TL;DR
A machine learning-assisted geophysical–geotechnical framework that integrates Electrical Resistivity Tomography, Seismic Refraction Tomography, and borehole-derived Standard Penetration Test data to improve subsurface characterization and engineering site assessment is presented.
The presence of near-surface cavities poses a significant geohazard due to potential ground subsidence and structural collapse. To mitigate threats to urban stability, this study presents an integrated geophysical framework to locate and characterize abandoned mining galleries and exploitation voids near Linares (Jaén, Spain). The approach combines four complementary techniques: electrical resistivity tomography (ERT), ground-penetrating radar (GPR), frequency-domain electromagnetics (FDEM), and microgravity. The resulting multi-physics responses were cross-referenced with visible surface subsidence features and archival mine plans. Air-filled galleries and shafts generated highly pronounced high-resistivity anomalies. Shallow voids detected at depths of 2–5 m were undocumented in 19th-century mining maps, suggesting older historical origins, whereas deeper ERT profiles and structural disturbance trends (up to 30 m) correlated well with historical records. Within this framework, FDEM provided high-resolution lateral mapping, GPR excelled at resolving ultra-shallow structural boundaries, and ERT characterized deep gallery networks. Crucially, microgravity mitigated inversion non-uniqueness by directly confirming physical mass deficits over the anomalies. This integrated workflow overcomes individual resolution limits, offering a practical tool for land-use planning and early geohazard risk assessment in collapse-susceptible areas.
Accurate delineation of gold mineralization in deeply weathered tropical terrains requires the integration of indirect geophysical data with direct subsurface observations. Electrical resistivity imaging (ERI) and induced polarization (IP) are widely used in mineral exploration; however, interpretation is often limited by non-uniqueness without borehole validation. This study evaluates the reliability of combined resistivity and induced polarization (Res-IP) methods for identifying sulphide-associated gold mineralization in the Central Belt of Peninsular Malaysia (CBPM) through correlation with borehole lithological data. The investigation was conducted in Jeli, Kelantan, a known gold-bearing region, using two intersecting 400 meters survey lines (SL-1 and SL-2) acquired with 61 electrodes at 5 to 10 meters Schlumberger spacing to generate two-dimensional (2D) Res-IP models. Boreholes Y (BH-Y) and Z (BH-Z) were drilled along SL-1 to validate geophysical anomalies, with detailed geological logging performed at 1.5 meters intervals to document lithology, alteration, quartz veining, and sulphide occurrence. The results reveal a consistent geophysical signature characterized by low resistivity (<100 Ωm) and high chargeability (>100 msec), indicative of disseminated to vein-hosted sulphides. These signatures correlate directly with quartz–sulphide vein intersections at depths of 70 to 72 meters in BH-Z and 102 to 102.6 meters in BH-Y, confirming subsurface structural and mineralogical continuity. This study establishes calibrated resistivity–chargeability thresholds for sulphide-bearing formations in tropical Malaysian geology, improving ERI interpretive confidence in highly weathered environments. The findings highlight the value of integrated geophysical and geological approaches for enhancing drill targeting and reducing exploration uncertainty in gold-prospective regions of Malaysia.
Accurate prediction of reservoir properties such as porosity, permeability, and lithofacies distribution is essential for reliable hydrocarbon reserve estimation, well placement, and field development planning. Conventional deterministic methods for relating seismic response to reservoir properties are limited by the nonlinear and multivariate nature of the seismic–petrophysical relationship, and single-attribute correlations frequently fail to capture the complexity of clastic reservoir systems. This study presents a machine learning-assisted workflow that integrates multi-attribute seismic analysis with wireline log data to predict porosity and classify lithofacies within a clastic hydrocarbon reservoir. A suite of seismic attributes including root-mean-square (RMS) amplitude, sweetness, instantaneous frequency, coherence, envelope amplitude, and acoustic impedance derived from post-stack inversion was extracted and calibrated against log measurements at control wells. Feature selection based on correlation ranking and recursive feature elimination identified the most informative attributes, and three supervised learning models random forest (RF), support vector regression (SVR), and a feed-forward artificial neural network (ANN) were trained for porosity prediction. A separate classification stage using random forest and gradient boosting was applied for lithofacies discrimination. The results demonstrate that the ANN and RF models substantially outperform single- and multi-attribute linear regression, achieving a coefficient of determination (R²) of 0.86–0.88 for porosity prediction at blind-test wells, while the ensemble classifier reached an overall lithofacies classification accuracy of 88%. Integrating machine learning with multi-attribute seismic analysis reduces prediction uncertainty and produces spatially continuous reservoir property volumes that support more reliable reservoir characterization and lower-risk exploration decision-making.
Earthquake-induced landslides and liquefaction often occur within the same seismic event but may be controlled by different combinations of ground shaking, topography, hydrology, site conditions, and tectonic setting. This study develops an interpretable remote-sensing framework to map and compare these secondary hazards after the 2025 Dingri Ms 6.8 earthquake in the southern Tibetan Plateau. Multi-source optical and SAR indicators, seismic variables, and geo-environmental factors were integrated using ensemble machine-learning models. A peak ground acceleration (PGA)-guided multivariate Gaussian mixture model was used to stratify hazard samples into PGA-associated statistical regimes, and regime-wise XGBoost models were interpreted using SHAP. The results show that landslide predictions are mainly associated with shaking intensity and topographic conditions, whereas liquefaction predictions show stronger regime-dependent associations with site conditions, hydrological proximity, geomorphic setting, and near-fault effects. The proposed framework provides a practical way to compare model-inferred feature associations across different shaking and environmental backgrounds. The results highlight that coseismic secondary hazards in high-altitude tectonic regions are not governed by a single uniform relationship, but by spatially heterogeneous combinations of seismic and environmental conditions associated with hazard occurrence.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
