Jaewook Lee

Because pore fluids have less influence on shear waves, S-wave velocity (VS) directly measures a rock frame’s stiffness. However, direct VS measurements are scarce in the vintage wells of the Delaware Basin. Although empirical and traditional machine-learning methods have been used to predict VS from conventional well logs, their estimates often depend on specific geologic formations or boundary conditions. In this study, we develop a self-attention, bidirectional long short-term memory (BiLSTM) model with depth-ordered sequences and automated hyperparameter optimization to overcome these limitations, requiring no prior geologic information. Trained on P-wave velocity, density, total porosity, and gamma-ray logs from 123 wells, the model captures nonlinear relationships in the data, achieving an R2 of 0.85 and surpassing existing empirical and regression-based approaches. Further validation in blind tests with eight wells in different counties demonstrates accurate VS predictions throughout the basin where direct VS logs are missing. To interpret the model’s “black box,” we compute Shapley values to quantify each input’s contribution to VS predictions. In addition, the BiLSTM’s superior performance extends to geomechanical properties, such as bulk modulus (K), shear modulus (G), Young’s modulus (E), and Poisson’s ratio (ν), highlighting its practical utility in seismic applications such as energy exploration and development, geothermal energy, carbon and hydrogen storage, and induced seismicity monitoring.

S‐wave velocity plays a crucial role in various applications but often remains unavailable in vintage wells. To address this practical challenge, we propose a machine learning framework utilizing an enhanced bidirectional long short‐term memory algorithm for estimating S‐wave sonic logs from conventional logs, including P‐wave sonic, gamma ray, total porosity, and bulk density. These input logs are selected based on traditional rock physics models, integrating geological and geophysical relations existing in the data. Our study, encompassing 34 wells across diverse formations in the Delaware Basin, Texas, demonstrates the superiority of machine learning models over traditional methods like Greenberg–Castagna equations, without prior geological and geophysical information. Among these machine learning models, the enhanced bidirectional long short‐term memory model with self‐attention yields the highest performance, achieving an R‐squared value of 0.81. Blind tests on five wells without prior geologic information validate the reliability of our approach. The estimated S‐wave velocity values enable the creation of a basin‐scale S‐wave velocity model through interpolation and extrapolation of these prediction models. Additionally, the bidirectional long short‐term memory model excels not only in predicting S‐wave velocity but also in estimating S‐wave reflectivity for seismic amplitude variation with offset applications in exploration seismology. In conclusion, these S‐wave velocity estimates facilitate the prediction of further elastic properties, aiding in the comprehension of petrophysical and geomechanical property variations within the basin and enhancing earthquake hypocentral depth estimation.

An accurate estimation of the shale permeability is essential to understand heterogeneous organic-rich shale reservoir rocks and predict the complexity of pore fluid transport in the rocks. However, predicting the matrix permeability by traditional models is still challenging because they require information often measured from core measurements. First, Kozeny’s equation (Kozeny, 1927) uses porosity and specific surface area of solid grains. However, it is difficult to characterize the specific surface area values or grain sizes from the logs. Second, Herron’s method (Herron, 1987) has been used for predicting permeability based on the mineral contents provided by well log data in conventional sandstone reservoirs. However, the predictive accuracy is low due to the different pore network structures of the shales. In this study, we estimate shale matrix permeability by a combined exploratory data analysis (EDA) and nonlinear regression estimation from the wireline logs. First, we conduct a bivariate correlation analysis for permeability and rock properties in core measurements. According to the correlation and Shapley value sensitivity test, we find that permeability change has a significant effect on the variation in porosity. Also, we investigate a nonlinear behavior between porosity and permeability. Second, we derive a nonlinear polylogarithmic estimation function of porosity to permeability, comparing it to the multivariate linear regression of porosity and clay volume fraction. As a result, a cubic logarithmic function of porosity significantly improves the fitting performance of the permeability values, better than the traditional methods. Moreover, we generate the permeability logs from the calibrated porosity logs, and they imply better shale permeability prediction as well. Since we can invert the porosity distribution from seismic data, this approach can provide a more accurate permeability estimation and reliable fluid flow modeling for shale and mudrock.

An accurate estimation of the shale permeability is essential to understand heterogeneous organic-rich shale reservoir rocks and predict the complexity of pore fluid transport in the rocks. However, predicting the matrix permeability by traditional models is still challenging because they require information often measured from core measurements. First, Kozeny’s equation (Kozeny, 1927) uses porosity and specific surface area of solid grains. However, it is difficult to characterize the specific surface area values or grain sizes from the logs. Second, Herron’s method (Herron, 1987) has been used for predicting permeability based on the mineral contents provided by well log data in conventional sandstone reservoirs. However, the predictive accuracy is low due to the different pore network structures of the shales. In this study, we estimate shale matrix permeability by a combined exploratory data analysis (EDA) and nonlinear regression estimation from the wireline logs. First, we conduct a bivariate correlation analysis for permeability and rock properties in core measurements. According to the correlation and Shapley value sensitivity test, we find that permeability change has a significant effect on the variation in porosity. Also, we investigate a nonlinear behavior between porosity and permeability. Second, we derive a nonlinear polylogarithmic estimation function of porosity to permeability, comparing it to the multivariate linear regression of porosity and clay volume fraction. As a result, a cubic logarithmic function of porosity significantly improves the fitting performance of the permeability values, better than the traditional methods. Moreover, we generate the permeability logs from the calibrated porosity logs, and they imply better shale permeability prediction as well. Since we can invert the porosity distribution from seismic data, this approach can provide a more accurate permeability estimation and reliable fluid flow modeling for shale and mudrock.

It is important to perform the quantitative interpretation of the continental margin lithosphere for a more accurate and comprehensive understanding of its tectonic behavior. In this study, we derived the seismic and elastic properties from the 2D seismic data recorded by the Korea Meteorological Administration (KMA) in 2014. In general, most of the previous researches have been based on travel time tomography or Full Waveform Inversion (FWI) methods. However, these methods are not robust to directly apply to the seismic ambient noise data due to its low signal-to-noise ratio (SNR). Therefore, we conducted L2-norm model-based impedance inversion to not only delineate the local geological structures but also suppress the meaningless footprints in the observed data. Moreover, we used FWEA18 models (Tao et al., 2018) as initial velocity and density models to create the inversion result more stable and accurate. Then, we interpreted the lithosphere and asthenosphere from the inverted P-impedance model, which is more obvious than the interpretation of the pre-existing data. The average depths of Moho and Lithosphere-Asthenosphere-Boundary (LAB) are 30 km and 80km, respectively. Furthermore, we estimated the change of bulk density as well as P- and S-wave velocities of the crust, lithospheric mantle, and asthenospheric mantle. Also, we predicted four elastic properties of each layer from the inverted seismic properties, such as bulk modulus, shear modulus, Young’s modulus, and Poisson’s ratio. These model results can help to understand the physical state and elastic behavior variations of the lithospheric and asthenospheric mantle as well as local lithospheric structures beneath the peninsula.