gamma). These ​​Thomsen's parameters​​ measure the deviation from a simple isotropic background. If they are all zero, we are back in our perfect, isotropic world. If they are small, non-zero numbers, they tell us exactly how the world is anisotropic.

Let's meet this remarkable trio:

• ​​Epsilon ($\epsilon$)​​: This is the most straightforward of the parameters. It describes the anisotropy of the primary or compressional wave (P-wave). Specifically, $\epsilon$ is the fractional difference between the P-wave stiffness (and approximately, the velocity) in the horizontal and vertical directions. It is defined as $\epsilon = \frac{c_{11}-c_{33}}{2c_{33}}$. A positive $\epsilon$ means that P-waves travel faster parallel to the material's "grain" or layering than across it. This is the first and most obvious thing one might want to know about an anisotropic rock.

• ​​Gamma ($\gamma$)​​: This parameter tells a similar story, but for one of the two types of shear waves, the SH-wave (where particles move horizontally). It is defined as $\gamma = \frac{c_{66}-c_{44}}{2c_{44}}$ and quantifies the fractional difference in SH-wave velocity between the horizontal and vertical directions.

• ​​Delta ($\delta$)​​: This is the most subtle and, in many ways, the most important of the three. Unlike $\epsilon$ and $\gamma$, $\delta$ is not defined by the speeds exactly along the horizontal and vertical axes. Instead, it governs the P-wave's behavior at intermediate angles, especially near the vertical direction. Its definition, $\delta = \frac{(c_{13}+c_{44})^2-(c_{33}-c_{44})^2}{2c_{33}(c_{33}-c_{44})}$, involves the "off-axis" stiffness $c_{13}$. This makes $\delta$ the key parameter controlling the shape of the P-wavefront for waves that are traveling nearly vertically. Since much of seismic exploration involves sending waves down and listening for near-vertical echoes, $\delta$ has an outsized importance in getting the seismic image right. It tells us whether the wavefront is more elliptical or more boxy than we would otherwise expect.

These parameters form a bridge between the abstract underlying physics (the stiffness constants) and the things we can actually measure, like wave speeds in different directions. They give us a simple, powerful formula for the P-wave phase velocity, $v_P$, as a function of the angle $\theta$ from the vertical symmetry axis:

Here, $V_{P0}$ is the vertical P-wave velocity. This elegant expression is the key that unlocks the strange and beautiful world of anisotropic wave propagation.

Here is where the story takes a fascinating turn. In an isotropic medium, wave energy travels in the same direction that the wavefronts are advancing. If you drop a pebble in a still pond, the circular ripples expand outwards, and the energy of each part of the ripple travels radially outwards, perpendicular to the ripple itself. The direction of energy flow (​​group velocity​​) and the direction normal to the wavefront (​​phase velocity​​) are one and the same.

In an anisotropic medium, this is no longer true. The direction of energy flow can diverge from the wavefront normal. Imagine a column of soldiers marching across a muddy field, but the mud is thicker on one side than the other. To keep their line straight (the "wavefront"), the soldiers on the less muddy side must slow down. The line of soldiers will appear to pivot, and the direction the whole column advances will not be the same as the direction each soldier is facing.

This difference between the group angle $\theta_g$ (the direction of the energy path, or ray) and the phase angle $\theta$ (the direction of the wavefront normal) is not just a qualitative idea; it's a direct and calculable consequence of the anisotropy described by Thomsen's parameters. For weak anisotropy, this difference, $\Delta(\theta) = \theta_g - \theta$, can be approximated by a simple relationship:

This tells us that the ray path bends away from the wavefront normal by an amount that depends on the interplay between $\epsilon$ and $\delta$. To first order, the speed of the energy packet is the same as the phase speed, but its direction is different. This is a critical insight for things like travel-time tomography, where we measure the time it takes for energy to travel between two points. The path it took is governed by the group velocity, not the phase velocity.

A powerful way to visualize this is through the ​​slowness surface​​—a plot of the inverse of phase velocity ($s = 1/v$) for all directions. In an isotropic medium, this surface is a perfect sphere. In an anisotropic one, it is a distorted shape. The magic of this surface is that the group velocity vector—the direction of energy flow—is always perpendicular to the slowness surface. If the surface is a sphere, the normal at any point is along the radius, so group and phase directions align. If the surface is not a sphere, the normal vector will generally not be radial, and the two directions will diverge.

What happens if the anisotropy becomes stronger? The P-wave slowness surface, governed by the interplay of $\epsilon$ and $\delta$, can become distorted in dramatic ways. If the conditions are right, the surface can develop "dents" or regions of concave curvature.

This is where the physics gets truly wild. A dent in the slowness surface means there are multiple different phase directions (points on the surface) that share the same normal direction. Physically, this means that energy can travel from a source to a single receiver along several different paths, arriving at different times. This phenomenon is called a ​​travel-time triplication​​. Imagine a single earthquake or explosion, but a seismometer records the P-wave arriving three separate times. This creates enormous confusion for seismic imaging, like trying to take a photograph with a lens that creates triple images.

The boundaries of these triplication zones on a wavefront are points of intense energy focusing, known as ​​cusps​​. They are sharp, horn-like features on the wavefront.

Is there a way to know if this beautiful but chaotic behavior will happen? Again, Thomsen's parameters provide the answer. The key is the ​​anellipticity parameter​​, $\eta$ (eta), defined as:

This single parameter measures the departure of the P-wave slowness surface from a simple ellipsoid.

• If $\eta = 0$ (meaning $\epsilon = \delta$), the anisotropy is called ​​elliptical​​. The slowness surface is a perfect ellipsoid. It is everywhere convex (like the surface of an egg), and triplications cannot occur. The relationship between phase and group velocity is simple and unique.
• If $\eta$ is small, the surface is a small perturbation of an ellipsoid and remains convex. Wave propagation is still well-behaved.
• However, if $\eta$ becomes too large (either positive or negative), the surface develops concave regions, and the door opens to the complexity of triplications and cusps.

This is a remarkable piece of physics: a simple relationship between two parameters, $\epsilon$ and $\delta$, dictates whether the wavefield is simple and uniquely defined or a complex web of multi-pathing arrivals.

Thomsen's parameters are far more than a convenient shorthand. They represent a profound shift in perspective. They provide a bridge connecting the microscopic physics of a rock's elastic stiffness tensor to the macroscopic, measurable behavior of waves traveling through it. They distill the bewildering complexity of anisotropic wave propagation into a few intuitive principles, allowing us to predict, model, and understand a world that is not as simple as a bowling ball. They reveal the hidden relationships between wave speed, energy paths, and wavefront shapes, showing them to be different facets of the same underlying physical reality. This, in essence, is the beauty and power of physics: to find the simple, unifying laws that govern a complex world.

In our previous discussion, we became acquainted with Thomsen's parameters, $\epsilon$, $\delta$, and $\gamma$. We saw them as a wonderfully intuitive language for describing how the very fabric of rock can have a "grain," causing seismic waves to travel at different speeds in different directions. This might seem like a subtle, almost academic, point. But as we are about to see, this subtlety is the key to a world of applications, transforming these simple parameters from a mere description into powerful tools for discovery. Armed with this language, we can begin to peer into the Earth's crust with a clarity that would otherwise be impossible.

Imagine you are trying to create a map of a city, but your ruler shrinks and stretches depending on whether you point it north-south or east-west. Your map would quickly become a distorted mess! This is precisely the problem geophysicists face when mapping the Earth's subsurface. We time how long it takes for seismic waves to travel from a source, bounce off a rock layer, and return to a receiver. If we assume the wave speed is the same in all directions (isotropy) when it is not, our depth calculations will be wrong.

This is not a hypothetical worry. Many of the rocks that form reservoirs for oil, gas, and water, such as shales, are composed of fine, flat particles that settled into horizontal layers over geological time. This layered structure creates a prominent vertical transverse isotropy (VTI), where waves travel faster horizontally than they do vertically. If we use the vertical travel speed to calculate the depth of a layer far from the source—where the wave has traveled a significant horizontal distance—we will place the layer deeper than it actually is. Thomsen's parameters allow us to quantify this effect precisely and correct our maps.

The consequences of ignoring anisotropy become even more apparent when we use sophisticated imaging techniques like seismic migration. Migration is a computational process that takes the jumble of recorded seismic signals and focuses them back to their origin points, creating a cross-sectional image of the subsurface. When we perform this migration assuming the Earth is isotropic, but it is in fact VTI, the resulting image is not sharp. On what are called angle-domain common-image gathers (ADCIGs)—a diagnostic plot where a reflector's depth is plotted against the wave's angle—a single flat rock layer will appear curved. It will either "frown" (curve downwards) or "smile" (curve upwards) away from the center.

This curvature is not just an error; it's a clue! The shape of this curve is a direct signature of the anisotropy. A classic result shows that the imaged depth, $z_{\mathrm{im}}$, as a function of angle $\theta$, relates to the true depth $z_r$ by an expression of the form:

This beautiful formula tells us that the initial curvature at small angles is governed by $\delta$, while the deviation from that simple parabolic shape at wider angles is controlled by the combination $\epsilon - \delta$. So, by measuring the "frown" in our migrated data, we can directly estimate the Thomsen parameters of the rock and use them to re-migrate the data correctly, collapsing the curve into the flat line it should be and bringing the subsurface into sharp focus.

The journey of a seismic wave is written in more than just its travel time. The energy it carries—and how much of that energy reflects off a boundary—tells its own story. The way a reflection's amplitude, or "brightness," changes with the source-receiver distance (offset) is a powerful technique known as Amplitude Variation with Offset (AVO) analysis.

Anisotropy adds a fascinating new chapter to this story. When a wave hits a boundary between two different VTI rocks, the strength of the reflection depends not only on the contrast in their isotropic properties (like density and vertical velocity) but also on the contrast in their Thomsen parameters. In a linearized approximation, the anisotropic part of the reflection coefficient, $R_{PP, aniso}(\theta)$, can be expressed in terms of the changes in $\delta$ and $\epsilon$ across the boundary, $\Delta\delta$ and $\Delta\epsilon$:

This equation reveals that the seismic amplitudes are encoded with information about the rock's fabric. By analyzing how the reflection brightness changes with angle, we can start to distinguish different types of rock interfaces. For example, we might be able to differentiate a porous, gas-filled sandstone from a non-porous one, because the presence of gas can alter the rock's anisotropic response. Thomsen's parameters give us the decoder ring to read this subtle, yet crucial, information.

We have seen that Thomsen's parameters leave their fingerprints all over our seismic data. This naturally leads to the "inverse problem": can we look at the data and deduce the parameters? This is where the true art and science of geophysics lie. It's not as simple as it sounds, because the effects of different parameters can become entangled.

Imagine trying to determine both $\epsilon$ and $\delta$ from travel times. As it turns out, your ability to tell them apart depends critically on your experiment's design. If you only use data from small offsets, where waves travel nearly vertically, the travel times are almost entirely sensitive to $\delta$. You have no leverage to see $\epsilon$. If you only use data from very large offsets, the effects of $\epsilon$ and $\delta$ become mixed up in a way that is difficult to untangle. To uniquely determine both, you need a broad range of offsets, from near to far. This demonstrates a profound principle: what we can know about the world is inextricably linked to how we choose to observe it.

This challenge becomes even greater in modern methods like Full-Waveform Inversion (FWI), which attempts to create a high-resolution model of the Earth by fitting every wiggle of the recorded seismic waveforms. This is a monumentally complex optimization problem. Trying to solve for the background velocity ($v_p$), $\epsilon$, and $\delta$ all at once is a recipe for failure. Instead, a hierarchical strategy is used, which is beautifully guided by the physics of wave scattering. We first solve for the parameter that controls the large-scale, low-wavenumber features of the model. This is typically the velocity, $v_p$, which is most sensitive to the small-angle scattering data that is most robustly recorded. Once we have a good background model, we then solve for the parameter sensitive to the next range of angles, which is $\delta$. Finally, we attempt to resolve $\epsilon$, whose signature is strongest at the widest, most difficult-to-record angles. This strategy—of building a model from coarse to fine—is a general principle for solving complex inverse problems far beyond geophysics.

Furthermore, some things may be simply unknowable with a given experiment. The parameter $\gamma$, which describes the anisotropy of horizontally-polarized shear waves (SH-waves), has essentially no influence on the travel times or amplitudes of P-waves or vertically-polarized SV-waves in a VTI medium. This means that no amount of P-wave data can ever constrain $\gamma$; it is "invisible" to our experiment. To measure it, we must design a different experiment, one that uses S-waves.

A good measurement is not just a number; it's a number with an uncertainty. After we run our sophisticated inversion algorithms, how confident are we in the resulting values of $\epsilon$ and $\delta$? This is the domain of uncertainty quantification.

A powerful way to think about this is through the lens of Bayesian inference. We start with a prior belief about our parameters, which might come from geological knowledge or measurements from a nearby well. This prior has a large uncertainty. We then combine this prior belief with the information from our seismic data. The data serves to reduce our uncertainty, leading to a posterior belief that is more sharply peaked around the true values.

This framework allows us to quantify how different data types contribute to our knowledge. For instance, using only P-wave data at a limited range of angles might leave us with a high degree of uncertainty and a strong "trade-off" (posterior correlation) between $\epsilon$ and $\delta$. But adding just one measurement of a horizontally traveling P-wave, or a single SH-wave measurement, can dramatically shrink the uncertainty and break the correlation, giving us much higher confidence in our final model. This formalizes the intuition that more data, and more varied data, leads to better knowledge.

So far, our story has been about wave speed. But as waves travel through rock, they also lose energy and their amplitude decays—a process called attenuation. Just as speed can be anisotropic, so too can attenuation. A wave traveling along the "grain" of a rock might lose energy at a different rate than a wave traveling across it.

Thomsen's framework can be elegantly extended to handle this phenomenon. By allowing the parameters $\epsilon$ and $\delta$ to be complex numbers, their imaginary parts can be used to describe direction-dependent energy loss. This connects the geophysical model to the deep physics of viscoelasticity and material science. The quality factor $Q$, a measure of how efficiently a material transmits wave energy, now becomes $Q(\theta)$, a function of direction. Its angular variation is controlled by the imaginary parts of $\epsilon$ and $\delta$. This is a beautiful example of how a robust physical parameterization can grow to encompass more complex, real-world physics.

Thomsen's parameters, therefore, are far more than a simple notational convenience. They are a physical language that allows us to understand and correct distortions in our seismic images, to read new information about rock properties from seismic amplitudes, to intelligently design experiments and inversion strategies, to formally quantify the limits of our knowledge, and to connect wave speed to the more subtle physics of energy loss. They are a cornerstone of modern quantitative geophysics, helping to illuminate the dark and complex structure of the planet beneath our feet.