Next Article in Journal
The Wind Parks Distorted Development in Greek Islands—Lessons Learned and Proposals Toward Rational Planning
Next Article in Special Issue
Experimental Evaluation of Residual Oil Saturation in Solvent-Assisted SAGD Using Single-Component Solvents
Previous Article in Journal
A LightGBM-Based Power Grid Frequency Prediction Method with Dynamic Significance–Correlation Feature Weighting
Previous Article in Special Issue
A Review of Enhanced Methods for Oil Recovery from Sediment Void Oil Storage in Underground Salt Caverns
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Shear Wave Velocity in Geoscience: Applications, Energy-Efficient Estimation Methods, and Challenges

by
Mitra Khalilidermani
1,
Dariusz Knez
1,* and
Mohammad Ahmad Mahmoudi Zamani
2
1
Department of Drilling and Geoengineering, Faculty of Drilling, Oil, and Gas, AGH University of Krakow, 30-059 Krakow, Poland
2
Iranian Mining and Industry Organization, Ahvaz, Iran
*
Author to whom correspondence should be addressed.
Energies 2025, 18(13), 3310; https://doi.org/10.3390/en18133310
Submission received: 11 April 2025 / Revised: 6 June 2025 / Accepted: 18 June 2025 / Published: 24 June 2025
(This article belongs to the Special Issue Enhanced Oil Recovery: Numerical Simulation and Deep Machine Learning)

Abstract

Shear wave velocity (Vs) is a key geomechanical variable in subsurface exploration, essential for hydrocarbon reservoirs, geothermal reserves, aquifers, and emerging use cases, like carbon capture and storage (CCS), offshore geohazard assessment, and deep Earth exploration. Despite its broad significance, no comprehensive multidisciplinary review has evaluated the latest applications, estimation methods, and challenges in Vs prediction. This study provides a critical review of these aspects, focusing on energy-efficient prediction techniques, including geophysical surveys, remote sensing, and artificial intelligence (AI). AI-driven models, particularly machine learning (ML) and deep learning (DL), have demonstrated superior accuracy by capturing complex subsurface relationships and integrating diverse datasets. While AI offers automation and reduces reliance on extensive field data, challenges remain, including data availability, model interpretability, and generalization across geological settings. Findings indicate that integrating AI with geophysical and remote sensing methods has the potential to enhance Vs prediction, providing a cost-effective and sustainable alternative to conventional approaches. Additionally, key challenges in Vs estimation are identified, with recommendations for future research. This review offers valuable insights for geoscientists and engineers in petroleum engineering, mining, geophysics, geology, hydrogeology, and geotechnics.

1. Introduction

Surface waves, which are generated near Earth’s boundaries, propagate parallel to the surface. These waves can be categorized based on the orientation of ground movement [1,2]. The velocity of shear waves is a crucial feature that characterizes the mechanical features and stiffness of subsurface formations. The velocity of shear waves serves as a pivotal factor in gaining insight into the response of underground formations across diverse geological and geotechnical contexts [3]. Shear waves, alternatively termed S-waves, represent seismic waves that travel horizontally or transversely through the Earth, contributing significantly to our comprehension of subsurface dynamics [4]. They travel slower than primary waves (P-waves) but faster than the surface waves. Vs is essential for elucidating the elastic characteristics of rocks. If we consider an isotropic elastic medium, the relationship between Vs, shear modulus (G), and density ( ρ ) could be expressed as [5]:
V S = G ρ
Based on the above equation, it can be seen that the Vs is directly related to the square root of shear modulus and inversely related to the square root of density.
Shear wave velocity is indeed a critical factor within geoscience, with lots of applications [6,7,8,9,10,11,12,13,14,15,16,17,18]. Earthquake engineering [6,7], characterization of geotechnical sites [8,9], reservoir evaluation [10,11], and analysis of groundwater resources [12,13] are conventional applications of shear wave velocity. Furthermore, new applications of Vs data encompass exploration for geothermal energy [14], assessment of landslide hazards [15], utilization in carbon capture and storage (CCS) projects [16], evaluation of geohazards in offshore environments [17], and exploration into the depths of the Earth’s subsurface [18]. In this research, a concise overview has been given to those applications.
Despite the numerous applications of Vs, there are some crucial challenges in predicting shear wave velocity in geoscience. For instance, issues such as a lack of adequate data, spatial and temporal variability, non-unique relationships, anisotropy and inhomogeneity, difficulty in integration of multiple data sources, different machine learning and AI techniques, interpretability, uncertainty quantification, and the need for standardization and collaboration can be mentioned. Those challenges are also elaborated in this research.
Precise determination of Vs holds paramount importance in both geoscience and geotechnical engineering. Conventional approaches to predicting Vs, such as laboratory analyses, tend to be labor-intensive, costly, and frequently unfeasible for expansive studies [19]. Therefore, researchers have been exploring energy-efficient and cost-effective ways of predicting shear wave velocity, taking advantage of advancements in geophysical and geotechnical data analysis techniques. For this purpose, geophysical methods, remote sensing techniques, and machine learning algorithms have been utilized to measure/predict the Vs [20,21,22].
In this research, an inclusive overview has been given to the different applications, energy-saving estimation techniques, and challenges related to accurate Vs prediction in geoscience and geoengineering fields. The objective of the research is to collate and integrate the most important aspects of Vs application and prediction from a multidisciplinary viewpoint. Thus, in this research, the different Vs applications and potential challenges have been elaborated in several disciplines, including petroleum engineering, geotechnics, hydrogeology, earthquake engineering, geology, mining engineering, etc.

2. Conventional Applications

The significance of shear wave velocity spans different disciplines within geoscience and geotechnical engineering, with its applications ranging from earthquake engineering to groundwater resource assessment. In earthquake engineering, Vs serves as a cornerstone in seismic hazard evaluation and the designation of earthquake-resistant structures [6,23]. Studies have explored its correlation with ground motion amplification, facilitating ground motion prediction and site response analysis [24]. Vs also aids in seismic site classification, crucial for assigning sites with appropriate design parameters [25]. In geotechnical site characterization, Vs plays a pivotal role in evaluating soil stiffness, liquefaction potential, and seismic site response [26]. It assists in soil characterization for design purposes, informing foundation design and slope stability analysis [27,28,29,30].
Reservoir characterization benefits significantly from Vs, particularly in hydrocarbon exploration, where it provides insights into lithology of formations. Vs aids in identifying lithological variations within reservoirs [31], estimating porosity [32], evaluating fluid content [33], and analyzing mechanical properties [34]. Shear wave velocity is vastly utilized in groundwater resource assessment to determine the properties and characteristics of aquifers [11]. It provides valuable information about the subsurface conditions and helps in understanding groundwater potential, flow behavior, and storage capacity [12]. Groundwater resource assessment utilizes Vs for aquifer characterization, groundwater flow modeling, seawater intrusion assessment, and groundwater resource mapping. Vs helps in assessing the properties of aquifers, like porosity and permeability [35] and hydraulic conductivity [36], facilitating groundwater flow modeling and mapping groundwater resources in different geological formations [12].
Conventional approaches of Vs prediction are commonly impractical for large-scale investigations due to their time-consuming and expensive nature. Therefore, there is a growing need for alternative approaches to estimate Vs accurately and efficiently, especially in the context of expansive geotechnical and geological investigations [6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36].

3. New Applications

The emerging applications of shear wave velocity (Vs) prediction span various domains within geoscience and geotechnical engineering, showcasing its versatility and significance in contemporary research. In geothermal energy exploration, Vs prediction serves crucial roles in reservoir characterization [37], fracture detection [38], resource potential mapping [39], and enhanced geothermal system (EGS) development [40]. By estimating Vs, researchers can assess the mechanical properties, fluid content, and heat transfer capabilities of subsurface formations, aiding in optimizing geothermal well design and operation [37,38,39,40].
Landslide hazard assessment benefits from Vs prediction, enabling landslide susceptibility mapping [41], early warning systems [42], slope stability analysis [43], and modeling of landslide-induced ground motion [44]. Vs estimation facilitates the evaluation of mechanical properties and slope stability, contributing to effective mitigation strategies and risk management [41,42,43,44].
In carbon capture and storage (CCS) projects, Vs prediction finds utilizations in storage formation characterization [45], caprock integrity assessment [46], reservoir monitoring, leakage detection [47], site selection, and risk assessment [48]. By estimating Vs, researchers can evaluate subsurface properties, containment integrity, and potential CO2 leakage pathways, enhancing the safety and efficacy of CCS initiatives [45,46,47,48].
Geohazard assessment in offshore environments relies on Vs prediction for seabed stability assessment [49], identification of submarine landslide hazards [50], offshore infrastructure design [51], and seismic hazard analysis [52]. Vs estimation aids in evaluating mechanical properties, detecting geohazards, and optimizing offshore infrastructure performance and safety [49,50,51,52].
Deep Earth exploration benefits from Vs prediction in mantle convection studies [53], seismic discontinuity identification [54], anisotropy studies [55], and core structure investigations [56]. Researchers can infer the seismic structure, understand mantle dynamics, and investigate phenomena, such as seismic anisotropy and core–mantle coupling, using shear wave velocity estimation [53,54,55,56].

4. Energy-Efficient and Cost-Effective Approaches

In this section, a concise description is given to the most energy-efficient and cost-effective methods of Vs measurement or prediction. As already mentioned, those methods include the geophysical methods, remote sensing technique, and machine learning algorithms. In what follows, those methods are described.

4.1. Geophysical Methods

Geophysical methods have emerged as valuable tools for shear wave velocity estimation, providing economical and environmentally friendly alternatives to conventional laboratory experiments and field measurements. In what follows, the applicable geophysical methods used for Vs prediction are elaborated.

4.1.1. Seismic Refraction Techniques

Originally developed for large-scale seismic studies, such as earthquake and seismotectonic research, seismic refraction is now extensively used for near-surface applications. Controlled seismic signals help map subsurface structures, supporting hydrogeological, engineering, and environmental studies [44,57].
Seismic refraction is a commonly applied geophysical technique for estimating Vs via analyzing the times of travel of S-waves and P-waves in the subsurface. In other words, both of these waves are determined by analyzing how fast they travel through different subsurface layers, based on the arrival times recorded at various points. This technique enables the construction of depth-dependent Vs profiles, aiding in geological and geotechnical investigations [58]. For instance, Sarkar et al. (2021) applied multichannel analysis of surface waves (MASW) alongside seismic refraction tomography (SRT) to achieve high-resolution Vs profiling in the Bhutan Himalaya, demonstrating its efficiency and cost-effectiveness [58].
As it was mentioned before, seismic waves include P-waves, which propagate faster and travel through both solids and liquids, and S-waves, which move more slowly and only through solids. While P-waves are more commonly used in refraction surveys due to their higher velocity and ease of detection, S-waves provide valuable insights into unconsolidated materials [45,59]. This method measures the duration for seismic waves to propagate through subsurface layers and reach multiple receivers at different distances from the source [59]. As it can be seen in Figure 1, the data obtained can identify velocity discontinuities, which help in predicting geological boundaries and material types.

4.1.2. Surface Wave Analysis

Surface wave analysis is a widely used geophysical technique for estimating Vs by measuring the dispersion characteristics of surface waves, particularly Rayleigh waves (R-waves). These waves are sensitive to subsurface Vs variations and can be measured at the surface without requiring intrusive methods. This makes surface wave analysis an affordable and non-destructive approach for characterizing the sites of projects. Nazarian was of the first who introduced the spectral analysis of surface waves (SASW) technique, which has since been commonly applied [60]. Boore (2006) reviewed progress in surface wave analysis and emphasized its potential for Vs estimation, while also highlighting challenges in extracting reliable dispersion curves and selecting appropriate inversion methods [61]. The technique typically involves three key stages: initially, seismic data are acquired, followed by dispersion curve estimation and, finally, inversion to derive the Vs profile.
One of the main advantages of R-wave-based techniques is that these waves dominate when generated by a vertical source, allowing effective Vs estimation. Steady-state testing was the initial approach, in which a monochromatic surface wave was produced by a vertically acting vibrator. Early studies determined wavelength and phase velocity by measuring phase differences between receivers [54,55,62,63]. However, early methods were limited by monochromatic sources and depth approximation assumptions. The introduction of spectral methods, such as SASW, enabled broader frequency analysis, improving depth resolution but requiring complex data processing.
Despite its advantages, surface wave analysis has several limitations. The inversion process is highly complex and influenced by the ambiguity of the solution, meaning different models can represent the data with equal accuracy. Additional uncertainties arise from experimental data errors, model assumptions (e.g., 1D isotropic elastic models), and parameterization choices. While a single best-fitting Vs profile may suffice for some applications, deeper layers are more challenging to resolve due to decreasing sensitivity at greater depths [2].
Lateral variations in seismic properties also pose challenges, as surface wave methods typically assume a 1D subsurface structure. These methods are most effective when applied to sites with minimal lateral heterogeneity and relatively flat ground surfaces. Additionally, higher modes in surface wave propagation can complicate the analysis. If not properly accounted for, they can be misinterpreted as fundamental modes, leading to errors in the velocity profile. However, simultaneous inversion of fundamental and higher-order modes can enhance the reliability of results by incorporating additional independent constraints [2]. Since no standardized procedures exist for handling higher modes, customized analyses by experienced professionals are often required [2,56].
Given these considerations, careful data acquisition and expert interpretation are essential for accurate Vs estimation using surface wave analysis. While commercially available software simplifies processing, improper use by non-experts can lead to significant errors, reducing trust in non-destructive techniques within parts of the earthquake engineering community [2].

4.1.3. Seismic Tomography

Seismic tomography is a robust imaging technique that reconstructs three-dimensional models of the Earth’s inner structure through seismic wave travel times and waveforms analysis. Similar to medical CT scans, this method utilizes seismic waves produced by earthquakes or explosions to infer subsurface structures. In a homogeneous Earth, seismic rays would propagate along straight paths; however, due to the planet’s complex layering, waves undergo reflection and refraction at material boundaries, altering their paths [64]. By recording seismic wave energy at multiple stations, researchers can detect anomalies in wave velocity, revealing variations in subsurface properties (Figure 2).
This technique has been widely applied to shear wave velocity estimation. Travel time tomography and waveform inversion are commonly used to develop detailed subsurface velocity models. For instance, Li et al. (2012) employed a simultaneous inversion of surface wave dispersion data together with receiver functions to construct a high-resolution three-dimensional Vs model, demonstrating seismic tomography’s effectiveness in capturing velocity variations in complex geological settings [65].
Other studies have also highlighted the capability of seismic tomography and related geophysical techniques in Vs estimation. Pilz et al. (2012) [66] presented a one-step inversion method to derive 3D S-wave structures using high-frequency correlation functions from ambient seismic noise. The approach, tested through simulations and real-world data, successfully mapped local subsurface heterogeneities. Manu-Marfo et al. (2019) [67] used ambient noise tomography to develop a 3D model of Vs and Moho topography for the Tyrrhenian Basin. Results revealed a broad low-velocity zone in the uppermost mantle, reflecting past tectonic evolution. The findings support an exhumed mantle basement beneath the Vavilov Basin and ongoing Africa–Eurasia convergence.
In essence, while all three techniques involve the use of seismic waves to probe the Earth’s subsurface, they differ in their methods of wave generation, propagation, and analysis, leading to different types of insights into Earth’s structure and properties.
These advancements in geophysical methods underscore their practical applications in geoscience and geotechnical engineering. Techniques such as seismic refraction, seismic tomography, and surface wave analysis provide efficient, non-invasive, and environmentally friendly means for Vs estimation. However, each method has its limitations, which must be considered when selecting the most suitable approach. Table 1 summarizes the key advantages and disadvantages of these techniques in Vs prediction. In this table, we can also see the depth range and spatial resolution of each method. Surface wave analysis offers intermediate spatial resolution for estimating shear wave velocity. It is not as high as some other techniques, like seismic refraction [68,69]. Surface wave analysis is most commonly used for estimating shear wave velocities in the near-surface environment, typically within the first 10 to 100 m [70]. Seismic tomography’s spatial resolution for estimating Vs can vary depending on the data acquisition and processing methods. Generally, it is not very high in comparison with other methods. However, it can still provide a useful representation of the subsurface structure. The depth of seismic tomography typically ranges from a few meters to a few kilometers. It is also used to image deeper structures, even to the Earth’s core, but the resolution and detailed imaging capabilities generally decrease with the depth [71].
Recent advancements in geophysical methods for estimating Vs are significantly enhancing subsurface characterization. For example, advancements in seismic interferometry have allowed for the reconstruction of subsurface velocity structures using passive seismic noise. This technique reduces the need for active sources and enhances the resolution of shallow subsurface imaging [72]. In addition, the compilation of uniformly processed global shear-wave-splitting data has provided a more consistent and extensive database for interpreting shear wave velocity structures, aiding in the assessment of subsurface anisotropy and heterogeneity [73].
Key future directions include the following: (1) Combining machine learning algorithms with established geophysical techniques, such as surface wave analysis and seismic tomography, is improving the accuracy and efficiency of Vs estimations. For instance, hybrid models that integrate full-waveform inversion with deep learning are being developed to enhance seismic imaging and subsurface characterization [74]. (2) Development of advanced inversion techniques: Innovative inversion methodologies, such as probabilistic neural network tomography, are enabling more accurate and rapid estimation of shear wave velocities from surface wave dispersion data, facilitating large-scale subsurface imaging [75]. (3) Enhanced data acquisition technologies: The deployment of advanced geophysical tools, including high-frequency surface wave methods and continuous monitoring systems, is improving the spatial resolution and temporal coverage of shear wave velocity measurements, leading to more detailed subsurface models [76]. These innovations are poised to transform geotechnical and seismological practices by offering more accurate, efficient, and non-invasive methods for subsurface characterization.
Seismic interferometry and comprehensive global datasets can also be mentioned as the two main breakthroughs of geophysical methods for estimating shear wave velocity. Advancements in seismic interferometry have allowed for the reconstruction of subsurface velocity structures using passive seismic noise [77]. This technique reduces the need for active sources and enhances the resolution of shallow subsurface imaging [72]. The compilation of uniformly processed global shear-wave-splitting data has provided a more consistent and extensive database for interpreting shear wave velocity structures, aiding in the assessment of subsurface anisotropy and heterogeneity [73].

4.2. Remote Sensing

Remote sensing is a technique for detecting and analyzing an area’s physical properties from a distance by measuring reflected and emitted radiation, typically using satellite or airborne sensors. This method captures large-scale images of the Earth’s surface, providing visibility beyond what is possible from ground-based observations. For example, satellite cameras can record ocean temperature variations, while sonar systems on ships generate images of the ocean floor without direct access. Figure 3 illustrates the concept of remote sensing.
A major benefit of remote sensing is its application in geological mapping. Unlike conventional methods that are often constrained by limited coverage, remote sensing allows for large-scale data acquisition through satellite imagery and satellite gravity measurements [78,79,80,81,82,83]. Satellite gravity maps the distribution of gravity anomalies to estimate rock density, functioning similarly to terrestrial gravimeters but deployed in orbit [84,85,86,87]. Meanwhile, satellite imagery provides high-resolution and frequently updated surface condition data, offering a dynamic alternative to relatively static gravity-based mapping [88]. When combined with Vs measurements at depths of up to 30 m (Vs30), remote sensing enhances geological assessments by correlating surface conditions with subsurface properties [89]. Since Vs30 serves as an indicator of rock stiffness, it can be theoretically linked to rock density derived from gravity measurements, improving the accuracy of geological models [89].
Beyond geological mapping, remote sensing techniques have proven valuable in geotechnical applications, particularly in estimating shear wave velocity (Vs) [21]. By analyzing surface features and soil composition captured through airborne or satellite imagery, researchers can establish correlations between spectral characteristics, reflectance patterns, and Vs values. For example, Wróbel et al. (2022) integrated surface wave analysis and borehole measurements with airborne imagery, demonstrating the cost-effectiveness and accuracy of this approach in Vs estimation [90]. Similarly, Holland et al. (2005) developed a methodology for estimating sediment density and Vs gradients in subsurface formations by formulating mathematical relationships between remote sensing data and geotechnical parameters [91]. These researchers highlighted the potential of remote sensing for rapid and cost-efficient Vs assessments over large spatial extents.
The benefits of remote sensing in Vs prediction include its non-intrusive nature, reduced need for extensive fieldwork, and capability to effectively span extensive regions [21]. Additionally, remote sensing facilitates temporal monitoring, allowing researchers to analyze changes in Vs over time. However, the precision of Vs estimations is influenced by the resolution and quality of the imagery, as well as the availability of ground truth data for calibration. Atmospheric and geometric distortions may introduce uncertainties, and correlations between surface features and Vs can vary depending on regional geological conditions, necessitating site-specific validation. Table 2 summarizes the key benefits and limitations of remote sensing in Vs forecasting.
Recent advancements in remote sensing techniques for estimating shear wave velocity have enhanced subsurface characterization. Key breakthroughs include seismic interferometry: Utilizing passive seismic noise correlations, seismic interferometry reconstructs subsurface velocity structures without the need for active sources. This method has been applied to image seismic scattering and velocity structures, particularly in volcanic regions [72]. Ambient noise tomography: By analyzing background seismic signals, this technique maps underground geological structures, providing insights into subsurface velocity variations [77]. Looking ahead, the future of remote sensing for Vs estimation is poised to benefit from the integration of multimodal data fusion and advanced machine learning algorithms, enabling more accurate and comprehensive subsurface imaging. Frequency–velocity convolutional neural networks (CNNs): These networks enable rapid, non-invasive 2D Vs imaging of near-surface geomaterials by analyzing normalized dispersion images. They offer flexibility in experimental configurations and have demonstrated effectiveness in imaging soil-over-bedrock interfaces [92].
Deep neural networks (DNNs) integrated with rock physics modeling: This approach combines theoretical rock physics models with DNNs to predict Vs across various frequency bands. It has shown superior accuracy compared to traditional empirical formulas, with prediction errors within 200 m/s in real field data applications [93].
For example, one of latest advancements in remote sensing and AI for Vs is the investigation conducted by Meng et al. Their study highlighted the growing role of artificial intelligence in enhancing the analysis of remote sensing data and emphasized how techniques like SAR altimetry and 3D imaging radar altimetry, combined with AI tools, have improved the detection and characterization of oceanic phenomena, like internal solitary waves. The integration of AI-driven approaches with remote sensing technologies represents a promising direction for future research in geophysics, as similar methodologies could be applied to Vs estimation by incorporating more advanced machine learning models and deep learning architectures. These developments align with the need for transformer-based models and physics-informed graph neural networks (GNNs) to further enhance the accuracy and reliability of shear wave velocity predictions across diverse geological conditions [94].

4.3. Machine Learning and Data-Driven Techniques

Lately, there has been growing emphasis on machine learning and data-driven methods for their potential in predicting shear wave velocity in various geotechnical and geophysical applications [95]. These methods provide economical and environmentally friendly substitutes for traditional techniques. This section reviews some major previous studies that utilized ML and data-driven approaches for Vs prediction, highlighting their methodologies, performance, and notable findings.
In 2007, Rezaee and colleagues [11] utilized neuro-fuzzy systems, artificial neural networks, and fuzzy logic to predict Vs based on log data from conventional wells in a sandstone reservoir located in the Carnarvon Basin, Australia. The models were trained on two wells and validated on a third, demonstrating high reliability. The similar performance of different intelligent methods confirmed their effectiveness for complex reservoir characterization. Rajabi (2010) [95] applied genetic algorithms and fuzzy logic together with neuro-fuzzy techniques to predict compressional, shear, and Stoneley wave velocities from well log data in the Sarvak carbonate reservoir, Iran. Using 3030 data points, the models were trained and tested, with fuzzy logic showing the lowest prediction errors. The results demonstrated the reliability of intelligent techniques for reservoir characterization. Asoodeh and Bagheripour (2012) [96] introduced a hybrid committee machine approach to forecast sonic wave velocities using log data of conventional wells. This approach combines fuzzy logic, artificial neural networks (ANNs), and neuro-fuzzy algorithms to improve predictive capabilities. Additionally, by incorporating a genetic algorithm combined with a pattern search technique, the model’s accuracy in predictions was further enhanced. Applied to the Asmari Formation in Iran, the approach outperformed individual intelligent systems in reliability and effectiveness.
Maleki et al. (2014) utilized the support vector machine (SVR) for predicting the Vs values in the southern part of Iran [97]. Their methodology utilized seismic data, geological attributes, and information from previous borehole logs. The SVR model exhibited promising accuracy and demonstrated the potential of data-driven approaches in capturing Vs variations. Bagheripour et al. (2015) [98] proposed a method based on the support vector regression (SVR) approach to predict Vs utilizing well log data, addressing limitations in direct measurements. By leveraging structural risk minimization, SVR outperformed neural networks and empirical models in accuracy. The method was successfully applied to carbonate reservoirs in Iran’s gas fields, offering a fast and cost-effective solution. In 2015, other scholars also presented a novel ant colony–fuzzy inference system (ACOFIS) for Vs prediction, integrating fuzzy reasoning with ant colony optimization. Applied to the Cheshmeh–Khosh oilfield, the method outperformed conventional approaches, improving prediction accuracy. ACOFIS also showed potential for estimating other reservoir rock properties [99]. Ataee et al. (2018) [100] studied the correlation between Vs and various soil index parameters, including SPT, soil moisture, fine content, depth, and liquid limit, by employing the ANN method and multiple regression analysis. The study, based on data from Mashhad, Iran, proposed new prediction equations for Vs and found that the ANN method provided more accurate predictions with lower estimation errors and higher correlation coefficients compared to multiple regression analysis.
Anemangely et al. (2019) [101] developed a model for Vs prediction in the Ahvaz field using well log data and feature selection via the Non-Dominated Sorting Genetic Algorithm II (NSGA-II). Chosen optimization algorithms (PSO, GA, and COA) together with the least square support vector machine (LSSVM) showed the LSSVM-COA model to be the most accurate and reliable, outperforming empirical and regression models. After validation of the model in the Ab-Teymour field, its generalizability and high accuracy for Vs estimation were demonstrated. Wang and Peng (2019) [102] proposed another method for estimating Vs using the mean impact value (MIV) and extreme learning machines (ELMs), applied to well log data of two wells located in China’s Ordos Basin. The method identified the most significant logs for Vs estimation and outperformed traditional models like ANN-LM with regard to the precision of prediction and computation time. The proposed ELM-MIV model was shown to be a highly efficient and precise tool for estimating shear wave velocity, with potential for integration into software systems for rapid log acquisition.
Azadpour et al. (2020) [103] combined machine learning and rock physics to improve Vs prediction in some carbonate reservoirs. Through adjustments to Gassmann’s fluid substitution model and introducing a factor exponent, the approach enhanced accuracy. Using inversion-based estimation and Gaussian process regression, the method outperformed conventional models, reducing prediction errors significantly. Lian et al. (2020) [104] integrated machine learning with rock physics modeling to improve the seismic interpretation accuracy. By generating synthetic well logs and testing RF, MLP, and SVR algorithms, the approach enhanced P- and S-wave velocity predictions. The proposed method outperformed conventional rock physics modeling, significantly increasing prediction accuracy. In another study, machine learning techniques were applied, particularly artificial neural networks, to predict Vs for geomechanical modeling in wells lacking direct measurements. The approach outperformed traditional linear models, especially in hydrocarbon-bearing intervals. It was then used to calculate in situ stresses and safe mud weight windows, with predictions closely matching actual drilling events [105]. Wang et al. (2020) [106] presented a hybrid model combining long short-term memory (LSTM) together with particle swarm optimization (PSO) networks to estimate Vs using traditional logs. By employing grey relational analysis to select sensitive logs and optimizing the LSTM model’s hyperparameters, the approach enhanced prediction accuracy and robustness. Tested with real petrophysical data, the model outperformed conventional methods, offering a solution for Vs prediction in oil and gas reservoir studies.
Zhang et al. (2021) [107] utilized the DNN method to estimate the Vs spatial distribution to assess the liquefaction of soil. The hybrid approach showed improved performance, thereby highlighting the potential benefits of integrating ML in Vs estimation. Jeong et al. (2021) [108] introduced a method based on conditional variational autoencoders (CVAEs) to reconstruct shear slowness data from other well logs, like neutron porosity, compressional slowness, gamma ray, and bulk density. Compared to bi-LSTM and LSTM models, the CVAE method provided superior results. Olayiwola and Sanuade (2021) [109] compared three models, including LSSVM, adaptive neuro-fuzzy inference system (ANFIS), and ANN, for predicting Vp and Vs by using wireline log data. R2, or coefficient of determination, can be calculated by subtracting the unexplained variation from the total variation, then dividing the result by the total variation. R2 tells us how much better the model is at predicting the data compared to just using the average of the data. This indicator was used in that research. The LSSVM model outperformed the others, achieving high accuracy, with R2 or coefficient of determination values of 0.9706 for Vp and 0.9991 for Vs. Based on their research, the approach can aid geoscientists and engineers in reservoir characterization and drilling operations. Miah (2021) [110] developed a data-driven model using coupled simulated annealing (CSA) and LSSVM to predict Vs. The study identified the Vp as the most influential parameter for estimating Vs, followed by shale volume, bulk density, and porosity. The proposed model, with a high correlation coefficient (R2 = 0.96), outperformed existing correlations and could be applied for accurate prediction of rock mechanical properties together with wellbore failure analysis, improving drilling safety and reducing exploration costs.
Kim (2022) [111] introduced a hybrid ML approach for generating synthetic shear sonic (DTS) logs, addressing the limited availability of DTS data. The approach combined supervised learning models (LSTM, SVR, and DNN) with data clustering and PSO to enhance prediction accuracy and optimize hyperparameters. The results showed that this integrated method outperformed non-clustered models, providing a more reliable and efficient solution for generating synthetic DTS logs. Wong et al. (2022) [112] explored the use of the ML technique for estimating sonic wave travel time in hydrocarbon exploration, addressing obstacles and benchmarks. They compared regression methods, including curve-fitting artificial neural and multiple linear regression (MLR) networks, for effective travel time prediction. The aim was to connect machine learning experts with oil and gas engineers to enhance prediction accuracy.
Mehrad et al. (2022) [113] proposed a high-accuracy model for predicting Vs in carbonate reservoirs using well log data. By applying the CNN and multilayer extreme learning machines (MELMs) with optimization algorithms, the MELM-COA model showed better generalizability and faster learning times compared to CNN. The proposed method outperformed traditional empirical equations and is recommended for similar fields with larger datasets. Laalam et al. (2022) [114] estimated the Vs values in the Williston Basin by using the random forest regression. For this purpose, they used a combination of remotely sensed geophysical data along with geological information. The model achieved promising results, demonstrating the potential of ML techniques in inferring Vs values. Zhang et al. (2022) [115] predicted the Vs values by the convolutional neural networks (CNNs) technique for a study conducted in Ordos Basin. They trained the model using a large dataset comprising seismic data, geological information, and well-logging data. The CNN-based model showed better performance than the traditional approaches, indicating the effectiveness of deep learning techniques in capturing complex relationships within the data. Jiang et al. (2022) [116] presented a deep learning approach to predict shear wave velocity from conventional logging data in tight sandstone formations, offering insights into efficient subsurface characterization.
Feng et al. (2023) [93] enhanced shear wave velocity prediction by integrating theoretical rock physics models into deep learning. By generating a comprehensive labeled dataset through forward simulation, the model improved physical interpretability and generalization. The proposed deep neural network outperformed traditional empirical formulas, achieving prediction errors below 5% in real-world applications. Zhang et al. (2023) [117] introduced a Gaussian process regression (GPR) approach to estimate the Vs in sandstone reservoirs by utilizing Vp and geological information of the reservoir. Unlike deep learning, GPR effectively handles small datasets while also quantifying uncertainty. Field tests showed that GPR outperformed conventional methods, providing more accurate and reliable S-wave velocity predictions. Kheirollahi et al. (2023) [118] focused on estimating Vs in carbonate reservoirs, where conventional S-wave logs usually were unavailable due to high costs. Various predictive models, including MLR, ANN, and ELM, were tested using conventional well logs. The feed-forward neural network model, with an optimal design found through grid search optimization, achieved the greatest precision, with R-values of 0.99 for training and 0.96 for testing datasets, and is proposed for Vs prediction in other wells. Cova and Liu (2023) [119] introduced a new method for Vs prediction utilizing the bidirectional gated recurrent unit (GCN-BiGRU) combined with a graph convolutional network. Their model captured both spatial and temporal dependencies in well log data and enhanced prediction accuracy by incorporating a self-learning graph neural network and feature generation techniques. The GCN-BiGRU outperformed other machine learning methods and empirical formulas, such as Castagna’s velocity formula, LSTM, and SVR, in predicting S-wave velocity based on the North Sea open dataset.
Akinyemi et al. (2023) [120] applied seven machine learning algorithms (LR, RFR, KNN, SVR, BPANN, CatBoost, and XGBoost) to estimate the shear and compressional sonic logs (DTS and DTP) from conventional wireline information from the Niger Delta Basin. The best performance for DTP prediction was achieved with CatBoost (R2 = 90%) and for DTS prediction with CatBoost (R2 = 95%). The models showed strong performance, with R2 values exceeding 80% for DTP and 90% for DTS, and blind testing validated their accuracy, confirming the potential for estimating geomechanical parameters in wells lacking sonic log data. Mustafa et al. (2023) [121] developed ML models to predict Vs in shale formations using wireline logs, addressing the high cost of direct measurements. Six machine learning techniques were tested, with ANN outperforming the others, achieving the highest accuracy (R2 = 0.96) and lowest errors. The proposed models offer a fast, reliable, and cost-effective solution for estimating S-wave velocity, aiding in the derivation of rock mechanical properties. Khalilidermani and Knez (2023) [122] improved Vs estimation for hydrocarbon reservoirs using well log data from the Aboozar limestone oilfield in Iran. Various methods, including empirical correlations, linear regression, and machine learning (multivariate linear regression and gene expression programming), were tested against actual Vs measurements. The gene expression programming (GEP) model delivered the most accurate predictions (R2 = 0.972), outperforming traditional empirical methods. The study provided a new correlation for precise Vs estimation, applicable to the Aboozar oilfield and similar geological formations.
Fu et al. (2024) introduced a neural network method for Vs prediction in locations lacking this data, aiming to enhance the accuracy of subsurface oil and gas distribution models [123]. By incorporating an attention module to assign weights to input data, followed by processing through a LSTM network, the approach achieved a mean absolute error (MAE) of 38.89 in Well B, outperforming the standard LSTM network’s MAE of 45.35. Gomaa et al. (2024) [124] presented an ANN model for accurately estimating Vs utilizing widely available data, like porosity, bulk density, Vp, and gamma ray. The model, developed using 2350 data points, demonstrated exceptional precision, with a strong coefficient of determination of 0.958, making it a practical tool for well design without the need for specialized software. This method provided a more efficient and budget-friendly solution to traditional methods of estimating Vs.
Dehghani et al. (2024) evaluated the effectiveness of different ML techniques in estimating the shear wave transit time, highlighting the potential of these techniques in geophysical applications [125]. Joshi et al. (2024) [126] developed an artificial intelligence model, VelProfES, designed to predict subsurface Vs profiles without relying on traditional geophysical techniques. The model, trained on bore log and Vs data from Japan’s Kyoshin network, used simple parameters, like layer thickness, SPT-N values, and soil type. The VelProfES model, enhanced with synthetic data from CTGANs, provided accurate predictions with minimal error, in contrast to traditional polynomial and empirical models. Yilmaz et al. (2024) presented a deep-learning-based approach for predicting Vs30 at strong motion station locations using three-channel earthquake records [127]. Leisi and Shad Manaman (2024) described an innovative seismic and well-logging data integration workflow to predict the shear velocity volume using an artificial neural network [128].
These studies underscore the importance of integrating machine learning techniques with traditional geophysical methods to enhance the accuracy and efficiency of S-wave velocity estimation, thereby advancing our understanding of subsurface formations. Estimating Vs is crucial for accurately characterizing subsurface formations, particularly in clastic rocks. Traditional methods often rely on empirical relationships, such as Castagna’s equation [129], which correlates Vp to Vs. Recent advancements have integrated ML methods with traditional geophysical models to enhance the precision of Vs predictions. In the context of Vs estimation, machine learning approaches have shown promise. For instance, combining LSTM with attention mechanisms has enhanced feature extraction from well log data, leading to more precise estimations [130]. Incorporating geographical information, like latitude and longitude, into ML models has also shown the improvement in Vs predictions. This approach accounts for regional geological variations, enhancing the model’s robustness and accuracy [131]. Another latest breakthrough of Vs prediction by AI techniques is the application of XAI techniques, such as SHapley Additive exPlanations (SHAP), which has provided insights into the decision-making processes of ML models. This transparency is crucial for validating model predictions and understanding the influence of different input features on Vs estimation [131].
Looking ahead, the future of AI-based methods in Vs prediction is poised to see further advancements, driven by continued improvements in model complexity, data integration, and computational power, potentially leading to even more accurate and efficient prediction capabilities. For example, combining ML models with FWI techniques is one of the promising methods that aims to leverage the strengths of both data-driven and physics-based approaches for more accurate and efficient subsurface characterization [74]. Additionally, future models will likely incorporate uncertainty quantification methods to assess the reliability of Vs predictions. This development will aid in risk assessment and decision-making processes in geotechnical engineering and seismic hazard analysis [132].
As we can see, according to the literature review, the interest of utilizing ML approaches in Vs prediction has increased in recent years.
Table 3 summarizes the performance of various machine learning models used for predicting shear wave velocity across the studies mentioned in the literature review. The table presents the error estimation metrics (R2, MSE, MAE, and RMSE) for each model, providing insights into the technological evolution of AI-based approaches in Vs estimation. According to this table, the performance of AI models in shear wave velocity estimation has improved significantly over the years. Early studies used relatively simpler methods, such as neuro-fuzzy systems and ANN, with performance metrics like R2 values around 0.94 and MSE between 0.008 and 0.015. However, modern techniques, such as LSSVM, XGBoost, CNN, and hybrid models, show much stronger results, with R2 values approaching 0.9991 and MAE and RMSE values decreasing significantly, reflecting technological advancements in model design, data quality, and algorithm optimization.
Numerous studies have investigated methods for estimating Vs, evaluating their effectiveness through real data, and offering important findings for petroleum studies. However, despite these progressions, challenges remain in selecting appropriate ML algorithms and interpreting their results. ML models often depend on statistical relationships, which may not always align with geological realities, leading to potential inaccuracies. Additionally, the performance of these models is highly sensitive to data quality and the selection of relevant features. To address these challenges, future studies should aim at creating sophisticated predictive models that prioritize key geological features. Incorporating domain-specific knowledge into ML algorithms can enhance their interpretability and reliability. Moreover, employing robust validation techniques and integrating multiple data sources can improve the generalizability of these models across different geological settings. In Table 4, the major merits and demerits of these techniques in predicting shear wave velocity are presented.
Therefore, we can conclude that while machine learning offers promising avenues for S-wave velocity estimation in clastic rocks, careful consideration of algorithm selection, feature importance, and model validation is essential to achieve accurate and reliable predictions.

5. Challenges of Vs Measurement

Predicting shear wave velocity in geoscience is an important task with various challenges. Based on the conducted research, a number of major challenges were detected and categorized by the authors. In what follows, those major challenges are discussed.

5.1. Data Availability and Quality

A significant hurdle in estimating Vs lies in the adequacy and reliability of data. Obtaining extensive data collections of shear wave velocity measurements from diverse geological settings can be difficult due to factors such as limited accessibility, cost, and logistical constraints [113]. To improve data availability, platforms like IRIS [133] and OpenTopography [134] can facilitate data sharing among researchers and institutions. Additionally, OSGeo [135] offers open-source tools for managing and analyzing geospatial data, which can be critical for enhancing the quality and accessibility of data used in shear wave velocity studies. Furthermore, standardizing measurement techniques and data reporting is essential for ensuring consistency and comparability across studies [113]. Initiatives like the IRIS Data Management Center and databases like the ShearWave Velocity Database could play a key role in promoting data standardization. These efforts, coupled with the establishment of standardized reporting guidelines, are helpful for improving the quality and availability of shear wave velocity data [136].

5.2. Spatial and Temporal Variability

Shear wave velocity exhibits significant spatial and temporal variations due to geological heterogeneity, soil compaction, weathering, and other factors. These variations pose challenges in accurately predicting shear wave velocity across different locations and time periods. Capturing the inherent variability of Vs and understanding its effect on prediction accuracy are critical for reliable assessments of site-specific seismic hazards [65]. Integrating high-resolution spatial data from platforms like LiDAR or UAV-based surveys (e.g., OpenTopography Data) can help capture fine-scale surface variations [134]. Temporal variability can be monitored by conducting periodic seismic surveys (e.g., MASW) to track changes due to seasonal soil moisture variations. Additionally, applying geostatistical methods (e.g., Kriging) for spatial interpolation and leveraging machine learning models can improve predictions of Vs across variable terrains and time periods [137].

5.3. Non-Unique Relationships

The relation between the Vs parameter and other geophysical or geological parameters is often not unique. Multiple combinations of parameters can result in similar shear wave velocity values, leading to challenges in accurately predicting Vs based on limited data inputs. Resolving the non-uniqueness of these relationships requires the integration of diverse data and robust development of inversion and modeling techniques [11]. To address this, various techniques focus on reducing the uncertainty and finding more reliable solutions. These include incorporating additional constraints, using different inversion approaches, and refining models based on statistical analysis [138].

5.4. Anisotropy and Inhomogeneity

Geological formations commonly exhibit anisotropy (directional dependence) and inhomogeneity (variations within a region). These characteristics significantly affect shear wave velocity predictions. Anisotropic properties can result from sedimentary layering, preferred orientation of minerals, or stress-induced fabric [139], while inhomogeneity can arise from variations in grain size, compaction, or saturation. Accounting for anisotropy and inhomogeneity requires advanced models and methods to accurately predict shear wave velocity in complex geological settings [110,140]. This issue can be addressed by a combination of advanced modeling techniques, including joint inversion of seismic data, and consideration of geological and rock physics properties is essential [141].

5.5. Integration of Multiple Data Sources

The incorporation of multiple data sources is crucial for improving shear wave velocity predictions. Combining geotechnical measurements, seismic data, remote sensing, and geological information enhances the understanding of subsurface conditions and improves the reliability of shear wave velocity estimation. However, effectively integrating different data types and extracting their complementary information remains a challenge, requiring sophisticated data fusion and integration algorithms [110]. To address this challenge, the main solution can be using data fusion techniques.

5.6. Interpretability and Uncertainty Quantification

Interpreting and communicating the results of Vs predictions are crucial for decision-making in geoscience. Additionally, understanding the uncertainties linked to the predictions and quantifying their influence on risk assessments are critical for accurate hazard evaluation. Efforts to develop interpretable models and establish robust uncertainty quantification frameworks are necessary to enhance the practical applicability of shear wave velocity predictions [142]. The main solution for this problem is to apply machine learning models with built-in interpretability tools, such as SHAP or LIME, to explain model predictions. Additionally, Bayesian approaches or Monte Carlo simulations can be used to quantify uncertainty in predictions by incorporating prior knowledge and propagating errors through the model, providing more robust and reliable Vs estimates [132].

5.7. Standardization and Collaboration

Establishing standardized protocols for shear wave velocity measurement, data collection, and reporting is crucial for improving data quality, comparability, and reproducibility. Collaboration among researchers, industry professionals, and policymakers is essential for addressing common challenges, sharing knowledge and resources, and advancing the field collectively [143]. The solution can be developing and implementing standardized protocols for Vs measurement, data collection, and reporting. This can be achieved through collaboration among researchers, industry professionals, and policymakers to establish common guidelines, improve data comparability, and ensure reproducibility across studies and projects.

5.8. Scale Effects

The Vs predictions may be affected by the scale at which the analysis is conducted. Extrapolating results from laboratory-scale measurements to field-scale conditions or vice versa can introduce uncertainties [144]. Therefore, it is important to employ multi-scale modeling approaches that integrate both laboratory-scale measurements and field-scale conditions. By calibrating laboratory measurements with field data and applying appropriate scaling techniques, we can reduce the uncertainties that arise when extrapolating results between different scales. This approach helps ensure more reliable predictions across varying spatial scales.
As mentioned in the above paragraphs, the Vs predictions in the subsurface formations encounter diverse challenges, especially stemming from the uncertainties in the geological, geotechnical, and structural properties of rocks and soils. Thus, for future works, it is recommended to develop strategies capable of mitigating such challenges.

6. Discussion

The Vs prediction in geoscience and geoengineering is a crucial and challenging task with substantial implications for various applications, including reservoir characterization, seismic hazard analysis, and geotechnical engineering. Despite significant progress, challenges remain in achieving high-accuracy and reliable Vs predictions across diverse geological settings [145]. In this discussion, we will delve into the key findings from the conducted research.
Conventional techniques, such as laboratory-based experiments and field measurements, have been foundational in understanding the physical properties of subsurface materials. These methods, although accurate, are time-consuming, labor-intensive, and often impractical for large-scale studies. However, they remain indispensable in certain contexts, especially for calibrating alternative methods or for specific high-precision tasks. The limitations of these traditional techniques have driven the need for more efficient and cost-effective approaches, with geophysical methods, remote sensing, and machine learning algorithms emerging as the primary alternatives.
Geophysical techniques, such as seismic refraction, surface wave analysis, and seismic tomography, have revolutionized the way Vs is estimated in geoscience. These methods offer significant advantages, including non-intrusiveness, high spatial resolution, and relatively low environmental impact. Seismic refraction, for example, is highly effective in constructing depth-dependent Vs profiles, which are essential for understanding subsurface structures [59]. Surface wave analysis, while traditionally challenging due to the complexity of data inversion, offers an efficient way to estimate shear wave velocity in near-surface environments [61]. Seismic tomography, especially in its advanced forms, like ambient noise tomography, has enabled high-resolution imaging of subsurface velocity variations, even in areas with limited access [65].
However, these methods are not without their drawbacks because they are sensitive to lateral variations in subsurface properties, and their accuracy diminishes with increasing depth and heterogeneity [71]. Seismic tomography, while powerful, demands complex data acquisition and processing, which can be computationally expensive and requires expert interpretation [64]. The evolution of hybrid models that integrate machine learning with geophysical techniques holds promise for overcoming some of these limitations. By combining the physics-based strength of geophysical methods with the data-driven flexibility of ML models, we can achieve more robust and accurate Vs predictions that account for both geological complexities and data uncertainties.
Remote sensing techniques, including satellite and airborne geophysical methods, have proven to be invaluable for large-scale Vs estimation. These techniques offer a non-intrusive and cost-effective way to acquire geophysical data across vast spatial extents. Remote sensing has been particularly useful for estimating Vs at depths of up to 30 m (Vs30) and for correlating surface features with subsurface properties [89]. The ability to collect real-time or near-real-time data without the need for fieldwork makes remote sensing an attractive option for environmental monitoring, resource exploration, and geohazard assessment. However, these techniques are highly sensitive to the quality of data acquired, the resolution of imaging, and the calibration to ground truth data. Furthermore, correlations between surface conditions and subsurface properties vary significantly depending on regional geological conditions, which presents a challenge for applying remote sensing methods universally across different geological terrains.
Machine learning and deep learning have fundamentally transformed the landscape of shear wave velocity prediction [146,147,148,149,150,151,152]. AI-based models have demonstrated exceptional capabilities in capturing complex, non-linear relationships within large datasets, providing more accurate and efficient predictions than traditional models [123]. The ability of ML algorithms to learn from diverse data sources, such as seismic measurements, geological logs, and remote sensing data, has led to substantial improvements in Vs estimation. For instance, hybrid models that combine machine learning techniques, such as LSTM networks with FWI, have shown promise in enhancing the precision and speed of seismic imaging and subsurface characterization [74]. Furthermore, AI techniques, such as CNNs, have outperformed classical regression models, providing more reliable and accurate predictions across different geological formations [115].
Nonetheless, AI-based methods face several critical challenges. The most pressing issue is the quality and availability of labeled training data, as ML models often require large, high-quality datasets to achieve reliable predictions. The lack of comprehensive and standardized datasets in geoscience is a significant hurdle, especially in remote or inaccessible regions [113]. Furthermore, while AI models can achieve high predictive accuracy, their interpretability remains a major concern. Geoscientists and engineers must understand how models arrive at their predictions, particularly in applications where decisions may have critical consequences, such as in seismic hazard assessment or infrastructure design. Approaches like XAI and model interpretability techniques such as SHAP and LIME are essential for increasing the transparency and trustworthiness of ML models in geoscience applications [131].
Additionally, the problem of overfitting, where models perform well on training data but fail to generalize across different geological settings, remains a significant challenge. Overcoming overfitting requires robust validation techniques, including cross-validation with diverse datasets, and the integration of domain-specific knowledge into AI models. Moreover, uncertainty quantification is an area that requires further research. As Vs estimation plays a crucial role in decision-making processes for infrastructure development and disaster preparedness, incorporating uncertainty quantification into AI models will enable more reliable risk assessments and better decision-making under uncertainty [132].
The future of Vs prediction lies in the continued integration of AI with traditional geophysical methods and the development of more robust, multi-source predictive models. Combining geophysical techniques, such as seismic tomography and surface wave analysis, with AI algorithms offers a powerful hybrid approach that can deliver both high accuracy and computational efficiency. Furthermore, the advent of new inversion algorithms and data fusion techniques will likely enhance the reliability of Vs predictions in geologically complex areas.
AI and machine learning also have a significant role in the optimization of data acquisition technologies, enhancing the efficiency of data collection in real time, reducing the need for expensive and labor-intensive fieldwork, and improving the quality of predictions with sparse data. Advancements in remote sensing and seismic interferometry will also complement AI-driven techniques, providing richer datasets for training and improving prediction accuracy [72]. The integration of PINNs and other hybrid models that combine domain knowledge with machine learning will enhance both the interpretability and accuracy of Vs estimation in complex geological settings.
Regarding the energy consumption, geophysical techniques for estimating shear wave velocity generally have low energy consumption because they rely on passive ambient noise or low-energy active sources, e.g., sledgehammers or small vibrators [153]. AI techniques can initially require substantial energy for training, but they may offer efficiencies in subsequent data processing and analysis [154]. Remote sensing methods, especially those deployed in space, can be energy-efficient during operation but may have higher initial deployment costs.
However, remote sensing techniques generally consume more energy per unit time, especially due to the power requirements of satellites, aircraft, and data processing. However, remote sensing can be more efficient overall, as it avoids the need for repeated, energy-intensive field campaigns and can cover large areas in one go. The tradeoff is that remote sensing might have higher upfront energy costs but can be more sustainable for large-scale or long-term studies, especially when considering repeated field campaigns that ground-based geophysical methods require [155].
In summary, the advancements in geophysical methods, remote sensing, and AI have the potential to revolutionize Vs prediction in geoscience and geoengineering. By addressing the challenges related to data availability, model interpretability, and uncertainty quantification, these technologies can offer more accurate, energy-efficient, and cost-effective solutions for subsurface characterization. Moving forward, multidisciplinary collaboration and the integration of advanced data fusion techniques will be crucial in overcoming existing challenges and enhancing the practical applications of Vs prediction.

7. Conclusions

Shear wave velocity (Vs) measurement and prediction are crucial for understanding subsurface formations in geoscience and geotechnical engineering. This review explored the advancements, methodologies, and challenges in predicting Vs. Geophysical methods offer cost-effective and eco-friendly options, though they face challenges related to sensitivity, data accessibility, and the need for controlled seismic sources. Remote sensing has proven valuable for estimating Vs over large areas, but data quality, uncertainty in correlations, and regional variability must be addressed for more reliable results.
Machine learning techniques, such as random forest regression, convolutional neural networks, and support vector machines, have shown promise in capturing complex relationships and improving prediction accuracy. However, issues such as data availability, overfitting, and interpretability need further refinement.
Energy efficiency plays a crucial role in Vs estimation methods. Geophysical techniques, using passive noise or low-energy sources, generally consume less energy. While AI techniques require substantial energy for training, they offer efficiencies in subsequent data analysis. Remote sensing methods, although energy-intensive at the outset, can be more sustainable for large-scale or long-term studies compared to ground-based methods requiring repeated fieldwork.
Key challenges in Vs prediction include data quality, spatial and temporal variability, non-unique relationships, and anisotropy. Solutions involve the development of standardized protocols, advanced modeling techniques, and integration of diverse data sources. Effective collaboration among researchers, industry professionals, and policymakers, along with robust uncertainty quantification, is vital for enhancing the practical applications of Vs prediction.
In summary, advancing Vs prediction requires continued multidisciplinary collaboration to address these challenges. By refining existing methods and exploring new approaches, significant progress can be made in improving geoscientific models, contributing to safer infrastructure development and seismic hazard mitigation.

Author Contributions

Conceptualization, formal analysis, methodology, writing—review and editing, writing—original draft preparation, M.K., D.K., and M.A.M.Z.; investigation, M.K. and M.A.M.Z.; supervision, validation, project administration, D.K. All authors have read and agreed to the published version of the manuscript.

Funding

The project was supported by the AGH University of Krakow, subsidy 16.16.190.779.

Data Availability Statement

All used data are accessible in the context of the article.

Conflicts of Interest

Author Mohammad Ahmad Mahmoudi Zamani was employed by the company Iranian Mining and Industry Organization. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Abbreviations

ACOFISAnt Colony–Fuzzy Inference System
AdaBoostAdaptive Boosting
AIArtificial Intelligence
ANFISAdaptive Neuro-Fuzzy Inference System
ANN-LMArtificial Neural Network–Levenberg–Marquardt
ANNArtificial Neural Network
Bi-LSTMBidirectional Long Short-Term Memory
BPNNBackpropagation Neural Network
BPANNBackpropagation Artificial Neural Network
BRRBayesian Ridge Regression
CatBoostCategorical Boosting
CCSCarbon Capture and Storage
CNNConvolutional Neural Network
COACuckoo Optimization Algorithm
CSACoupled Simulated Annealing
CTGANConditional Generative Adversarial Network
CTComputed Tomography
CVAEConditional Variational Autoencoders
DLDeep Learning
DNNDeep Neural Network
DTSDelta-T Shear (Shear Wave Slowness)
DTPDelta-T Compressional (Compressional Wave Slowness)
EGSEnhanced Geothermal System
ELMExtreme Learning Machine
FLFuzzy Logic
FWIFull-Waveform Inversion
GAGenetic Algorithm
GEPGene Expression Programming
GNNGraph Neural Networks
GCN-BiGRUGraph Convolutional Network with Bidirectional Gated Recurrent Units
GPRGaussian Process Regression
IRISIncorporated Research Institutions for Seismology
KNNK-Nearest Neighbors
LBLaplacian Boosting
LiDARLight Detection and Ranging
LIMELocal Interpretable Model-Agnostic Explanations
LRLinear Regression
LSSVMLeast Squares Support Vector Machine
LSTMLong Short-Term Memory
LSSVM-COALeast Squares Support Vector Machine–Cuckoo Optimization Algorithm
LSSVM-CSALeast Squares Support Vector Machine–Coupled Simulated Annealing
LSSVM-GALeast Squares Support Vector Machine–Genetic Algorithm
LSSVM-PSOLeast Squares Support Vector Machine–Particle Swarm Optimization
MAEMean Absolute Error
MELMMultilayer Extreme Learning Machine
MELM-COAMultilayer Extreme Learning Machine–Cuckoo Optimization Algorithm
MLMachine Learning
MLPMultilayer Perceptron
MLRMultiple Linear Regression
MIVMean Impact Value
MASWMultichannel Analysis of Surface Waves
MSEMean Squared Error
NFNeuro-Fuzzy
NFSNeuro-Fuzzy System
NNNeural Network
NSGA-IINon-Dominated Sorting Genetic Algorithm II
OSGeoOpen-Source Geospatial Foundation
PINNsPhysics-Informed Neural Networks
PSOParticle Swarm Optimization
R2Coefficient of Determination
RFRandom Forest
RFRRandom Forest Regression
RMSERoot Mean Squared Error
SASWSpectral Analysis of Surface Waves
SHAPSHapley Additive exPlanations
SPT-NStandard Penetration Test—N Value
SRTSeismic Refraction Tomography
SVRSupport Vector Regression
UAV-basedUnmanned Aerial Vehicle-based
VelProfESVelocity Profile Estimation System
VsShear Wave Velocity
VpCompressional Wave Velocity
XAIExplainable Artificial Intelligence
XGBoostExtreme Gradient Boosting

References

  1. Moss, R.E.S. Quantifying measurement uncertainty of thirty-meter shear-wave velocity. Bull. Seismol. Soc. Am. 2008, 98, 1399–1411. [Google Scholar] [CrossRef]
  2. Foti, S.; Hollender, F.; Garofalo, F.; Albarello, D.; Asten, M.; Bard, P.Y.; Comina, C.; Cornou, C.; Cox, B.; Di Giulio, G.; et al. Guidelines for the good practice of surface wave analysis: A product of the InterPACIFIC project. Bull. Earthq. Eng. 2018, 16, 2367–2420. [Google Scholar] [CrossRef]
  3. Shafiee, A.; Azadi, A. Shear-wave velocity characteristics of geological units throughout Tehran City, Iran. J. Asian Earth Sci. 2007, 29, 105–115. [Google Scholar] [CrossRef]
  4. Incorporated Research Institutions for Seismology (IRIS). S-Wave Motion—Animation. Available online: https://www.iris.edu/hq/inclass/animation/swave_motion (accessed on 19 February 2025).
  5. L’Heureux, J.S.; Long, M. Relationship between shear-wave velocity and geotechnical parameters for Norwegian clays. J. Geotech. Geoenvironmental Eng. 2017, 143, 04017013. [Google Scholar] [CrossRef]
  6. Sundararajan, N.; Seshunarayana, T. Shear wave velocities in the estimation of earthquake hazard over alluvium in a seismically active region. J. Geol. Soc. India 2018, 92, 259–264. [Google Scholar] [CrossRef]
  7. Parolai, S.; Richwalski, S.M.; Milkereit, C.; Fäh, D. S-wave velocity profiles for earthquake engineering purposes for the Cologne area (Germany). Bull. Earthq. Eng. 2006, 4, 65–94. [Google Scholar] [CrossRef]
  8. Jamiolkowski, M. Role of geophysical testing in geotechnical site characterization. Soils Rocks 2012, 35, 117–137. [Google Scholar] [CrossRef]
  9. Anbazhagan, P.; Sitharam, T.G. Site characterization and site response studies using shear wave velocity. J. Seismol. Earthq. Eng. 2008, 10, 53–67. [Google Scholar]
  10. Li, X.Y.; Zhang, Y.G. Seismic reservoir characterization: How can multicomponent data help? J. Geophys. Eng. 2011, 8, 123. [Google Scholar] [CrossRef]
  11. Rezaee, M.R.; Ilkhchi, A.K.; Barabadi, A. Prediction of shear wave velocity from petrophysical data utilizing intelligent systems: An example from a sandstone reservoir of Carnarvon Basin, Australia. J. Pet. Sci. Eng. 2007, 55, 201–212. [Google Scholar] [CrossRef]
  12. Mourad, M.; Tsuji, T.; Ikeda, T.; Ishitsuka, K.; Senna, S.; Ide, K. Mapping aquifer storage properties using s-wave velocity and insar-derived surface displacement in the Kumamoto area, Southwest Japan. Remote Sens. 2021, 13, 4391. [Google Scholar] [CrossRef]
  13. Pugin, A.J.M.; Pullan, S.E.; Hunter, J.A.; Oldenborger, G.A. Hydrogeological prospecting using P-and S-wave landstreamer seismic reflection methods. Near Surf. Geophys. 2009, 7, 315–328. [Google Scholar] [CrossRef]
  14. Hedtmann, N.; Alber, M. Investigation of water-permeability and ultrasonic wave velocities of German Malm aquifer rocks for hydro-geothermal energy. In Proceedings of the ISRM EUROCK, Ostrava, Czech Republic, 20–22 June 2017. [Google Scholar]
  15. Sharifi-Mood, M.; Olsen, M.J.; Gillins, D.T.; Mahalingam, R. Performance-based, seismically-induced landslide hazard mapping of Western Oregon. Soil Dyn. Earthq. Eng. 2017, 103, 38–54. [Google Scholar] [CrossRef]
  16. Shukla, R.; Ranjith, P.; Haque, A.; Choi, X. A Review of Studies on CO2 Sequestration and Caprock Integrity. Fuel 2010, 89, 2651–2664. [Google Scholar] [CrossRef]
  17. Peuchen, J.; De Ruijter, M.R.; Hospers, B.; Assen, R.L. Shear wave velocity integrated in offshore geotechnical practice. In Proceedings of the SUT Offshore Site Investigation and Geotechnics, London, UK, 26–28 November 2002. [Google Scholar]
  18. Hosseini, K.; Matthews, K.J.; Sigloch, K.; Shephard, G.E.; Domeier, M.; Tsekhmistrenko, M. SubMachine: Web-based tools for exploring seismic tomography and other models of Earth’s deep interior. Geochem. Geophys. Geosystems 2018, 19, 1464–1483. [Google Scholar] [CrossRef]
  19. Nejad, M.M.; Momeni, M.S.; Manahiloh, K.N. Shear wave velocity and soil type microzonation using neural networks and geographic information system. Soil Dyn. Earthq. Eng. 2018, 104, 54–63. [Google Scholar] [CrossRef]
  20. Lee, M.W. A simple method of predicting S-wave velocity. Geophysics 2006, 71, F161–F164. [Google Scholar] [CrossRef]
  21. Putra, D.P.N.; Utama, W.; Garini, S.A.; Indriani, R.F. Geological Mapping Optimisation Using Satellite Gravity, Satellite Imagery, and Seismic Shear Velocity (Vs30) to Monitor Lithological Condition in Geothermal Area of Mount Salak, West Java. BIO Web Conf. 2024, 89, 02001. [Google Scholar] [CrossRef]
  22. Wang, J.; Cao, J. Data-driven S-wave velocity prediction method via a deep-learning-based deep convolutional gated recurrent unit fusion network. Geophysics 2021, 86, M185–M196. [Google Scholar] [CrossRef]
  23. Boore, D.M.; Joyner, W.B.; Fumal, T.E. Equations for estimating horizontal response spectra and peak acceleration from western North American earthquakes: A summary of recent work. Seismol. Res. Lett. 1997, 68, 128–153. [Google Scholar] [CrossRef]
  24. Dong, Y.; Lu, N. Dependencies of shear wave velocity and shear modulus of soil on saturation. J. Eng. Mech. 2016, 142, 04016083. [Google Scholar] [CrossRef]
  25. Thitimakorn, T.; Channoo, S. Shear wave velocity of soils and NEHRP site classification map of Chiang Rai City, Northern Thailand. Electron. J. Geotech. Eng. 2012, 17, 2891–2904. [Google Scholar]
  26. Uma Maheswari, R.; Boominathan, A.; Dodagoudar, G.R. Use of surface waves in statistical correlations of shear wave velocity and penetration resistance of Chennai soils. Geotech. Geol. Eng. 2010, 28, 119–137. [Google Scholar] [CrossRef]
  27. Baxter, C.D.; Bradshaw, A.S.; Green, R.A.; Wang, J.H. Correlation between cyclic resistance and shear-wave velocity for providence silts. J. Geotech. Geoenvironmental Eng. 2008, 134, 37–46. [Google Scholar] [CrossRef]
  28. Yunmin, C.; Han, K.; Ren-peng, C. Correlation of shear wave velocity with liquefaction resistance based on laboratory tests. Soil Dyn. Earthq. Eng. 2005, 25, 461–469. [Google Scholar] [CrossRef]
  29. Rodriguez-Marek, A.; Bray, J.D.; Abrahamson, N.A. A geotechnical seismic site response evaluation procedure. In Proceedings of the 12WCEE 2000: 12th World Conference on Earthquake Engineering, Auckland, New Zealand, 30 January–4 February 2000. [Google Scholar]
  30. Oh, T.M.; Bang, E.S.; Cho, G.C.; Park, E.S. Estimation of undrained shear strength for saturated clay using shear wave velocity. Mar. Georesources Geotechnol. 2017, 35, 236–244. [Google Scholar] [CrossRef]
  31. Oloruntobi, O.; Butt, S. The shear-wave velocity prediction for sedimentary rocks. J. Nat. Gas Sci. Eng. 2020, 76, 103084. [Google Scholar] [CrossRef]
  32. Kováčik, J.; Emmer, Š. Correlation between shear wave velocity and porosity in porous solids and rocks. J. Powder Technol. 2013, 1, 643167. [Google Scholar] [CrossRef]
  33. Korneev, V. Slow waves in fractures filled with viscous fluid. Geophysics 2008, 73, N1–N7. [Google Scholar] [CrossRef]
  34. Hussien, M.N.; Karray, M. Shear wave velocity as a geotechnical parameter: An overview. Can. Geotech. J. 2015, 53, 252–272. [Google Scholar] [CrossRef]
  35. Azhar, A.S.B.; Latiff, A.H.A.; Lim, L.H.; Gödeke, S.H. Groundwater investigation of a coastal aquifer in Brunei Darussalam using seismic refraction. Environ. Earth Sci. 2019, 78, 220. [Google Scholar] [CrossRef]
  36. Jarvis, K.D.; Knight, R.J. Aquifer heterogeneity from SH-wave seismic impedance inversion. Geophysics 2002, 67, 1548–1557. [Google Scholar] [CrossRef]
  37. Sayers, C.M. Elastic wave velocities in a granitic geothermal reservoir. Geophys. Prospect. 2022, 71, 114–124. [Google Scholar] [CrossRef]
  38. Lou, M.; Rial, J.A. Characterization of geothermal reservoir crack patterns using shear-wave splitting. Geophysics 1997, 62, 487–494. [Google Scholar] [CrossRef]
  39. Wawerzinek, B.; Buness, H.; von Hartmann, H.; Tanner, D.C. S-wave experiments for the exploration of a deep geothermal carbonate reservoir in the German Molasse Basin. Geotherm. Energy 2021, 9, 6. [Google Scholar] [CrossRef]
  40. Mendrinos, D.; Karytsas, C.; Karytsas, S.; Poletto, F.; Farina, B.; Barison, E. Correlations of Seismic Velocities and Elastic Moduli with Temperature in Superhot and Enhanced Geothermal Systems. Clean Technol. 2022, 4, 440–457. [Google Scholar] [CrossRef]
  41. Del Gaudio, V.; Muscillo, S.; Wasowski, J. What we can learn about slope response to earthquakes from ambient noise analysis: An overview. Eng. Geol. 2014, 182, 182–200. [Google Scholar] [CrossRef]
  42. Chen, Y.; Irfan, M.; Uchimura, T.; Wu, Y.; Yu, F. Development of elastic wave velocity threshold for rainfall-induced landslide prediction and early warning. Landslides 2019, 16, 955–968. [Google Scholar] [CrossRef]
  43. Qureshi, M.U.; Yamada, S.; Towhata, I. A simplified technique for slope stability assessment based on in-situ S-wave velocity measurement. In Earthquake-Induced Landslides: Proceedings of the International Symposium on Earthquake-Induced Landslides, Kiryu, Japan, 2012; Springer: Berlin/Heidelberg, Germany, 2013; pp. 871–881. [Google Scholar]
  44. Feng, Z.Y.; Lu, Y.R.; Shen, Z.R. A numerical simulation of seismic signals of coseismic landslides. Eng. Geol. 2021, 289, 106191. [Google Scholar] [CrossRef]
  45. Fjaer, E.; Holt, R.M.; Horsrud, P.; Raaen, A.M. Petroleum Related Rock Mechanics; Elsevier: Amsterdam, The Netherlands, 2008. [Google Scholar]
  46. Blake, O.O.; Faulkner, D.R.; Worden, R.H.; Armitage, P.J.; Espie, A.A. Effect of thermal shock on the permeability and seismic wave velocity of the caprock and reservoir during CO2 injection. Int. J. Greenh. Gas Control 2022, 118, 103691. [Google Scholar] [CrossRef]
  47. Gholami, R.; Raza, A.; Iglauer, S. Leakage risk assessment of a CO2 storage site: A review. Earth-Sci. Rev. 2021, 223, 103849. [Google Scholar] [CrossRef]
  48. Ringrose, P.S.; Mathieson, A.S.; Wright, I.W.; Selama, F.; Hansen, O.; Bissell, R.; Saoula, N.; Midgley, J. The In Salah CO2 storage project: Lessons learned and knowledge transfer. Energy Procedia 2013, 37, 6226–6236. [Google Scholar] [CrossRef]
  49. Nadim, F.; Kvalstad, T.J. Risk assessment and management for offshore geohazards. In Proceedings of the ISGSR2007 First International Symposium on Geotechnical Safety & Risk, Shanghai, China, 18–19 October 2007. [Google Scholar]
  50. Ten Brink, U.S.; Lee, H.J.; Geist, E.L.; Twichell, D. Assessment of tsunami hazard to the US East Coast using relationships between submarine landslides and earthquakes. Mar. Geol. 2009, 264, 65–73. [Google Scholar] [CrossRef]
  51. Zhao, M.; Gao, Z.; Wang, P.; Du, X. Response spectrum method for seismic analysis of monopile offshore wind turbine. Soil Dyn. Earthq. Eng. 2020, 136, 106212. [Google Scholar] [CrossRef]
  52. Brithish Geological Survey. Offshore Seismic Hazard in UK Waters. Available online: https://www.earthquakes.bgs.ac.uk/hazard/hazard_offshore.html (accessed on 24 January 2025).
  53. Gao, W.; Grand, S.P.; Baldridge, W.S.; Wilson, D.; West, M.; Ni, J.F.; Aster, R. Upper mantle convection beneath the central Rio Grande rift imaged by P and S wave tomography. J. Geophys. Res. Solid Earth 2004, 109, B03305. [Google Scholar] [CrossRef]
  54. Zhang, Z.; Yao, H.; Yang, Y. Shear wave velocity structure of the crust and upper mantle in Southeastern Tibet and its geodynamic implications. Sci. China Earth Sci. 2020, 63, 1278–1293. [Google Scholar] [CrossRef]
  55. Mainprice, D.; Tommasi, A.; Couvy, H.; Cordier, P.; Frost, D.J. Pressure sensitivity of olivine slip systems and seismic anisotropy of Earth’s upper mantle. Nature 2005, 433, 731–733. [Google Scholar] [CrossRef]
  56. Butler, R.; Tsuboi, S. Antipodal seismic reflections upon shear wave velocity structures within Earth’s inner core. Phys. Earth Planet. Inter. 2021, 321, 106802. [Google Scholar] [CrossRef]
  57. Mussett, A.E.; Khan, M.A. Looking into The Earth: An Introduction to Geological Geophysics; University Press: Cambridge, NY, USA, 2000; 470p. [Google Scholar]
  58. Sarkar, R.; Kolathayar, S.; Drukpa, D.; Choki, K.; Rai, S.; Tshering, S.T.; Yuden, K. Near-surface seismic refraction tomography and MASW for site characterization in Phuentsholing, Bhutan Himalaya. SN Appl. Sci. 2021, 3, 394. [Google Scholar] [CrossRef]
  59. U.S. Environmental Protection Agency (EPA). Seismic Refraction. Available online: https://www.epa.gov/environmental-geophysics/seismic-refraction (accessed on 18 April 2024).
  60. Nazarian, S. In Situ Determination of Elastic Moduli of Soil Deposits and Pavement Systems by Spectral-Analysis-of-Surface-Waves Method (Shear Velocity, Propagation, Liquefaction, Non-Destructive, Earthquake). Ph.D. Thesis, The University of Texas, Austin, TX, USA, 1984. [Google Scholar]
  61. Boore, D.M. Determining subsurface shear-wave velocities: A review. In Proceedings of the Third International Symposium on the Effects of Surface Geology on Seismic Motion, Grenoble, France, 30 August–1 September 2006; pp. 67–85. [Google Scholar]
  62. Jones, R. In-situ measurement of the dynamic properties of soil by vibration methods. Geotechnique 1958, 8, 1–21. [Google Scholar] [CrossRef]
  63. Abbiss, C.P. Shear wave measurements of the elasticity of the ground. Geotechnique 1981, 31, 91–104. [Google Scholar] [CrossRef]
  64. SAGE. Available online: https://www.iris.edu/hq/files/programs/education_and_outreach/aotm/7/SeismicTomography_Background.pdf (accessed on 11 September 2024).
  65. Li, Y.; Wu, Q.; Pan, J.; Sun, L. S-wave velocity structure of northeastern China from joint inversion of Rayleigh wave phase and group velocities. Geophys. J. Int. 2012, 190, 105–115. [Google Scholar] [CrossRef]
  66. Pilz, M.; Parolai, S.; Picozzi, M.; Bindi, D. Three-Dimensional Shear Wave Velocity Imaging by Ambient Seismic Noise Tomography. Geophys. J. Int. 2012, 189, 501–512. [Google Scholar] [CrossRef]
  67. Manu-Marfo, D.; Aoudia, A.; Pachhai, S.; Kherchouche, R. 3D Shear Wave Velocity Model of the Crust and Uppermost Mantle beneath the Tyrrhenian Basin and Margins. Sci. Rep. 2019, 9, 3609. [Google Scholar] [CrossRef]
  68. Louie, J.N. Faster, better: Shear-wave velocity to 100 meters depth from refraction microtremor arrays. Bull. Seismol. Soc. Am. 2001, 91, 347–364. [Google Scholar] [CrossRef]
  69. Ellefsen, K.J.; Tuttle, G.J.; Williams, J.M.; Lucius, J.E. S-wave refraction survey of alluvial aggregate (No. 2005-5012). Open-File Rep. Kansas Geol. Surv. 2005, 36, 11. [Google Scholar]
  70. Boiero, D. Surface Wave Analysis for Building Shear Wave Velocity Models. Ph.D. Thesis, Politecnico di Torino, Torino, Italy, 2009. [Google Scholar]
  71. Fichtner, A.; Kennett, B.L.; Tsai, V.C.; Thurber, C.H.; Rodgers, A.J.; Tape, C.; Rawlinson, N.; Borcherdt, R.D.; Lebedev, S.; Priestley, K.; et al. Seismic tomography 2024. Bull. Seismol. Soc. Am. 2024, 114, 1185–1213. [Google Scholar] [CrossRef]
  72. Wikipedia Contributors. Seismic Interferometry. Wikipedia. Available online: https://en.wikipedia.org/wiki/Seismic_interferometry?utm_source=chatgpt.com (accessed on 23 May 2025).
  73. Wolf, J.; Becker, T.W.; Garnero, E.; Liu, K.H.; West, J.D. Comprehensive global data set of uniformly processed shear-wave splitting measurements. Geophys. J. Int. 2025, 241, 863–875. [Google Scholar] [CrossRef]
  74. Zerafa, C.; Galea, P.; Sebu, C. Synergizing Deep Learning and Full-Waveform Inversion: Bridging Data-Driven and Theory-Guided Approaches for Enhanced Seismic Imaging. arXiv 2025, arXiv:2502.17585. [Google Scholar] [CrossRef]
  75. Earp, S.; Curtis, A.; Zhang, X.; Hansteen, F. Probabilistic neural network tomography across Grane field (North Sea) from surface wave dispersion data. Geophys. J. Int. 2020, 223, 1741–1757. [Google Scholar] [CrossRef]
  76. Miller, R.D.; Bradford, J.H.; Holliger, K. (Eds.) Advances in Near-Surface Seismology and Ground-Penetrating Radar; Society of Exploration Geophysicists, American Geophysical Union, Environmental and Engineering Geophysical Society: Tulsa, OK, USA, 2010. [Google Scholar]
  77. Wikipedia contributors. Subsurface mapping by ambient noise tomography. Wikipedia. Available online: https://en.wikipedia.org/wiki/Subsurface_mapping_by_ambient_noise_tomography?utm_source=chatgpt.com (accessed on 4 June 2025).
  78. U.S. Geological Survey (USGS). What Is Remote Sensing and What Is It Used For? Available online: https://www.usgs.gov/faqs/what-remote-sensing-and-what-it-used (accessed on 31 December 2024).
  79. Yusroni, S.A.; Kemalasari, V.T.; Putra, D.P.N. Identification of Land Cover Changes from Landsat 8 Oli Satellite Imagery Using Normalized Difference Vegetation Index (NDVI) Method (Study Case: Surabaya). J. Mar. Sci. Technol. 2021, 2, 5–10. [Google Scholar] [CrossRef]
  80. Avtar, R.; Komolafe, A.A.; Kouser, A.; Singh, D.; Yunus, A.P.; Dou, J.; Kumar, P.; Gupta, R.D.; Johnson, B.A.; Minh, H.V.T.; et al. Assessing sustainable development prospects through remote sensing: A review. Remote Sens. Appl. Soc. Environ. 2020, 20, 100402. [Google Scholar] [CrossRef] [PubMed]
  81. Ejemeyovwi, D.O.; Achima, B.T.; Ogwu, C. Remote Sensing Satellite Systems and Capabilities in Mapping Environmental Resources. KIU J. Soc. Sci. 2023, 9, 117–124. [Google Scholar] [CrossRef]
  82. Retallack, A.; Finlayson, G.; Ostendorf, B.; Clarke, K.; Lewis, M. Remote sensing for monitoring rangeland condition: Current status and development of methods. Environ. Sustain. Indic. 2023, 19, 100285. [Google Scholar] [CrossRef]
  83. Eskandari, A.; Hosseini, M.; Nicotra, E. Application of Satellite Remote Sensing, UAV-Geological Mapping, and Machine Learning Methods in the Exploration of Podiform Chromite Deposits. Minerals 2023, 13, 251. [Google Scholar] [CrossRef]
  84. Rizki, M.K.; Lubis, A.M.; Handayani, L.; Mukti, M.M.; Anggono, T.; Gaol, K.L.; Sudrajat, Y. Gravity Anomaly Evaluation of the Geothermal Field Distribution Around the Seismo-Volcanic Area of Toba Lake. J. Phys. Conf. Ser. 2022, 2377, 012042. [Google Scholar] [CrossRef]
  85. Setyawan, A.; Khusna, L.M.; Suseno, J.E.; Rina, D.I.; Yulianto, T.; Aribowo, Y. Detecting hot spring manifestations based on gravity data satellite on mountain Lawu. J. Phys. Conf. Ser. 2021, 1943, 012034. [Google Scholar] [CrossRef]
  86. Flechtner, F.; Reigber, C.; Rummel, R.; Balmino, G. Satellite Gravimetry: A Review of Its Realization. Surv. Geophys. 2021, 42, 1029–1074. [Google Scholar] [CrossRef]
  87. Noeker, M.; Karatekin, Ö. The wedge-pentahedra method (WPM): Topographic reduction of local terrain in the context of solar system surface gravimetry and robotic exploration. Front. Space Technol. 2022, 3, 982873. [Google Scholar] [CrossRef]
  88. Kadhim, N.; Mourshed, M.; Bray, M. Advances in remote sensing applications for urban sustainability. Euro-Mediterr. J. Environ. Integr. 2016, 1, 7. [Google Scholar] [CrossRef]
  89. Ghione, F.; Köhler, A.; Dichiarante, A.M.; Aarnes, I.; Oye, V. Vs30 and depth to bedrock estimates from integrating HVSR measurements and geology-slope approach in the Oslo area, Norway. Front. Earth Sci. 2023, 11, 1242679. [Google Scholar] [CrossRef]
  90. Wróbel, M.; Stan-Kłeczek, I.; Marciniak, A.; Majdański, M.; Kowalczyk, S.; Nawrot, A.; Cader, J. Integrated Geophysical Imaging and Remote Sensing for Enhancing Geological Interpretation of Landslides with Uncertainty Estimation—A Case Study from Cisiec, Poland. Remote Sens. 2022, 15, 238. [Google Scholar] [CrossRef]
  91. Holland, C.W.; Dettmer, J.; Dosso, S.E. Remote sensing of sediment density and velocity gradients in the transition layer. J. Acoust. Soc. Am. 2005, 118, 163–177. [Google Scholar] [CrossRef]
  92. Abbas, A.; Vantassel, J.P.; Cox, B.R.; Kumar, K.; Crocker, J. A frequency-velocity CNN for developing near-surface 2D vs images from linear-array, active-source wavefield measurements. Comput. Geotech. 2023, 156, 105305. [Google Scholar] [CrossRef]
  93. Feng, G.; Zeng, H.H.; Xu, X.R.; Tang, G.Y.; Wang, Y.X. Shear wave velocity prediction based on deep neural network and theoretical rock physics modeling. Front. Earth Sci. 2023, 10, 1025635. [Google Scholar] [CrossRef]
  94. Meng, J.; Zhang, H.; Sun, L.; Wang, J. Remote sensing techniques for detecting internal solitary waves: A comprehensive review and prospects. IEEE Geosci. Remote Sens. Mag. 2024, 12, 46–78. [Google Scholar] [CrossRef]
  95. Rajabi, M.; Bohloli, B.; Ahangar, E.G. Intelligent approaches for prediction of compressional, shear and Stoneley wave velocities from conventional well log data: A case study from the Sarvak carbonate reservoir in the Abadan Plain (Southwestern Iran). Comput. Geosci. 2010, 36, 647–664. [Google Scholar] [CrossRef]
  96. Asoodeh, M.; Bagheripour, P. Prediction of compressional, shear, and Stoneley wave velocities from conventional well log data using a committee machine with intelligent systems. Rock Mech. Rock Eng. 2012, 45, 45–63. [Google Scholar] [CrossRef]
  97. Maleki, S.; Moradzadeh, A.; Riabi, R.G.; Gholami, R.; Sadeghzadeh, F. Prediction of shear wave velocity using empirical correlations and artificial intelligence methods. NRIAG J. Astron. Geophys. 2014, 3, 70–81. [Google Scholar] [CrossRef]
  98. Bagheripour, P.; Gholami, A.; Asoodeh, M.; Vaezzadeh-Asadi, M. Support vector regression-based determination of shear wave velocity. J. Pet. Sci. Eng. 2015, 125, 95–99. [Google Scholar] [CrossRef]
  99. Nourafkan, A.; Kadkhodaie-Ilkhchi, A. Shear wave velocity estimation from conventional well log data by using a hybrid ant colony–fuzzy inference system: A case study from Cheshmeh–Khosh oilfield. J. Pet. Sci. Eng. 2015, 127, 459–468. [Google Scholar] [CrossRef]
  100. Ataee, O.; Hafezi Moghaddas, N.; Lashkari Pour, G.R.; Abbari Nooghabi, M.J. Predicting shear wave velocity of soil using multiple linear regression analysis and artificial neural networks. Sci. Iran. 2018, 25, 1943–1955. [Google Scholar] [CrossRef]
  101. Anemangely, M.; Ramezanzadeh, A.; Amiri, H.; Hoseinpour, S.A. Machine learning technique for the prediction of shear wave velocity using petrophysical logs. J. Pet. Sci. Eng. 2019, 174, 306–327. [Google Scholar] [CrossRef]
  102. Wang, P.; Peng, S. On a new method of estimating shear wave velocity from conventional well logs. J. Pet. Sci. Eng. 2019, 180, 105–123. [Google Scholar] [CrossRef]
  103. Azadpour, M.; Saberi, M.R.; Javaherian, A.; Shabani, M. Rock physics model-based prediction of shear wave velocity utilizing machine learning technique for a carbonate reservoir, southwest Iran. J. Pet. Sci. Eng. 2020, 195, 107864. [Google Scholar] [CrossRef]
  104. Lian, J.; John, P.C.; Brian, R.; Pablo, G. Rock physics modelling using machine learning. In SEG Technical Program Expanded Abstracts 2020; Society of Exploration Geophysicists: Houston, TX, USA, 2020; pp. 2530–2534. [Google Scholar] [CrossRef]
  105. Khatibi, S.; Aghajanpour, A. Machine learning: A useful tool in geomechanical studies, a case study from an offshore gas field. Energies 2020, 13, 3528. [Google Scholar] [CrossRef]
  106. Wang, J.; Cao, J.; Yuan, S. Shear wave velocity prediction based on adaptive particle swarm optimization optimized recurrent neural network. J. Pet. Sci. Eng. 2020, 194, 107466. [Google Scholar] [CrossRef]
  107. Zhang, Y.; Xie, Y.; Zhang, Y.; Qiu, J.; Wu, S. The adoption of deep neural network (DNN) to the prediction of soil liquefaction based on shear wave velocity. Bull. Eng. Geol. Environ. 2021, 80, 5053–5060. [Google Scholar] [CrossRef]
  108. Jeong, J.; Park, E.; Emelyanova, I.; Pervukhina, M.; Esteban, L.; Yun, S.T. Application of conditional generative model for sonic log estimation considering measurement uncertainty. J. Pet. Sci. Eng. 2021, 196, 108028. [Google Scholar] [CrossRef]
  109. Olayiwola, T.; Sanuade, O.A. A data-driven approach to predict compressional and shear wave velocities in reservoir rocks. Petroleum 2021, 7, 199–208. [Google Scholar] [CrossRef]
  110. Miah, M.I. Improved prediction of shear wave velocity for clastic sedimentary rocks using hybrid model with core data. J. Rock Mech. Geotech. Eng. 2021, 13, 1466–1477. [Google Scholar] [CrossRef]
  111. Kim, J. Synthetic shear sonic log generation utilizing hybrid machine learning techniques. Artif. Intell. Geosci. 2022, 3, 53–70. [Google Scholar] [CrossRef]
  112. Wong, W.K.; Nuwara, Y.; Juwono, F.H.; Motalebi, F. Sonic waves travel-time prediction: When machine learning meets geophysics. In Proceedings of the 2022 International Conference on Green Energy, Computing and Sustainable Technology (GECOST), Sousse, Tunisia, 2–4 October 2022; pp. 159–163. [Google Scholar] [CrossRef]
  113. Mehrad, M.; Ramezanzadeh, A.; Bajolvand, M.; Hajsaeedi, M.R. Estimating shear wave velocity in carbonate reservoirs from petrophysical logs using intelligent algorithms. J. Pet. Sci. Eng. 2022, 212, 110254. [Google Scholar] [CrossRef]
  114. Laalam, A.; Mouedden, N.; Ouadi, H.; Chemmakh, A.; Merzoug, A.; Boualam, A.; Djezzar, S.; Aihar, A.; Berrehal, B.E. Prediction of Shear Wave Velocity in the Williston Basin Using Big Data Analysis and Robust Machine Learning Algorithms. In Proceedings of the 56th U.S. Rock Mechanics/Geomechanics Symposium, Santa Fe, NM, USA, 26–29 June 2022. [Google Scholar] [CrossRef]
  115. Zhang, Y.; Zhang, C.; Ma, Q.; Zhang, X.; Zhou, H. Automatic prediction of shear wave velocity using convolutional neural networks for different reservoirs in Ordos Basin. J. Pet. Sci. Eng. 2022, 208, 109252. [Google Scholar] [CrossRef]
  116. Jiang, R.; Ji, Z.; Mo, W.; Wang, S.; Zhang, M.; Yin, W.; Wang, Z.; Lin, Y.; Wang, X.; Ashraf, U. A Novel Method of Deep Learning for Shear Velocity Prediction in a Tight Sandstone Reservoir. Energies 2022, 15, 7016. [Google Scholar] [CrossRef]
  117. Zhang, Y.; Wang, D.; Li, H.; Gao, J. S-wave velocity prediction using physical model-driven Gaussian process regression: A case study of tight sandstone reservoir. Geophysics 2023, 88, D85–D93. [Google Scholar] [CrossRef]
  118. Kheirollahi, H.; Manaman, N.S.; Leisi, A. Robust estimation of shear wave velocity in a carbonate oil reservoir from conventional well logging data using machine learning algorithms. J. Appl. Geophys. 2023, 211, 104971. [Google Scholar] [CrossRef]
  119. Cova, D.; Liu, Y. Shear wave velocity prediction using bidirectional recurrent gated graph convolutional network with total information embeddings. Front. Earth Sci. 2023, 11, 1101601. [Google Scholar] [CrossRef]
  120. Akinyemi, O.D.; Elsaadany, M.; Siddiqui, N.A.; Elkurdy, S.; Olutoki, J.O.; Islam, M.M. Machine learning application for prediction of sonic wave transit time-A case of Niger Delta basin. Results Eng. 2023, 20, 101528. [Google Scholar] [CrossRef]
  121. Mustafa, A.; Tariq, Z.; Iqbal, A.; Naeem, M. A Data-Driven Intelligent Approach to Predict Shear Wave Velocity in Shale Formations. In Proceedings of the 57th U.S. Rock Mechanics/Geomechanics Symposium, Atlanta, GA, USA, 25–28 June 2023. [Google Scholar] [CrossRef]
  122. Khalilidermani, M.; Knez, D. Comparing Artificial Intelligence Algorithms with Empirical Correlations in Shear Wave Velocity Prediction. Appl. Sci. 2023, 13, 13126. [Google Scholar] [CrossRef]
  123. Fu, X.; Wei, Y.; Su, Y.; Hu, H. Shear Wave Velocity Prediction Based on the Long Short-Term Memory Network with Attention Mechanism. Appl. Sci. 2024, 14, 2489. [Google Scholar] [CrossRef]
  124. Gomaa, S.; Shahat, J.S.; Aboul-Fotouh, T.M.; Khaled, S. Neural Network Model for Predicting Shear Wave Velocity Using Well Logging Data. Arab. J. Sci. Eng. 2024, 50, 4721–4730. [Google Scholar] [CrossRef]
  125. Dehghani, M.; Jahani, S.; Ranjbar, A. Comparing the performance of machine learning methods in estimating the shear wave transit time in one of the reservoirs in southwest of Iran. Sci. Rep. 2024, 14, 4744. [Google Scholar] [CrossRef] [PubMed]
  126. Joshi, A.; Raman, B.; Mohan, C.K.; Cenkeramaddi, L.R. A new machine learning approach for estimating shear wave velocity profile using borelog data. Soil Dyn. Earthq. Eng. 2024, 177, 108424. [Google Scholar] [CrossRef]
  127. Yılmaz, B.; Türkmen, M.; Meral, S.; Akagündüz, E.; Tileylioglu, S. Deep Learning-Based Average Shear Wave Velocity Prediction Using Accelerometer Records. arXiv 2024, arXiv:2408.14962. [Google Scholar] [CrossRef]
  128. Leisi, A.; Shad Manaman, N. Three-Dimensional Shear Wave Velocity Prediction by Integrating Post-Stack Seismic Attributes and Well Logs: Application on Asmari Formation in Iran. J. Pet. Explor. Prod. Technol. 2024, 14, 2399–2411. [Google Scholar] [CrossRef]
  129. Castagna, J.P.; Batzle, M.L.; Eastwood, R.L. Relationships between compressional-wave and shear-wave velocities in clastic silicate rocks. Geophysics 1985, 50, 571–581. [Google Scholar] [CrossRef]
  130. Rajabi, M.; Hazbeh, O.; Davoodi, S.; Wood, D.A.; Tehrani, P.S.; Ghorbani, H.; Mehrad, M.; Mohamadian, N.; Rukavishnikov, V.S. Predicting shear wave velocity from conventional well logs with deep and hybrid machine learning algorithms. J. Petrol. Explor. Prod. Technol. 2023, 13, 19–42. [Google Scholar] [CrossRef]
  131. Entezari, I.; Sharp, J.; Mayne, P.W. A data-driven approach to predict shear wave velocity from CPTu measurements. In Cone Penetration Testing 2022; CRC Press: Boca Raton, FL, USA, 2022; pp. 374–380. [Google Scholar]
  132. Chala, A.T.; Ray, R. Uncertainty Quantification in Shear Wave Velocity Predictions: Integrating Explainable Machine Learning and Bayesian Inference. Appl. Sci. 2025, 15, 1409. [Google Scholar] [CrossRef]
  133. IRIS. Available online: https://ds.iris.edu/ds/nodes/dmc/ (accessed on 4 June 2025).
  134. OpenTopography. Available online: https://portal.opentopography.org/dataCatalog (accessed on 4 June 2025).
  135. OSGeo Foundation. Available online: https://www.osgeo.org/ (accessed on 1 June 2025).
  136. Ribeiro, M.T.; Singh, S.; Guestrin, C. “Why should I trust you?” Explaining the predictions of any classifier. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 1135–1144. [Google Scholar] [CrossRef]
  137. Cressie, N. Statistics for Spatial Data; Wiley: New York, NY, USA, 1993. [Google Scholar]
  138. Tarantola, A. Inverse Problem Theory and Methods for Model Parameter Estimation; Society for Industrial and Applied Mathematics: Philadelphia, PA, USA, 2005. [Google Scholar]
  139. Rai, C.S.; Hanson, K.E. Shear-wave velocity anisotropy in sedimentary rocks: A laboratory study. Geophysics 1988, 53, 800–806. [Google Scholar] [CrossRef]
  140. Valcke, S.L.A.; Casey, M.; Lloyd, G.E.; Kendall, J.M.; Fisher, Q.J. Lattice preferred orientation and seismic anisotropy in sedimentary rocks. Geophys. J. Int. 2006, 166, 652–666. [Google Scholar] [CrossRef]
  141. Kaderali, A.; Stanoev, T. On anisotropy and inhomogeneity parameter estimation using traveltimes. arXiv 2020, arXiv:2012.03393. [Google Scholar] [CrossRef]
  142. Toro, G.R. Uncertainty in shear-wave velocity profiles. J. Seismol. 2022, 26, 713–730. [Google Scholar] [CrossRef]
  143. Bernard, R.W.; Jason, T.D.; Thomas, S. Guidelines for Estimation of Shear Wave Velocity Profiles; PEER Report 2012/08; Pacific Earthquake Engineering Research Center: Berkeley, CA, USA, 2012. [Google Scholar]
  144. Sehizadeh, M.; Nasseri, M.H.; Young, R.P. The scaling effect on rock behaviour by comparing compressional wave velocities measured in the lab and field. Adv. Geosci. 2018, 45, 167–176. [Google Scholar] [CrossRef]
  145. Khalilidermani, M.; Knez, D. Shear Wave Velocity Applications in Geomechanics with Focus on Risk Assessment in Carbon Capture and Storage Projects. Energies 2024, 17, 1578. [Google Scholar] [CrossRef]
  146. Bekdaş, G.; Aydın, Y.; Işıkdağ, U.; Nigdeli, S.M.; Hajebi, D.; Kim, T.-H.; Geem, Z.W. Shear Wave Velocity Prediction with Hyperparameter Optimization. Information 2025, 16, 60. [Google Scholar] [CrossRef]
  147. Nath, F.; Asish, S.M.; Ganta, D.; Debi, H.R.; Aguirre, G.; Aguirre, E. Artificial Intelligence Model in Predicting Geomechanical Properties for Shale Formation: A Field Case in Permian Basin. Energies 2022, 15, 8752. [Google Scholar] [CrossRef]
  148. Al-Anazi, A.; Gates, I.D. A Support Vector Machine Algorithm to Classify Lithofacies and Model Permeability in Heterogeneous Reservoirs. Eng. Geol. 2010, 114, 267–277. [Google Scholar] [CrossRef]
  149. Zhou, Y.; Gu, Z.; He, C.; Yang, J.; Xiong, J. An Improved Decline Curve Analysis Method via Ensemble Learning for Shale Gas Reservoirs. Energies 2024, 17, 5910. [Google Scholar] [CrossRef]
  150. Zhao, J.; Jin, L.; Yu, X.; Azzolina, N.A.; Wan, X.; Smith, S.A.; Bosshart, N.W.; Sorensen, J.A.; Ling, K. Progress of Gas Injection EOR Surveillance in the Bakken Unconventional Play—Technical Review and Machine Learning Study. Energies 2024, 17, 4200. [Google Scholar] [CrossRef]
  151. Ali, N.; Fu, X.; Chen, J.; Hussain, J.; Hussain, W.; Rahman, N.; Iqbal, S.M.; Altalbe, A. Advancing Reservoir Evaluation: Machine Learning Approaches for Predicting Porosity Curves. Energies 2024, 17, 3768. [Google Scholar] [CrossRef]
  152. Rezaee, R.; Ekundayo, J. Permeability Prediction Using Machine Learning Methods for the CO2 Injectivity of the Precipice Sandstone in Surat Basin, Australia. Energies 2022, 15, 2053. [Google Scholar] [CrossRef]
  153. Zhang, H.; Mi, B.; Xi, C.; Liu, Y.; Guan, B.; Ning, L. Extraction of surface-wave phase velocities from ambient noise in the presence of local noise sources based on matched-field processing. J. Appl. Geophys. 2022, 204, 104755. [Google Scholar] [CrossRef]
  154. García-Martín, E.; Rodrigues, C.F.; Riley, G.; Grahn, H. Estimation of energy consumption in machine learning. J. Parallel Distrib. Comput. 2019, 134, 75–88. [Google Scholar] [CrossRef]
  155. Anand, A.; Deb, C. The potential of remote sensing and GIS in urban building energy modelling. Energy Built Environ. 2023, 5, 957–969. [Google Scholar] [CrossRef]
Figure 1. Diagram showing how seismic waves travel through different ground layers at different velocities and reflect back to space.
Figure 1. Diagram showing how seismic waves travel through different ground layers at different velocities and reflect back to space.
Energies 18 03310 g001
Figure 2. An animation image shows 10 seismic stations recording tremors from 2 tremors. By analyzing data gathered from stations that show decreased wave speeds, an anomaly is detected.
Figure 2. An animation image shows 10 seismic stations recording tremors from 2 tremors. By analyzing data gathered from stations that show decreased wave speeds, an anomaly is detected.
Energies 18 03310 g002
Figure 3. Illustration of remote sensing.
Figure 3. Illustration of remote sensing.
Energies 18 03310 g003
Table 1. Advantages and disadvantages of different geophysical methods.
Table 1. Advantages and disadvantages of different geophysical methods.
TechniqueAdvantagesDisadvantagesDepth RangeSpatial
Resolution
Seismic RefractionProvides depth-dependent shear wave velocity profile
High-resolution estimation of Vs
Cost-effective and efficient
Non-intrusive and minimal site disturbance
Requires controlled seismic sources
Limited to near-surface applications
Sensitive to geophone spacing and array design
Under 100 mhigh
Surface Wave AnalysisEstimates shear wave velocity variations with depth
Cost-effective and efficient
Minimal disturbance to the site
Requires long-wavelength surface waves
Extracting reliable dispersion curves is tough
Appropriate inversion methods are crucial
10–100 mintermediate
Seismic TomographyProvides detailed subsurface velocity models
Captures Vs variations in complex settings
High-resolution estimation of Vs
Requires seismic data from multiple receivers
Computational complexity
Sensitivity to initial model assumptions
Meters to a few kilometerslow
Table 2. Advantages and disadvantages of the remote sensing technique.
Table 2. Advantages and disadvantages of the remote sensing technique.
TechniqueAdvantagesDisadvantages
Remote SensingCost-effective and energy-efficient
Provides spatially extensive information
Non-intrusive and reduces fieldwork requirements
Enables monitoring of temporal variations
Relies on correlations with surface features
Requires ground truth data for calibration and validation
Uncertainties due to atmospheric and geometric effects
Correlations may vary with regional geological conditions
Table 3. Error estimation metrics comparison of AI models for shear wave velocity prediction.
Table 3. Error estimation metrics comparison of AI models for shear wave velocity prediction.
ReferenceAI ModelError Estimation
Rezaee et al. (2007) [11]NFSMSE = 0.001
Rajabi (2010) [95]GA, FL, NFMSE = 0.0153, 0.0084, 0.0184
Asoodeh and Bagheripour (2012) [96]FL, ANN, NFMSE = 0.0081, 0.0068, 0.0078
Maleki et al. (2014) [97]SVR, BPNNR2 = 0.97, 0.94
Bagheripour et al. (2015) [98]SVRR2 = 0.9716
Nourafkan et al. (2015) [99]ACOFISMSE = 0.0033, R2 = 0.9590
Anemangely et al. (2019) [101]LSSVM-COA, LSSVM-PSO, LSSVM-GAR2 = 0.929, 0.877, 0.868
Wang and Peng (2019) [102]ELM, ANNRMSE = 0.0795, 0.0913
Azadpour et al. (2020) [103]MLR2 = 0.941
Lian et al. (2020) [104]SVM, RF, MLPAverage R2 for ML models = 0.64
Khatibi and Aghajanpour (2020) [105]NNR2 = 0.9555
Olayiwola and Sanuade (2021) [109]LSSVMR2 = 0.9991
Miah (2021) [110]LSSVM-CSAR2 = 0.96
Wong et al. (2022) [112]ANN, MLRR2 = 0.86 (ANN), 0.36 (MLR)
Mehrad et al. (2022) [113]CNN, MELM, LSSVMR2 = 0.825, 0.826, 0.815
Laalam et al. (2022) [114]XGBoost, RFR, LR, AdaBoost, BRRR2 = 0.55 to 0.92, XGBoost outperforms others
Zhang et al. (2022) [115]CNNsR2 = 0.957
Kheirollahi et al. (2023) [118]MLR, ELM, ANNR2 = 0.99 (ANN), 0.96 (MLR, ELM)
Cova and Liu (2023) [119]GCN-BiGRUR2 = 0.947
Akinyemi et al. (2023) [120]LR, KNN, SVR, RFR, XGBoost, CatBoost, BPANNR2 = 0.92 to 0.94 (CatBoost had best result)
Khalilidermani and Knez (2023) [122]GEPR2 = 0.972
Mustafa et al. (2023) [121]ANNR2 = 0.96
Fu et al. (2024) [123]NNMAE = 38.89, LSTM MAE = 45.35
Gomaa et al. (2024) [124]ANNR2 = 0.58
Dehghani et al. (2024) [125]ANN, LR, RF, LB, SVMR2 = 0.8780, 0.8471, 0.8470, 0.9495, 0.8583, 0.7975
Leisi and Shad Manaman (2024) [128]ANNRMSE = 0.94
Table 4. Advantages and disadvantages of machine learning and data-driven approaches.
Table 4. Advantages and disadvantages of machine learning and data-driven approaches.
TechniqueAdvantagesDisadvantages
ML and Data-Driven ApproachesCaptures complex relationships in data
Offers high prediction accuracy
Can handle non-linear relationships
Utilizes available data effectively
Allows for incorporation of multiple data points
Requires large, labeled datasets
May suffer from overfitting
Interpretability of results may be challenging
Performance dependent on data quality
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Khalilidermani, M.; Knez, D.; Zamani, M.A.M. Shear Wave Velocity in Geoscience: Applications, Energy-Efficient Estimation Methods, and Challenges. Energies 2025, 18, 3310. https://doi.org/10.3390/en18133310

AMA Style

Khalilidermani M, Knez D, Zamani MAM. Shear Wave Velocity in Geoscience: Applications, Energy-Efficient Estimation Methods, and Challenges. Energies. 2025; 18(13):3310. https://doi.org/10.3390/en18133310

Chicago/Turabian Style

Khalilidermani, Mitra, Dariusz Knez, and Mohammad Ahmad Mahmoudi Zamani. 2025. "Shear Wave Velocity in Geoscience: Applications, Energy-Efficient Estimation Methods, and Challenges" Energies 18, no. 13: 3310. https://doi.org/10.3390/en18133310

APA Style

Khalilidermani, M., Knez, D., & Zamani, M. A. M. (2025). Shear Wave Velocity in Geoscience: Applications, Energy-Efficient Estimation Methods, and Challenges. Energies, 18(13), 3310. https://doi.org/10.3390/en18133310

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop