Statistical Analysis of the Depth-Velocity Trade-off in Reflection Seismology

Rafael da Silva Garcia, Francisco Márcio Barboza · 2026

Read Statistical Analysis of the Depth-Velocity Trade-off in Reflection Seismology by Rafael da Silva Garcia, Francisco Márcio Barboza (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-04328.txt.

Introduction

Rafael da Silva Garcia and Francisco Marcio Barboza quantify a classic ambiguity in reflection seismology in a 2026 arXiv paper. A layer's depth and its wave speed trade off against each other, so quite different underground models can produce nearly identical travel times for reflected signals. Interpreters need to know how wide that ambiguity really is.

Their approach is statistical and computational: thousands of synthetic models with random depth-velocity pairs are run through Monte Carlo simulation, and each one's travel times are scored against a reference model. The results confirm broad regions of parameter space where the error stays nearly flat, meaning the data alone cannot choose between the models. A two-layer geological example closes the paper, tying the measured ambiguity back to a concrete structure.

Catalog record

CREATORRafael da Silva Garcia, Francisco Márcio Barboza
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE12 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-04328

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Statistical Analysis of the Depth-Velocity Trade-off in Reflection Seismology Rafael da Silva Garcia, Francisco Márcio Barboza Abstract Seismic interpretation is strongly influenced by the relationship between subsurface layer depth and velocity. Small variations in these parameters can produce almost identical responses, characterizing the depth-velocity trade-off phenomenon. This work proposes a statistical and computational approach to evaluate the extent of this effect through Monte Carlo simulations of thousands of synthetic models. The analysis involves generating random depth–velocity pairs, computing travel times, and calculating the root-mean-square error (RMSE) relative to a base model. Results show that multiple parameter combinations yield nearly indistinguishable travel times, confirming the presence of wide ambiguity regions. Finally, a two-layer geological velocity model is presented to illustrate the structural–velocity relationship. K eywords Reflection Seismology \cdot Depth-Velocity Trade-off \cdot Ambiguity \cdot Python 1 Introduction Reflection seismology is an essential technique for characterizing the Earth’s subsurface. Its basic principle involves measuring the travel times of seismic waves reflected at geological interfaces, allowing the estimation of layer depths and velocities [ 10 , 1 ] . However, this estimation is not unique: different combinations of depth ( z ) and velocity ( v ) can generate similar travel times, a phenomenon known as the Depth-Velocity trade-off . Such ambiguity can lead to erroneous interpretations, especially when auxiliary data (such as well logs or seismic inversion) are unavailable [ 2 , 8 ] . Global inversion and optimization studies show that the resolution of the seismic inverse problem is often non-unique, making it necessary to incorporate probabilistic constraints and statistical modeling [ 9 , 5 , 7 ] . This work aims to statistically quantify the ambiguity associated with this trade-off . Through random simulations and error analysis (RMSE), we evaluated the frequency and distribution of models that produce responses nearly indistinguishable from a base scenario. 2 Methodology The proposed methodology is based on numerical simulations implemented in the Python language, widely used in geosciences due to its flexibility and robustness for data analysis and scientific visualization [ 6 , 4 ] . The analysis of temporal similarity is based on the root mean square error (RMSE), a concept widely used in seismic inversion studies and geological model fitting [ 9 , 5 ] . 2.1 Base Model The base model consists of a horizontal reflector at 400 m depth and a velocity of 2000 m/s. The travel time is calculated according to: T(x)=\sqrt{\left(\frac{2z_{0}}{v_{0}}\right)^{2}+\left(\frac{x}{v_{0}}\right)^{2}} (1) where x represents the offset between source and receiver. This type of formulation is traditional in velocity analysis and forms the basis of the hyperbolic approximation used in seismic processing [ 10 ] . 2.2 Random Model Generation 5000 pairs of (z,v) were generated with values uniformly distributed in the following intervals: • Depth: 200–600 m • Velocity: 1000–3000 m/s The z/v ratio was used as an indicator of ambiguity, as models with values close to that of the base model ( z_{0}/v_{0}=0.2 s) tend to generate equivalent travel times. This type of statistical approach reflects the concept of non-uniqueness in probabilistic inversion [ 9 ] . 2.3 RMSE Calculation For each model, the travel time T_{i}(x) was calculated and compared with the travel time of the base model T_{0}(x) . The root mean square error is given by: RMSE=\sqrt{\frac{1}{N}\sum_{j=1}^{N}(T_{i}(x_{j})-T_{0}(x_{j}))^{2}} (2) Low RMSE values indicate ambiguous models, i.e., different (z,v) pairs capable of reproducing the same observable response [ 9 ] . 2.4 Monte Carlo Simulation Algorithm To formalize the analysis process, the workflow is summarized in Algorithm 1 . This pseudocode describes the iterative logic for model generation and error quantification, which is the basis of the Monte Carlo simulation. Algorithm 1 Monte Carlo Simulation for Trade-off Analysis 0: Z_{\text{range}} , V_{\text{range}} , N_{\text{samples}} , X_{\text{vector}} , Z_{0} , V_{0} . 0: RMSE_{\text{vector}} 1: Calculate T_{\text{base}} using Z_{0} , V_{0} over X_{\text{vector}} (Eq. 1). 2: for i=1 to N_{\text{samples}} do 3: V_{i}\leftarrow Generate uniform number in V_{\text{range}} . 4: Z_{i}\leftarrow Generate uniform number in Z_{\text{range}} . 5: T_{i}\leftarrow Calculate travel time with Z_{i} and V_{i} (Eq. 1). 6: \text{DIF}\leftarrow T_{i}-T_{\text{base}} . 7: RMSE_{i}\leftarrow Calculate RMSE(\text{DIF}) (Eq. 2). 8: Add RMSE_{i} to the vector RMSE_{\text{vector}} . 9: end for 10: return RMSE_{\text{vector}} 2.5 Computational Tools and Optimization The implementation of Algorithm 1 utilized the NumPy library for performance optimization. Instead of using the explicit iterative loop described in the pseudocode, the actual code was rewritten using the vectorization and broadcasting technique of NumPy. This optimization (seen in the Python code through the manipulation of z_{\text{col}} , v_{\text{col}} , and x_{\text{row}} dimensions) allows the calculation of the travel time matrix ( T_{\text{models}} ) to be performed in a parallel and efficient manner. This design consideration was fundamental to ensure the speed and robustness of the Monte Carlo simulation with 5000 samples. 3 Results and Discussion The results obtained from the Monte Carlo simulation not only confirm the existence of the depth-velocity trade-off but also quantify the extent of the ambiguity region in the parameter space (v,z) , using the RMSE error as a metric of indistinguishability. 3.1 Geological Model and Conceptual Ambiguity The synthetic Geological Model presented in fig. 1 serves as a structural reference for the analysis. It is composed of a homogeneous upper layer of 2000\text{ m/s} (defining the v_{0} of the Base Model) and a lower layer of 5000\text{ m/s} , separated by the reflector at 400\text{ m} . It is important to note that the trade-off analysis in Section 3 is performed only on the parameters (z,v) of the upper layer, as per Equation 1, since the velocity of the lower layer does not influence the reflection travel time in a single-layer formulation. Figure 1: Synthetic two-layer geological model, with interface at 400 m. The z/v ratio is a direct indicator of the travel time at zero offset ( T(0) ). Figure fig. 2 shows the distribution of this ratio. Models with z/v between 0.19 and 0.21 (shaded region), which encompass the base value of z_{0}/v_{0}=0.2 s, were classified as potentially ambiguous, representing about 9% of the samples. Figure 2: Distribution of the z/v ratio with emphasis on ambiguous models close to 0.2 s. 3.2 Parameter Space and the Nature of the Trade-off Figure fig. 3 presents the distribution of the tested models in the (v,z) space. The blue band represents the ambiguous combinations according to the z/v criterion, which form a nearly linear band around the ratio z/v=0.2 . This distribution demonstrates that, to maintain an approximately constant reflection travel time, depth ( z ) and velocity ( v ) must be directly proportional, confirming the strong linear correlation inherent to the depth-velocity trade-off [ 2 ] . Figure 3: Distribution of the generated models in the (v,z) space, highlighting the conceptual ambiguity region. 3.3 The Effect in the Time Domain: Hyperbola Overlap Figure fig. 4 illustrates the impact of ambiguity in the observable domain. An overlap is observed between the hyperbolic curve of the base model (red) and dozens of ambiguous models (blue), demonstrating the difficulty in distinguishing the scenarios based only on the measured reflection times. Figure 4: Hyperbola overlap of the base model and ambiguous models ( z/v criterion). 3.4 Error Distribution and Indistinguishability Criterion To quantify ambiguity more rigorously,

Read the complete text