Subsurface Geological Structures Research Articles

Graph convolutional network for lithological classification and mapping using stream sediment geochemical data and geophysical data

Lithological mapping plays a vital role in acquiring a comprehensive understanding of subsurface rocks and geological structures. Traditional lithological mapping relies heavily on manual field surveys, with the accuracy of the mapping greatly constrained by the expertise of mappers, and thus this process is commonly slow and costly. In the present study, the graph convolutional network (GCN) method was proposed for efficient and automatic lithological mapping, using various geological survey data (e.g. geochemical and geophysical data). The GCN is not limited by the spatial distribution of sampling data, and hence it can be directly applied to the integrated resampling data that do not adhere to the regular spatial distribution patterns of Euclidean space. To evaluate the potential application of the GCN method in lithological classification and mapping, a case study was conducted in the Maqin region (China). Geochemical and geophysical data were utilized as the basis for lithological classification, as they can provide valuable information on geochemical compositions and physical properties of geological bodies, respectively. The study process was primarily divided into three parts. First, basic pre-processing was conducted on the geochemical data and the geophysical data, including centred log-ratio transformation and resampling. Subsequently, the resampled points were divided into training and testing sets in the proportion of 5.5:1. In the training set, 20% of the points are randomly selected as the validation set. Ultimately, using each resampled point as a node and the integrated geochemical and geophysical data as features, a GCN model and an undirected graph were constructed based on an adjacency distance of 1 km. The graph was employed by the constructed GCN model for lithological classification. The results reveal that the majority of lithological units can be accurately classified with an overall accuracy of 86%, indicating that the GCN model exhibits good applicability in lithological classification.

Deep Blind Fault Activity—A Fault Model of Strong Mw 5.5 Earthquake Seismogenic Structures in North China

North China is one of the high-risk areas for destructive and strong earthquakes in mainland China and has experienced numerous strong historical earthquakes. An earthquake of magnitude MW 5.5 struck Pingyuan County, Dezhou city, in Shandong Province, China, on 6 August 2023. This earthquake was the strongest in the eastern North China Craton since the 1976 Tangshan earthquake. Since the earthquake did not produce surface ruptures, the seismogenic structure for fault responsible for the Pingyuan MW 5.5 earthquake is still unclear. To reveal the subsurface geological structure near the earthquake epicenter, this study used high-resolution two-dimensional (2D) seismic reflection profiles and constructed a three-dimensional (3D) geometric model of the Tuqiao Fault by interpreting the faults in the seismic reflection profiles. This study further combined focal mechanism solutions, aftershock clusters, and other seismological data to discuss the seismogenic fault of the Pingyuan MW 5.5 earthquake. The results show that the Tuqiao Fault is not the seismogenic fault of the MW 5.5 earthquake. The actual seismogenic structure may be related to the NE-oriented high-angle strike-slip blind fault developed in the basement. We further propose three possible fault models for the strong seismogenic structure in North China to discuss the potential seismotectonics in this region.

Fine Shear-Wave Velocity Structures of Subsurface beneath the Guangdong–Hong Kong–Macao Greater Bay Area with Dense Seismic Array and SPAC Method

Abstract We apply the spatial autocorrelation (SPAC) method to construct the 3D subsurface shear-wave velocity structure model using the short-period dense seismic array (containing 725 nodal geophones) located at the Guangdong–Hong Kong–Macao Greater Bay area (GBA). We first divided the dense array into numerous subarrays, with each subarray consisting of nine nodal geophones, and obtained 562 subarrays that can provide 1D VS profiles of the same quantity. Then, the SPAC method and genetic algorithm are utilized to extract the dispersion curve of the Rayleigh wave from the raw microtremor data and invert VS structure, respectively. Finally, a 3D VS structure model from the surface to 3.3 km depth is derived by combining all 1D VS structures. Relatively low-velocity anomalies above 700 m are considered unconsolidated shallow sediments as well as relatively high-velocity anomalies beneath 1100 m are attributed to consolidated granite bedrock. Meanwhile, low-velocity anomalies that are identified through the vertical VS profile at a depth of about 900–3000 m can be contributed to the fractured zone, and striped low-velocity anomalies in the horizontal VS maps reveal the location of the deeply buried faults in the study area. The results also mean that the SPAC method combined with the records of short-period dense seismic array can be effectively applied to image subsurface structures in high-populated urban area. The development of this noise-resistance and environment-friendly geophysical technique provides a reliable and effective way to explore the complicated subsurface geological structures, which is of great significance to urban engineering construction and earthquake disaster reduction work in densely populated urban agglomerations.

Deep learning for high-resolution seismic imaging

Seismic imaging techniques play a crucial role in interpreting subsurface geological structures by analyzing the propagation and reflection of seismic waves. However, traditional methods face challenges in achieving high resolution due to theoretical constraints and computational costs. Leveraging recent advancements in deep learning, this study introduces a neural network framework that integrates Transformer and Convolutional Neural Network (CNN) architectures, enhanced through Adaptive Spatial Feature Fusion (ASFF), to achieve high-resolution seismic imaging. Our approach directly maps seismic data to reflection models, eliminating the need for post-processing low-resolution results. Through extensive numerical experiments, we demonstrate the outstanding ability of this method to accurately infer subsurface structures. Evaluation metrics including Root Mean Square Error (RMSE), Correlation Coefficient (CC), and Structural Similarity Index (SSIM) emphasize the model's capacity to faithfully reconstruct subsurface features. Furthermore, noise injection experiments showcase the reliability of this efficient seismic imaging method, further underscoring the potential of deep learning in seismic imaging.

Tomographic imaging of South Rewa basin, Central India: Implications of Gondwana rifting and Late Cretaceous volcanism

We have imaged thick (3.0–5.0 km) sub-trappean Gondwana sediments deposited in a typical rift-environment of the south Rewa basin in Central India masked by the Deccan basalts. The basalts intrude through the basin forming several dykes, sills, and other intrusions along with large-scale undulations of the basement showing complex geology and tectonic settings due to Late Cretaceous volcanism (∼65 Ma). To delineate complex subsurface geological structures and understand the tectonic settings, we use long-offset seismic reflection data along the N-S trending 127 km Kuvari-Shahdol profile of the basin. The conventional stack and pre-stack time migration techniques fail to image the complex subsurface structures dominated by steeply dipping faults. The steeply dipping faults with small-scale structures are influenced by strong lateral and vertical velocity variations that cause problems in imaging the basin. Hence, a robust tomographic inversion with pre-stack depth migration technique is adopted to image fine-scale subtle subsurface geological structures with smooth velocity variations. The tomographic velocity model and pre-stack depth migration seismic image could decipher basaltic trap thickness, numerous dyke intrusions, alternate horsts and grabens with deposition of thick hydrocarbon-bearing sub-trappean Gondwana rocks along with the basement configuration. The deep basinal faults with highly distorted basement structures represent the intense tectonic activity of this Gondwana rift-basin. The deep-seated hydrocarbon reservoir is discovered with an excellent trapping mechanism forming both pre-rift and syn-rift settings of the rift-basin. The seismic image is further constrained and corroborated by the inversion of residual Bouguer gravity anomaly data to obtain corresponding density model derived from tomographic velocity model. The large basement faults, horsts and grabens, several intrusives like dyke, sill, laccolith, and thick sub-trappean Gondwana formations with Deccan trap cover indicate complex geology and tectonic settings of south Rewa basin forming the Late Cretaceous Volcanic Province of India.

Geological Structure and Possible Mineralized Zones in Ikungi – Manyoni Area, Tanzania: Delineated from High-resolution Aeromagnetic Data

Understanding geological structures is important in mineral exploration. Delineating subsurface geological structures, structural intersections, and complexities of potential geological settings for the formation of hydrothermal mineral deposits, such as gold requires using magnetic data, particularly, high-resolution aeromagnetic data. To define possible targets for mineral exploration, this study used linear detection and the Centre for Exploration Targeting (CET) techniques on high-resolution aeromagnetic data for structural delineation and interpretation. The study's conclusions indicate that the majority of geological features, such as faults, shears, fault intersections, and fracture systems, trend NE-SW, N-S, and occasionally NW-SE. These structures create potentially large areas that have a strong chance of becoming mineralized. The CET grid analysis technique defined mineralized zones, particularly the structural intersections which are potential targets in mineral exploration. The effectiveness of the CET technique in mineral prospecting was demonstrated by the correlation between the results and existing mineral occurrences (such as Shanta Gold Mine, abandoned pits, and the operating pits of small-scale miners in Muhintiri). Based on the findings reported herein, any sites that share characteristics with those that have demonstrated gold mineralization should be prioritized for ground follow-up

Near surface resistivity investigation of regolith of Middle Benue Trough: implications on road pavement along Lafia-Shendam Axis, North Central Nigeria

ABSTRACT Electrical characteristics and subsurface structures associated with the regolith of Middle Benue Trough have been investigated. Vertical electrical sounding (VES) and constant separation traversing (CST) techniques were employed for data acquisition along Lafia-Shendam Road, Lafia, North Central Nigeria. Forty (40) Schlumberger sounding stations and four (4) Wenner resistivity profiles were occupied along opposite sides of the road to delineate and decipher the subsurface geological structures. Second-order Dar Zarrouk parameters (longitudinal and transverse resistivities, mean resistivity and coefficient of anisotropy) computed from VES parameters were also used to evaluate the pseudo-anisotropy characteristics of the regolith cover. Geoelectric sequences delineated within the study area include thin topsoil, alluvium deposit, weathered layer (laterite and saprolite) and fresh basement. The regolith is hosted by a weathered sedimentary rock at an average depth which varies from 10.2 to 37.3 m. The regolith structure is not complex and has a greater proportion of conductive laterite. The saprolite associated with the host sedimentary rock, masked in the VES technique, is revealed as a very thin zone with moderate resistivity in the CST 2-D structures. The predominance of laterite in the regolith is also corroborated by the range of values of the coefficient of anisotropy and mean resistivity.

Electrical Resistivity Tomography (ERT) for investigating subsurface structures in the karst area of the Ek Leuntie Cave, Aceh Province

Ek Leuntie Cave, in the southern city of Banda Aceh, Aceh Province, is recognized as an educational media based on geopark sites because it holds a 7400-year paleotsunami track record. As a karst area, Ek Leuntie Cave is vulnerable to the impact of geological disasters such as earthquakes, sinkholes and subsidence. As an effort to reduce the negative impact of geological disasters, geophysical studies have been applied in the Ek Leuntie Cave Karst area. The study aims to identify the subsurface geological structure of the Karst area. The identification was carried out using the Electrical Resistivity Tomography (ERT) method of Wenner-Schlumberger configuration applied to two measurement trajectories that have South-North and West-East directions in front of the Ek Leuntie Cave door. ERT data acquisition results were then modeled in inversion using Res2Dinv software to produce a 2D model of the subsurface geological structure of the Ek Leuntie Cave karst area based on variations in rock resistivity values. Interpretation of subsurface geological structure modeling is validated with shallow drilling data (Hand Auger). The inversion modeling results show that the rock resistivity value in the Ek Leuntie Cave area reaches a depth of 30 m which has a resistivity value of 0 - 1000 Ωm. The results of the interpretation of resistivity values and validation of drill data show that the Ek Leuntie Cave area is covered by a layer of sand with fine to coarse texture and the layer below is a layer of limestone at a depth of ± 0.4 meters with a resistivity value of 50 - 250 Ωm. Limestone generally shows porosity with the possibility of caves or pores in the rock formation at Ek Leuntie Cave, making it vulnerable to damage from earthquakes, landslides, and subsidence. The results of the study are expected to provide updated information in efforts to develop the Ek Leuntie Cave geopark area.

3D seismic analysis of mine planning using Aczel–Alsina aggregation operators based on T-spherical fuzzy information

3D seismic attributes analysis can help geologists and mine developers associate subsurface geological features, structures, faults, and ore bodies more precisely and accurately. The major influence of this application is to evaluate the usage of the 3D seismic attributes analysis in gold mine planning. For this, we evaluate the novel theory of complex T-spherical hesitant fuzzy (CTSHF) sets and their operational laws. Furthermore, we derive the CTSHF Aczel–Alsina weighted power averaging (CTSHFAAWPA) operator, CTSHF Aczel–Alsina ordered weighted power averaging (CTSHFAAOWPA) operator, CTSHF Aczel–Alsina weighted power geometric (CTSHFAAWPG) operator, and CTSHF Aczel–Alsina ordered.com weighted power geometric (CTSHFAAOWPG) operator. Some properties are also investigated for the above operators. Additionally, we evaluate the problems of 3D seismic attributes analysis to mine planning under the consideration of the proposed operators, for this, we illustrate the problem of the multi-attribute decision-making (MADM) technique for the above operators. Finally, we demonstrate some examples for making the comparison between prevailing and proposed information to improve the worth of the derived operators.

Application of ERT, IP and VLF-EM Methods to Investigate Landslide-Prone Structures at Archaeological Sites in Lamreh, Aceh Besar, Indonesia

Since the 21st century, Lamreh, Aceh Besar Regency, Indonesia played an important role during the maritime silk route at the gate of the Malacca strait. This article investigates the subsurface structure of landslide-prone areas in cultural heritage based on 2D resistivity, chargeability and current density models of ERT, IP and VLF-EM methods, respectively. Field data measurements were carried out on 2 crossing profiles along the cliff suspected of experiencing landslides. The length of each profile is 220 with 4 m distance between stations. The 2D models reveal that the subsurface geological structure of Lamreh is composed of a mixed layer of clastic sediment and volcanic material at the top, followed by a layer of calcareous sandstone, and volcanic breccia at the bottom. The 3 layers are most easily distinguished in the resistivity model. The topmost layer is permeable but dry, i.e., characterized by a more resistive layer in the models. While the second layer is characterized by an intermediate conductivity and the bottom layer is highly conductive. The conductivity in these 2 layers is influenced by the degree of water content within the rocks. The chargeability models derived from IP data can distinguish between the dry layer on the surface and the saturated layer below. Meanwhile, the current density models obtained from VLF-EM data have proven the presence of fractures and faults along the profiles due to weathering as also seen in the resistivity models. HIGHLIGHTS Investigation of subsurface structure in landslide-prone area of Lamreh, Aceh Besar Regency, Indonesia, a cultural heritage site with historical significance in the maritime silk route. Utilization of 2D resistivity, chargeability and current density models (ERT, IP and VLF-EM) to analyze the subsurface structure. Field data measurements conducted on 2 crossing profiles along the suspected landslide cliff, with a profile length of 220 and 4 m station distance. Subsurface geological structure of Lamreh identified as a mixed layer of clastic sediment and volcanic material at the top, followed by calcareous sandstone and volcanic breccia at the bottom. Resistivity models most effectively distinguish the 3 layers, highlighting permeable but dry top layer, intermediate conductivity in the second layer, and high conductivity in the bottom layer influenced by water content. Chargeability and current density models confirm surface dryness and presence of fractures and faults due to weathering GRAPHICAL ABSTRACT

Re-evaluation of a geothermal system model based on high-resolution gravity field inversion: A case study of the Maritaing geothermal field, Alor Regency, East Nusa Tenggara Province, Indonesia

The Maritaing region in East Nusa Tenggara Province, Indonesia, offers significant geothermal potential for direct power generation and use. This study aims to re-evaluate the geothermal system model in the Maritaing geothermal field using high-resolution gravity inversion. High-resolution gravimetric data were collected using satellite observations and data processing techniques, allowing a better understanding of the subsurface geological structure and density distribution of rocks associated with the geothermal reservoir. This study applied data corrections to acquire complete Bouguer anomalies, gravity gradient analysis to detect geological structures, and inversion modeling of residual gravity anomalies to predict subsurface rock density distribution. The complete Bouguer anomalies exhibit positive values ranging from 70 to 140 mGal and have a northwest–southeast orientation, indicating the existence of andesitic lava and pyroclastic volcanic rocks. Gravity gradient analysis determines the geological structures, such as faults or fractures, used as conduits to create Kura Hot Springs. The model provides useful insights into geological occurrences, such as faults, rock layers, and reservoir accumulations that support geothermal fluids. Based on the present study’s findings, we provide recommendations for developing and managing geothermal resources in the Maritaing area.

The effectiveness of very low-frequency electromagnetics (VLF-EM) method in detecting buried targets at a controlled site

The ever-increasing anthropogenic activities that pose a significant threat to environmental security and sustainability have spurred geophysicists to review enhance geophysical techniques for shallow geophysical investigations, especially in identifying illegal buried materials. This article applied very low-frequency electromagnetic (VLF-EM) at an experimental geophysical site (EGS) to examine the geophysical response over various buried targets. The VLF-EM data acquired on the site with and without buried targets demonstrate the nature of the anomalies and the characteristic signals of the buried targets. There are significant variations in the anomalies source-bodied between the site with and without buried targets. The result of the pre-burial investigation shows no major contrast in the equivalent current density values across the site without targets. Minors and major anomalies were encountered after burying the targets corresponding to the buried targets. Some signals become stronger over a large buried target. However, there were a few undetected targets and some cases of unsatisfied results, which were also discussed. The performance of the VLF-EM method in terms of depth estimation was also examined. A significant variation was noted due to the presence of the buried targets and it was noted that the current density seems to always emerge maximally and minimally around the conductor targets and non-conductive targets, respectively. The failure of the VLF-EM to detect the buried pipes in various orientations was examined. The VLF-EM method is more consistent at small spacing and it can be very useful for identifying underground metallic and non-metallic targets. The study successfully provides useful information to complement the complexity of the use of geophysical methods while enhancing the subsurface information and understanding of VLF-EM anomalies or responses generated by various targets such as subsurface geological structures, buried waste contaminants and underground utilities to boost environmental studies and engineering investigations.

Reflection seismic imaging of subsurface geologic structures in the Bayan Obo mining area, China

A dense 2D seismic reflection survey was conducted at the Bayan Obo rare-earth-elements-Nb-Fe deposit in China to explore its geologic features and mineral endowment. A composite processing workflow was established for prestack depth imaging of this data set, which involved static correction to compensate for shifts caused by lateral inhomogeneities in near-surface layers and rugged topography. Standard methods were used to attenuate different types of noise and improve the signal-to-noise ratio of the data. A velocity model in the depth domain was obtained from refraction tomography and velocity analysis, and then used in migration. The preliminary interpretation of the final image revealed the mapping of mineralization at its true position in depth, indicating a relatively deep extension (approximately1.5 km). The results demonstrate the potential of seismic imaging as a complementary exploration tool to gravity and electromagnetic methods for mineral exploration purposes in the Bayan Obo mining area, providing a high-resolution seismic image and allowing for depth characterization of the mineral deposits. In addition, the images provide important information for guiding the selection of drilling borehole locations for deep exploration.

Enhancement of the balanced total horizontal derivative of gravity data using the power law approach

Mapping the boundaries of subsurface geological structures constitutes a fundamental aspect of interpreting gravity data. Various methods relying on gravity derivatives have been introduced to delineate these boundaries. However, these approaches are not without their limitations, often leading to the generation of false edges or resulting in divergent outcomes. This article introduces a novel method that utilizes the power law approach and balanced total horizontal derivative for mapping gravity data edges. To assess the robustness of the proposed technique, we conducted tests on model examples and applied it to a real case study in northwest Vietnam. The results demonstrate that our method can accurately and clearly map edges in comparison to existing methodologies, while also preventing the generation of spurious edges in the output maps.

Interpretation of vertical electrical soundings by means of ISM, PHT and C-N, case of a ditch in Al-Sharquieh phosphate mine in Syria

The main objective of this paper is basically to calibrate and compare the different geoelectrical results obtained by the three interpretative techniques of the inverse slope method (ISM), Pichgin and Habibuleav (PHT) and concentration-number (C-N) fractal modeling with the well lithology of an existing known ditch in the Al-Sharquieh phosphate mine in Syria. Those interpretative techniques are applied herein to interpret the vertical electrical soundings (VES) points measured along a parallel profile to the ditch. ISM reveals the presence of four to five geoelectrical layers in the study region. The PHT allows the determination of the tectonic subsurface features of the study region, by analyzing the non homogeneity points and their distributions along a given geoelectrical profile. The fractal C-N modeling provides the different apparent resistivity ranges dominating in the study region. Good agreements are obtained between the direct observed field lithological descriptions of both wall sides of this ditch, and the results obtained by those three interpretative geoelectrical techniques. The high radioactive readings exceeding four hundred count per second (cps) are due to the presence of secondary uranium accompanied with fractures, cracks and joints. The secondary uranium occurrences accompanied by phosphate deposits are found in the study area in the fractured zones, as determined by the geoelectrical PHT technique. The three mentioned geoelectrical methodologies are therefore used as indirect techniques for delineating the radioactive zones related to uranium prospecting. It is a successful application of the joint use of those three techniques, that could be applied in mining to delimit the subsurface geological structure. The reliability and efficacy of (ISM), (PHT), and (C-N) fractal modeling techniques are consequently well confirmed, and it is therefore recommended to use and apply them for characterizing and describing similar lithological phosphatic sequences, and for several shallow mining, environmental and civil engineering applications.

GEOLOGICAL INTERPRETATION OF NORTH MADURA OFFSHORE SEDIMENTARY BASIN BASED ON GRAVITY DATA ANALYSIS

The area of north Madura offshore is one of the sedimentary basins within the North East Java Basin which has interesting paleogeographical developments. These developments are the result of various sedimentation processes and changes in the depositional environments. The purpose of this study is to identify subsurface geological structures, delineate sedimentary sub-basins, and estimate sediment thickness. For that, we do spectrum analysis, optimum upward continuation filtering, spectral decomposition analysis, 2D forward modeling and 3D inversion of gravity data. The results of the spectral analysis show that the average thickness of sedimentary rocks in the study area is around 3.788 km. From the gravity data analysis, 10 sedimentary sub-basins were delineated, indicating structural patterns of basement high, grabens, and faults. 2D and 3D modeling revealed four layers of stratigraphy, in order from youngest to oldest are Tertiary-Neogene sedimentary rocks with a density value of 2.1 gr/cc, the Tertiary-Neogene sedimentary rocks with a density of 2.3 gr/cc, the Tertiary - Paleogene sedimentary rocks with a density of 2.45 gr/cc, and the basement layer identified as volcanic igneous rock with a density of 2.67 gr/cc.

Delineation of geological structures of Arta geothermal prospect in Djibouti based on the gravity data analysis and interpretation

The Arta geothermal prospect is located in the southern part of the Gulf of Tadjoura, 45 km west of Djibouti city. This study uses 398 gravity data stations collected by the Djiboutian Office for Geothermal Energy Development (ODDEG) from February to May 2018. The geothermal activity of the Arta area is revealed by surface thermal manifestations, such as near-boiling fumaroles (98 °C). This study aims to delineate the subsurface geological structures responsible for surface geothermal manifestations and define the depth and geometry of the geothermal reservoir in the study area. A Complete Bouguer anomaly was obtained using a Bouguer density of 2.55 g/cm3, derived as an average density from the F-H, G-H and CVUR methods. A regional and residual anomaly trend separation method was applied to focus on the structures of interest in the study area, and a residual Bouguer anomaly map was obtained. Edge detection filters, including Horizontal Derivatives (HD) and an Improved Normalized Horizontal tilt angle (INH), were used to examine the gravity data for edge identification. The density was then modelled in three-dimensional (3-D), utilizing the VOXI Earth Modeling Tool performed in Geosoft Oasis Montaj. The model was constrained using estimated lower (-0.35 g/cm3) and upper (0.65 g/cm3) bound density values from the expected density values of the local geology. The Bouguer and residual anomaly maps show a high-density body in the central part, surrounded by low-anomaly structures. The high-density body is interpreted as an intrusive body covering the basement of the study area. The anomalies imaged from HD and INH indicate a significant number of high gradient anomalies trending N-S, NE-SW, and NW-SE, corresponding to the trend of the geological structures of the Arta geothermal prospect. These high-gradient anomalies can be interpreted as faults, contact zones, or geological boundaries. The 3-D model imaged a high-density anomaly trending N-S in the central part at 2 and 3 km below sea level with a volume of about 126 km3 and an average block density of 2.66 g/cm3, interpreted as magma intrusion or intrusive volcanic rocks. The 3-D vertical cross-sections revealed that the intrusion is visible at depth and that some intrusions could be seen at the surface and near surface. The findings of this gravity study will contribute to the overall and future geothermal exploration and development of the Arta geothermal prospect.

An ensemble-based machine learning solution for imbalanced multiclass dataset during lithology log generation

The lithology log, an integral component of the master log, graphically portrays the encountered lithological sequence during drilling operations. In addition to offering real-time cross-sectional insights, lithology logs greatly aid in correlating and evaluating multiple sections efficiently. This paper introduces a novel workflow reliant on an enhanced weighted average ensemble approach for producing high-resolution lithology logs. The research contends with a challenging multiclass imbalanced lithofacies distribution emerging from substantial heterogeneities within subsurface geological structures. Typically, methods to handle imbalanced data, e.g., cost-sensitive learning (CSL), are tailored for issues encountered in binary classification. Error correcting output code (ECOC) originates from decomposition strategies, effectively breaking down multiclass problems into numerous binary subproblems. The database comprises conventional well logs and lithology logs obtained from five proximate wells within a Middle Eastern oilfield. Utilizing well-known machine learning (ML) algorithms, such as support vector machine (SVM), random forest (RF), decision tree (DT), logistic regression (LR), and extreme gradient boosting (XGBoost), as baseline classifiers, this study aims to enhance the accurate prediction of underground lithofacies. Upon recognizing a blind well, the data from the remaining four wells are utilized to train the ML algorithms. After integrating ECOC and CSL techniques with the baseline classifiers, they undergo evaluation. In the initial assessment, both RF and SVM demonstrated superior performance, prompting the development of an enhanced weighted average ensemble based on them. The comprehensive numerical and visual analysis corroborates the outstanding performance of the developed ensemble. The average Kappa statistic of 84.50%, signifying almost-perfect agreement, and mean F-measures of 91.04% emphasize the robustness of the designed ensemble-based workflow during the evaluation of blind well data.

An attempt to use convolutional neural network to recover layered-earth structure from electrical resistivity tomography survey

The Electrical Resistivity Tomography (ERT) method is a geophysical prospecting technique for determining subsurface structure in terms of electrical resistivity distribution. Direct current is injected into the ground through a pair of electrodes, and the potential difference is measured using another pair of electrodes. The estimated resistivity model of the subsurface is then obtained using a mathematical optimization algorithm called the inversion algorithm, which incorporates multiple measured potential differences from different electrode configurations as the input data. The inverted resistivity model serves as a resource for the interpretation process of the survey. Specialists employ the model and the prior geological information in the area to extract the subsurface geological structure under the data measurement profile. The inverted model can be treated as a two-dimensional image in which each element represents resistivity. The image can be visualized as a blurred or distorted version of the actual resistivity image of the subsurface. We, therefore, developed a post-inversion image enhancement code using a neural network aiming to recover the true subsurface model with predetermined features. We employed the convolutional neural network (CNN) that uses the inverted models as input and returns an enhanced model as output. We tested our developed neural network with a case study that utilized the proposed method to recover the layered earth structure which is often found in real surveys. The electrode spacing was 5 meters in the Schlumberger array configuration. The true subsurface resistivity models of training data and validating data for evaluating the neural network were randomly generated based on the predetermined feature of the layered structure. We then generated their respective inverted models from those data through simulated inversion routine. As a result, 94.75% of the enhanced models in the validation dataset exhibited a lower loss metric compared to the original inverted models in the same validation dataset. We also applied the proposed method to practical field data to recover the subsurface layers and compared them with stratigraphic information from a borehole in each area. The result showed that the proposed method gives well-resolved structures in the enhancement process after the typical inversion.

Structural mapping of the west central Arabian Shield (Saudi Arabia) using downward continued magnetic data

In this study, we have applied a stable downward continuation technique to magnetic anomalies from the global geomagnetic field model MAG2V3 to improve the resolution of magnetic data of the west central Arabian Shield, Saudi Arabia. Then, advanced interpretation methods such as the tilt angle of the horizontal gradient (TAHG), horizontal gradient of a modified tilt angle (HGSTDR), and enhanced horizontal gradient (EHGA) have been applied to downward continued magnetic dataset to highlight the main structural features in the region. The observed geologic structures from the magnetic analysis revealed NW-SE, NE-SW, and ENE-WSW orientations in the west central Arabian Shield. The depths of these structures were computed by the analytic signal (AS) technique and the obtained result shows depths varying from 1.2 to 9.2 km. This study provided a new map of the subsurface geological structures for a better understanding of the tectonic evolution of the west central Arabian Shield.