The Ms7.0 Jiuzhaigou earthquake on August 8, 2017, struck the Jiuzhaigou Valley, a World Natural Heritage site. This event represents another significant seismic impact on Northwest Sichuan following the 2008 Wenchuan earthquake, the first within a Chinese heritage site. Listed for its "Outstanding Universal Value" (OUV) based on exceptional natural beauty, Jiuzhaigou faced an unprecedented challenge: managing regional-scale restoration within a World Heritage site after a major earthquake-A task with no prior domestic or international parallel. Based on a comprehensive survey of eco-geological environment elements, an analysis of geohazard distribution patterns, and a systematic diagnosis of ecological issues, this study focuses on the critical needs of post-disaster geohazard prevention and geological environment protection, integrating Nature-based Solutions (NbS). This study obtains the following results: (1) Developed a water environment-friendly glutinous rice lime travertine cementitious material that achieves a compressive strength of > 12 MPa after 60 days of underwater curing; established an ecological restoration technology for damaged travertine dams using "vibro-replacement gravel piles for foundation reinforcement-glutinous rice mortar dam construction-carbonized bamboo anchors for slope protection-native plant greening-fractal landscape design", successfully preserving the authenticity of travertine lake clusters. (2) Innovated an ecological approach for managing collapse hazards in steep rock slopes affected by earthquakes, combining structural reinforcement, ecological soil stabilization, and bioinduction; proposed a niche reconstruction technique involving "barrier construction-vegetation establishment-crust formation" for large-scale loose debris landslides and soil-water disasters; with rainfall erosion resistance > 150 mm/h and 90% soil loss reduction, achieving effective integration between geological hazard control and ecological recovery. (3) Put forward a multi-level regulation technology for geohazards across " site-unit-watershed" scales, realizing synergistic effects between geohazard prevention, ecological functional recovery, and eco-geological environment protection, and after treatment, soil organic matter, nutrients, and biomass reached 90% of undisturbed levels. (4) Constructed a multi-dimensional evaluation method, incorporating 7 primary, 13 secondary, and 17 tertiary objectives, for the effectiveness of ecological geohazard mitigation by interdisciplinary integration of geological stability, soil and water conservation, ecological restoration, and landscape harmony. This research provides a scientific trajectory for disaster prevention, habitat quality enhancement, and biodiversity conservation in Jiuzhaigou, offering theoretical references and technical support for the protection and sustainable development of eco-geological environments in World Natural Heritage sites.

In western mountainous regions, factors involving active tectonics, steep terrain, frequent hazards, variable climate, and fragile ecosystems render suitable spoil disposal sites scarce under multi-factor coupling, making site selection highly challenging. To address the practical challenges of large-scale transportation projects in western mountainous areas characterized by substantial and spatially dispersed spoil volumes and difficult site selection, this study systematically identifies key siting challenges and screens critical influencing factors, proposing a hierarchical multi-factor coupled siting framework comprising extent delineation, site screening, and suitability zoning. A risk-based safety suitability model is developed to quantify the site safety of spoil disposal areas and the potential losses from secondary hazards; an economic suitability model centered on transport costs integrates haul distance and access conditions; and an ecological suitability model uses ecosystem vulnerability as a proxy, assesses three dimensions of the pattern, quality, and service. Entropy weighting and the Analytic Hierarchy Process (AHP) are combined to derive composite weights, yielding integrated suitability zoning and prioritized candidate site lists. This research provides an efficient, quantitative, and generalizable methodological framework for spoil disposal siting in complex mountainous transportation projects, substantially enhancing siting rigor while minimizing safety risks, disposal costs, and ecological restoration complexity at the source.

To address technology-environment mismatch, fragmented implementation of individual measures, and insufficient basis for technology adaptation in the ecological prevention and control of hazards in geologically disturbed areas, an integrated multi-technology configuration method is developed. Based on the coupling relationships among the engineering, hazard, ecological, and geological subsystems, a framework for scenario attribute identification and eco-geological environmental problem diagnosis is established. According to their dominant functions, treatment technologies are classified into six technical units: mechanical reinforcement, hydrothermal regulation, soil improvement, vegetation restoration, landscape integration, and water-fertilizer supply. Subject to safety redlines and engineering applicability constraints, candidate technologies within each technical unit are evaluated for suitability using combined weighting and comprehensive scoring methods, and multi-technology configuration schemes are generated according to prerequisite dependencies, functional complementarity, compatibility, and potential conflicts among the technical units. A large post-earthquake collapse deposit in Jiuzhaigou is taken as a case study. Under the combined constraints of intense rainfall disturbance, steep and loose deposits, coarse and infertile substrates, and high landscape sensitivity, a configuration scheme centered on local slope stabilization, soil and water process regulation, soil habitat reconstruction, near-natural vegetation restoration, and targeted maintenance is developed. The proposed method can transform ecological prevention and control technology selection from experience-based judgment to systematic adaptation under scenario constraints, thereby providing methodological support for technology configuration in the ecological treatment of hazards in complex geologically disturbed areas.

Engineering disturbances accelerate the degradation of alpine meadows, but restoring degraded areas through ecological remediation helps maintain regional ecosystem stability. This study investigates changes in soil nutrients and microbial community characteristics over three years of ecological restoration, using soil from the ecologically restored roadside slopes in the Wangbuqu Meadow as the research subject. Results indicate that organic carbon, total nitrogen, ammonium nitrogen, and available potassium content in post-restoration soil significantly increased by 344.6%, 95.8%, 139.9%, and 174.1%, respectively. Microbial community diversity progressively enhanced, with rising temporal trends in the abundance of Acidobacteria, Chloroflexi, Bacteroidetes, and Gemmatimonadetes. Concurrently, soil urease, protease, and sucrase activities markedly intensified. Partial Least Squares-Path Modeling (PLS-PM) analysis reveals that remediation duration served as the primary driver for soil organic carbon and nutrient recovery, facilitating synergistic restoration of soil and microbial communities by influencing microbial community structure and soil enzyme activity. This study provides empirical support for soil recovery outcomes evaluation in disturbed alpine meadow ecosystems.

This study aims to evaluate the effects of plant-microbe combined remediation on soil stabilization and erosion resistance of TBM muck slopes on the Qinghai-Xizang Plateau, where slope materials are loose in structure, deficient in nutrients, and difficult to revegetate. Indoor pot experiments and field slope trials were conducted. A control (CK) and five microbial treatments were established, including Bacillus subtilis (K), Bacillus licheniformis (DY), Bacillus amyloliquefaciens (JD), Aspergillus niger (H), and Trichoderma harzianum (HC). Soil nutrients, enzyme activities, plant growth, and slope hydrodynamic parameters were measured. The results show that the plant-microbe combined treatments improved root development, enhanced enzyme activities, and ameliorated nutrient conditions, thereby exerting integrated effects on slope hydrodynamic characteristics and ultimately increasing slope erosion resistance. Runoff power was reduced by up to 97.36%, while the resistance coefficient increased by up to 486.26%. The findings provide an effective technical approach for the ecological restoration and soil and water conservation of TBM waste-muck slopes in alpine and high-altitude regions.

To investigate the impact of large-scale engineering on the sources of soil organic carbon (SOC) on the Qinghai-Xizang Plateau, this study was conducted along a tunnel corridor in the region. Tunnel-affected (TC) and unaffected (CK) sites at three elevations (3 240, 3 420 and 3 600 m). Soil samples were collected from depths of 0-10 and 10-30 cm to determine aggregate structure and the contributions of microbial-derived carbon, indicated by amino sugars, and plant-derived carbon, indicated by lignin phenols, to SOC in the particulate (＞0.053 mm) and mineral-associated (＜0.053 mm) aggregate fractions. The results show that tunnel construction had no significant effect on glucosamine, galactosamine, total amino sugars, fungal necromass carbon or total lignin phenols in soil aggregates (P＞0.05). However, it significantly reduced muramic acid, mannosamine, and bacterial residue carbon. In the particulate fraction, amino sugar contents did not differ significantly among elevations at 0-10 cm. There was no significant difference in amino sugars in soil particulate aggregates across different elevations at 0-10 cm soil depth. However, at 10-30 cm soil depth, amino sugar content at 3 420 m was significantly lower than that at 3 240 m and 3 600 m. For mineral-associated aggregates, amino sugar content at 3 420 m was lower than those at the other elevations. Meanwhile, in the mineral-bound aggregates of the tunnel-affected area, the average contributions of microbial residue carbon and lignin phenols to aggregate organic carbon were 26.36% and 20.65%, respectively, which were significantly lower than those in the control area (P＜0.05). The study indicates that tunnel construction reduces the stability of soil aggregates, especially that of mineral associated aggregates. Specifically, under the interactive effects of tunnel construction, altitude, and soil depth, the microbial residue carbon content of mineral-bound aggregates and their contribution rate to soil organic carbon decrease significantly, whereas microbial- and plant-derived carbon contributions to organic carbon in particulate aggregates were not significantly affected by tunnel construction. The findings highlight the importance of protecting and enhancing soil carbon pools in areas disturbed by large-scale engineering activities on the Qinghai-Xizang Plateau, with implications for strengthening the plateau's ecological security barrier and advancing China's "carbon neutral" goal.

Road construction and operation, as major forms of human activities, have become important sources of disturbance to natural ecosystems. It is essential for biodiversity conservation and ecological restoration along transportation corridors to accurately understand the effects of roads on plant communities and their habitats, to identify how vegetation and soil properties vary with road age and distance from roads, and to assess the associated extinction risks of narrow-range species within road-affected areas. In this study, we investigated the domestic section of the Sino-Nepal Highway, a major road infrastructure project extending approximately 1 000 km across the Qinghai-Xizang Plateau. Based on field surveys of vegetation and soil conducted at nine herbaceous sites and 19 shrub sites located within the indirect road-disturbance zone, 10-150 m from both sides of roads of different ages, we evaluated road-induced ecological effects at both the plant community and species levels. Linear mixed-effects models (LMMs) were used to comprehensively assess the effects of road age and distance from roads on plant community characteristics and soil physicochemical properties. Redundancy analysis (RDA) and Mantel tests were applied to examine relationships between plant community characteristics and soil properties in herbaceous and shrub communities. In addition, the maximum entropy (MaxEnt) model was used to predict the potential suitable habitats and extinction risks of four narrow-range species within the corridor. The results show that the indirect disturbance zone within 10-150 m of roads represents a key area of ecological change in road-affected ecosystems. Roads influenced plant community characteristics primarily through disturbance-induced changes in soil physicochemical properties. Changes in the composition and diversity of herbaceous communities were mainly associated with variations in soil moisture content (SMC) and pH, whereas shrub communities were influenced by a broader set of soil variables, including SMC, pH, soil total nitrogen (STN), soil total phosphorus (STP), and soil organic carbon (SOC), with the magnitude of these effects varying with distance from the road. Within a 5 km buffer on both sides of the road corridor, Taxus wallichiana exhibited the highest extinction risk, followed by Rhodiola crenulata and Rhodiola wallichiana, whereas Meconopsis horridula showed the lowest risk. The findings contribute to a deeper understanding of the ecological impacts of road infrastructure on the Qinghai-Xizang Plateau and facilitate the future restoration of road-disturbed ecosystems and the conservation of narrow-range species.

To clarify the effective composition of the vegetation layer for ecological restoration of steep slopes and the evolution patterns of vegetation layer properties under freezing-thawing conditions, the changes in soil cohesion characteristics under freezing-thawing conditions were clarified through direct shear tests, combined with the attachment test on the steep slope surface, vegetation test and the optimized vegetation layer ratio. The results show that biochar can improve the vegetation performance of the soil; the addition of polyacrylamide is not conducive to plant growth after adding binder and water-retaining agent. The optimal vegetation layer composition is 0.8% seaweed polysaccharide, 1% carboxymethyl cellulose, and 1.5% biochar. After 20 freeze-thaw cycles, the loss rate of soil cohesion in the substrate soil is 1.27-2.93 times that of the reconstructed vegetation layer. Low water content conditions are conducive to maintaining the soil cohesion characteristics during the freeze-thaw process. The cohesive structure formed by the binder and water-retaining agent effectively maintains the stability of the reconstructed vegetation layer, and its anti-freeze-thaw deterioration ability is significantly better than that of the substrate soil. The findings can provide a theoretical basis for the material ratio of ecological restoration vegetation layer and engineering protection for steep slopes in the seasonally frozen areas.

This paper proposes a combined modification method based on enzyme-induced magnesium carbonation (EIMC), integrating bio-crust formation, organic matter regulation, and the Mg²⁺ slow-release effect from EIMC solidified balls. Through a standardized indoor erosion experiment, the laws and mechanisms of the three working together to enhance the hydro-mechanical stability of external-soil used in ecological slope restoration were comparatively studied. Carbonation degree, erosion rate, and cumulative erosion amount were quantified to assess both the soil modification and its anti-erosion performance; micro-structural features and micro-scale mechanical properties were characterized using scanning electron microscopy (SEM) coupled with nanoindentation. Experimental results show that the combined modification of the three achieved the best effect. The erosion rate was 77.9% lower than that of the single EIMC treatment and 82.4% lower than that of the blank group. Under the condition of 261 mm/24h rainfall and 100 minutes of erosion, the cumulative erosion amount was the lowest (9.84 g). Sweet potato starch improves carbonation degree by 2.5% and elastic modulus by 43.4% via nucleation regulation, pore structure optimization, and stabilization of the aqueous chemical environment. EIMC solidified ball enables sustained Mg²⁺ release, enhancing deep-layer carbonation degree by 3.6% and elastic modulus by 132.6%. Spray application facilitates rapid in situ formation of a dense, continuous bio-crust layer on the slope surface-effectively suppressing surface rain erosion.

Strong earthquake-induced landslides often cause extensive vegetation removal, soil disturbance, and habitat fragmentation, making their ecological impacts and natural recovery important considerations in post-earthquake ecological restoration and hazard-impact assessment. The 2022 Luding earthquake triggered numerous landslides in the alpine gorge region of Mount Gongga. To investigate vegetation-greenness damage and early recovery of earthquake-induced landslides in a densely vegetated mountain environment, this study used Sentinel-2 imagery from 2019 to 2026 and landslide-unit boundaries to analyze the pre-earthquake vegetation-greenness baseline, post-earthquake damage, recovery process, and associated factors on the eastern slope of Mount Gongga based on the normalized difference vegetation index (NDVI). Mean NDVI across landslide units decreased from 0.785 before the earthquake in 2022 to 0.488 in 2023, representing a decline of 37.9%, with marked differences in damage among individual landslide units. Vegetation greenness subsequently showed rapid but nonlinear recovery. In 2024, 82.7% of landslide units exhibited increasing NDVI; recovery weakened in 2025 under the influence of regional phenological variability, followed by widespread greening again in 2026. By July 2026, the mean recovery ratio had reached 0.832. Based on the pre-earthquake baseline, post-earthquake anomalies, recovery continuity, and endpoint status, vegetation-greenness trajectories were classified into three types: recovery, fluctuating, and persistently low-greenness. The recovery type was dominant, accounting for 79.8% of landslide units, whereas 16.5% remained persistently low in greenness. Pre-earthquake NDVI and landslide surface area showed the clearest associations with vegetation-greenness damage and recovery, while the length-to-width ratio showed weaker associations, and mean slope and aspect showed no consistent relationships. Overall, earthquake-induced landslides on the eastern slope of Mount Gongga exhibited rapid but nonlinear early greening, with pronounced differences in recovery trajectories among individual landslide units, and some landslides still showing greenness deficits by the fourth observable post-earthquake growing season. This study reveals, at the landslide-unit scale, the transition of vegetation greenness from post-earthquake damage to differentiated recovery, providing a basis for identifying persistently low-greenness landslides and prioritizing areas for post-earthquake ecological restoration.

With further advancement of national ecological civilization construction, traditional geological hazard control projects focusing on rigid protection, have presented increasingly prominent limitations in ecological protection and restoration, ecological environment quality improvement, and landscape restoration. The ecological functionality and durability of such projects are severely inadequate. Taking geological hazard risk mitigation and ecological benefit enhancement as core objectives, this study elucidates the coupling relationship between the ecological environment and control projects, based on the analysis of planning and design for ecological geological hazard control and the organic configuration of ecological control technologies. By comprehensively evaluating three dimensions-structural durability, control effectiveness durability, and landscape restoration durability-the study identifies specific indicators that objectively reflect ecological durability and establishes an assessment index system for the durability of eco-friendly geological hazard control projects. The proposed system includes 7 primary indicators, 13 secondary indicators, and 17 tertiary indicators, along with a clear quantification methodology. This research can promote the improvement of ecological benefits in geological hazard prevention and control and advance ecological disaster mitigation practices, providing scientific parameters and theoretical support for scheme formulation and effectiveness evaluation of ecological geological hazard control projects.

Uniaxial compressive strength (UCS) is a fundamental parameter for characterizing the mechanical behavior of rock masses which is essential for geotechnical stability analysis. However, the efficient and precise determination of UCS remains a significant challenge due to the complexity of rock materials. This study develops a robust predictive framework by integrating the Light Gradient Boosting Machine (LightGBM) with the Coati Optimization Algorithm (COA). Based on a comprehensive database of 273 rock samples, the COA is utilized to systematically optimize the hyperparameters of the LightGBM model. The results indicate that the proposed COA-LightGBM model exhibits exceptional predictive accuracy, achieving an RMSE of 20.48, an R² of 0.83, and an MAE of 15.24. Comparative analyses demonstrate that the COA-LightGBM model significantly outperforms conventional models such as K-Nearest Neighbors, Decision Trees, and Random Forests. Furthermore, the Shapley Additive Explanations (SHAP) method is introduced to elucidate the internal decision-making mechanism and bridge the gap between "black-box" models and physical interpretability. Feature analysis reveals that porosity (n) is the most critical factor governing rock strength. This research not only validates the superior optimization capability of COA but also provides a scientific and interpretable solution for the rapid assessment of rock mechanical parameters in engineering practice.

This study investigated the synergistic effects of Aspergillus niger spore powder and a novel water-retaining material (SJP) on iron tailings from the Awulale iron ore belt. An indoor pot experiment was conducted to determine plant growth, tailings physicochemical properties, soil enzyme activities, and rhizosphere microbial community characteristics, and a comprehensive evaluation was performed to identify the optimal amendment combination. The treatments significantly promoted alfalfa growth, with plant height increasing by 166.67% under the optimal treatment relative to the corresponding control. Tailings moisture content, nutrient concentrations, and enzyme activities increased by 203.61%, 87.92%-1 935.67%, and 186.74%-1 532.24%, respectively. The treatments also altered the rhizosphere microbial community structure, while functional prediction indicated significant enrichment of fundamental metabolic pathways, including glycolysis and amino acid biosynthesis. The optimal combination comprised an iron tailings-to-nutrient soil mass ratio of 1∶0, an SJP concentration of 1%, and 2.0 g of A. niger spore powder. These findings provide a technical reference for the ecological restoration of iron tailings in Xinjiang.

To evaluate the restorative potential of moss-algal biological soil crusts (biocrusts) on alpine construction spoil, we conducted a comparative field experiment at a spoil-disposal site. We established contrasting moss-algal assemblages and quantified their effects on soil physicochemical properties, enzyme activities, and microbial communities. Biocrust colonization increased soil ammonium nitrogen up to 2.68-fold, nitrate nitrogen up to 2.85-fold, available phosphorus up to 7.67-fold, and soil organic carbon to 26.11 g/kg. Cellulase, β-glucosidase, and phytase activities reached 2.00, 2.29, and 45.6 times the control levels, respectively. Additionally, biocrusts enhanced microbial α-diversity and evenness, and enabled the detection of rare phyla such as Synergistota and Dadabacteria. Overall, moss-algal biocrusts improved soil nutrient status and optimized the structure of soil microbial communities. The findings provide theoretical and practical guidance for applying biocrust technology to ecological restoration of alpine spoil sites.

Against the background of global climate change, increasingly frequent extreme rainfall events have intensified the risk of rainfall-induced landslides in mountainous regions, highlighting the importance of accurate landslide susceptibility assessment. The quality of training samples, particularly negative sample selection, is a key factor affecting the reliability of machine learning-based susceptibility models. This study focuses on a section of the Laibao Expressway severely affected by the Beijing "23·7" extreme rainfall event. DInSAR, NDVI, and optical remote sensing imagery were integrated to identify geological hazards. The coupled TRIGRS-Scoops3D model was then used to evaluate slope stability under rainfall infiltration conditions, and the results were used to optimize negative sample selection. Logistic regression (LR), random forest (RF), support vector machine (SVM), and XGBoost models were further developed for landslide susceptibility assessment. Their predictive performance was evaluated using receiver operating characteristic (ROC) curves and related performance metrics, while SHAP analysis was used to quantify the contributions of conditioning factors. The results show that the Beijing "23·7" extreme rainfall event triggered 466 geological hazards, including 345 landslides, mainly distributed on mountainous slopes with pronounced topographic relief and strong valley incision. The TRIGRS-Scoops3D model showed capability in identifying potentially unstable areas, with 90.43% of the observed landslides falling within unstable zones and a %LR_class value of 80.09%, representing improvements of 15.36% and 9.43%, respectively, over the TRIGRS model. After negative sample selection was optimized using the physics-based model, XGBoost achieved the best performance, with an AUC of 0.956, followed by SVM and RF with AUC values of 0.906 and 0.899, respectively, while LR yielded a relatively lower AUC of 0.861. SHAP analysis further indicated that slope gradient and rainfall were the dominant factors controlling rainfall-induced landslide susceptibility in the study area. The findings provide a useful reference for landslide susceptibility assessment in complex mountainous regions and for geological hazard prevention and mitigation along mountainous transportation corridors.

The Zhengchong lithium-rubidium-cesium deposit in Daoxian County, Hunan Province, is a large rare metal deposit being developed in the Nanling region. The ore-forming geological body is dominated by pervasively mineralized quartz-zinnwaldite-topaz porphyritic rocks. However, the genesis of the quartz-zinnwaldite-topaz porphyritic rocks in this deposit remains controversial: magmatic versus metasomatic origin. It is essential to elucidate their genesis for refining the metallogenic mechanisms of rare metal deposits, since such distinctive greisen-type rocks in the Nanling region are critical hosts for rare metal mineralization. Based on field geological investigations and laboratory analyses, it is concluded that the quartz-zinnwaldite-topaz porphyritic rocks are not products of greisenization. The principal lines of evidence are as follows: (1) the contact boundary between the quartz-zinnwaldite-topaz porphyritic rocks and granites (or granite porphyries) is marked by an abrupt change in mineral proportions, and granite xenoliths occur within the quartz-zinnwaldite-topaz porphyritic rocks; (2) the quartz-zinnwaldite-topaz porphyritic rocks display a typical porphyritic texture with mostly euhedral phenocrysts, in which mineral-melt inclusions are widely developed. Cathodoluminescence (CL) images of quartz phenocrysts reveal primary growth zoning; (3) Al contents in quartz are relatively homogeneous, with an average Ti content as high as 21.8 μg/g; (4) micas are dominated by zinnwaldite with scarce muscovite. The high F content of zinnwaldite and the low F content of coexisting topaz collectively support a magmatic origin; (5) monazite lacks hydrothermal alteration textures and alteration minerals, with magmatic geochemical characteristics; (6) whole-rock Y/Ho, Nb/Ta ratios, and TE_1-3 values (the deviation of the first and third tetrad of granite REE patterns from a hypothetical tetrad effect-free REE pattern) show characteristic behavior of trace elements in silicate melts, which differ from those of granitic rocks or greisens that have undergone significant magmatic-hydrothermal alteration. Geochemical modeling results indicate that the quartz-zinnwaldite-topaz porphyritic rocks are likely formed by multistage in situ crystallization of the Jinjiling complex.

Seismic image super-resolution reconstruction is of great significance for improving the accuracy of seismic interpretation and geological modeling. However, existing methods often suffer from large model sizes and high computational costs, and exhibit limited detail recovery capability under strong noise and complex geological structures. To address these issues, this paper proposes an enhanced residual feature distillation generative adversarial network (ERFDN-GAN) for seismic image super-resolution reconstruction. The network is built upon enhanced residual feature distillation blocks (ERFDBs), where a feature distillation mechanism is employed to reduce model complexity. Spatial attention and channel attention mechanisms are further introduced to enhance feature representation capability. Experimental results demonstrate that the proposed method achieves overall superior performance compared with representative methods in terms of PSNR and SSIM, while maintaining low model complexity and stable reconstruction performance under complex scenarios. The proposed ERFDN-GAN method effectively balances computational efficiency and reconstruction quality, providing technical support for seismic interpretation, geological modeling, and the intelligent acquisition and mining of multi-source geoscience data.

The Zhaishang-Liba mining district in the West Qinling Orogen is one of the key exploration areas for the National Strategic Action for Mineral Exploration Breakthrough in China. Due to the thick shallow surface cover, traditional exploration methods face severe challenges. Mineral prospectivity mapping driven by big data and artificial intelligence provides a new perspective for identifying weak mineralization anomalies in covered terrains. However, during the generation of positive samples, existing models rarely consider the spatial heterogeneity of deposit scales and metallogenic energy fields, which restricts the accuracy and reliability of the predictive models. Therefore, this paper proposes an innovative positive sample construction method dynamically constrained by the "resource-area" relationship. Integrating multi-source exploration data, including geological, geochemical, and remote sensing data, 17 predictor variables were extracted to construct a Random Forest (RF) predictive model, coupled with the SHAP (SHapley Additive exPlanations) method for geological interpretability analysis. The predictive model demonstrates excellent generalization and classification performance. Under a probability threshold of 0.52, it successfully captures 92% of the known gold deposits within merely 8% of the predicted area. SHAP analysis reveals that variables such as Au, Sn, Cu, Zn, fault density, Sb, As, and proximity to intrusions play dominant roles in the prediction. The predicted high-probability areas spatially exhibit a "linear" distribution pattern significantly controlled by the superposition of multi-stage faults, and an "annular" pattern surrounding the "Five Golden Flowers" composite intrusions, profoundly reflecting the metallogenic dynamic mechanism that gold mineralization in this area is strictly controlled by the spatiotemporal coupling of the "tectonic-magmatic" system. Based on the prediction results, 6 deep exploration targets were delineated. Notably, Target P1 was verified by a deep drill hole (ZK4-1), which successfully intercepted a gold mineralization body with a maximum grade of 1.24 g/t. The study demonstrates that the intelligent mineral prospectivity mapping framework constrained by deposit scale holds significant scientific indicative value and application prospects in covered areas.

To address the limitations of traditional binary discrimination diagrams, which arise from low dimensionality and difficulty in revealing nonlinear metallogenic mechanisms when processing zircon trace element data, this study introduces machine learning methods to enhance the accuracy and objectivity of porphyry copper mineralization potential assessment. Based on global zircon trace element data from porphyry deposits, we systematically compared the classification performance of five models-XGBoost, Random Forest, Support Vector Machine, Logistic Regression, and Neural Network- and interpreted the model decision-making mechanisms using the SHAP explainable framework. The results show that XGBoost achieves the highest classification accuracy (86.85%), followed by Random Forest (83.84%) and Logistic Regression (83.84%), while Support Vector Machine (78.90%) and Multilayer Perceptron (69.86%) exhibit relatively lower performance. SHAP analysis identifies Eu_N/Eu_N*, 10 000(Eu_N/Eu_N*)/Y, Ti, Y, △FMQ, and Gd/Dy as key indicators for discriminating mineralization potential. The models constructed in this study, such as XGBoost and Random Forest, can effectively identify porphyry copper mineralization potential and screen critical exploration indicators, offering not only new insights into intelligent mineral exploration but also important references for model selection and optimization in practical applications.

To address the problems of low accuracy, poor spatial continuity and high uncertainty in 3D geological modeling caused by sparse borehole data, unbalanced categories and complex geological structures, a phased joint modeling method integrating semi-supervised deep learning and multiple-point geostatistics is proposed in this paper. Firstly, a semi-supervised Spatial-Aware Attention Net (SAAN) based on neighborhood spatial context features is constructed, and the pseudo-label mechanism is utilized to improve the classification accuracy of strata and to generate the initial model; Subsequently, the multiple-point geostatistics Random Neighborhood Search (RNS) strategy is combined to iteratively optimize the areas with low confidence and geological semantic conflicts; Finally, the collaborative consistency entropy is proposed to quantify the two-stage prediction consistency and joint uncertainty. Taking the borehole data in the Chengdu Plain area as an example, the proposed method outperforms baseline methods such as Inverse Distance Weighting (IDW), Support Vector Machine (SVM), Random Forest (RF), Multi-layer Perceptron (MLP), and Kolmogorov-Arnold Networks (KAN) in terms of reproduction of stratum spatial distribution, boundary characterization, and structural continuity. In addition, the collaborative consistency entropy can effectively identify structural uncertainties and avoid misjudgments of high entropy values but consistent predictions. The proposed phased joint modeling framework can achieve high-precision, structurally reasonable and uncertainty controllable three-dimensional geological model reconstruction under the conditions of limited and unbalanced borehole data, providing effective technical support for the characterization and engineering application of underground structures in complex geological environments.

The differences in gas content and their controlling factors in the deep coalbed methane (CBM) reservoirs of the Carboniferous-Permian strata in the Dacheng Uplift of the Bohai Bay Basin remain critical challenges in exploration and development efforts. This study analyzes gas content data from existing wells in the Dacheng Uplift and conducts numerical simulations. The results demonstrate that the Dacheng Uplift exhibits a distinct "lower in the north and higher in the south" gas content distribution pattern in its Carboniferous-Permian coal seams, with gas contents in southern exploration wells generally exceeding 10 m³/t. Key controlling factors on coalbed gas content include reservoir temperature, pressure, water saturation, coal rank, and pore characteristics. As coal rank (R_o) increases, methane adsorption capacity strengthens while free gas content decreases. Pore volume shows a significant positive correlation with gas content, whereas total porosity negatively correlates with coal rank (R_o). Increased ash content also leads to reduced total porosity. This study further constructs isothermal adsorption and free gas content prediction models for deep coal seams under the influence of reservoir pressure, temperature, coal rank, and apparent density. Model validation demonstrates good correlation between predicted results and experimental data, validating their applicability for estimating methane adsorption capacity in deep coal seams. Calculations reveal that methane saturation adsorption reaches a critical value of 14 m³/t at approximately 1 100 m burial depth, while free gas content increases with depth before stabilizing. The critical burial depth for total gas content is around 1 200 m, where gas content reaches 16 m³/t. Considering gas content, gas occurrence state, and burial depth comprehensively, the optimal development depth range for the Dacheng Uplift is identified as 1 500-2 600 m, where total gas content ranges between 14-16 m³/t with free gas accounting for 25%-60%. This depth interval combines high gas content with mobile free gas advantages, representing the most favorable target zone for CBM exploitation in this area. The high gas content zones in the Dacheng Uplift are significantly influenced by coal rank and thermal maturity, with the gas generation capacity of the coal seams being the primary controlling factor for the enrichment of deep coalbed methane in this area.

With the acceleration of the global energy structure transformation, deep coalbed methane as an important unconventional natural gas has become a strategic replacement resource for China's natural gas production increase. As an important part of the Bohai Bay Basin, the Jizhong Depression has undergone multiple tectonic evolutions in its Carboniferous-Permian coal-bearing strata, presenting complex geological characteristics and coalbed methane enrichment patterns. Based on drilling, logging and experimental test data, this paper systematically analyzes the geological characteristics, main controlling factors of enrichment and accumulation models of deep coalbed methane in the Jizhong Depression, and proposes a favorable exploration area selection scheme. The research shows that the deep coal seams in the Jizhong Depression are thick (the maximum single layer of the Taiyuan Formation is 25 meters), with significant burial depth differences (1 000-5 600 meters), and the vitrinite reflectance (R_o) of the coal rocks generally ranges between 0.65% and 0.95%, with vitrinite as the main maceral component (46.7%-97.4%). The thermal maturity is significantly affected by volcanic baking, and the local vitrinite reflectance (Ro) is as high as 4.6%. The gas content of the coal seam (6-18 m³/t) is closely related to the coal thickness, pressure coefficient and preservation conditions. The weak overpressure environment (pressure coefficient ＞1.15) and stable cap rock (breakthrough pressure of carbonaceous mudstone 15-29 MPa) in the deep part promote gas enrichment. The study reveals that anomalous heat flow triggered by volcanic activity is a key geological factor enhancing coalbed methane generation, and the spatio-temporal configuration of secondary hydrocarbon generation and structural preservation conditions dominates the accumulation process. Through comprehensive evaluation, the middle to outer belt of the Wena Slope and the northeastern part of the Dacheng Uplift are classified as Class Ⅰ favorable areas, with advantages such as large coal seam thickness (6-8 meters), high gas generation intensity (＞110 m³/t), and large Langmuir volume (average 11.39-12.42 m³/t). The research results provide theoretical support for the exploration and development of deep coalbed methane in the Jizhong Depression and offer a reference for the evaluation of coalbed methane in similar residual faulted basins.

The Taiyuan Formation in the Linxing Middle Zone, eastern margin of the Ordos Basin, serves as the key layer for the collaborative exploration and development of tight sandstone gas and coalbed methane (CBM). To clarify the enrichment patterns of coal measure gas in this region, this study finely characterizes the sedimentary evolution characteristics of large tidal sand bodies in the lower section of the Taiyuan Formation, clarifying the spatial stacking relationship of tidal sand bodies and their relationship with gas-bearing genesis, on the basis of the division of sequence stratigraphy, and in combination with seismic, logging, core, and outcrop data. It is suggested that the lower section of the Taiyuan Formation is a tidal flat sedimentary system, and tidal channels and sand flats are the main reservoir types of the Taiyuan Formation gas reservoir in this area. The large tidal sand bodies are distributed in a tongue-shaped pattern with thickness greater in the north and less in the south, and in an east-west lens shape in the northern part of the study area, showing a vertical evolution law from foundation development to peak development to decline and disappearance, which has a strong scouring effect on the underlying 8+9# coal seam. The stacking types of tidal sand bodies mainly include multi-stage stacking, vertical stacking, lateral stacking, and isolated types. Among them, multi-stage and vertical stacking sand bodies have excellent connectivity, greater thickness, and better gas content, making them potential high-quality reservoirs in the area. Comprehensive analysis suggests that the primary controlling factor for the distribution of gas-bearing units is sedimentary microfacies, followed by sand body stacking styles (multi-stage and vertical stacking styles are most favorable), while the physical conditions inside tidal channels are the essential factors resulting in differences in gas-bearing properties. The research results can provide a theoretical basis and technical support for the collaborative and efficient exploration and development of deep coalbed methane and tight sandstone gas in the Linxing Middle Block.