The underlying public dataset, called the Synthetic Oil Reservoir Prediction Dataset, contained seven original predictors: rock type, porosity, permeability, trap type, seismic score, proximity to an oil field and estimated reservoir depth. Its binary outcome recorded whether oil was present or absent. The full dataset included 2,004 negative records, or 72.19%, and 772 positive records, or 27.81%. Before modeling, the researchers used a stratified split with random seed 42 to create a development set of 2,220 records and a held-out test set of 556, preserving class proportions in both subsets. The test records were not used for imputation, feature engineering, class balancing, feature selection, hyperparameter tuning, model selection or probability-threshold selection.

That separation was central to the study’s leakage-controlled design. Within each cross-validation fold, preprocessing parameters were estimated only from the training portion. Missing numerical values were replaced with the training-fold median, while missing categorical values were filled with the training-fold mode. Data-consistency checks examined duplicated records, non-finite values, negative permeability, porosity outside the dataset’s admissible range and invalid category labels. Numerical anomalies were screened with Isolation Forest, and flagged training observations were winsorized using the first and 99th percentiles estimated from that fold; flagged validation and test records were not automatically removed. The analysis was also repeated after excluding flagged training observations as a sensitivity analysis.

SMOTE-NC was then applied only to the training portion of each cross-validation fold, after imputation and feature generation but before final one-hot encoding. It created additional minority-class examples using nearest minority-class neighbors, with synthetic categorical values assigned from the modal category among those neighbors. Neither validation folds nor the held-out test set were oversampled. Consequently, all reported validation and test metrics were calculated using the original class distributions. The final model was fitted to the complete development set only after preprocessing, feature selection and hyperparameter choices had been made, and the held-out set was evaluated once as a final test.

The researchers also constructed target-independent engineered descriptors before fitting the model. These included a reservoir-quality index based on porosity and permeability, log-transformed permeability, depth-derived temperature and pressure proxies, and interaction terms involving rock type, porosity and permeability. Other indices combined geological, seismic and spatial variables into quantities named mobility ratio, sweep efficiency, biosurfactant efficiency, polymer efficiency, enhanced oil recovery potential and synergistic EOR effect. Despite those operational-sounding names, they are computational composites of the available synthetic predictors, not measurements of oil recovery, chemical concentration, viscosity, interfacial tension, injected-fluid performance or displacement efficiency. The authors explicitly warn that their feature rankings should not be interpreted as evidence of field-scale enhanced oil recovery performance.

Recursive feature elimination was used within each training fold to identify a compact subset of predictors. L2-regularized logistic regression served as the ranking estimator, and candidate subsets were compared using mean validation ROC-AUC. The final model received 13 selected predictors. To learn different aspects of each reservoir record, the network used three parallel branches. A probabilistic branch based on prior-data-fitted tabular classification produced posterior probabilities, predictive entropy and probability margin. A contextual branch treated each feature as a token, embedded it with trainable feature identities and processed those tokens through three Transformer blocks with four attention heads. A sequential branch used TabNet-style sparse masks across four decision steps, encouraging the model to select different features for different records.

The three branches were not simply averaged. Each representation was first projected into a common 64-dimensional latent space. A gating network then assigned each record its own three branch weights, allowing the model to emphasize uncertainty estimates, contextual relationships or sparse feature selection according to the input. Sparsity regularization encouraged selective feature use, while a diversity term discouraged the branches from learning identical representations. The fused output passed through dense layers and a sigmoid to produce an oil-presence probability. Hyperparameters were selected with 60 trials of a tree-structured Parzen Estimator using Hyperband pruning, optimizing mean validation ROC-AUC and using mean F1 as a secondary criterion. The selected configuration used AdamW, a batch size of 64, an initial learning rate of 3 × 10^-4 and early stopping after 12 epochs without validation-loss improvement.

Across five repeats of five-fold stratified cross-validation, the complete network produced 25 paired validation estimates. Mean accuracy was 98.60%, with a standard deviation of 0.27 percentage points and a 95% confidence interval of 98.49% to 98.71%. Mean ROC-AUC was 0.9809 and mean precision-recall AUC was 0.9831. The mean Brier score was 0.0138. At the pre-specified classification threshold of 0.50, held-out accuracy was 98.74%, with a bootstrap 95% confidence interval of 97.84% to 99.64%. Sensitivity was 98.06%, specificity was 99.00%, the F1 score was 97.75% and the Matthews correlation coefficient was 0.9688. The observed positive prevalence in the test set was 27.88%, while the model predicted a positive rate of 28.06%.

Calibration results also indicated that the predicted probabilities were broadly consistent with the observed test outcomes, although the evaluation sample was modest. The held-out Brier score was 0.0124, expected calibration error was 0.0160, the calibration slope was 1.08 and the intercept was −0.10. The reported ROC-AUC and precision-recall AUC bootstrap intervals extended to 1.0000, reflecting the small number of errors in the 556-record test set. Statistical comparisons used paired Wilcoxon signed-rank tests across the 25 cross-validation estimates, with Holm adjustment for seven baseline comparisons. The model exceeded k-nearest neighbours, support vector machines, multilayer perceptrons, an autoencoder-classifier combination, random forest, a deep-and-wide neural network and XGBoost in ROC-AUC. XGBoost was the strongest baseline, reaching 93.53% held-out accuracy and a ROC-AUC of 0.9497; its mean paired cross-validation ROC-AUC was 0.0320 below the proposed model. All seven comparisons remained statistically significant after Holm adjustment, with adjusted p values from 2.98 × 10^-5 to 4.17 × 10^-7.

Ablation experiments indicated that the complete configuration was not merely benefiting from the addition of several neural networks. The sequential-attention branch was the strongest individual branch, with a cross-validation ROC-AUC of 0.9431, while the contextual and sequential branches together reached 0.9638. Unweighted concatenation produced a ROC-AUC of 0.9689, lower than the full model’s 0.9809. Removing diversity regularization reduced the score to 0.9756, and removing sparsity regularization reduced it to 0.9741. The complete architecture also had the lowest Brier score among the tested variants. Attention analysis ranked five engineered proxies as the most influential: the final EOR score, synergistic EOR effect, sweep efficiency, mobility ratio and enhanced oil recovery potential. Together they accounted for 0.592 of normalized importance and were selected in all 25 folds, but their importance reflects predictive association within synthetic data rather than demonstrated physical or operational effects.

The central achievement is therefore a carefully validated multi-branch architecture for heterogeneous tabular reservoir descriptors, with sample-specific fusion showing stronger internal performance than the evaluated alternatives. The architecture also illustrates how uncertainty, contextual feature relationships and selective feature use can be combined without forcing them into a single representation. Yet the evidentiary boundary is equally important. The study used one synthetic dataset and did not include production histories, well-log sequences, pressure-transient measurements, fluid compositions, injection schedules or reservoir-management variables. Independent testing on field data from different lithologies, operating conditions, class prevalences and acquisition systems would be required to establish generalizability and practical value. Until then, the model’s near-perfect scores demonstrate consistency within a controlled synthetic benchmark, not proof that it can forecast oil presence or measure recovery efficiency in a working reservoir.

Subject of Research: Evaluation of a multi-branch deep-learning model for oil-presence classification in synthetic reservoir data

Article Title: Integrated analysis of geological and seismic parameters for oil presence prediction in reservoirs using the synergistic oil recovery-net model

Article References: Li, D. (2026). Integrated analysis of geological and seismic parameters for oil presence prediction in reservoirs using the synergistic oil recovery-net model. Frontiers in Earth Science, 14, Article 1900399. https://doi.org/10.3389/feart.2026.1900399

Image Credits: AI Generated

DOI: 10.3389/feart.2026.1900399

Keywords: oil-presence prediction, reservoir characterization, deep learning, attention mechanisms, petrophysics, seismic data, synthetic data, feature selection, model validation, Integrated, analysis, geological

The research, led by Xiaoxia Dong and Nanxi Zhang of Sinopec Southwest Oil & Gas Company together with colleagues at Chengdu University of Technology, focused on the first member of the Lower Silurian Longmaxi Formation, one of China’s most important shale gas targets. The team collected 27 shale samples from seven core drilling wells that span a striking range of structural settings: quiet, gently folded synclines; fault-bounded uplifted blocks; clamped fault blocks squeezed between reverse faults; low-amplitude anticlines; and intensely deformed anticlines. Because the samples come from the same stratigraphic interval, differences in their microscopic fabric can be attributed largely to deformation intensity rather than to original depositional variation, making the dataset an unusually clean test of how tectonics sculpts a reservoir.

The analytical toolkit combined several complementary techniques. X-ray diffraction quantified the mineral composition of each sample, petrographic thin sections and field-emission scanning electron microscopy revealed pore types at scales down to a few nanometers, overburden porosity and permeability measurements characterized the bulk flow properties, and low-temperature nitrogen adsorption experiments mapped the distribution of pore sizes. On top of these measurements, the researchers applied the Frenkel-Halsey-Hill fractal model to the adsorption data, extracting two fractal dimensions that describe, respectively, the roughness of pore surfaces at low relative pressure and the complexity of the pore network at high relative pressure. Fractal dimensions near 2 indicate smooth, regular pore systems, while values approaching 3 signal rough, tortuous, highly heterogeneous networks.

The geochemical results confirm that the Yongchuan shale is a genuinely world-class source rock. Total organic carbon content ranges from 2.8 to 6.2 percent, averaging 4.9 percent, and vitrinite reflectance values between 2.65 and 3.02 percent place the organic matter firmly in the over-mature dry gas window. The rock is dominated by quartz, with contents spanning 29.0 to 72.8 percent, and contains relatively modest amounts of clay minerals, between 7.8 and 37.0 percent. Brittle minerals such as quartz and feldspar average 60.5 percent of the whole rock, yielding a brittleness index that ranges from 55.4 to 89.0 percent. In practical terms, this is a stiff, glassy shale that fractures readily under stress, which is precisely the property that allows tectonic forces to leave such a deep imprint on its internal architecture.

Under the scanning electron microscope, the contrast between weakly and strongly deformed areas is dramatic. In the syncline areas, where deformation is minimal, the pore system is dominated by honeycomb-like organic matter pores, typically a few nanometers to tens of nanometers across, with spherical and elliptical shapes and only sparse, small laminar cracks along bedding planes. In the anticlines, fault uplift areas, and clamped fault blocks, the picture changes completely. Organic pores are less abundant and often deformed or compacted, while inorganic pores, clay-mineral interlayer pores, dissolution pores, and a rich suite of microfractures take center stage. Three crack types appear: laminar cracks along bedding, abnormally high-pressure cracks opened by overpressured fluids, and tectonic cracks generated directly by structural stress. The abundance of tectonic cracks rises in step with deformation intensity, mirroring the fractures visible in the drill cores.

Quantitative measurements reinforce the visual story. Overall porosity ranges from 2.13 to 5.28 percent and permeability from 0.0002 to 0.4209 millidarcys, marking a classic low-porosity, low-permeability reservoir. Yet the ranking of structural units is revealing: the clamped fault block area shows the highest average porosity and by far the highest permeability, while the gently deformed synclines sit at the opposite end of the permeability spectrum. Nitrogen adsorption curves are type II with H3 and H4 hysteresis loops, indicating wedge-shaped, slit-like, and parallel-plate pores. Pore size distributions are double-peaked, with a dominant peak in the micropore range around 1 to 2 nanometers. In the synclines, micropores average 28.7 percent of the pore volume and macropores a mere 2.9 percent; in the anticlines, macropores balloon to an average of 17.4 percent while micropores shrink to 12.9 percent, clear evidence that compression has squeezed small pores into larger ones.

The fractal analysis delivers the study’s most elegant insight. Across all samples, the surface fractal dimension D1 ranges from 2.5528 to 2.7158 and the structural fractal dimension D2 from 2.8219 to 2.9285, with D2 consistently higher, meaning the larger seepage pores are more structurally complex than the tiny adsorption pores. The fault uplift area records the roughest surfaces and the most intricate pore networks, while the low-amplitude anticline area is the most homogeneous. Crucially, the factors controlling fractal dimension shift with deformation style. In synclines, pore complexity is largely decoupled from organic carbon and clay content, whereas in anticlines it correlates positively with both, and in fault blocks the relationships invert again. Structural deformation, in other words, does not just damage pores one by one; it rewrites the statistical rules governing the whole pore system.

These microscopic differences translate directly into gas behavior. Total gas content across the study area spans 2.57 to 7.86 cubic centimeters per gram, and the ranking by unit is striking: the low-amplitude anticline area averages 7.2 cubic centimeters per gram, more than double the syncline average of 3.51. Correlation analysis shows that gas content tracks organic carbon positively in the synclines, fault blocks, and low-amplitude anticlines, but negatively in the intense anticlines, where the authors infer that free gas escapes along open fractures. The balance between adsorbed gas held on pore surfaces and free gas stored in larger voids also flips with deformation: free gas fractions rise steadily from syncline to anticline as fractures and macropores proliferate, while adsorbed gas declines. Moderate deformation therefore boosts storage and flow, but excessive deformation opens escape routes that drain the reservoir, which the authors suggest may explain disappointing production from some anticlinal wells.

Synthesizing these observations, the team proposes a differential evolution model that reads almost like a metamorphic pathway for pore systems. As deformation intensity increases, the dominant pore type shifts from organic matter pores to an organic-clay composite pore-fracture system; macropore proportions climb, micropore proportions fall, and the gas phase migrates from adsorbed to free. The authors argue that this coupling between deformation intensity and pore evolution is not a local curiosity but a general principle, citing consistent patterns from the Qiongzhusi Formation in Yunnan, the Permian Basin in India, and the Newark Basin in the United States. For explorers, the practical message is to map deformation intensity as a primary reservoir variable and to target the sweet spot where deformation is strong enough to enhance permeability and free gas storage, but not so strong that the gas has already slipped away through fractures.

Subject of Research: Effects of differential structural deformation on pore structure, fractal characteristics, and gas enrichment in Longmaxi Formation shale reservoirs

Article Title: Influence of differential structural deformation on pore structure and fractal characteristics of shale reservoirs: a case study of the Longmaxi Formation in the Yongchuan area, South Sichuan Basin

Article References: Dong, X., Zhang, N., Wang, B., Wang, T., Fu, M., & Ouyang, J. (2026). Influence of differential structural deformation on pore structure and fractal characteristics of shale reservoirs: a case study of the Longmaxi Formation in the Yongchuan area, South Sichuan Basin. Frontiers in Earth Science, 14, Article 1901814. https://doi.org/10.3389/feart.2026.1901814

Image Credits: AI Generated

DOI: 10.3389/feart.2026.1901814

Keywords: shale gas, Longmaxi Formation, Sichuan Basin, structural deformation, pore structure, fractal dimension, nitrogen adsorption, scanning electron microscopy, organic matter pores, microfractures, gas content, reservoir characterization

The scale of the new resource is striking. The repository currently holds approximately 25,700 data points covering up to 34 distinct parameters, measured on more than 8,200 samples drawn from over 3,400 sampling locations across eight German federal states. These numbers represent the first time that such a broad and statistically robust body of rock property information has been brought together in a single, unified framework for the country. Until now, systematic datasets covering the major crystalline and metamorphic basement units and their physiomechanical characteristics were scarce, fragmented, or lacked the metadata needed for rigorous evaluation. The AMPEDEK project was designed specifically to close this gap, and its first release marks a milestone for subsurface research in Germany.

Building the database required an unusual combination of archival detective work and fresh laboratory science. The team compiled, evaluated, and integrated existing data from more than 40 published studies, alongside material contributed by geological surveys. To fill the remaining holes in coverage, the researchers conducted their own sampling and measurement campaigns, retrieving approximately 650 cylindrical plugs from drill cores and outcrop samples and analyzing them systematically for their mineralogical, geochemical, petrophysical, and rock mechanical properties. Particular emphasis was placed on sampling a broad spectrum of lithologies, ensuring that the major crystalline units of the German basement are represented rather than only the rock types that happen to be well studied or commercially fashionable.

What makes the database more than a simple pile of numbers is its structure. Every entry is linked through unique sample identifiers that connect metadata, geochemistry, petrography, and physiomechanical rock properties. This means a user can trace a single measurement back to its parent sample, its location, its rock type, and the conditions under which it was measured. In an era when machine learning models and numerical simulations are only as good as the data fed into them, that kind of traceability is essential. It allows researchers to assess data quality, compare results across studies, and avoid the silent errors that arise when heterogeneous datasets are merged without consistent documentation.

The practical motivation behind the project is rooted in some of the most consequential engineering challenges Germany faces. Deep geothermal energy, which taps heat from hot crystalline rocks far below the surface, depends on knowing how those rocks conduct heat, how easily fluids move through their fracture networks, and how they deform under stress. The safe storage of high-radioactive nuclear waste, meanwhile, demands an intimate understanding of the mechanical strength, mineralogy, and long-term stability of the host formations. For both applications, a comprehensive and statistically robust understanding of rock characteristics is essential for reservoir characterization of the deep subsurface, accurate planning of exploration campaigns, parameterization of predictive three-dimensional geological models, interpretation of geophysical data, and execution of numerical thermo-hydraulic-mechanical-chemical simulations. Better data translate directly into reduced economic risks and lower project uncertainties.

The parameters captured in the database span the full range of properties that matter for such work. Petrophysical measurements characterize how rocks respond to heat, electricity, magnetism, and fluid flow, providing the bridge between laboratory samples and the geophysical signals recorded in the field. Mechanical tests quantify strength, stiffness, and deformation behavior, which govern whether a fractured reservoir will hold open pathways for circulating water or seal itself shut under pressure. Geochemical and mineralogical data reveal the composition and history of each rock unit, offering clues about alteration, fracturing, and the processes that shaped the basement over hundreds of millions of years. With up to 34 parameters available per sample, the compilation allows researchers to examine how these properties co-vary, a critical step for building predictive models that can estimate conditions at sites where no direct measurements exist.

The open-access nature of the release amplifies its value. Hosted through the TU Darmstadt data repository, the database is freely available to researchers, industry, and public authorities, in line with the principles of findable, accessible, interoperable, and reusable research data. For early-stage exploration, this means companies and geological surveys can perform preliminary reservoir assessments without commissioning expensive new drilling programs. For the scientific community, the compilation provides a robust empirical foundation for validating analytical measurement methods, since laboratories can benchmark their results against a large, standardized reference population. And for modelers, it supplies the ground truth needed to test whether simulated rock behavior matches reality at the scale of real geological formations.

The authors are explicit that this first release is a beginning rather than an end. Future efforts will focus on expanding the database toward regions with low data density, ensuring that the coverage of Germany’s crystalline basement becomes more geographically even. They also plan high-temperature and high-pressure experiments designed to improve understanding of how rock parameters vary under in situ conditions. This latter point is scientifically crucial: laboratory measurements made at room temperature and ambient pressure may not faithfully represent the behavior of rocks kilometers underground, where they are subjected to intense heat and confining stress. By characterizing these variations experimentally, the team aims to make the database even more directly applicable to real subsurface conditions.

For Germany’s energy transition, the timing of the release could hardly be more significant. The country is actively pursuing deep geothermal energy as a low-carbon heat and power source, and it continues to search for a permanent repository for its nuclear waste, a process in which the crystalline basement is among the candidate host rock environments under consideration. Both endeavors require exactly the kind of systematic, quality-controlled property data that AMPEDEK now provides. A drilling campaign that costs tens of millions of euros can succeed or fail on the accuracy of the assumptions made before the drill bit ever turns, and every improvement in the underlying data reduces the gamble.

The AMPEDEK database also sets a template that other countries could follow. Crystalline basement rocks underlie much of central and northern Europe, and comparable data fragmentation problems exist almost everywhere. By demonstrating how literature compilation, geological survey cooperation, and targeted new measurements can be woven into a single, sample-linked, open-access framework, the German team has shown a practical path from scattered data to national reference resource. As the database grows through future releases, it is likely to become an indispensable tool not only for German geoscientists but for the international community studying the deep critical zone where energy, waste, and water challenges converge beneath our feet.

Subject of Research: A standardized open-access database of geochemical, petrophysical, and mechanical rock properties of Germany's crystalline basement

Article Title: Atlas of geochemical, petrophysical and mechanical rock properties of the crystalline basement in Germany – first release

Article References: Weydt, L. M., Ramírez-Guzmán, Á. A., Muhl, L., Deon, F., Schleicher, A. M., & Sass, I. (2026). Atlas of geochemical, petrophysical and mechanical rock properties of the crystalline basement in Germany – first release. https://doi.org/10.5194/essd-2026-555

Image Credits: AI Generated

DOI: 10.5194/essd-2026-555

Keywords: crystalline basement, rock properties, petrophysics, geochemistry, geothermal energy, nuclear waste storage, AMPEDEK database, Germany, reservoir characterization, open-access data, rock mechanics, subsurface exploration

The research, led by Olawale Johnson Allo of the University of Lagos with colleagues from Mountain Top University and the Federal University of Petroleum Resources, focused on the Kay Field in the Coastal Swamp depobelt of the onshore Niger Delta. The team integrated a 3D post-stack seismic volume covering about 55.52 square kilometers with five composite well log suites, including sonic, density, gamma ray, resistivity, and porosity measurements. Their target was reservoir Sand E, one of five major sandstone intervals identified in the wells, and specifically the striking velocity contrast between wells designated K4 and K5.

The first analytical step was to determine what kind of rock fabric the reservoirs possess. The researchers superimposed theoretical rock physics model curves onto cross-plots of compressional velocity against porosity for the hydrocarbon-bearing intervals. Both reservoir sands aligned closely with the Dvorkin-Nur friable sand model, which describes uncemented grain packs in which porosity is reduced by poorly sorted material deposited away from grain contacts rather than by diagenetic cement bridging the contacts. This diagnosis carries real physical meaning: in an uncemented sand, grain-to-grain stiffness is governed primarily by confining pressure, not by cementation. It also implies the rock is mechanically weak and prone to sanding during production, a practical concern for reservoir engineers.

Because cementation could be ruled out, the team turned to composition. Prior experimental work had shown that even tiny quantities of clay at grain boundaries, on the order of one percent, can soften a sandstone matrix and suppress compressional velocity, while small amounts of contact cement do the opposite and stiffen the frame. To quantify the shale effect in the Kay Field sands, the researchers stratified their well log observations into five shale volume intervals, from 0 to 10 percent up to 41 to 50 percent, and ran multilinear regressions of compressional velocity against porosity and shale volume within each interval. The resulting twenty empirical equations, with correlation coefficients reaching 62 percent in the cleanest sands, provide localized predictive relations linking reservoir composition directly to seismic velocity.

The statistical picture that emerged was decisively asymmetric. At well K5, 59.8 percent of the Sand E samples fall within the higher shale volume range of 21 to 50 percent, while only 40.2 percent occupy the cleaner 0 to 20 percent range. Well K4 shows the mirror image: 77.5 percent of its samples sit in the low shale volume range, with just 22.5 percent in the clay-rich bins. This compositional asymmetry, the authors conclude, is the principal driver of the velocity suppression at K5. Notably, the effect is cumulative across the shale volume bins, and it operates independently of depth. Sand E at K5 is actually buried about 149 meters deeper than at K4, yet its velocities are lower, demonstrating that burial depth exerts a negligible influence compared with clay content in these uncemented sands.

The analysis also revealed instructive limitations in the regression framework. Two equations, derived from small samples of only 38 and 52 observations in narrow porosity and shale volume windows, produced anomalously high coefficients, while others suffered from wide spreads in porosity and shale volume values that mapped to the same velocity, dragging correlation coefficients down to 28 and 33 percent. The authors attribute these spreads partly to minor data spikes from acquisition artifacts. The lesson is methodological as much as geological: the predictive power of rock property relations depends not only on the physics but on the range and density of the observations used to constrain them.

The crucial test came when the team connected these well-scale findings to the seismic domain. Using checkshot data to establish time-depth calibration, they extracted root-mean-square amplitude attributes around both well locations at the Sand E and Sand C reservoir tops. In Sand C, the amplitude distributions at the two wells tracked each other closely, consistent with the absence of a significant velocity contrast. In Sand E, however, the amplitude spectra diverged sharply. Well K4, with its faster, cleaner sand, displayed a predominantly high amplitude spectrum, while the clay-rich interval at K5 produced a subdued response. The mechanism is straightforward: higher P-velocity in Sand E at K4 enhances the acoustic impedance contrast with the encasing shale, generating anomalously bright reflections.

That brightness is precisely the trap. Both wells penetrate the same hydrocarbon-bearing sand, confirming that the amplitude difference is a lithological signature rather than a direct hydrocarbon indicator. The high amplitudes at K4 reflect a geologically cleaner, quartz-dominated framework deposited in a high-energy distributary channel environment with good sorting and limited clay infiltration, whereas the muted response at K5 points to a more heterolithic interval, likely deposited in a lower-energy tidal flat margin or inter-distributary transition zone where shale laminae degrade the impedance contrast. An interpreter relying on bright-spot analysis alone could easily read the K4 anomaly as evidence of a larger hydrocarbon accumulation and overestimate volumes, when in fact the amplitude simply measures the sand’s shale content.

The implications extend well beyond one field. Deltaic clastic reservoirs worldwide are built from stacked channel, mouth bar, and tidal deposits that vary laterally over distances of hundreds of meters, exactly the scale separating wells K4 and K5. The study demonstrates that shale volume variation at such short spatial scales can modulate seismic amplitudes by magnitudes comparable to genuine fluid-driven anomalies, making unconstrained amplitude interpretation inherently risky. The authors argue that shale volume diagnostics, derived from routine gamma ray and porosity logs and stratified through rock physics analysis, should be embedded in amplitude-based reservoir characterization workflows before amplitude anomalies are attributed to pore fluids. In the Coastal Swamp depobelt, as in analogous basins, the difference between a bright spot and a shale-out may be only a few percent of clay, and only rock physics can tell them apart.

Subject of Research: Effects of shale volume on compressional velocity and seismic amplitude in friable sandstone reservoirs of the Niger Delta

Article Title: Rock physics evaluation of shale volume effects on compressional velocity and seismic amplitude in friable sandstone reservoirs, Coastal Swamp, Niger Delta

Article References: Allo, O. J., Ayolabi, E. A., Ameloko, A. A., & Uchegbulam, O. (2026). Rock physics evaluation of shale volume effects on compressional velocity and seismic amplitude in friable sandstone reservoirs, Coastal Swamp, Niger Delta. Discover Geoscience, 4(1), Article 288. https://doi.org/10.1007/s44288-026-00667-3

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00667-3

Keywords: rock physics, Niger Delta, shale volume, compressional velocity, seismic amplitude, friable sandstone, reservoir characterization, Dvorkin-Nur model, porosity, multilinear regression, Coastal Swamp depobelt, bright spots

The team’s answer matters because the stakes of seismic inversion extend far beyond academic curiosity. The three elastic parameters at issue — P-wave velocity, S-wave velocity, and density — are the raw ingredients from which geophysicists derive indicators of fluid content, porosity, and hydrocarbon saturation. Traditional deterministic inversion methods, such as least-squares approaches, deliver a single best-fitting model but say little about how confident one should be in it. Bayesian inference offers a different philosophy: instead of one answer, it produces an entire probability distribution over plausible underground models, blending prior geological knowledge with the information contained in the seismic data itself. The catch is that this posterior distribution cannot be written down analytically for realistic problems, so researchers must sample from it using Markov Chain Monte Carlo methods — and the choice of sampler shapes both the quality of the answer and the size of the computing bill.

The four contenders in this statistical showdown represent distinct generations of sampling technology. The oldest and simplest is the Metropolis-Hastings algorithm, developed in the 1950s and 1970s, which explores the model space by random walks: it proposes a random perturbation of the current model and accepts or rejects it according to a probability ratio. It requires no gradient calculations, making it cheap per step, but in high-dimensional spaces its drunken-wander behavior means samples are highly correlated and convergence is slow. At the other extreme sits Hamiltonian Monte Carlo, borrowed from molecular dynamics, which endows each model with an auxiliary momentum variable and simulates frictionless motion across an energy landscape defined by the negative logarithm of the posterior. By following the gradient of that landscape, HMC proposes bold, physically informed moves that traverse the parameter space far more efficiently — provided the gradients are computed correctly and the step sizes are tuned with care.

Between these poles lie two Langevin-based methods rooted in the mathematics of Brownian motion, the theory Paul Langevin formulated in 1908 as a simplification of Einstein’s treatment of jittering particles. The Metropolis-Adjusted Langevin Algorithm, or MALA, discretizes a stochastic differential equation whose drift term pulls each proposal toward regions of higher posterior probability, then applies a Metropolis acceptance step to remove discretization bias. Its newer variant, Lip-MALA, goes further: it adapts the step length at every iteration according to a Lipschitz condition, automatically shrinking steps where the posterior landscape changes abruptly and enlarging them where it is smooth. This dynamic tuning, the authors note, improves stability and mixing in the high-dimensional posteriors that seismic inversion generates, where the number of unknowns scales with the number of layers in the model.

The forward physics connecting models to data is the Aki-Richards approximation, the workhorse equation of amplitude-versus-offset analysis that emerged in the early 1980s. It expresses the reflection coefficient at each rock interface as a linearized function of the contrasts in P-wave velocity, S-wave velocity, and density, weighted by angle-dependent coefficients. The researchers worked in a logarithmic parameterization — inverting for the logarithms of the three elastic parameters — which guarantees physically sensible positive values and dovetails naturally with the structure of the Aki-Richards equations. Working at three incidence angles, they assembled a matrix operator that converts any candidate model into synthetic seismic traces, then measured how well those traces match the observed data through a Gaussian likelihood combined with a Gaussian prior built from well logs and low-frequency velocity models.

The first arena was a synthetic test: noise-free seismic traces generated from real well-log values at angles of 9, 18.5, and 27.5 degrees, with the true answer known in advance. Each algorithm ran a Markov chain of 10,000 accepted realizations after a burn-in period, and the team judged performance through posterior means, standard deviations, correlation coefficients, root-mean-square errors, acceptance rates, execution times, and a multivariate effective sample size diagnostic to confirm convergence. The results drew a clear dividing line. For velocity estimation, the three gradient-based methods all outperformed Metropolis-Hastings, with HMC achieving the highest correlation and the lowest error. For density, the picture inverted in an unexpected way: MH and HMC both struggled, with correlations below 0.29, while MALA and Lip-MALA delivered correlations above 0.60 and errors roughly a third smaller.

The computational ledger told an equally important story. Metropolis-Hastings was the fastest of the four, despite its statistical clumsiness, because each of its steps is so cheap. HMC was the slowest, its long Hamiltonian trajectories and repeated gradient evaluations exacting a heavy price in wall-clock time — and it also recorded the lowest acceptance rate. Lip-MALA, by contrast, posted the highest acceptance rate of all four algorithms, a sign that its adaptive step control consistently proposes moves the posterior is willing to accept. MALA ran quickly as well. The trade-off, in short, was stark: gradient information buys accuracy, but the bill arrives in computing hours.

Synthetic benchmarks, however, can flatter an algorithm. The decisive test came from real data: an oil field in eastern Venezuela’s Eastern Basin, where alternating sandstones and shales host brine and oil, and where partial-stack seismic traces at angles of 19, 24, and 29 degrees had to be tied to upscaled well logs — a correlation of 0.55 that reflects the messiness of field data. Here the same hierarchy largely held. Metropolis-Hastings again showed the weakest velocity correlations, while Lip-MALA achieved the highest and the lowest errors, and the Langevin pair again dominated density estimation, with correlations above 0.48 where MH and HMC fell below 0.28. The team then scaled the problem to two dimensions, inverting a grid of 351 time samples by 136 horizontal cells over a 340-meter span, using a prior model built from interpreted horizons and faults. In this more realistic setting, HMC redeemed itself: its long trajectories navigated the rougher, more heterogeneous posterior landscape better than local Langevin steps, producing the lowest posterior standard deviations for all three parameters — though at the highest computational cost, with Lip-MALA offering the best compromise between precision and efficiency.

The practical guidance that emerges is refreshingly concrete. When speed matters most — rapid screening of frontier areas with poor well control, or near-real-time decisions during drilling — Metropolis-Hastings or MALA remain defensible choices despite their statistical limitations. When fidelity is paramount, as in mature fields where drilling and production decisions hinge on accurate elastic properties, HMC earns its runtime, particularly in two-dimensional and higher-dimensional problems where its ability to traverse multi-basin posterior landscapes reduces global uncertainty. Lip-MALA stands out as the algorithm for practitioners who want gradient-informed accuracy without HMC’s price tag, and as the most reliable option for the stubborn problem of density estimation. The authors also offer a methodological caution: researchers should report both pointwise fit and uncertainty measures, because an algorithm that wins on one metric can lose on another, and choosing the wrong yardstick can make the best tool look like the worst.

Beyond the rankings, the study demonstrates something conceptually larger: that the full Bayesian machinery — priors, likelihoods, posterior sampling, and rigorous convergence diagnostics — can be carried from idealized synthetic exercises all the way to geologically complex field data, delivering not just pictures of the subsurface but honest probability distributions over what lies there. The framework is not locked to its simplifying assumptions of one-dimensional layered models and linearized reflectivity; the authors point out that more sophisticated forward modeling and petrophysical relationships can be slotted into the same statistical skeleton. As exploration moves into ever more challenging terrain, and as computing resources grow, the quiet contest between random walks, Hamiltonian trajectories, and Langevin diffusion may well determine how clearly the industry can see what it is drilling toward.

Subject of Research: Comparison of Markov Chain Monte Carlo algorithms for Bayesian prestack seismic inversion of elastic properties

Article Title: Bayesian inference based on algorithms: MH, HMC, MALA and Lip-MALA for prestack seismic inversion

Article References: Perez-Roa, R., Infante, S., Barragan, G., & Manzanilla, R. (2026). Bayesian inference based on algorithms: MH, HMC, MALA and Lip-MALA for prestack seismic inversion. Nonlinear Processes in Geophysics, 33(2), 173-195. https://doi.org/10.5194/npg-33-173-2026

Image Credits: AI Generated

DOI: 10.5194/npg-33-173-2026

Keywords: seismic inversion, Bayesian inference, Markov Chain Monte Carlo, Metropolis-Hastings, Hamiltonian Monte Carlo, MALA, Lip-MALA, AVO, elastic properties, reservoir characterization, Venezuela, geophysics

The study is notable less for any single measurement than for its integrated design. Previous investigations of the Nigerian tar sands, which stretch across the coastal Dahomey Basin into Togo, Benin, and southeastern Ghana, relied overwhelmingly on a single geophysical method, usually electrical resistivity. That approach works because bitumen is electrically resistive: hydrocarbons displace conductive pore water, so bitumen-saturated sands light up as high-resistivity anomalies. But resistivity alone is notoriously ambiguous. Dry sand, laterite, and other resistive materials can mimic the signature of oil sand, and a single dataset cannot resolve whether a resistive blob is a reservoir or an artifact. The new work tackles that ambiguity head-on by pairing resistivity imaging with seismic refraction tomography, which senses an entirely different physical property: the stiffness of the rock, expressed as the speed at which compressional waves travel through it.

The field campaign covered six traverses, each roughly 157 meters long, using an ABEM SAS1000 resistivity meter with 64 electrodes spaced 2.5 meters apart in a dipole-dipole configuration. Along the very same lines, the team deployed an ABEM Terraloc Mark 8 seismograph with matching 2.5-meter geophone spacing, striking the ground with a 10-kilogram hammer to generate seismic energy. This co-located geometry meant that every resistivity section could be compared directly, meter by meter, with its seismic counterpart. The resistivity data were inverted with a smoothness-constrained least-squares algorithm in RES2DINV over five iterations, converging to a root-mean-square error of about 3.5 percent, while the seismic first arrivals were picked in SeisImager and tomographically inverted in the Plotrefa module.

The electrical results were striking. Across all six profiles, a consistent three-layer pattern emerged: a thin, low-resistivity topsoil, a spectacularly resistive second layer, and locally a deeper clayey sand. The interpreted oil sand horizon showed resistivity values generally exceeding 3000 ohm-meters and locally soaring past one million ohm-meters, far above the 1000 to 100,000 ohm-meter range typically reported for oil sands elsewhere in the basin. On profiles 1 through 3, this resistive unit sits beneath just 1 to 4.5 meters of topsoil; on profiles 4 through 6 it begins at depths of roughly 1.5 to 3.5 meters and reaches thicknesses of about 8 to 12 meters. In several places the resistive body extends beyond the depth of investigation, hinting at a deposit considerably larger than the survey could fully image.

The seismic data told a matching story in a different physical language. Every profile revealed a three-layer velocity structure: near-surface topsoil at around 300 meters per second, an intermediate layer at roughly 1900 to 2300 meters per second interpreted as bitumen-bearing sand, and a basal unit exceeding 2800 to 3000 meters per second corresponding to consolidated, clay-interbedded sand. The intermediate velocities make physical sense. Bitumen stiffens unconsolidated sand relative to loose topsoil but leaves the layer mechanically softer than fully cemented rock, so the oil sand occupies a distinctive velocity window between its neighbors. Crucially, the intermediate-velocity zones aligned laterally with the high-resistivity anomalies on profiles 1, 3, and 4, with profile 4 showing a shallow resistive body coinciding with a seismic layer of about 2100 meters per second and 22 to 26 meters thick. Two independent physical measurements pointing to the same body at the same location sharply reduce the chance that either anomaly is a phantom.

The two methods did not agree perfectly on thickness, and the discrepancy is itself informative. Electrical resistivity tomography suggested oil sand thicknesses of roughly 5 to 30 meters, while seismic refraction yielded 12 to 34 meters for the same horizons. The authors attribute this to the fundamental physics of each technique. Resistivity represents a volumetric average sensitive to porosity, fluid saturation, and clay content, so gradational boundaries and partially saturated sands tend to be smoothed or underestimated. Seismic refraction responds to elastic stiffness and acoustic impedance, making it more sensitive to compaction and mechanical boundaries, and thus more likely to detect deeper, more continuous velocity contrasts. Differences in inversion strategy, smoothness-constrained least squares versus tomographic inversion, add further resolution differences. Rather than undermining the interpretation, the mismatch quantifies the uncertainty inherent in indirect imaging and underscores why no single method should be trusted alone.

To ground the geophysics in actual rock, the team collected three outcrop samples, labeled ON-1 through ON-3, stripped of weathered surfaces and treated with gasoline in the laboratory to dissolve the bitumen away from the sand grains. Sieve analysis showed the cleaned sands to be medium-grained, with mean grain sizes between 1.07 and 1.38 phi, and moderately sorted, with sorting values from 1.207 to 1.342. Skewness and kurtosis statistics revealed nearly symmetrical, leptokurtic distributions in two samples, suggesting a predominantly unimodal sediment source, while one sample was strongly fine-skewed and mesokurtic, hinting at mixed sediment supply. Moderate sorting of medium sand is exactly the texture that favors porosity and permeability, which the authors interpret as indicating moderate reservoir quality, adequate storage capacity though potentially heterogeneous flow behavior.

Heavy mineral analysis added a provenance detective story. Under the microscope, the samples yielded assemblages dominated by zircon, rutile, and tourmaline, the three ultra-stable heavy minerals that survive weathering and transport when almost everything else has been destroyed. Zircon was the most abundant, averaging 17.67 grains per sample, followed by rutile at 21.33 and tourmaline at 16.0, with significant contributions from staurolite, garnet, apatite, and epidote. The resulting Zircon-Tourmaline-Rutile index ranged from 57 to 67 percent, placing the sediments in the sub-mature category, above the 50 percent threshold for immaturity but below the 76 percent that marks fully mature, extensively recycled sand. That combination points to a proximal source in felsic igneous and metamorphic rocks, most plausibly the Precambrian Basement Complex adjacent to the basin, with only moderate transport distances and limited reworking. Garnet, staurolite, and epidote are diagnostic of medium- to high-grade metamorphic terrains, reinforcing the interpretation.

Assembled together, the evidence sketches a coherent geological history. The sediments were shed from nearby crystalline highlands, carried moderate distances by rivers, and deposited in a fluvio-marine transitional environment consistent with the established setting of the Late Cretaceous Afowo Formation, whose sandy strata are known to be tar-bearing and whose interbedded shales are rich in organic matter. The basin itself was born in the Cretaceous when Africa and South America rifted apart, and its eastern arm holds more than 2500 meters of Cretaceous and younger sediment above the metamorphic basement. The oil sand of the Afowo Formation sits within this fill, and the new surveys show it occurring at depths of just 0 to 3 meters, thick enough to reach roughly 30 meters, laterally continuous across all six profiles, and with the low clay content and favorable grain size that reduce processing complexity and enhance recovery efficiency.

The authors are careful to flag the limits of their work. All geophysical interpretations remain indirect, and without borehole cores or laboratory petrophysical measurements, the lithological assignments are indicative rather than definitive. The sedimentological sample set of just three outcrops cannot capture the full spatial variability of the deposit, and questions of lateral heterogeneity and fluid saturation remain open. The team calls for borehole validation, three-dimensional geophysical imaging, and expanded sampling as the next steps. Even so, the study makes a compelling methodological argument: when electrical and elastic images of the subsurface are acquired along the same lines and interpreted together with the mineralogy of the rocks themselves, the ambiguity that has long hampered single-method exploration largely dissolves. For a country whose conventional oil reserves are gradually declining, and whose tar sands have been estimated among the largest in the world, that integrated framework may prove to be the most valuable discovery of all.

Subject of Research: Integrated geophysical and sedimentological characterization of oil sand deposits in the Eastern Dahomey Basin, Nigeria

Article Title: Integrated geophysical and geological characterization of oil sand in Eastern Dahomey Basin Southwestern, Nigeria

Article References: Adekoya, S. A., Coker, J. O., Ikhane, P. R., Oladunjoye, H. T., & Adenuga, O. A. (2026). Integrated geophysical and geological characterization of oil sand in Eastern Dahomey Basin Southwestern, Nigeria. Discover Geoscience, 4(1), Article 340. https://doi.org/10.1007/s44288-026-00633-z

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00633-z

Keywords: oil sand, bitumen, Dahomey Basin, Nigeria, electrical resistivity tomography, seismic refraction tomography, sedimentology, heavy minerals, ZTR index, reservoir characterization, geophysics, Afowo Formation

The research, conducted by a team from GEOPIC, the research arm of Oil and Natural Gas Corporation in Dehradun, along with collaborators at Pandit Deendayal Energy University and Maharashtra Institute of Technology World Peace University, focused on the gas-bearing tight sandstones of the Jhuran Formation. These rocks, deposited during the syn-rift and early post-rift phases that followed the fragmentation of Gondwana, are anything but uniform. Permeability values range from a near-impermeable 0.01 millidarcies to a still-modest 10 millidarcies, and the formation is sliced into parallel fault blocks by northwest-southeast trending structures that complicate any attempt to extrapolate reservoir quality between wells. For an emerging hydrocarbon province like the Kachchh Offshore Basin, that heterogeneity is a formidable barrier to exploration.

The conventional toolkit offers little comfort. Classic permeability estimation leans on empirical porosity-permeability transforms calibrated to laboratory core measurements. In homogeneous reservoirs such relationships perform adequately, but in tight, clay-rich sandstones the link between porosity and permeability weakens dramatically, because flow is governed by pore geometry, grain packing, clay distribution and cementation rather than by pore volume alone. Compounding the problem, core plugs are expensive and sparse; in this study, direct core measurements existed for just one well, and only fifteen of them. The team therefore turned to machine learning, training four supervised algorithms on wireline log data from four wells: Random Forest, Support Vector Regression, a Multi-Layer Perceptron neural network, and Extreme Gradient Boosting.

The input features were the workhorses of petrophysics: gamma ray, deep resistivity, bulk density, neutron porosity and sonic transit time, supplemented by engineered variables such as volume of clay, total and effective porosity, and a density-neutron-resistivity combination parameter. Nuclear magnetic resonance logging provided an independent permeability benchmark in two wells, derived through the Schlumberger-Doll Research equation, which relates permeability to porosity and the logarithmic mean of the NMR T2 relaxation distribution. Crucially, the NMR-derived permeability served only as the target variable, never as an input, a design choice that prevents data leakage and ensures the models can be applied to wells lacking NMR coverage. Spearman rank correlation analysis confirmed that no input pair approached the severe multicollinearity threshold, supporting their joint inclusion.

Validation was deliberately stringent. Wells A, C and D supplied the training data, split 70-30 into training and testing subsets, with five-fold cross-validation used for hyperparameter tuning. Well B, which contained the fifteen core measurements, was held out entirely and used only once, after model finalisation, as a blind validation well. On that single held-out well, Random Forest achieved a coefficient of determination of 0.77 against the core and NMR benchmarks, the best cross-well generalisation of the four algorithms, and posted the lowest prediction variance, with a residual standard deviation of 1.41. Support Vector Regression, by contrast, fit its training data beautifully, reaching a training R-squared of 0.92, but its validation performance collapsed, the classic signature of overfitting.

The contrast between algorithms carried a practical lesson. The Multi-Layer Perceptron and XGBoost both achieved near-perfect alignment on training data yet generalised poorly, with XGBoost showing the largest gap between training and blind-validation performance and the highest prediction uncertainty in Well A, where residual standard deviations reached 7.81. Random Forest, averaging hundreds of decision trees each trained on random subsets of the data, proved the most stable performer across all wells, with residual standard deviations as low as 0.38 in Well C. The authors suggest a heuristic for similarly data-starved settings: Random Forest as the default choice, Support Vector Regression where smooth low-variance surfaces are needed, and neural networks or gradient boosting only with strong regularisation and blind-well checks, or once more wells become available.

What elevates the study beyond a routine algorithm bake-off is its use of SHAP, or Shapley Additive Explanations, a technique borrowed from game theory that decomposes each prediction into the contributions of individual input features. The SHAP analysis revealed a clear hierarchy of geological control. Effective porosity exerted the strongest first-order influence on permeability, followed by sonic transit time, which acts as a proxy for compaction. Most strikingly, shale volume displayed strongly negative SHAP values beyond a threshold of roughly 35 to 40 percent, marking a permeability collapse in clay-rich intervals. In other words, permeability in the Jhuran sandstones is governed by nonlinear interactions among shale content, porosity and compaction, not by porosity alone, precisely the kind of behaviour that linear empirical transforms cannot capture.

The team was candid about the limits of its evidence base. Four wells with uneven log and core availability is a small dataset, and the NMR calibration constants, drawn from standard literature values for water-wet clastic systems, were checked only qualitatively against the fifteen core samples from Well B rather than formally recalibrated for the Jhuran Formation. The reported R-squared of 0.77 reflects a single blind well, and the authors explicitly caution that basin-wide generalisation requires confirmation with additional held-out wells and more core control. Well D, lacking both core and NMR data, relied on a facies-derived pseudo-benchmark, an independent consistency check but not a direct measurement. These caveats frame the work honestly as a proof of concept rather than a finished predictive product.

Even so, the implications reach well beyond one basin. Frontier exploration provinces worldwide face the same predicament: geologically complex reservoirs, a handful of wells, and sparse core data, all while drilling decisions carry enormous financial risk. The study demonstrates that a transparent, reproducible machine learning workflow, built on open-source libraries and validated against independent physical measurements, can extract usable permeability predictions from ordinary wireline logs under exactly those constraints. By pairing ensemble learning with explainability, the researchers have shown that machine learning need not be a black box; it can double as an analytical instrument that tells geoscientists which geological processes matter most. As exploration accelerates along India’s northwestern continental margin, that combination of prediction and insight may prove as valuable as the predictions themselves, offering a transferable template for characterising tight reservoirs wherever data is scarce and heterogeneity is the rule.

Subject of Research: Interpretable machine learning prediction of permeability in tight Jhuran Formation sandstones of the Kachchh Offshore Basin, India

Article Title: Data driven permeability modelling in tight Jhuran formation sandstones of the Kutch offshore Basin India

Article References: Jyotsna, G. D. D., Anees, M., Mavani, S. H., Thakur, S., Saxena, A., Singh, S. K., Desai, D. B. S. G., & Singh, D. A. (2026). Data driven permeability modelling in tight Jhuran formation sandstones of the Kutch offshore Basin India. Discover Geoscience, 4(1), Article 366. https://doi.org/10.1007/s44288-026-00724-x

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00724-x

Keywords: permeability prediction, machine learning, tight sandstone, Jhuran Formation, Kachchh Offshore Basin, Random Forest, SHAP analysis, well logging, NMR logging, reservoir characterization, petrophysics, explainable AI