Seismoacoustic profiling and rock-magnetic analyses provide a coherent basis for reconstructing the depositional history of Lake Bolshoy Kisegach (Southern Urals, Russia). We combined seismoacoustic sections with Accelerator Mass Spectrometry (AMS) 14C dating of bulk organic carbon from five core sites (258 to 540 cm) and measurements of magnetic susceptibility, natural remanent magnetization, coercivity spectra, differential thermomagnetic curves, and scanning electron microscopy (SEM) observations. The basin has heterogeneous bottom morphology that controls sediment distribution; sediment thickness reaches 7.5 m in the lake’s center. Two units are distinguished: Late Pleistocene strata (~3 m) overlain by Holocene sapropels (4.5 m). Calibrated radiocarbon ages indicate continuous accumulation since at least 12600 cal. yr BP and suggest lake initiation ~23000 cal. yr BP. Sedimentation rates vary from 0.2 to 1.24 mm·yr−1. Initial magnetic susceptibility (χ) ranges from 0.01 to 3.99 × 10−6 m3·kg−1, and natural remanent magnetization (NRM) ranges from 4.3 × 10−8 to 8.5 × 10−5 A/m. The magnetic fraction comprises mixtures of single-domain, multidomain, and superparamagnetic grains. Variations in magnetic parameters track recognized Late-Glacial and Holocene climate phases: the Younger Dryas marks the coldest interval, whereas the Subatlantic (~2000 cal. yr BP to recent) is the most humid. Lake regressions occurred during the Allerød interstadial, the Younger Dryas and the Boreal. These results demonstrate the value of integrating seismoacoustic guidance for coring with magnetic proxies to resolve lake-basin development.
The rice-crayfish coculture system involves growing rice and crayfish together in dedicated fields, providing important economic and ecological benefits. Accurate and timely instance-level monitoring of rice-crayfish fields (RCFs) is essential for tracking rice growth, estimating yield, and managing water resources. Previous deep learning-based approaches, relying on semantic segmentation and extra post-processing for RCF instance isolation, which can introduce propagate errors, leading to boundary inaccuracies and reduced reliability in real-world applications. To address these issues, we propose EMC-YOLO, an end-to-end instance segmentation framework that integrates an efficient vision transformer with a multi-scale, contour-guided YOLO architecture. By unifying global context modeling, multi-scale representation, and boundary refinement within a single pipeline, the proposed method performs direct instance-aware mask prediction. This design eliminates error-prone post-processing and enables precise delineation of individual RCFs across diverse scales, shapes, and crop stages. Experimental results highlight that the proposed method achieves a box mAP (mean average precision) of 0.954 and a mask mAP of 0.952, surpassing YOLOv8n-seg by 1.6 and 2.5 percentage points, respectively. This robust performance enables reliable extraction of RCF instances under variable field conditions, establishing EMC-YOLO as a powerful solution for high-precision, fine-scale mapping of rice-crayfish coculture systems and providing actionable technical support for smart agriculture implementation.
The issue of potentially toxic elements (PTEs) pollution in river sediments has a focal point of global concern in the environmental field. This study employed a receptor model and multiple indices to analyze the PTEs in river sediments from a dry-hot valley. The results showed that the average concentrations of PTEs within the river sediments followed the order: Pb > Zn > Cr > Cu > Ni > As > Cd > Hg, with elevated contents mainly observed in the middle and upper reaches. Three sources of PTEs were determined using the positive matrix factorization (PMF) model. Among these, mineral exploitation and transportation activities in the upper reaches mainly contribute to As, Cd, Hg, Pb, and Zn, accounting for the largest proportion (46.41%). This is followed by a natural source characterized by Cr and Ni, representing basalt weathering (27.40%), while agricultural activities (26.19%) in the middle and lower reaches mainly contribute to Hg and Cu. The outcomes of the geo-accumulation index (Igeo), potential ecological risk index (PERI), and environmental capacity based on the linear mass balance model suggest an overall slight to moderate pollution level (53.8%) and low ecological risk (43.6%), with high risks and pollution concentrated in the upper reaches. A total of 41.03% and 5.13% of the samples showed an unacceptable non-carcinogenic risk for children and adults. Cd and As were identified as the leading elements to the ecological and human health risks, respectively. While mining activities are the dominant contributor to both ecological and human health risks. Most areas in the upper reaches are environmental capacity overloaded zones, and the existing environmental capacity gradually increases from the upper reaches to the lower reaches. Moreover, the dynamic environmental capacity fluctuates and decreases over time, while the overloaded zones gradually recover. Overall, this study quantitatively identified three sources of PTEs in river sediments and conducted multi-dimensional evaluations of PTEs, confirming Cd and Pb pollution in the upstream area caused by mining activities.
The Boxing subsag of the Bohai Bay Basin is estimated to contain a significant volume of shale oil resources; however, the present-day in situ stress distributions are poorly understood, which limits further shale oil production. To address the above issue, a one-dimensional mechanical earth model and three-dimensional geomechanical model were established to analyze the stress distribution within shale oil reservoirs of the Boxing subsag. The results indicate that the vertical (Sv), horizontal maximum (SHmax) and horizontal minimum (Shmin) principal stress magnitudes follow the relationship of SHmax > Sv > Shmin, indicating a dominant strike-slip faulting stress regime. Laterally, the distributions of SHmax and Shmin are similar, with higher and lower stress values in the north-western and southern parts of the study area, respectively. The calculated horizontal and vertical stress difference coefficients suggest that hydraulic fractures in shale oil reservoir of the Boxing subsag may be complex. In addition, based on interpretations from imaging logs, the SHmax and natural fracture orientations are ~E–W-trending and ~ENE–WSW-trending, respectively. The angles between the SHmax orientation and fracture strike are relatively small, indicating that natural fractures are highly mechanically effective under the present-day in situ stress state. Horizontal wells are better drilled along the Shmin orientation; therefore, hydraulic fractures may be optimal for contacting more reservoir rocks. The results provide geological and engineering references for shale oil development in the Boxing subsag.
The frequency of extreme flood and rainfall disasters worldwide has seriously impacted socioeconomic development in recent years, highlighting the critical nature of accurate flood simulations and quantitative forecasts. With rapid advances in hydraulic engineering, regional hydrological conditions, particularly underlying surface characteristics and runoff generation mechanisms, have been significantly altered, thereby influencing flood simulations and forecasts. In this study, the Liuxihe model, a next-generation distributed physical hydrological model, was employed to quantitatively simulate flood events in the Liuxihe Reservoir Basin and Huanglongdai Reservoir Basin. The model incorporates a reservoir module to assess the effect of water engineering on flood forecasting. By comparing simulations with and without the reservoir module for 14 flood events, 8 for parameter calibration and 6 for validation, the role of reservoirs in flood prediction was evaluated. The results revealed that simulations integrating the reservoir module were highly consistent with the observed data. Key performance metrics, including the Nash coefficient, peak flow error, peak time error, process relative error and correlation coefficient, were significantly improved. Specifically, the inclusion of the reservoir module increased the average Nash coefficient by 70.63% and the correlation coefficient by 6.29% and reduced the peak flow error, process relative error, and peak time error by 79.45%, 53.03%, and 23.53%, respectively. These satisfactory flood simulation findings indicate that water engineering, particularly reservoirs, substantially influences the simulation accuracy of the Liuxihe hydrological model. Therefore, when hydrological forecasting is conducted in river basins with reservoirs, it is necessary to account for reservoir regulation in hydrological forecasting to minimize uncertainties. Moreover, this study provides critical technical support for increasing flood prediction accuracy, offering theoretical guidance for flood mitigation and water resource management.
The Junggar Basin exemplifies a “source–sink” system in deep time, characterized by complex sedimentary processes. In this study, we provide a comprehensive characterization of the source-sink system and reconstruct the quantitative evolution of sediment supply, constrained by full-bank discharge data. Drawing on field outcrops, core samples, and seismic reflection profiles, we delineate the key sedimentary features of the Upper Wuerhe Formation. Incised valleys were identified through seismic reflection terminations and sedimentological analyses of well logs, mud logs, and cores. To quantify variations in sediment discharge driven by changes in the source area, we employed a modified fulcrum method. Valley geometries were measured directly from core data, with additional parameters—such as valley width (Wv) and paleo-slope (S)—estimated using empirical equations. Finally, we characterized shifts in sedimentary pathways and source-area dynamics based on the paleogeomorphology and depositional signatures within the sink region. More importantly, we use modified fulcrum approach to quantify the response of sand transport to the changes in the source area, and the valley data were used to estimate the full bank discharge (), full bank bed load () and full bank suspension load () of the river. Then the sediment supply in the source area can be restored. The results show: 1) the Upper Wuerhe Formation in the Upper Permian corresponds to a complete third-order sequence which can be further divided into a lowstand system tract (LST), a transgressive system tract (TST) and a highstand system tract (HST). The main transportation pathways in the study area are valleys, which predominantly developed during the LST and TST phases, and primarily infilled during the LST phase. 2) In terms of source area, light minerals, heavy mineral suite, and U–Pb zircon data are consistent and all show that the source comes from the surrounding mountains, and a small part comes from the uplift inside the basin. 3) The quantitative calculation results of the modified fulcrum approach indicate a decreasing trend in both the full bank bed load () and full bank suspension load () from LST to TST, suggesting a gradual reduction in the overall sand transport volume of the river. This finding aligns with the outcomes of paleo-geomorphology and paleoclimate analysis, supporting the inference that the LST period exhibits strong source area characteristics and abundant material supply. Furthermore, the sedimentary features in the sink area further corroborates this result. 4) There are significant changes of “source–sink” system in two stages. The different characteristics of the source area, transfer system and sedimentary paleogeomorphology between the LST stage and TST stage ultimately led to significantly different sedimentary characteristics in the two stages. In LST stage, the source area produces a sufficient sediment discharge, and the powerful sediment discharge forms the valleys characterized by large numbers and large depth. The sediments are mainly coarse grained. In TST stage, the more humid climate narrows the source area, and the number and depth of valleys become less and shallower. The sediment volume in the sink is smaller than that in LST stage. On the whole, the evolution of the source region controls these factors and ultimately determines the evolution of the whole system. The results of quantitative reconstruction of sedimentary path using valley data are consistent with the results of U–Pb data, which can be used as a new method for quantitative study of “source–sink”.
Multi-scale mechanical research has become an effective approach for analyzing the behavior of complex materials, enabling the prediction of macroscopic properties from microscale characteristics. Pyrite, a key mineral affecting the mechanical properties of shale, has attracted considerable attention due to its brittleness and distinctive microstructure. However, the specific effects of pyrite’s microscopic features on shale mechanics have not been systematically investigated. In this study, Python-based machine learning techniques are employed to establish an automated framework for quantitatively analyzing pyrite’s microscopic characteristics. The influence of pyrite content, particle size, shape factor, and distribution on shale mechanical properties and fracture network formation is examined using a random forest regression algorithm and numerical simulations. The results show that pyrite content exerts the strongest influence on shale mechanics, followed by shape factor, while particle size and distribution mode have comparatively weaker effects. Compressive strength and fracture behavior are particularly sensitive to elevated pyrite content and shape factor. All four factors (pyrite content, particle size, shape factor, and particle distribution) affect mechanical properties through stress concentration, with shape factor additionally governing particle interlocking. Distribution mode further modulates the mechanical response by influencing the formation of force chain networks. Numerical simulations reveal that fracture network development is optimized when pyrite content is 3.00% and the shape factor is 0.81, thereby enhancing the fracturing effect. This study provides theoretical support for hydraulic fracturing in shale reservoirs and introduces a novel perspective on the role of pyrite’s microscopic characteristics in governing shale mechanical behavior.
The increasing impacts of tropical cyclones (TCs) under climate change necessitate a deeper understanding of boundary layer turbulence dynamics, particularly in urban landscapes dominated by high-rise building clusters in the coastal megacities. This study investigates the turbulent characteristics within the TC boundary layer (within 400 m, TCBL) of Typhoon Muifa (2022) during its landfall in Shanghai, China, using synchronized Doppler wind lidar observations from two sites ~18 km apart: one site in Lujiazui, the dense downtown with dozens of skyscrapers, and the other in Baoshan, a rural countryside area. Key findings of this research reveal distinct turbulence regimes in the two environments. Maximum wind speeds were similar in the downtown (about 35 m s−1) and the countryside (about 37 m s−1), with downdraft magnitudes exceeding 5 m s−1 (Downtown) and 6.8 m s−1 (Countryside), respectively. The turbulence kinetic energy (TKE) profiles diverged significantly: urban TKE exhibited a non-monotonic vertical structure, peaking at ~250 m before declining, whereas rural TKE decreased uniformly with height. Analysis of the estimable TKE budget terms identified shear production as the largest measurable source term of enhanced turbulence in the urban environment. This finding is consistent with the wind speed disparities and complex airflow interactions induced by building clusters. Meanwhile, the vertical advection term also was identified as a significant contributor in sustaining and developing turbulence. These results highlight the critical influence of urban morphology on TCBL dynamics, with implications for refining numerical simulations of wind fields, improving disaster resilience strategies for coastal megacities, and informing wind-resistant design standards for high-rise structures in typhoon-prone regions.
This study presents an integrated framework combining a real probability-based approach with automated machine learning (AutoML) to quantitatively assess the landslide risk of the southern segment of the Beijing–Guangzhou Railway under heavy rainfall. A total of 6660 landslides triggered by a single rainfall event were analyzed, incorporating twelve key influencing factors, including accumulated rainfall, topography, and geological conditions. A sample database was constructed based on the real probability criterion, and three models, namely XGBoost, Random Forest (RF), and Extra Trees (ET), were efficiently optimized using the FLAML framework. Among them, XGBoost yielded the best performance, achieving the highest AUC and demonstrating the strongest spatial agreement between predicted high-risk zones and observed landslide occurrences. Feature importance analysis identified accumulated rainfall, strata, and elevation as the dominant contributing factors. The results reveal that the overall risk of the railway section during this event was primarily classified as extremely low or very low. Notably, areas experiencing intense precipitation still pose significant risks to railway operations due to potential rapid-onset landslides. This research provides a data-driven, scalable tool for quantitative disaster risk assessment and offers a foundation for the development of a dynamic early warning platform by incorporating real-time rainfall forecasts.
As a significant coal resource base in China, the Turpan-Hami Basin hosts substantial coal reserves with notable potential for trace element enrichment and mineralization. This study investigates the restrictions imposed by the sedimentary environment on trace element distribution through geochemical analysis of coal and depositional settings in the No. 3-2 coal seam of the XX Coal Mine. The results indicated that the coal in the seam is rich in vitrinite and is classified as a low-rank coal with ultralow ash content, high volatility, and moderate fixed carbon. The average contents of vitrinite, moisture, ash, volatile matter and fixed carbon are 83.54%, 4.57%, 4.11%, 41.65%, and 55.96%, respectively. The average random vitrinite reflectance is 0.487%. Only the values for V, Co, Ni, Zn, Sr, In, and Ba are close to the average values for worldwide coals. The trace elements in coal are generally present at low concentrations and are depleted. The contents of major and trace elements in the roof of the vertical upward coal seam are generally greater than those in the coal. The rare earth elements and yttrium in coal are generally characterized by LREY enrichment, with negative Eu and Ce anomalies and positive Y anomalies. The predominant sediment source in the study area is identified as acid rock. The coal-forming plants consisted primarily of herbaceous plants. The coal-forming environment was interpreted as a freshwater phase, with moderate weathering, a dry and hot paleoclimate, and an oxidizing environment. Sediment source is identified as the dominant factor controlling trace element depletion, while insufficient hydrodynamic conditions suppress terrigenous debris input, further restricting element enrichment.
This study analyzed the present-day in situ stress field characteristics in the Longmaxi Formation of Luzhou area using methods of comprehensive logging calculations, numerical simulations, etc. Further, this paper analyzed and quantitatively evaluated the mechanical effectiveness of natural fractures in the target layer based on the relationship between in situ stresses and natural fracture development. The results reveal that the dominant orientation of present-day in situ stress in the Layer S1l1-1 is WNW-ESE-trending in the Luzhou area, affecting by local structures. The horizontal maximum principal stress in the Layer S1l1-1 ranges from 70 MPa to 125 MPa, with the majority between 85 MPa and 110 MPa. The horizontal minimum principal stress varies from 50 MPa to 100 MPa, predominantly falling between 75 MPa and 90 MPa. The stress distribution exhibits strong heterogeneous characteristics, significantly influenced by tectonic activity. Under the present-day in situ stress state, mechanically effective natural fractures constitute a small portion of the total fractures. The effectiveness is mainly influenced by factors such as fracture inclination, fault zone development, and the orientation of stress relative to fracture strike. The results are expected to support shale gas exploration and development in the Longmaxi Formation of Luzhou area, Sichuan Basin.
Low-permeability reservoirs are globally significant oil and gas resources and their recovery can commonly be enhanced by water injection technology. However, water-induced fracture (WIF) is caused by water-rock interactions during long-term water injection. Their rapid propagation leads to premature water breakthroughs between injection and production wells. Traditional research has often ignored water-rock reactions, leading to an insufficient theoretical foundation for controlling water breakthrough risks. Based on damage mechanics theory, this study established a comprehensive numerical model coupling two-phase flow, water-rock interactions, rock deformation, and fracture propagation. By introducing a rock hydration coefficient (Kw) to quantitatively characterize the effect of hydration on rock mechanical properties, the mechanism of WIF propagation and its controlling factors are systematically revealed. The results demonstrate that: 1) Water-rock interactions exacerbate reservoir heterogeneity, altering geomechanical properties such as Young's modulus and leading to stress redistribution. The stress at the fracture tip can be reduced by up to 18.2%; 2) Water-rock reactions significantly influence WIF propagation. Compared to low hydration states, The propagation rate of WIF increases by 12 times under high hydration conditions (Kw>0.5); 3) In order to avoid rapid WIF propagation, fracture spacing between production and injection wells should exceed 180 m, injection pressure should be 12.5% lower than the minimum horizontal stress, and the viscosity of injection fluid should be increased to over 10 cp. These measures effectively control fracture connectivity risks and avoid water breakthrough.
Storm surges are among the most destructive coastal hazards, and accurate prediction is vital for disaster preparedness. This paper reviews storm surge modeling/prediction methods and applications from early two-dimensional hydrodynamic models such as the Sea, Lake, and Overland Surges from Hurricanes (SLOSH) and the Advanced Circulation Model (ADCIRC) to fully coupled atmosphere-wave-ocean models, which represent key interactions among winds, waves, and circulation. To resolve dynamics in complex coastal settings, the fully coupled models substantially improve storm surge modeling/prediction accuracy. More recently, artificial intelligence (AI) has introduced a new generation of methods: machine learning and deep learning surrogates offer rapid emulation of physics-based models, and sequence models provide efficient time-series forecasting. The newly developed AI method of the physics-informed neural networks (PINN) that combine data-driven flexibility with hydrodynamic equations may not only provide fast predictions but also reveal more dynamical mechanisms. These developments highlight clear evolutions of storm surge modeling and prediction from simplified operational models toward hybrid physics-based AI frameworks, aimed at delivering faster, more accurate, and more reliable storm surge predictions.