Search Results (73)

Side-scan sonar (SSS) serves as the primary perceptual instrument for Autonomous Underwater Vehicles (AUVs) in large-scale marine search and rescue (SAR) operations. However, the detection of critical targets is frequently hindered by severe hydro-acoustic noise, the spatial discontinuity of wreckage, and the weak visual signatures of small targets. To surmount these challenges, this paper presents WPG-DetNet. First, we introduce a Wavelet-Embedded Residual Backbone (WERB) to reconstruct the conventional downsampling paradigm. By substituting standard pooling with the Discrete Wavelet Transform (DWT), this architecture explicitly disentangles high-frequency noise from structural information in the frequency domain, thereby achieving the adaptive preservation of edge fidelity for large human-made targets while filtering out speckle interference. Then, addressing the distinct challenge of discontinuous aircraft wreckage, the framework further incorporates a Debris Graph Reasoning Module (D-GRM). This module models scattered fragments as nodes in a topological graph to capture long-range semantic dependencies, transforming isolated instance recognition into context-aware scene understanding. Finally, to bridge the gap between AI and underwater physics, we design a Shadow-Aided Decoupling Head (SADH) equipped with a physics-informed geometric loss. By enforcing mathematical consistency between target height and acoustic shadow length, this mechanism establishes a rigorous discriminative criterion capable of distinguishing weak-echo human bodies from seabed rocks based on shadow geometry. Experiments on the SCTD dataset demonstrate that WPG-DetNet achieves a mean Average Precision ($mA{P}_{50}$) of 97.5% and a Recall of 96.9%. Quantitative analysis reveals that our framework outperforms the classic Faster R-CNN by a margin of 12.8% in $mA{P}_{50}$ and surpasses the Transformer-based RT-DETR-R18 by 5.6% in high-precision localization metrics ($mA{P}_{50:95}$). Simultaneously, WPG-DetNet maintains superior efficiency with an inference speed of 62.5 FPS and a lightweight parameter count of 16.8 M, striking an optimal balance between robust perception and the real-time constraints of AUV operations. Full article

This work presents a novel deep learning workflow for estimating porosity and permeability from combined data, where numerical variables such as high-resolution bulk density (RHOB) and photoelectric factor (PEF) data are integrated with X-ray computed tomography (X-CT) image data, using a dual-energy X-CT approach (DECT). Convolutional neural networks (CNNs) were calibrated with routine core analysis (RCAL) laboratory measurements from one well from Sinú-San Jacinto Basin (Colombia). The CNN architecture combines two main branches: An image branch, in which a CNN extracts spatial features from normalized X-CT sections using 3 × 3 convolution layers, ReLU activation, batch normalization, and maxPooling, and a numerical branch, which processes the input vectors corresponding to RHOB and PEF using fully connected dense layers and dropout regularization. Both branches are concatenated in a fusion layer, from which the model’s final predictions are made. Results indicate a strong correlation between porosity, permeability, RHOB and PEF logs, and CT images. The porosity model achieved excellent predictive performance, with an R² = 0.996, MAE = 3.96 × 10⁻³, MSE = 3.82 × 10⁻⁵, and 0.064 maximum error. The permeability model also performed well, with a linear R² = 0.983, though metrics reflected the wide dynamic range of permeability. Consequently, artificial neural networks (ANNs) can accurately predict porosity and permeability at various depths where no corresponding laboratory data exists, demonstrating excellent predictive capabilities over several rock intervals, in a high vertical resolution because of X-CT data scale (0.625 mm). Full article

Coal and rock recognition is one of the key technologies in mining production, but traditional methods have limitations such as single-feature representation dimension, insufficient robustness, and unbalanced performance in lightweight design under noise interference and complex feature conditions. To address these issues, an Efficient Channel Attention Reparameterized Network (ECA-RepNet) based on recurrence plot and Efficient Channel Attention mechanism is proposed. The one-dimensional vibration signal is mapped to the two-dimensional image space through a recurrence plot (RP), which retains the dynamic characteristics of the time series while capturing the complex patterns in the signal. Multi-scale feature extraction and lightweight design are achieved through the reparameterized large kernel block (RepLK Block) and the depthwise separable convolution (DSConv) module. The ECA module is introduced to embed multiple convolutional layers. Through global average pooling, one-dimensional convolution, and dynamic weight allocation, the modeling ability of inter-channel dependencies is enhanced, the model robustness is improved, and the computational overhead is reduced. Experimental results demonstrate that the ECA-RepNet model achieves 97.33% accuracy, outperforming classic models including ResNet, CNN, and MobileNet in parameter efficiency, training time, and inference speed. Full article

Enhanced silicate rock weathering (ESRW) has been proposed as a promising carbon dioxide removal strategy, yet its carbon sequestration pathways, durability, and ecosystem dependence remain incompletely understood. Here, we synthesize evidence from field experiments, observational studies, and modeling to compare ESRW-induced carbon dynamics across forest and cropland ecosystems using a unified SOC–SIC dual-pool framework. Across both systems, ESRW operates through shared geochemical processes, including proton consumption during silicate dissolution and base cation release, which promote atmospheric CO₂ uptake. However, carbon fate diverges markedly among ecosystems. Forest systems, characterized by high biomass production, deep rooting, and strong hydrological connectivity, primarily favor biologically mediated pathways, enhancing net primary productivity and mineral-associated organic carbon (MAOC) formation, while facilitating downstream export of dissolved inorganic carbon (DIC). In contrast, intensively managed croplands more readily accumulate measurable soil inorganic carbon (SIC) and soil DIC over short to medium timescales, particularly under evapotranspiration-dominated or calcium-rich conditions, although SOC responses are often moderate and variable. Importantly, only a subset of ESRW-driven pathways—such as MAOC formation and secondary carbonate precipitation—represent durable carbon storage on decadal to centennial timescales. By explicitly distinguishing carbon storage from carbon transport, this synthesis clarifies the conditions under which ESRW can contribute to climate change mitigation and highlights the need for ecosystem-specific deployment and monitoring strategies. Full article

To address the high-dilution issues arising from efficient mining methods such as medium-deep drilling for underground thin veins of precious metals, detecting raw rock fragments after blasting for subsequent sorting has become a cutting-edge research focus. With the continuous advancement of artificial intelligence, deep learning offers novel applications for rock detection. Accordingly, this study employs an improved lightweight YOLOv8n model to detect two typical thin-vein precious metal ores: gold ore and wolframite. In consideration of the computational resource constraints in underground environments, a triple optimization strategy is proposed. First, GhostConv and C2f-Ghost modules were introduced into the backbone network to reduce redundant computations while preserving feature representation capabilities. Second, the VoVGSCSP module was incorporated into the neck to further decrease model parameters and computational load. Finally, the ECA mechanism was embedded before the SPPF pooling layer to enhance feature extraction for ores and rocks, thereby improving detection accuracy. The results demonstrate that the GVE-YOLOv8 model contains only 2.28 million parameters—a 24.3% reduction compared to the original YOLOv8n. FLOPs decrease from 8.1 G to 5.6 G, and the model size reduces from 6.3 MB to 4.9 MB, while detection accuracy improves to 98.3% mAP50 and 95.3% mAP50-95. This enhanced model meets the performance requirements for accurately detecting raw ore and rock fragments after underground blasting, thereby providing a novel research method for thin-vein mining. Full article

Karst regions cover about 12% of Earth’s land surface and exhibit high uncertainty in soil organic carbon (SOC) pools due to strong spatial heterogeneity. This study quantifies the association between rock fissure network morphology and SOC pools across three karst rocky habitat types in the Maolan National Nature Reserve (Guizhou, China): Type I (predominantly sub-horizontal and weakly connected fissures), Type II (oblique and moderately connected fissures), and Type III (predominantly subvertical and highly connected fissures). Fissure network morphology was characterized using quantitative network morphology metrics, and SOC pools (content, density, and stock) were measured from field samples (with long-term sequestration estimated). Type I habitats showed the highest SOC content, density, stock, and sequestration estimates, whereas Type III habitats consistently showed the lowest values. Across habitats, SOC density and stock were negatively associated with metrics reflecting steeper fissure orientation, greater spatial heterogeneity, and higher network connectivity, while SOC content was positively associated with fissure network complexity. These findings highlight fissure network morphology as an important structural dimension for explaining SOC variability in karst rocky habitats and suggest incorporating fissure information into SOC assessment and habitat-specific soil and vegetation management in karst landscapes. Full article

Background and Clinical Significance: Simultaneous, multiple infections coinfections caused by zoonotic or fecal-orally transmitted diseases are common in tropical and subtropical regions. Published data report that leptospirosis may coexist with other infections, complicating the clinical presentation and trajectory due to overlapping symptoms and leading to more severe clinical progress. Case Presentation: We describe a clinical case of a 34-year-old female diagnosed with a triple infection caused by Leptospira, Hepatitis A Virus, and Hepatitis E Virus. To our knowledge, this is the first case described in the literature in a non-endemic area without travel history to tropical or subtropical regions. The patient presented with one-week history of influenced clinical status, myalgia, headaches, nausea, high fever, bloody diarrheas, and abdominal pain. During the last two days, she also developed jaundice. Swimming in the rock pools of the island where she lives was indicated as the source of the infection. The laboratory tests revealed increased values of inflammatory markers, thrombocytopenia, and severe abnormalities of liver function. Serological and PCR tests for a wide range of pathogens proved an acute infection caused by Leptospira interogans, Hepatitis A virus, and Hepatitis E Virus. The patient received intravenous fluids and antibiotic treatment with ceftriaxone for seven days leading to gradual clinical improvement and normalization of liver function tests with subsequent reduction in jaundice within 30 days. Conclusions: This case report suggests that clinical suspicion and laboratory investigation should include the probability of coinfections even in non-endemic areas based on medical history of the patients. An early diagnosis of a zoonotic disease and other infective agents of acute hepatitis are vital in order to choose the appropriate treatment option and avoid severe complications. Full article

Huhehu Sag is a sag with high exploration degree in Hailar Basin. With large sedimentary thickness, complete stratigraphic development and excellent oil generation conditions, it is the main oil- and gas-producing sag in Hailar Basin. The primary source rocks are the Nantun Formation, with the Tongbomiao and Damoguaihe Formations as secondary sources. Hydrocarbon accumulation periods in the sag were comprehensively analyzed using methodologies including source rock hydrocarbon generation-expulsion history, authigenic illite dating of reservoirs, and fluid inclusion homogenization temperature analysis. Results reveal two major accumulation stages: Stage 1 (125–90 Ma), corresponding to the depositional period of the Yimin Formation, represented the peak paleo-geothermal regime and the primary hydrocarbon accumulation phase. Intensive hydrocarbon generation and expulsion, coupled with robust migration dynamics, facilitated large-scale oil and gas pooling. Stage 2(65 Ma-now), from the deposition of Qingyuangang Formation to the present, uplift and denudation reduce the burial depth of source rocks, the hydrocarbon generation intensity is weakened. This phase involved secondary adjustments of pre-existing reservoirs and continued charging of newly generated hydrocarbons. The Huhehu Sag is a typical half-graben structure. Fault-block and fault-lithologic reservoirs dominate, distributed zonally along gentle and steep slopes. Lithologic reservoirs primarily occur near or within the central hydrocarbon-generating sub-sags. The most favorable hydrocarbon accumulation zones are located in the sub-sag centers and adjacent areas with high-quality reservoirs. Full article

The precise mapping of rock joint traces is fundamental to the design and safety assessment of foundations, retaining structures, and underground cavities in building and civil engineering. Existing deep learning approaches either impose prohibitive computational demands for on-site deployment or disrupt the topological continuity of subpixel lineaments that govern rock mass behavior. This study presents BATNet-Lite, a lightweight encoder–decoder architecture optimized for joint segmentation on resource-constrained devices. The encoder introduces a Boundary-Aware Token-Mixing (BATM) block that separates feature maps into patch tokens and directionally pooled stripe tokens, and a bidirectional attention mechanism subsequently transfers global context to local descriptors while refining stripe features, thereby capturing long-range connectivity with negligible overhead. A complementary Multi-Scale Line Enhancement (MLE) module combines depth-wise dilated and deformable convolutions to yield scale-invariant responses to joints of varying apertures. In the decoder, a Skeletal-Contrastive Decoder (SCD) employs dual heads to predict segmentation and skeleton maps simultaneously, while an InfoNCE-based contrastive loss enforces their topological consistency without requiring explicit skeleton labels. Training leverages a composite focal Tversky and edge IoU loss under a curriculum-thinning schedule, improving edge adherence and continuity. Ablation experiments confirm that BATM, MLE, and SCD each contribute substantial gains in boundary accuracy and connectivity preservation. By delivering topology-preserving joint maps with small parameters, BATNet-Lite facilitates rapid geological data acquisition for tunnel face mapping, slope inspection, and subsurface digital twin development, thereby supporting safer and more efficient building and underground engineering practice. Full article

As a globally renowned alpine gorge region and seismically active zone, the eastern margin of the Qinghai–Tibet Plateau (QTP) is highly prone to landslide dam formation. Considering unstable landslide dams often pose catastrophic risks to downstream areas, current research on landslide dams along QTP primarily focuses on the breach mechanisms of unstable dams, while studies on the formation mechanisms of stable landslide dams—which can provide multiple benefits to downstream regions—remain limited. This paper selected the Conaxue Co landslide dam on the eastern margin of the QTP as one case example. Field investigation, sampling, numerical simulation, and comprehensive analysis were carried out to disclose its formation mechanisms. Field investigation shows that the Conaxue Co landslide dam was formed by a high-speed long-runout landslide blocking the river, with its structure exhibiting a typical inverse grading pattern characterized by coarse-grained rock overlying fine-grained layers. The inverse grading structure plays a critical role in the stability of the Conaxue Co landslide dam. On one hand, the coarse, hard rock boulders in the upper dam mitigate fluvial erosion of the lower fine-grained sediments. On the other hand, the fine-grained layer in the lower dam acts as a relatively impermeable aquitard, preventing seepage of dammed lake water. Additionally, the step-pool system formed in the spillway of the Conaxue Co landslide dam contributes to the protection of the dam structure by dissipating 68% of the river’s energy (energy dissipation rate η = 0.68). Understanding the formation mechanisms of the Conaxue Co landslide dam can provide critical insights into managing future landslide dams that may form in the QTP, both in emergency response and long-term strategies. Full article

Discovering, analyzing, and finding a key to understanding the physiological and biochemical responses that Vitis vinifera L. undertakes against drought stress is of fundamental importance for this profitable crop. Today’s considerable climatic fluctuations force researchers and farmers to focus on this issue with solutions inclined to respect the ecosystem. In this academic work, we focused on describing the drought stress consequences on several parameters of secondary metabolites on Vitis vinifera leaves (quercetins, kaempferol, resveratrol, proline, and xanthophylls) and on some ecophysiological characteristics (e.g., water potential, stomatal conductance, and leaf temperature) to compare the answers that diverse agronomic management techniques (i.e., irrigation with and without zeolite, pure zeolite and no application) could instaurate in the metabolic pathway of this important crop with the aim to find convincing and thought-provoking responses to use this captivating and versatile mineral, the zeolite known as the “magic rock”. Stressed grapevines reached 56.80 mmol/m²s gs at veraison and a more negative stem Ψ (+10.63%) compared to plants with zeolite. Resveratrol, in the hottest season, fluctuated from 0.18–0.19 mg/g in zeolite treatments to 0.37 mg/g in stressed vines. Quercetins were inclined to accumulate in response to drought stress too. In fact, we recorded a peak of quercetin (3-O-glucoside + 3-O-glucuronide) of 11.20 mg/g at veraison in stressed plants. It is interesting to note how the pool of metabolites was often unchanged for plants treated with zeolite and for plants treated with water only, thus elevating this mineral to a “stress reliever”. Full article

Morphological problems for distinguishing between glacier ice, glacier ice with a debris cover (debris-covered glaciers), and rock glaciers are outlined with respect to recognising and mapping these features. Decimal latitude–longitude [dLL] values are used for geolocation. One model for rock glacier formation and flow discusses the idea that they consist of ‘mountain permafrost’. However, signs of permafrost-derived ice, such as flow features, have not been identified in these landsystems; talus slopes in the neighbourhoods of glaciers and rock glaciers. An alternative view, whereby rock glaciers are derived from glacier ice rather than permafrost, is demonstrated with examples from various locations in the mountain domain, 𝔻𝕞. A Google Earth and field examination of many rock glaciers shows glacier ice exposed below a rock debris mantle. Ice exposure sites provide ground truth for observations and interpretations stating that rock glaciers are indeed formed from glacier ice. Exposure sites include bare ice at the headwalls of cirques and above debris-covered glaciers; additionally, ice cliffs on the sides of meltwater pools are visible at various locations along the lengths of rock glaciers. Inspection using Google Earth shows that these pools can be traced downslope and their sizes can be monitored between images. Meltwater pools occur in rock glaciers that have been previously identified in inventories as being indictive of permafrost in the mountain domain. Glaciers with a thick rock debris cover exhibit ‘hidden ice’ and are shown to be geomorphological units mapped as rock glaciers. Full article

The Dead Sea is one of the most saline terminal lakes on Earth, and few organisms survive in this harsh environment. In some onshore spring pools, active and diverse microbial communities flourish. In the geological past, microbial-rich environments left their marks in the form of stromatolites. Stromatolites are studied to better understand the appearance of life on Earth and potentially on other planets. Hyperspectral methodologies have been shown to be useful for detecting structures in stromatolites. In an effort to characterize the biosignatures and chemical composition inherent to stromatolites, we created a spectral classification scheme for distinguishing between stromatolites and their bedrock environment—typically carbonatic rocks, mostly dolomites. The overarching aim comprises the development of an automated hyperspectral reflectance method for detecting the presence of stromatolites. We collected and measured 82 field samples with an ASD spectrometer and used our spectral dataset to train three machine learning algorithms (linear regression, K-Nearest Neighbor, XGBoost). The results show the successful detection of stromatolites, with all three prediction methods giving high accuracy rates (stromatolite > 0.9, bedrock dolomite > 0.8). The continuum removal and spectral ratio technique results identified two significant spectral regions, ~1900 nm (water) and ~2310–2320 nm (carbonates), that allow one to differentiate between stromatolites and dolomites. This study establishes the grounds for the automated detection of a fossilized livable environment in a carbonatic terrain based on its hyperspectral reflectance data. The results have significant implications for future mapping efforts and emphasize the feasibility of automated mapping, extending the data acquisition to airborne or satellite-based hyperspectral remote sensing technologies to detect life forms in extreme environments. Full article

With global climate change and the deterioration of the ecological environment, the safety of hydraulic engineering faces severe challenges, among which soil-dwelling termite damage has become an issue that cannot be ignored. Reservoirs and embankments in China, primarily composed of earth and rocks, are often affected by soil-dwelling termites, such as Odontotermes formosanus and Macrotermes barneyi. Identifying soil-dwelling termite damage is crucial for implementing monitoring, early warning, and control strategies. This study developed an improved YOLOv8 model, named MCD-YOLOv8, for identifying traces of soil-dwelling termite activity, based on the Monte Carlo random sampling algorithm and a lightweight module. The Monte Carlo attention (MCA) module was introduced in the backbone part to generate attention maps through random sampling pooling operations, addressing cross-scale issues and improving the recognition accuracy of small targets. A lightweight module, known as dimension-aware selective integration (DASI), was added in the neck part to reduce computation time and memory consumption, enhancing detection accuracy and speed. The model was verified using a dataset of 2096 images from the termite damage survey in hydraulic engineering within Hubei Province in 2024, along with images captured by drone. The results showed that the improved YOLOv8 model outperformed four traditional or enhanced models in terms of precision and mean average precision for detecting soil-dwelling termite damage, while also exhibiting fewer parameters, reduced redundancy in detection boxes, and improved accuracy in detecting small targets. Specifically, the MCD-YOLOv8 model achieved increases in precision and mean average precision of 6.4% and 2.4%, respectively, compared to the YOLOv8 model, while simultaneously reducing the number of parameters by 105,320. The developed model is suitable for the intelligent identification of termite damage in complex environments, thereby enhancing the intelligent monitoring of termite activity and providing strong technical support for the development of termite control technologies. Full article

Furnas volcano, one of the three active central volcanoes of São Miguel (the Azores archipelago), hosts mineral waters with significant special variations, divided into hyperthermal (89.4–95.4 °C), thermal (29.9–70.0 °C), and cold (14.2–21.4 °C) waters. Groundwaters are classified as Na-HCO₃, with a neutral to slightly acidic pH, except one SO₄-Na acidic sample. The major elements are primarily influenced by rock leaching and volcanic input, patterns also reflected in the trace elements, including the rare earth elements. The major cations, along with lithium, iron, aluminum, rubidium, and strontium, indicate the influence of water–rock interactions. Some samples depict a higher influence in this input, shown by the similar REE behavior between them and the local rock behavior. The volcanic input is distinguished into two environments: an acid sulfate boiling pool, formed by steam heating, and neutral HCO₃-Cl waters, where bicarbonate-rich waters mix with a neutral chloride fluid from a deep reservoir. The deeper reservoir also provides boron, arsenic, antimony, and tungsten, also seemingly associated with a positive spike in europium due to rock dissolution at temperatures above 250 °C or a reducing environment. This interpretation is corroborated by the stability of the strontium isotopes between samples. Full article