The Middle Jurassic is considered a potential interval of climatic transition. Although most climate records of this period come from marine archives, continental climate variations also deserve equal attention. In this study, we use major and trace element geochemistry of fine-grained clastic rocks with grain-size end-member analysis to reconstruct chemical weathering intensity and climate changes from the early to late Middle Jurassic in the Yunyang section, located in the northeastern Sichuan Basin. Our results show a gradual decline in weathering intensity and a transition toward drier conditions from the early Middle Jurassic (Aalenian-Bajocian) to the late Middle Jurassic (Bathonian-early Callovian). This climate change aligns with shifts in terrestrial biota. These changes might have been caused by regional tectonic processes and changes in atmospheric circulation patterns. The findings provide a theoretical basis for understanding the paleoenvironment associated with a newly discovered dinosaur assemblage in the basin, following the well-known Zigong dinosaur fauna, and offer insights into terrestrial climate changes in the Sichuan Basin during the Middle Jurassic.

As an essential unconventional oil and gas resource, shale oil is of great significance to energy replacement and socio-economic development. Total organic carbon (TOC) and pyrolyzed hydrocarbon (S1), as key parameters for hydrocarbon reservoir evaluation, are important guides for practical exploration. Usually, the high precision determination of TOC and S1 requires sample collection and laboratory analysis, but is often compromised due to the cost and the limitation of coring continuity. With the advent of the digital age, increasingly intelligent methods are being employed in this field, such as Δlog R, support vector regression (SVR), and backpropagation neural network (BPNN). However, Δlog R has low performance, SVR does wrong in feature extraction, and BPNN is prone to local optimum. The coefficients of determination (R2) for TOC prediction using the three methods registered values of 0.25, 0.69, and 0.74. In contrast, the R2 values for S1 prediction were 0.23, 0.54, and 0.58. Thus, a low-cost, intelligent, and high-precision method to predict TOC and S1 is needed. This paper proposes a new model for predicting TOC and S1 in shale reservoirs based on an improved Deep learning network model (Encoder-ECA) based on the Transformer. With nearly 3 000 rock samples selected from the Jurassic Da’anzhai Member of the Sichuan Basin, China, and sedimentary facies variations in the study area, our results show that the Encoder-ECA model achieves an R2 of 0.86 for TOC content prediction and an R2 of 0.82 for S1 content prediction. In addition, the Encoder-ECA model was successfully applied to the recently implemented exploratory well evaluations in the study area, and the prediction results were used to optimize the sweet spot section, with a combined daily production of 22.4 mcf of oil and 33 600 mcf of gas. Simultaneously, data from different basins will be utilized to validate the applicability range of the model. This research demonstrates the great potential of deep learning technology in unconventional resource evaluation. It confirms the application of the Encoder-ECA model in the exploration practice of lacustrine facies shale oil and gas.

During the ice-free Cretaceous, the tropics have been proposed as key drivers of global climate, yet direct mid- to high-latitude evidence remains absent. Here, we present centennialresolution geochemical records of hydroclimatic variability from early Campanian (83.62- 83.14 Ma) midlatitude lacustrine successions. The records exhibit significant sub-precessional cycles at 10-12 k.y. and 5-5.4 k.y., and other millennial-scale cycles, primarily at -3.6 k.y., -2.6-2.5 k.y., and -2.3 k.y. High-precision chronology, phase locking of different proxies, and eccentricity-modulated amplitudes identify these sub-precessional signals as semi- and quarter-precession cycles. As these signals are most consistent with a tropical origin, their footprints in midlatitude hydroclimate records highlight the tropical control on climate during this greenhouse interval. Such low- to high-latitude teleconnection would provide efficient poleward heat and moisture transport, helping to reconcile the anomalously warm Cretaceous poles under a weak meridional temperature gradient. We further show that the -3.6-k.y. and -2.6-2.5 k.y. cycles arise from semi- and quarter-precession cycles via combination of tones and harmonics, whereas the -2.3 k.y. cycles are most likely linked to solar activity, underscoring their persistence throughout geological history.

The study of fractures constitutes a critical phase in oil and gas exploration and development, encompassing the analysis of their distribution, morphology, and density, which crucially influence subsequent exploration efforts. While core samples traditionally offer a direct reflection of well fracture characteristics, their high cost has led researchers to rely on Formation MicroImager (FMI) logging images increasingly. However, conventional methods of fracture analysis using FMI images typically require manual extraction and analysis by researchers, a process that is not only time-consuming and labor-intensive but also prone to subjective biases. To address these challenges, this study introduces a fracture segmentation model based on a two-stage decoupling strategy, named the TSNet (Two Stage Network). This model initially segments fractures accurately within the information-rich areas of the FMI image and subsequently connects disjointed fracture fragments through simulated generation in the blank bands, thereby effectively reconstructing the fracture's integrity and enhancing the continuity of segmentation. Furthermore, the TSNet model incorporates a deformable convolution operator that adaptively conforms to the varying morphologies of fractures, thus enhancing the segmentation of complex fracture structures and improving both the accuracy and efficiency of fracture analysis. Achieving a 72.41 % IoU in fracture segmentation accuracy with FMI image data from the Qaidam Basin, the TSNet model not only offers a novel approach to fracture segmentation but also provides researchers with more feasible options for conducting detailed fracture analyses.

Due to the great abundance of microfossils even in a small sample, they are ideal specimens for machine learning, which needs sufficient sample size. Taking the taxonomic controversy in a certain radiolarian lineage as a case study, a quantitative and objective approach and its advantage to fossil taxonomy is discussed in this study. Unsupervised machine learning algorithms are used to determine the species of radiolarians based on their morphological characteristics. K-Means, Agglomerative Clustering, and Meanshift are applied to build clustering models, with the centroid-based K-Means algorithm providing the most accurate classification results at a 92.26% accuracy. This method improves the efficiency of fossil identification and presents an accurate and objective method for assessing controversies associated with traditional methods. (c) 2025 Elsevier B.V. and Nanjing Institute of Geology and Palaeontology, CAS. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

Abstract The early‐middle Miocene was a critical period of climate transition, marked by significant carbon cycle perturbations and dynamic changes in the Antarctic Ice Sheet (AIS). However, relatively quantitative understandings of how pelagic sedimentary systems responded to and participated in these changes remain limited. To address this gap, mass accumulation rates (MARs) for a low‐latitude pelagic sequence recovered at International Ocean Discovery Program Site U1502 were calculated to evaluate sediment erosion, transport, and deposition. Enabled by a robust, high‐resolution age model established here, we quantified interactions between pelagic MARs and sea level through cross‐recurrence analyses. Integrating additional climate proxies such as atmospheric CO 2 and illite crystallinity, our results demonstrate that cryospheric evolution acted as a critical boundary condition in modulating the long‐term climatic response mode of low‐latitude pelagic sedimentation. During the Miocene Climatic Optimum (MCO) and Middle Miocene Climate Transition (MMCT), when the AIS underwent landward retreat, pelagic MARs responded positively to global changes. This means that higher pelagic MARs occurred during warmer and wetter periods, and vice versa. In contrast, during periods of significant AIS expansion before the MCO and after the MMCT, sea‐level fluctuations strongly modulated shelf accommodation, thereby buffering terrigenous inputs and leading to lower pelagic MARs despite warmer, wetter climate conditions. Furthermore, we found that pelagic sedimentation may mitigate positive benthic δ 13 C excursions in the middle Miocene, motivating future broader perspectives on this relationship.

Accurate fluid characterization is critical for reservoir development planning and typically relies on pressure-volume-temperature (PVT) experiments. However, in structurally complex reservoirs, fluid classification based solely on laboratory measurements can lead to misinterpretations. In the Gongshanmiao block of the Sichuan Basin, initial PVT analysis suggested that the reservoir was a condensate gas system. Subsequent field development revealed inconsistencies with this interpretation, including abnormal gas–oil ratios and atypical pressure build-up behavior that deviated from expected condensate gas reservoir performance. To resolve this discrepancy, this study proposes a diagnostic framework that integrates geoscience and engineering data, including fluid sampling, 3D structural modeling, production performance analysis, pressure build-up testing, and hydraulic fracturing data. The integrated analysis revised the initial PVT-based interpretation, and the results indicated that the reservoir is more accurately characterized as a saturated oil system with an overlying gas cap, rather than a condensate gas reservoir. Furthermore, the integrated interpretation clarifies the structural trapping mechanism and delineates the spatial extent of the gas cap. Overall, the proposed approach provides an integrated geoscience-engineering workflow for fluid reclassification in structurally complex reservoirs, which reconciles laboratory fluid analysis with field production behavior, offering a systematic framework for fluid interpretation in similar geological settings.

Microfossils play a crucial role in biostratigraphy and paleoenvironmental reconstruc-tions,as the first appearance datum(FAD)and last appearance datum(LAD)of specific microfos-sils enable precise stratigraphic correlations and age determinations.However,traditional identifica-tion methods are often time-intensive and heavily dependent on expert knowledge.To overcome these limitations,we propose a dual-path deep learning model,MicroViT,which integrates convolu-tional neural networks(CNNs)and vision transformers(ViTs)to automate the identification of Ceno-zoic ostracods(Microlimnocythere,Cyprideis,Qaidamocythere,Hemicyprinotus,Qaibeigouia,Aus-trocypris,and Candoniella)from the Qaidam Basin.MicroViT achieves an accuracy of 95.34％,demonstrating superior performance across all classification metrics.Furthermore,we utilized Gra-dient-weighted Class Activation Mapping(Grad-CAM)to visualize the decision-making process of the model,revealing that DL models focus on morphological features such as reticulation and hon-eycomb-like spots.We also investigated the potential for extending this approach to other microfos-sil groups,such as charophytes and sporopollen,as well as to diverse ostracod populations.These results highlight the significant potential of deep learning techniques for rapid and accurate micro-fossil classification,offering promising applications in micropaleontology and stratigraphic studies.

Around ~800-1200 ka, the transition of glacial-interglacial cycles from earlier ~40-kyr into later ~100-kyr cyclicities without obvious changes in orbital parameters, known as the Middle-Pleistocene Transition (MPT), suggests that Earth’s internal factors, in addition to external astronomical forcing, are also essential for the glacial cycles. However, it is still unclear how internal and external factors interact to lead to the MPT and the ~100-kyr cycle. Here, we statistically analyzed the power spectral relationship between the ~21-kyr, ~41-kyr, and ~100-kyr components within 57 paleoclimate archives and reconstructed the astronomical phase relative to the maximal changing rate of benthic foraminifer oxygen isotopes (δ18O) over the past 2700 ka to explore the role of astronomical forcings in driving glacial cycles and their relationship with internal factors. The statistical results show that the ~21-kyr power ratio complements the ~100-kyr power ratio. The precession phase covaries with pCO2-modulated glacial dynamics and exhibits a contrasting correlation with the precession power ratio of benthic δ18O before and after ~1500 ka. These findings suggest that pCO2-modulated latitudinal extension of the icesheets determined the glacial response to precession. Around 1500 ka, the response apparently shifted into a nonlinear mode, enabling the gradual extension of glacial cycles into ~100-kyr periodicities at the expense of precession power, which signified the onset of the ~100-kyr glacial cycles. Our study confirms the nonlinear precession origin of ~100-kyr glacial cycles, featuring the possible low- and high-latitude interplay at the precession band.

Millennial-scale climate cycles persist throughout greenhouse climates, yet the mechanisms remain unclear. Here, using proxy reconstructions, we present centennial-resolution geological records from early Campanian greenhouse deposits in both mid-latitude East Asia and low-latitude Southern Atlantic. These records document pronounced millennial ( ~ 1-6 kyr) wet-dry climate cycles. The amplitude modulation relationship between the most prominent ~4-5-kyr cycles (corresponding to the ¼ precession cycle) and eccentricity aligns perfectly with the theoretically calculated equatorial insolation cycles, demonstrating the climate effect of this predicted insolation forcing on global climate. Other millennial cycles primarily emerge from these ~4-5-kyr cycles via nonlinear amplitude modulation and combination tones. Proxy reconstructions and theoretical calculation thus converge to demonstrate that during this warm greenhouse period, precession can directly and indirectly stimulate millennial climate cycles. The deterministic link between astronomical parameters and millennial climate cycles implies that high-frequency climate oscillations may be predictable in future greenhouse-like climates, particularly under anthropogenic warming.

Earth's climate has been dominated by ~100-kyr glacial cycles over the past ~800 ka, yet the mechanism remains debated. Here, we present correlation analyses of spectral power ratios of global records spanning the past 2.7 Ma, revealing a persistent anticorrelation between ~21-kyr and ~100-kyr power ratios, but no significant relationship between ~41-kyr and ~100-kyr power ratios. This suggests that ~100-kyr climate cycles are more related to eccentricity-modulated precession than to obliquity. Phase analyses of benthic δ18O/ice volume and δ13C (carbon cycle) since Antarctic glaciation onset (~34 Ma) show that strong ~100-kyr cycles emerged only when these proxies were phase-coupled. Such coupling recurred at ~2.4-Myr eccentricity maxima during the unipolar regime (before 7.5 Ma) and minima during the bipolar regime (after 4 Ma), explaining the persistent ~21-kyr/~100-kyr anticorrelation because eccentricity modulates precession amplitude. We propose that internal carbon cycle dynamics and ~2.4-Myr eccentricity-modulated δ¹⁸O/ice volume-δ¹³C coupling amplified ~100-kyr climate cycles not only over the past ~800 ka but since 34 Ma. Given that eccentricity will remain low for the next 400 kyr, ~100-kyr periodicities may continue to dominate future climate variability, assuming Earth remains in a bipolar regime.

Subduction initiation, a cornerstone of plate tectonics, remains shrouded in complexity and mystery. This study reviews numerical models and natural occurrences of subduction initiation induced by mantle plumes on Earth and other terrestrial planets. We argue that mantle plumes, as non-plate tectonic agents, can initiate subduction in both ancient and contemporary settings without pre-existing lithospheric weaknesses. Our analysis of numerical simulations indicates that far-field stresses and the plume’s tail play pivotal roles in initiating subduction, while lithospheric weak zones and plate dynamics modulate the symmetry of subduction zones. The positioning of oceanic plateaus relative to the mantle plume significantly impacts subduction initiation. This paper also explores additional properties of mantle plumes crucial for subduction zone formation, fostering a more comprehensive understanding of plume-induced subduction initiation and its implications for global tectonics.

The current low-resolution chronostratigraphic framework for the early Ediacaran Period hampers a comprehensive understanding of potential trigger mechanisms for environmental upheavals and their connections to evolutionary innovation. Here, we establish a high-resolution astrochronological framework spanning ~57.6 million years of the early Ediacaran, anchored by the radioisotopic date of the Gaskiers glaciation onset, based on key sections from South China. Constrained by multiple radioisotopic dates, this framework precisely constrains the timing of the Marinoan deglaciation, Ediacaran Negative carbon isotope excursions 1 and 2 (EN1 and EN2), and key fossil assemblages (acanthomorphic acritarchs, Weng'an and Lantian biotas). These dates indicate the rapid termination of the Marinoan glaciation in South China within 106-107 years, while providing robust temporal evidence for the global synchroneity of EN1, EN2, and Marinoan deglaciation. The integrated chronology refines the age model for early Ediacaran biotic evolution, revealing that ecosystems gradually increased in complexity over multi-million-year timescales while global taxonomic diversity remained relatively stable, punctuated by rapid transitions to novel communities coinciding with biogeochemical perturbations.

Following sustained development, numerous palaeontology databases and datasets of various types have been created. However, the lack of a unified standard language to describe knowledge and unclear sharing mechanisms between different databases and datasets has limited the large-scale integration and application of paleontological data. The knowledge graph, as a key technology for semantic translation and data fusion, offers a possible solution to these challenges. Given the potential of knowledge graphs to overcome these obstacles, this paper presents a practical approach to express paleontological knowledge in a knowledge graph via the resource description framework language. By delving into the structured data associated with calcareous nannofossil biozones (the UC zone, CC zone and NC zone), we propose an ontology to describe the semantic units and logical relationships of paleontological biozones and species and then integrate relevant species records from unstructured research reports to construct a knowledge graph for calcareous nannofossils, that integrates multisource paleobiological data and knowledge reconstruction. Our focus lies in detailing the technical aspects of constructing a paleontological knowledge graph. The results demonstrate that knowledge graphs can integrate semistructured and unstructured paleontological data from various sources. This work aims to assist palaeontologists in building and utilizing knowledge graphs, serving as an initial effort for future paleontological knowledge reasoning.

Surface ruptures associated with large historical earthquakes provide essential insights into earthquake magnitudes and the kinematics of their seismogenic faults. In 1955, a major earthquake struck Zheduotang Village, Kangding City, producing surface ruptures along the Zheduotang fault, a segment of the southern Xianshuihe fault zone. The magnitude of this event remains a topic of debate, with estimates ranging from M6.6 to M7.5, primarily due to discrepancies in the interpretation of its associated surface ruptures. This study synthesizes previous research on the surface ruptures of the 1955 Zheduotang earthquake and presents new field data, including high-resolution unmanned aerial vehicle (UAV)-based topographic surveys, trenching, and lichenometry in the epicentral region. Analysis of surface rupture freshness, faulting event chronology from trenching, and lichen size measurements collectively support a similar to 55 km-long surface rupture zone, corresponding to a moment magnitude (M-w) of similar to 7.1 for the 1955 earthquake. Offset glacial landform analysis reveals a late Quaternary left-lateral slip rate of similar to 2.5-3.0 mm/yr in the southern segment of the Zheduotang fault, which is lower than the previously documented similar to 4 mm/yr in the northern segment. Deformed landforms and surface ruptures indicate that the fault trends NNW and exhibits predominantly left-lateral strike-slip motion in its northern segment, while the southern segment trends NW and includes a notable normal faulting component. Our findings suggest that the Zheduotang fault defines the southwestern boundary of the Bamei-Kangding releasing stepover zone within the southern Xianshuihe left-lateral strike-slip fault zone. These results enhance the understanding of seismic hazards and the tectonic kinematics along the eastern boundary of the Tibetan Plateau.

During the Toarcian, Early Jurassic, the Earth experienced a global carbon cycle perturbation associated with global warming and widespread oceanic anoxia, known as the Toarcian Oceanic Anoxic Event (T-OAE). Carbon isotope data from oceans and land around the world record a clear negative excursion (ranging from -2 %o to 8.5 %o). However, the factors controlling the differences in the magnitude of this negative carbon isotope excursion (CIE) are still unclear, and there are many different explanations for the different durations of the Toarcian negative CIE. This study compiled nine high-resolution marine boreholes and sections data from Western Tethys Ocean and two continental drill hole data from the Eastern Tethys region. Through cyclostratigraphic analysis, it was found that the duration of the negative CIE was -350-1800 kyr. In addition, by combining power decomposition analysis (PDA) with machine learning methods, it is believed that long eccentricity and precession affect the changes in carbon isotope characteristics during T-OAE, but there may be a threshold effect in the driving process. When the long eccentricity/precession is below the threshold, or even within a certain range, the stronger the effect of the orbital cycle, the longer duration of the negative CIE, and the greater the magnitude of the excursion. This study enriches the terrestrial records in T-OAE research and deeply explores the impact of different factors on the Early Jurassic oceanic anoxic event, which is of great significance for understanding the impact mechanism of oceanic anoxic events.

Neogene relative sea-level fluctuations in the Pearl River Mouth Basin are reconstructed through cyclostratigraphic analysis of the shelf-margin PY33-1-1 well, strategically selected based on seismic profile analysis confirming equilibrium between sedimentation and subsidence rates. This equilibrium state, evidenced by parallel strata geometries and uniform thickness trends in seismic reflectors, ensures minimal tectonic overprinting and optimal preservation of eustatic signals. Time-series analysis of gamma-ray (GR) logs reveals pervasive Milankovitch cycles throughout a similar to 23 Myr succession. An astronomically tuned age-depth model is established by aligning 405-kyr-filtered GR cycles to the global delta O-18 stack, anchored at 23.03 Ma near the Oligocene-Miocene boundary. Evaluation of different anchor point choices (22.90-23.11 Ma) confirms that the tuning framework is robust within this debated range, with minimal impact on stratigraphic interpretation.Dynamic noise after orbital tuning (DYNOT) applied to the tuned GR series identifies similar to 1.2-Myr sea-level fluctuations, reconstructing a relative sea-level change (GSLC) curve. This curve exhibits strong correlations (r = 0.68-0.82) with global delta O-18 records (5.98-13.3 Ma) and aligns well with composite sea-level proxies from 1.82-15.8 Ma. However, signal fidelity weakens notably between 15.8-23.03 Ma, especially within the high-energy tidal-flat facies of the Zhujiang Formation, due to lithological noise and sedimentary overprint. Comparative DYNOT analysis from a delta-front setting supports the strong facies-dependence of orbital signal preservation in marginal marine systems.Phase analysis reveals coherent 1.2-Myr obliquity modulation from 1.82-17.18 Ma, while significant phase deviations before 17.18 Ma reflect depositional regime reorganization. These changes are interpreted as an environmental transition from tidal flat to shallow marine settings, coupled with increased global ice volume following the Middle Miocene Climate Transition, which improved orbital signal fidelity and modulated the GR-climate linkage.This study demonstrates that shelf-margin successions characterized by sedimentation-subsidence equilibrium and depositional stability can effectively archive astronomically paced sea-level changes. The refined astronomical framework and reconstructed GSLC curve capture long-term eustatic trends coupled with orbital climate cycles, validating integrated cyclostratigraphy and noise modeling as powerful tools for basin-scale sea-level reconstructions.

Over the past six decades, the scientific ocean drilling (SOD) programs have collected vast and invaluable data for Earth history research. However, the scattered state of these data across multiple repositories, along with inconsistent standards, methodologies, and terminologies, have increased the complexity of data processing and posed barriers to its effective utilization. While several databases have been developed to consolidate and improve access to SOD data, each has limitations in scope. To address these challenges, we introduce the Sediment Spatial and Temporal Database (SedST, http://sedst.org), a new platform designed to aggregate chronological data from the extensive archives of SOD, including postcruise literature, and provide tools for data accessibility. SedST standardizes biostratigraphic, magnetostratigraphic, and radiometric dating records, aligning them with the latest Geologic Time Scale 2020, to ensure consistency and coherence. Employing the Bchron methodology, SedST establishes age-depth models for >1000 SOD boreholes. Presently, SedST encompasses 37,329 entries from 181 expeditions and 1300 holes, covering the world’s ocean basins and including records as old as the Late Jurassic. The platform provides powerful tools for sample ID to depth conversions and individual hole to composite site depth transformation. Accessible via user-friendly graphical interfaces, SedST simplifies data queries and allows the export of search results in CSV format. As a constantly developing platform under the umbrella of the Deep-time Digital Earth program, SedST is committed to enhancing the accessibility and discoverability of marine sediment data, fostering new insights into Earth’s geological history.

Marine nitrogen cycle perturbations during the Toarcian oceanic anoxic event (T-OAE, ca. 183 Ma) are recorded by the bulk nitrogen isotope compositions (δ15Nbulk) of sediments, which emphasize the impact of seawater redox changes on the biogeochemical nitrogen cycle. However, lacustrine nitrogen cycling during the T-OAE is unexplored. Here, we report the first δ15Nbulk data from the Anya section in the Ordos Basin (China) and establish a model of the lacustrine nitrogen cycle during the T-OAE. δ15Nbulk values covary with other redox proxies (Corg/P and FeHR/FeT ratios), and indicate varying phases of deposition linked to redox conditions. In particular, a positive shift in δ15Nbulk (ca. +3.1‰ to +7.4‰) prior to the T-OAE in black shale facies indicates nitrogen loss (via partial denitrification and/or anammox) under anoxic bottom water conditions. However, lower δ15Nbulk (mean = 3.1‰ ± 0.5‰) in the T-OAE itself likely indicates the predominance of nitrogen assimilation (nitrification and nitrate assimilation) and reduced nitrogen loss in oxygenated waters. The cause of redoxcline collapse may be linked to the destruction of a stable chemocline caused by enhanced hydrological cycling during T-OAE warming. The nitrogen cycle of the paleo-Ordos lake may have behaved differently to the nitrogen cycle in the ocean during the T-OAE. Our work illustrates that even during marked global warming, the regional nitrogen cycle is mainly controlled by localized depositional conditions. Our results suggest that redoxcline collapse played a significant role in biogeochemical nitrogen cycling in lacustrine systems, contributing to our understanding of the terrestrial response to the T-OAE.