Predominant-Mode Inversion of Surface Waves: Inherently Addressing Inconspicuous Low Frequency Mode Jumps

summary

The gist

Predominant-mode inversion offers a robust and practical approach for surface-wave inversion at sites with strong impedance contrasts.

In short

The predominant-mode inversion method automatically selects Rayleigh waves based on maximum vertical surface amplitude at each frequency, bypassing manual mode selection. This approach successfully recovered accurate velocity contrasts and interface depths from synthetic and real seismic data, outperforming fundamental-mode inversions which tended to overestimate velocities.

Key concepts

Predominant Mode
Instead of choosing a specific wave type manually, this method automatically picks the Rayleigh wave mode that has the largest vertical surface amplitude at every measured frequency. This eliminates errors caused by incorrectly assuming which mode is dominant, leading to more reliable results.
Forward Modeling (Thin Layer Method)
This mathematical technique is used to simulate how surface waves behave based on subsurface properties. It treats the Rayleigh wave dispersion problem as an eigenvalue problem, allowing researchers to calculate both the phase velocities and the actual shape of the wave modes simultaneously.
Mode Jump
This refers to sudden changes in how surface waves behave at different frequencies, often indicating a change in which mode is dominant. The predominant-mode framework is specifically designed to handle these jumps without needing prior knowledge of where they occur.
Dispersion Misfit (md)
This metric measures how poorly the modeled dispersion curve matches the actual observed experimental data. The inversion seeks to minimize this value, using a weighted average of phase velocity differences between the model and the observations to find the best subsurface parameters.

Terminology used across episodes

This episode discusses

The paper

Predominant-Mode Inversion of Surface Waves: Inherently Addressing Inconspicuous Low Frequency Mode Jumps · Read on arXiv

Mrinal Bhaumik, Brady R. Cox

Utah State University

DOI: 10.1016/j.enggeo.2026.108834

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We're looking at the paper: "Predominant-Mode Inversion of Surface Waves".

Mira: Here's the gist of the paper. Predominant-mode inversion offers a robust and practical approach for surface-wave inversion at sites with strong impedance contrasts.

Kai: The predominant-mode framework inherently addresses low-frequency mode jumps by automatically identifying, at each frequency, the Rayleigh-wave mode with the maximum vertical surface amplitude.

Mira: This method allows inversions to proceed without the need for explicit mode indexing by the inversion analyst or unvalidated assumptions about fundamental modal dominance.

Kai: Addressing Modal Ambiguity The study introduces a new strategy where, instead of extracting only phase velocity, complete mode shapes are computed from the forward model to examine their amplitude distribution at the free surface.

Paper summary: Kai: So, we’re diving into the details of this paper, "Predominant-Mode Inversion of Surface Waves: Inherently Addressing Inconspicuous Low Frequency Mode Jumps". To start, the authors explain the core issue they are solving: inversion of Rayleigh wave dispersion data is really tough at sites with strong impedance contrasts where modal energy transitions smoothly from one mode to another at low frequencies.

Mira: They point out that analysts often misinterpret that smooth transition as just a continuation of the fundamental mode, which causes them to overestimate shear wave velocity in deeper layers or get the bedrock depth wrong.

Lev: That’s a common pitfall because when higher modes arise due to strong impedance contrast, they tend to be excited at lower frequencies, making them hard to detect in the experimental dispersion data.

Kai: The paper introduces this predominant-mode inversion framework as a way around that by automatically selecting the Rayleigh wave mode with the maximum vertical surface amplitude for every frequency you test.

Mira: They claim that this selection ensures an incorrect mode will inevitably lead to higher misfit because its phase velocity simply won't match what you recorded in the dispersion behavior.

Lev: So, instead of making a guess upfront about which mode is active, the method lets the data tell it which one is most likely present at each point in frequency space.

Kai: The methodology for this involves using the Thin Layer Method to turn the Rayleigh wave dispersion problem into an eigenvalue problem so they can calculate both the eigenvalues, which are phase velocities, and eigenvectors, which are mode shapes simultaneously.

Mira: The theoretical surface amplitude associated with a specific mode is written as a function of wavenumber, angular frequency, and the mode index.

Lev: And then the predominant mode is defined simply as the one with the largest vertical amplitude at the surface for each frequency that you are analyzing.

Kai: This definition is what makes it robust because it doesn't depend on any prior knowledge about which mode is dominant or how many modes there are.

Mira: It’s a clever way to handle modal ambiguity, and they show this method works even when the modal energy transition is smooth and continuous due to a lack of spatial resolution in the data.

Lev: And they also mention that by focusing on removing poorly constrained transition-zone frequencies, you stabilize the inversion process because it prevents conflicting modal behavior from happening during fitting.

Kai: The paper goes on to show how this framework handles the target dispersion data well across the usable frequency range, even when most of the inversion error is concentrated at a few isolated frequencies.

Mira: So what we’re seeing here is that you can get good velocity structure recovery even if there are some localized errors, which is better than getting a globally biased result where everything is just off.

Lev: It's about stabilizing the inversion by avoiding those transition frequency bands where the data gets messy.

Kai: This sets up nicely for us to talk about what this actually means for application and how it compares to other methods.

Conclusion: Kai: So we’ve covered a lot of ground on this paper, "Predominant-Mode Inversion of Surface Waves: Inherently Addressing Inconspicuous Low Frequency Mode Jumps". The authors are Mrinal Bhaumik and Brady R. Cox from Utah State University.

Mira: They essentially argue that the main problem with traditional methods is the difficulty in identifying which mode is active at low frequencies when there’s a smooth modal transition due to strong impedance contrasts.

Lev: And they propose this predominant-mode framework as a way to avoid making assumptions about mode dominance upfront and instead lets the data dictate what's dominant at each frequency.

Kai: The implication is that for someone listening, it means you can expect more stable velocity profile recovery when dealing with complex subsurface structures.

Mira: It means that if you’re working on a challenging site, this method provides a robust tool because it minimizes the risk of getting systematically biased results from mode misidentification.

Lev: And since they made the MATLAB code public, it means we can actually try to implement this approach ourselves and see how it runs in practice on real hardware.

Kai: So in short, this paper suggests that by focusing on finding the predominant mode at every frequency is a smart way to get more reliable velocity structure estimates.

More episodes

← Home