Lithology identification is the process of determining which rocks occur at particular depths underground. It is central to petroleum exploration, reservoir modeling and the assessment of geological formations, because rocks with different grain sizes, mineral compositions and pore structures can behave very differently when they contain water, oil or gas. The primary data source is often the well log: a set of continuous measurements recorded as a drilling tool moves through a borehole. Natural gamma radiation, density and other physical properties produce curves that fluctuate with depth, creating a numerical signature of the formations crossed by the well. Traditionally, machine-learning models divide these curves into individual depth samples and classify each sample independently. That strategy is mathematically convenient and works with relatively small labeled datasets, but it can produce geologically implausible results, such as isolated one-point changes from sandstone to mudstone and back again.

The problem arises because a depth sample contains only part of the geological story. Sedimentary formations generally possess vertical continuity: a layer tends to extend over a finite thickness, and transitions between rock types often reflect depositional processes rather than random changes. Density, porosity and gamma-ray responses commonly vary coherently across adjacent measurements. Treating each point as an independent observation discards this structure. Fully sequence-based approaches, including recurrent neural networks, long short-term memory networks and Transformers, can model relationships across depth, but they are usually data-hungry and may be unreliable when labeled wells are limited. SClith was developed as a compromise. It retains the practical point-wise classification framework while adding engineered features that summarize local vertical organization before the supervised model makes its prediction.

The first stage uses simple non-iterative clustering, or SNIC, to segment a well-log curve into connected intervals. SNIC was originally developed for image segmentation, where it groups neighboring pixels into compact regions with similar visual properties. Unlike ordinary K-Means clustering, which groups observations according to numerical similarity alone, SNIC also accounts for spatial proximity and enforces connectivity. In the one-dimensional version used for well logs, each sample is represented by a feature vector—such as normalized natural gamma-ray and density values—together with its depth coordinate. The algorithm assigns samples to nearby cluster centers using a distance that combines feature differences with depth separation. A compactness parameter controls the balance: a lower emphasis on depth favors numerical similarity, while a stronger depth term encourages continuous intervals.

SNIC’s computational design is important because it avoids repeatedly recalculating all cluster assignments. K-Means alternates between assigning points to clusters and recomputing cluster centers until the solution stabilizes. SNIC instead grows connected segments from initialized seeds using a priority queue. At each step, the neighboring sample with the smallest combined feature-and-depth distance is added, and the cluster centroid is updated incrementally. For a one-dimensional log, initial seeds are distributed along the depth interval, with the approximate spacing determined by the number of samples divided by the desired number of clusters. Once all samples have been assigned, each segment can be represented by its mean or other summary attributes. These segmentation-derived values are then fused with the original log measurements, giving the classifier both the raw signal and a compact description of its local stratigraphic context.

The researchers tested the enhanced features with three supervised algorithms: random forest, XGBoost and a Transformer. Random forest combines the votes of many decision trees, each trained using randomized subsets of data and features. XGBoost builds trees sequentially, with each new tree focusing on errors left by earlier ones, while regularization limits overfitting. Its objective function combines a prediction-loss term with penalties related to tree complexity and leaf weights. The Transformer uses self-attention, projecting the input into query, key and value representations so that relationships within a sequence can be weighted during prediction. However, because the study’s main formulation still treated depth samples as point-wise inputs, the Transformer could not fully exploit its ability to construct long-range attention patterns. In this setting, the researchers found that XGBoost was generally the most effective and sample-efficient model.

The evaluation used data from the Yanchang Oilfield in the southeastern Yishan Slope of the Ordos Basin. The geological succession includes the Permian Shiqianfeng, Shihezi and Shanxi formations, along with the Carboniferous Taiyuan and Benxi formations. These units record a transition from marine–continental and shallow-marine environments toward deltaic, fluvial and lacustrine systems under increasingly continental conditions. The dataset included clastic rocks, coal, limestone, dolomite and gypsum. The clastic rocks were further divided according to grain size into conglomerate, coarse sandstone, medium sandstone, fine sandstone, siltstone and mudstone. Such diversity makes the dataset a demanding test: some rocks have distinctive log responses, while others overlap substantially, and minority lithologies may be represented by far fewer samples than dominant units.

Across repeated experiments, SClith consistently outperformed direct classification, in which the original log features were sent straight to a supervised model. It also surpassed schemes that appended cluster labels generated by K-Means. In representative comparisons, SNIC-enhanced inputs raised test accuracy over K-Means-enhanced inputs by 3.8 percentage points for random forest, 4.0 points for XGBoost and 9.3 points for the Transformer. The corresponding accuracies reached 85.1 percent, 94.2 percent and 92.1 percent, respectively. XGBoost delivered the strongest result among the tested classifiers, reaching 94.2 percent in that comparison. The researchers attribute the advantage to SNIC’s ability to preserve both connected intervals and boundaries, reducing the cross-layer confusion that can occur when a clustering algorithm responds only to numerical distance and local noise.

The improvements were not uniform across every rock type. Mudstone and limestone were the most stable and easiest to identify in sensitivity tests, with XGBoost F1-scores—an accuracy measure balancing precision and recall—typically around 70–80 percent for the strongest classes. Dolomite, coal and medium sandstone responded more strongly to SNIC’s parameters. The compactness parameter produced clearer separation near a value of 18, while balanced performance was obtained with an initial step size of roughly 14–18 samples. Fine sandstone, siltstone, conglomerate, coarse sandstone and gypsum remained difficult to distinguish, often recording much lower F1-scores. These weaknesses reflect both overlapping physical signatures and class imbalance, in which abundant lithologies can dominate the training process. The authors also observed that identification of transitional facies remains a significant challenge.

The researchers emphasize that SClith is not a replacement for genuinely sequence-based geological modeling, but a bridge toward it. Its SNIC stage has near-linear computational complexity and single-pass convergence, making it potentially practical for large collections of well logs. Hyperparameters were selected using the elbow method for clustering and Bayesian optimization for the integrated framework. Bayesian optimization uses a probabilistic model of the performance landscape to choose promising parameter settings without exhaustively testing every possibility. Future versions could incorporate geological priors such as sandstone thickness, depositional continuity and anisotropy, as well as seismic attributes, core descriptions and stratigraphic markers registered to the same depths. The study also proposes sequence-to-sequence models that use non-overlapping depth intervals, helping prevent data leakage and allowing attention-based networks to learn genuine vertical relationships. For now, the central finding is straightforward: when an algorithm is given not only what a log measures but also how neighboring measurements fit together, its interpretation of the rocks beneath our feet can become markedly more geological.

The study, titled “Multi-objective inversion of mining-induced subsidence and rock mass mechanical parameters based on degradation zoning,” treats the mine as a changing mechanical system rather than a uniform block of rock. Conventional subsidence models often simplify the ground into broad layers with constant properties, even though excavation can progressively damage the rock mass. Fractures may develop around the mine opening, existing joints can open or slide, and repeated loading and unloading can reduce stiffness and strength. By dividing the rock mass into degradation zones, the researchers seek to represent how its mechanical behavior evolves spatially as mining advances.

At the heart of the work is an inverse problem. In a forward geotechnical model, engineers assign values such as elastic modulus, cohesion, friction angle, density and deformation characteristics, then calculate how the ground should respond to excavation. In an inverse model, the direction is reversed: observed subsidence and other field information are used to estimate the underground parameters that produced the measured response. This is considerably more difficult because different combinations of rock properties can generate similar surface movements. A model may match one observation while giving an unrealistic description of the underground structure, making the problem mathematically underdetermined.

The Longwanggou study addresses that uncertainty through multi-objective inversion. Instead of optimizing a single target, such as minimizing the difference between predicted and observed subsidence, the approach considers several objectives at once. These may include the accuracy of surface displacement, the consistency of deformation patterns and the plausibility of inferred mechanical parameters. In technical terms, the method searches for a balance among competing error functions rather than forcing every aspect of the simulation into one score. The result is intended to be a set of parameter combinations that can reproduce the observed behavior while remaining compatible with the physical characteristics of a damaged rock mass.

The degradation-zoning concept supplies the physical structure needed for that optimization. Near an extracted panel, the rock may experience intense fracturing and lose part of its load-bearing capacity. Farther away, the rock can remain comparatively intact, although stress concentrations may still alter its response. Separating these areas into zones allows the inversion to assign different mechanical parameters to different parts of the model. This is more realistic than assuming that a single value of stiffness or strength applies throughout the strata. It also creates a bridge between numerical simulation and the mechanics of failure, because the zones can be interpreted as stages or regions of mining-induced deterioration.

Subsidence is not simply a downward movement at the surface. It is the visible outcome of a three-dimensional chain of events that begins when coal removal changes the stress field underground. The roof may bend, fracture or collapse into the mined-out space; adjacent strata may deform gradually; and the overburden may transmit those changes upward. Depending on geological conditions and mining geometry, the surface can develop a broad subsidence basin, steep gradients near its margins or localized features associated with faults and weak layers. Those gradients matter because they can impose tensile and compressive strains on roads, railways, pipelines, buildings and drainage systems, even when the total vertical displacement appears modest.

By linking subsidence inversion with the estimation of rock mass parameters, the researchers aim to make the model useful for more than reproducing a map of ground movement. Mechanical parameters determine how the mine and its surrounding strata respond to future excavation, altered panel layouts and changing support conditions. A calibrated model could therefore help engineers assess where deformation is likely to intensify, identify zones requiring closer monitoring and improve predictions of the long-term stability of underground workings. The broader significance lies in turning surface observations into a continuously refined picture of underground conditions, a capability that could support safer and more efficient resource extraction.

The research also reflects a wider transformation in mining science, where numerical modeling is increasingly connected to monitoring systems and optimization algorithms. In principle, displacement measurements from surveying, satellite radar, GNSS stations or other instruments can be compared with simulations and used to update model parameters. Yet no inversion method can eliminate uncertainty entirely. Measurement errors, simplified geological assumptions and incomplete knowledge of fractures all influence the result. Multi-objective strategies are valuable precisely because they can expose trade-offs between fitting the observations and preserving physical realism. For Longwanggou, the case study provides a test of how degradation-aware modeling can capture the complex relationship between mining operations, rock deterioration and surface response.

The study’s importance extends beyond one coal mine. Many mining regions face the same challenge: extracting resources while predicting how the ground will behave months or years after excavation. A framework that combines degradation zoning, multi-objective optimization and subsidence observations could eventually be adapted to other geological settings, provided it is recalibrated with local data. Its most compelling promise is not a single prediction, but a more responsive form of geotechnical decision-making—one in which models learn from the landscape as it changes. By treating subsidence as evidence of evolving underground mechanics, the Longwanggou research offers a pathway toward more transparent risk assessment and more scientifically grounded mine planning.

Subject of Research: Mining-induced subsidence and rock mass mechanical parameters using degradation zoning and multi-objective inversion.

Article Title: Multi-objective inversion of mining-induced subsidence and rock mass mechanical parameters based on degradation zoning: a case study of Longwanggou coal mine.

Article References: Zhang, S., Liu, H., Zhang, K. et al. “Multi-objective inversion of mining-induced subsidence and rock mass mechanical parameters based on degradation zoning: a case study of Longwanggou coal mine.” Scientific Reports (2026). https://doi.org/10.1038/s41598-026-64546-0

Image Credits: AI Generated

DOI: 10.1038/s41598-026-64546-0

Keywords: mining-induced subsidence, rock mass mechanics, degradation zoning, multi-objective inversion, numerical modeling, coal mining, geotechnical engineering, Longwanggou coal mine.