Numerical Solution of the 3-D Travel Time Tomography Problem

Michael V. Klibanova Jingzhi Li,Wenlong Zhang

DOI: https://doi.org/10.48550/arXiv.2209.09420

2022-09-20

Abstract:The first numerical solution of the 3-D travel time tomography problem is presented. The globally convergent convexification numerical method is applied.

Numerical Analysis

What problem does this paper attempt to address?

This paper attempts to solve the three - dimensional Travel Time Tomography Problem (3 - D Travel Time Tomography Problem, 3D - TTTP). Specifically, the objective of the research is to solve this problem by numerical methods, especially by applying the globally convergent convexification method.

### Problem Background

The Travel Time Tomography Problem (TTTP) is a Coefficient Inverse Problem (CIP), which involves recovering the wave velocity distribution inside the medium from the boundary measurement data. In this paper, the author focuses on the TTTP in the three - dimensional case and proposes the first numerical solution method.

### Specific Problem Description

Given a three - dimensional region \(\Omega\) and a source surface \(S\) outside \(\Omega\), wave sources are placed on a part \(L_{\text{src}}\subseteq S\). Let \(c(x)\) be the wave propagation velocity and \(m(x)=n^{2}(x)\), where \(n(x)\) is the refractive index. The propagation time of the wave from the source point \(x_{0}\in L_{\text{src}}\) to the point \(x\in\Omega\) is:

\[
\tau(x, x_{0})=\int_{\Gamma(x, x_{0})}\sqrt{m(y(s))}ds
\]

where \(\Gamma(x, x_{0})\) is the geodesic generated by the Riemannian metric. The Travel Time Tomography Problem can be formulated as: given the boundary measurement data \(g(x, x_{0})\), that is, \(\tau(x, x_{0}) = g(x, x_{0})\), where \(x\in\partial\Omega\) and \(x_{0}\in L_{\text{src}}\), solve for the distribution of the function \(m(x)\) within \(\Omega\).

### Main Challenges

1. **Non - linearity**: The governing equations (such as the Eikonal equation) are non - linear.
2. **Ill - posedness**: CIPs usually have ill - posedness, resulting in multiple local minima and valley phenomena in the least - squares cost function.
3. **Unknown geodesics**: In the actual solution process, the geodesic \(\Gamma(x, x_{0})\) is also unknown.

### Solution

To address the above challenges, the author introduced a globally convergent convexification method. This method constructs a weighted Tikhonov - like least - squares cost function \(J_{\lambda}\) with a Carleman Weight Function (CWF) and minimizes it on an appropriate convex set. When the parameter \(\lambda\) is large enough, this cost function is strictly convex and has a unique minimum point.

### Application and Significance

3D - TTTP has important applications in areas such as seismic wave propagation velocity recovery and non - destructive testing. For example, in geophysics, TTTP can be used to recover the seismic wave propagation velocity inside the earth; in remote building detection, time - resolved electromagnetic data transmitted are used for detection, etc.

In summary, this paper aims to solve the three - dimensional Travel Time Tomography Problem numerically for the first time and proposes a globally convergent convexification method to address the non - linearity and ill - posedness challenges in this problem.