Deep Flow Research Articles

Research on a Passenger Flow Prediction Model Based on BWO-TCLS-Self-Attention

In recent years, with the rapid development of the global demand and scale for deep underground space utilization, deep space has gradually transitioned from single-purpose uses such as underground transportation, civil defense, and commerce to a comprehensive, livable, and disaster-resistant underground ecosystem. This shift has brought increasing attention to the safety of personnel flow in deep spaces. In addressing challenges in deep space passenger flow prediction, such as irregular flow patterns, surges in extreme conditions, large data dimensions, and redundant features complicating the model, this paper proposes a deep space passenger flow prediction model that integrates a Temporal Convolutional Network (TCN) and Long Short-Term Memory (LSTM) network. The model first employs a dual-layer LSTM network structure with a Dropout layer to capture complex temporal dynamics while preventing overfitting. Then, a Self-Attention mechanism and TCN network are introduced to reduce redundant feature data and enhance the model’s performance and speed. Finally, the Beluga Whale Optimization (BWO) algorithm is used to optimize hyperparameters, further improving the prediction accuracy of the network. Experimental results demonstrate that the BWO-TCLS-Self-Attention model proposed in this paper achieves an R2 value of 96.94%, with MAE and RMSE values of 118.464 and 218.118, respectively. Compared with some mainstream prediction models, the R2 value has increased, while both MAE and RMSE values have decreased, indicating its ability to accurately predict passenger flow in deep underground spaces.

Stabilizing training of affine coupling layers for high-dimensional variational inference

Abstract Variational inference with normalizing flows is an increasingly popular alternative to MCMC methods. In particular, normalizing flows based on affine coupling layers (Real NVPs) are frequently used due to their good empirical performance. In theory, increasing the depth of normalizing flows should lead to more accurate posterior approximations. However, in practice, training deep normalizing flows for approximating high-dimensional posterior distributions is often infeasible due to the high variance of the stochastic gradients. 
In this work, we show that previous methods for stabilizing the variance of stochastic gradient descent can be insufficient to achieve stable training of Real NVPs. As the source of the problem, we identify that, during training, samples often exhibit unusual high values. 
As a remedy, we propose a combination of two methods: (1) soft-thresholding of the scale in Real NVPs, and (2) a bijective soft log transformation of the samples. We evaluate these and other previously proposed modification on several challenging target distributions, including a high-dimensional horseshoe logistic regression model. Our experiments show that with our modifications, stable training of Real NVPs for posteriors with several thousand dimensions and heavy tails is possible, allowing for more accurate marginal likelihood estimation via importance sampling. 
Moreover, we evaluate several common training techniques and architecture choices and provide practical advise for training Real NVPs for high-dimensional variational inference. Finally, we also provide new empirical and theoretical justification that optimizing the evidence lower bound of normalizing flows leads to good posterior distribution coverage.

Effect of the Variable Kuroshio Intrusion Pathway on Deep Flow in the Northern South China Sea

Abstract The effect of the variable Kuroshio intrusion pathway on deep flow in the northern South China Sea (NSCS) is investigated using observational and theoretical analyses. With significant seasonality in path variation, the leaking path is found to be most frequent, whereas the looping (leaping) path is transient and prefers to occur in winter (summer). Such multipath variability contributes greatly to the highest surface eddy kinetic energy in the NSCS, with the energy peaking for the looping path in winter, followed by the leaking and leaping paths. The information flow–based causality analysis suggests that the energetic surface perturbations in the Kuroshio intrusion region are causal to the intraseasonal variability in deep flow east of the Dongsha Islands, wherein the linkage is strongest within the Kuroshio Loop Current eddy shedding event related to the looping path. Therefore, the underlying dynamics for the vertical coupling is illustrated by this event, in which the eddy is highly compressed to the surface because of sloping topography and its energy is partially transferred downward through vertical pressure work to force the abyssal topographic Rossby waves. Moreover, the energetics and instability analyses reveal that barotropic instability dominates the energy transfer from background flow to eddies in the Kuroshio intrusion region, which energizes the surface perturbations most intensively for the looping path, followed by the leaking and leaping paths. The characteristic scales of the most unstable mode estimated by the reduced-gravity model are analogous to the observations and account for the rapid growth of surface perturbations in the Kuroshio intrusion region.

MORPHODYNAMIC CHARACTERISTICS OF A SANDY BED MEANDERING CHANNEL AT DIFFERENT DISCHARGES

Transportation of sediment is a critical issue in riverine environments, particularly in meandering channels where flow dynamics are complex. However, information regarding erosion and deposition processes in meandering channels is still limited. Therefore, an experimental investigation on the influence of discharge variation on the morphodynamics of a meandering channel has been carried out. The experimental investigation was conducted at the Hydraulic and Hydrology Laboratory, UTM Johor Bahru. The study investigates the flow profiles and the morphological changes of inbank flow conditions represented by shallow and deep flow depths. The findings revealed that Manning’s n in a deep flow depth was 60.12% higher than a shallow flow depth. Moreover, velocity at the channel bend was increased by 8.3% to 14.6% compared to the crossover along the channel. This indicates that the geometrical planform plays a crucial part on the variability of velocity in the channel. Therefore, the outcomes of this study could provide more effective river management strategies to resolve erosion and deposition issues in river engineering studies.

Advancements in Hydroponic Systems: A Comprehensive Review

Hydroponics, the practice of growing plants without soil using mineral nutrient solutions in water, represents a significant shift in agricultural methodologies. Modern hydroponic systems have evolved substantially since the early 20th century, with advancements in system design, automation, nutrient management, and environmental control enhancing efficiency, scalability, and applicability. This review examines recent innovations that have addressed traditional challenges and created new opportunities for hydroponic cultivation. Hydroponics offers advantages over traditional soil-based agriculture, including precise nutrient control, reduced water usage, and the elimination of soil-borne diseases and pests. These benefits are critical in addressing modern agricultural challenges such as soil degradation, water scarcity, and the need for increased food production for a growing global population. Hydroponic systems can be implemented in diverse environments, making them adaptable solutions for enhancing food security and sustainability. Key developments in various hydroponic systems are highlighted, including the Nutrient Film Technique (NFT), Deep Water Culture (DWC), Ebb and Flow, Aeroponics, Wick Systems, and Drip Systems. Innovations in these systems focus on optimizing nutrient delivery, oxygenation, and integrating sensors for precise control, improving overall performance and yield. Technological advancements, such as automation and control systems, sensors, and monitoring technologies, have revolutionized hydroponic farming by enabling real-time data collection and environmental management. The integration of the Internet of Things (IoT) and smart farming practices, combined with data analytics and machine learning, has further optimized system performance and decision-making. LED lighting and vertical farming techniques have maximized space utilization and improved crop yields, particularly in urban environments. Advances in nutrient solutions, disease management, and water quality have optimized plant health and resource efficiency. Economic and environmental considerations, including cost-benefit analyses and comparisons with traditional agriculture, highlight the potential for hydroponics to offer better returns on investment and reduce environmental impact. Despite the significant progress, challenges such as high initial setup costs and technical complexities remain. Future research opportunities and interdisciplinary collaborations are essential to address these challenges and drive further innovation in hydroponic farming.

Simulation-based brownfield factory planning using deep reinforcement learning

Factory layout planning aims at generating optimized layout variants under consideration of varying planning objectives and boundary conditions. Automated planning approaches are commonly used in the early phase of layout planning since planning manually is time-consuming. Recent developments propose deep Reinforcement Learning (RL) based approaches that are capable of generating high-quality solutions by learning the metrics of the underlying problem. However, the existing deep RL-based approaches only consider greenfield planning and neglect the more common case of brownfield planning. This paper addresses this research gap with a novel planning approach that combines deep RL and material flow simulation to solve brownfield planning problems.

Topographic Steering of the Upper Arctic Ocean Circulation by Deep Flows

Topographic Steering of the Upper Arctic Ocean Circulation by Deep Flows

Superhydrophobicity With Self-Adaptive Water Pressure Resistance and Adhesion of Pistia Stratiotes Leaf.

Superhydrophobic surfaces are promising for optimizing amphibious aircraft by minimizing water drag and adhesion. Achieving this involves ensuring these surfaces can resist high liquid pressure caused by deep water and fluid flow. Maximizing the solid-liquid contact area is a common strategy to improve liquid pressure resistance. However, this approach inevitably increases solid-liquid adhesion, making it challenging to guarantee a trade-off between the two wetting characteristics. Here, it is found that the Pistia stratiotes leaf exhibits superhydrophobicity with high water pressure resistance and low adhesion, attributed to its self-adaptive deformable microstructure with unique re-entrant features. Under pressure, these microstructures deform to increase the solid-liquid contact area, thereby enhancing water pressure resistance. The re-entrant features elevate the deformation threshold, enabling higher modulus microstructures to achieve adaptive response. This facilitates the recovery of deformed microstructures, restoring the air layer and maintaining low adhesion. Following these concepts, Pistia stratiotes leaf-inspired surfaces are fabricated, achieving an 183% improvement in water impact resistance and an ≈80% reduction in adhesion after overpressure compared to conventional superhydrophobic surfaces. The design principles inspired by Pistia stratiotes promise significant advancements in amphibious aircraft and other trans-media vehicles.

Justice & Joy: Transforming Healing Praxis in Counseling Psychology and Beyond

Amid today's polycrisis—marked by global violence, chaos, and uncertainty—there is an urgent imperative for healing. In this reimagined presidential address, I advocate for centering justice and joy as powerful forces for healing, which is a core value of the Society of Counseling Psychology (SCP). I share SCP's yearlong activities exploring personal and collective healing. Drawing from interdisciplinary research, critical dialogue, and my experiences co-teaching a course on this topic, I introduce a framework outlining six healing ingredients of joy: deep connection, existential recognition, flow, shared beliefs and actions, radical hope, and self-determination/freedom/liberation. Before presenting this framework, I examine Black joy as a counter to the epistemic violence that often erases the contributions of Black scholars. The proposed new framework portrays joy not only as a personal emotion, but as a collective assertion of humanity. I conclude with recommendations to advance healing practices in research, training, and practice.

Deep mantle plumes feeding periodic alignments of asthenospheric fingers beneath the central and southern Atlantic Ocean

High-resolution full waveform seismic tomography of the Earth's mantle beneath the south and central Atlantic Ocean brings into focus a series of asthenospheric low shear velocity channels, or "fingers" on both sides of the southern and central mid-Atlantic ridge (MAR), elongated in the direction of absolute plate motion with a spacing of [Formula: see text]1,800 to 2,000 km, and associated with bands of shallower residual seafloor depth anomalies that suggest channeled flow over thousands of kilometers. Each of the three most clearly resolved fingers on the African side of the MAR corresponds to a separate group of whole mantle plumes rooted in distinct patches at the core-mantle boundary, feeding hotspots, and volcanic lines with distinct isotopic signatures. Plumes of a given group appear to merge at the top of the lower mantle before separating again, suggesting interaction of deep mantle flow with a more vigorous mesoscale circulation in the upper mantle. The corresponding hotspots are generally offset from the location of the deep mantle plume roots. The distinct isotopic signatures of these hotspot groups are also detected in the mid-ocean ridge basalts at the location where the fingers meet the ridge. Meanwhile, at least some of the variability within each plume group could originate in the upper mantle and extended transition zone where plumes in a given group appear to merge and pond. This study also adds to mounting evidence that the African large low shear velocity province is not a uniform, unbroken pile of dense material rising high above the core-mantle boundary, but rather a collection of mantle plumes rooted in patches of distinct composition.

Morphological and Hydrogeological Controls of Groundwater Flows and Water Age Distribution in Mountain Aquifers and Streams

AbstractMountains are an essential source of the terrestrial component of the hydrological cycle, supplying high‐quality water to river networks and floodplain aquifers, especially during droughts. Traditionally, mountain hydrology has focused on shallow processes, overlooking the significance of deep‐seated rock formations due to characterization challenges. Recent field studies have revealed that fractured rock formations can host rich aquifers despite their low permeability. Nonetheless, it is unclear how deep flows interact with the overall hydrological functioning of mountain areas, how they contribute to the long‐term water budget, and how climate, morphology, and geology jointly control them. Through numerical simulations, we have gained new insights into mountain aquifers, addressing (a) the proportion of groundwater base flow and its age distribution, (b) water storage and its sensitivity to groundwater recharge, (c) the impact of long term mean recharge on the extent of the groundwater‐fed surface drainage network under various morphological and geological settings. We showed that subsurface travel times follow a Gamma distribution, whose parameters are modulated by recharge, hydraulic conductivity, and topography. High recharge and strong decay with depth of the hydraulic conductivity in a hilly topography lead to a shallow water table mimicking the surface topography and spatially distributed low‐intensity outflows that feed a dense drainage network. In rugged catchments, the groundwater contribution intensifies and concentrates in the downstream portion of the river network as recharge declines. These findings can help assess how a changing climate might impact hydrological regimes under various geomorphological conditions and identify sustainable water uses in mountain environments.

Alkalinity sources in the Dutch Wadden Sea

Abstract. Total alkalinity (TA) is an important chemical property that plays a decisive role in the oceanic buffering capacity with respect to CO2. TA is mainly generated by weathering on land as well as by various anaerobic metabolic processes in the water and sediments. The Wadden Sea, located in the southern North Sea, is hypothesized to be a source of TA for the North Sea, but quantifications are scarce. This study shows observations of TA, dissolved inorganic carbon (DIC), and nutrients in the Dutch Wadden Sea in May 2019. Surface samples were taken along several transects in order to investigate spatial distribution patterns and compare them with data from the late 1980s. A tidal cycle was sampled to further shed light on TA generation and potential TA sources. We identified the Dutch Wadden Sea as a source of TA and estimated an export of 6.6 Mmol TA per tide to the North Sea. TA was generated in the sediments, with deep pore water flow during low tide enriching the surface water. A combination of anaerobic processes and CaCO3 dissolution were potential TA sources in the sediments. We deduce that seasonality and the associated nitrate availability specifically influence TA generation by denitrification, which is low in spring and summer.

Deep Learning-Based Rapid Flow Field Reconstruction Model with Limited Monitoring Point Information

Deep Learning-Based Rapid Flow Field Reconstruction Model with Limited Monitoring Point Information

Impacts of Light Exposure and Soil Covering on Sweet Potato Storage Roots in a Novel Soilless Culture System

Soilless culture systems, which promote plant growth and enable the precise control of the root-zone environment, have yet to be fully established for sweet potatoes. In this study, we developed a soilless culture system and examined the effects of soil covering and light exposure on the storage roots of sweet potatoes. Sweet potato seedlings with induced storage roots were transplanted into five systems: a previously developed pot-based hydroponics system (Pot), an improved version with storage roots enclosed in a plastic box and covered with a soil sheet (SS), the SS system without the soil sheet (SD), the SD system with light exposure to storage roots after 54 days (SL), and a deep flow technique (DFT) hydroponics system. Our study enabled the time-course observation of storage root enlargement in the SS, SD, and SL systems. In the SL system, light exposure suppressed the storage root enlargement and reduced epidermal redness. No storage root enlargement was observed in the DFT system, even at 151 days after transplantation. Light exposure in the SL system increased the chlorophyll and total phenolic contents in the cortex beneath the epidermis, while the starch content was the lowest in this system. These findings indicate that the developed system can induce normal storage root enlargement without soil. Additionally, the observed changes in growth and composition due to light exposure suggest that this system is effective for controlling the root-zone environment of sweet potatoes.

Combined evaluation of visual field and optical coherence tomography angiography data in multiple sclerosis patients

Aims: To evaluate Humphrey visual field analyzer (HVFA) data and optical coherence tomography angiography (OCT-A) parameters in multiple sclerosis (MS) patients with optic neuritis (ON) attacks. Methods: 34 eyes of 34 patients with MS diagnosis who had ON attacks were included in the study. After detailed ophthalmoscopic examination, OCT-A and HVFA images were taken. Spearman correlation analysis was used in statistical analysis. Results: The mean age of the participants was 31.52±10.09 years. In the correlation analysis, a moderate correlation was found between mean deviation (MD) and foveal deep blood flow (DF) (rho=0.52), and a weak correlation was found between MD and foveal superficial blood flow (SF) (rho=0.40) and optic disc head blood flow (ODF) (rho=0.38). A weak negative correlation was found between pattern standard deviation (PSD) and DF (rho=-0.37) and ODF (rho=-0.40). Conclusion: This study showed that OCTA data were affected in a manner consistent with HVFA data. More practical and quickly applicable OCTA scans may facilitate the estimation of visual field findings in MS patients.

Formation mechanism and oil-bearing properties of gravity flow sand body of Chang 63 sub-member of Yanchang Formation in Huaqing area, Ordos Basin

Abstract Gravity flow sand body is an important reservoir for deep water deposition. The in-depth study of its formation mechanism and oil enrichment law can provide important guidance for oil and gas exploration in deep water areas. In this article, taking the deep-water gravity flow sand body of the Triassic Chang 63 sub-member in the Huaqing area as an example, the identification characteristics, formation mechanism, and oil-bearing characteristics of the gravity flow sand body are systematically discussed. The results show that sandy debris flow, turbidite flow, and mixed event flow present different superposition combination modes on vertical gravity flow deposition. The lithofacies of the gravity flow sandstone complex are mainly affected by lake level fluctuation, provenance supply, and hydrodynamic conditions. According to the formation conditions, sedimentary types. and distribution characteristics of deep water gravity flow sand bodies, a three-dimensional sedimentary mode of deep water gravity flow sand bodies is established. It is found that the physical and oil-bearing properties of the sandy debris flow sand body are better than those of the turbidity flow sand body. The reason is that the proportion of debris in the sand body of sandy debris flow is high, and the content of debris particles is generally greater than 70% according to the statistics of thin sections, which is conducive to the formation of pores. Second, sandy debris flow belongs to block transport, and the mixing of lithology is not sufficient in the process of block flow transport, so some pores will be preserved. Finally, gravity flows, which are dense at the beginning, become less dense later as detrital sediments unload, allowing them to be transported to distant regions. Therefore, the monolayer thickness of the sandy debris flow sand body is large (generally greater than 0.5 m), and the particle size of the debris ranges from 0.03 to 0.3 mm. Finally, the physical and oil-bearing properties of the sandy clastic flow sand body are better than those of the turbidity flow sand body.

Pore-Scale Simulation of Deep Shale Gas Flow Considering the Dual-Site Langmuir Adsorption Model Based on the Pore Network Model.

As a key resource in the oil and gas sector, shale gas development profoundly influences the advancement of the global energy industry. Deep shale gas reservoirs, typically found at depths exceeding 3500 m, represent a significant portion of total shale gas reserves. Currently, pore network models primarily simulate middle and shallow shale gas, insufficiently addressing the unique challenges posed by high-temperature and high-pressure conditions in deep shale gas formations. There is an urgent need to explore the microscale flow dynamics of deep shale gas. In this work, we use the pore network model to simulate the flow dynamics of deep shale gas, particularly under nanoconfinement conditions. The simulation integrates adsorption, slip flow, Knudsen diffusion, and bulk flow phenomena. Utilizing the dual-site Langmuir adsorption model tailored for high-temperature and high-pressure conditions enhances the accuracy of gas conductivity calculations for the adsorbed phase, thereby reflecting the specific characteristics of deep shale gas reservoirs. Moreover, this research investigates the flow characteristics of deep shale gas under varying sensitivity parameters, such as water saturation, temperature, and pressure. It explores methane flow dynamics, focusing on effective diffusion coefficients and permeability. The modified approach to calculating adsorption phase conductivity using the dual-site Langmuir adsorption model accurately captures the adsorption behaviors and characteristics of deep shale gas reservoirs.

Influence of Cow Dung Extract Complemented with Nutrient Solution on the Growth Performance of Lettuce (Lactuca sativa L)

Limited studies have focused on plant growth performance using organic-based solutions complemented with mineral elements in a hydroponic system. Thus, this study aimed to investigate the growth performance of lettuce (Lactuca sativa L.) as influenced by cow dung extract combined with a hydroponic nutrient solution. Treatments considered as four different levels of aerated cow dung extracts (C), viz., C1 = 50 g l-1, C2 =100 g l-1, C3 =150 g l-1 and C4 =200 g l-1 and four strengths of standard nutrient solution (S), viz, S1 = 30% of standard nutrient solution, S2 = 40% of standard nutrient solution, S3 = 50% of standard nutrient solution and S4 = 60% of standard nutrient solution. The experiment was carried out using a deep flow technique in a semi-greenhouse. Various growth and physiological parameters were measured in this experiment. The obtained data were subjected to statistical analysis with 4 replicates by analysis of variance (ANOVA) using SPSS. In the case of growth parameters, the tallest plant (23.54 cm), maximum number of leaves per plant (17.01) broadest leaf (13.32 cm), and the highest fresh weight (112.05 g/plant) were recorded from C3 while the lowest in C1. For hydroponic nutrient solution, the tallest plant (23.13 cm), the maximum number of leaves per plant (16.66), the widest leaf breath (14.17 cm), and the highest fresh weight (116.0 g/plant) were recorded from S4 while the lowest in S1. On the other hand, physiological traits viz. leaf area, net assimilation ratio, and relative growth rate were statistically higher in C3 and lowest in C1. In the nutrient solution, all physiological parameters were highest in S4 and the lowest in S1. In the case of the interaction effect, the highest fresh weight and almost all the parameters were found best in C3S4 and the lowest in C1S1. Therefore, the analysis showed that in terms of growth promotion properties hydroponic nutrient solution along with cow dung extract had a substantial impact and C3S4 was the most preferable treatment combination. Based on these findings, in a hydroponic system, an inorganic nutrient solution combined with organic liquid fertilizer derived from cow dung extract (as an alternative nutrient source) requires further improvements to achieve optimal growth and yield.

Spatial and temporal dynamics of groundwater biogeochemistry in the deep subsurface of a high-energy beach

Intertidal sandy beach systems are considered complex biogeochemical reactors. At beach sites that are subject to high tidal and wave energy, seawater circulation can reach tens of meters deep into the subsurface and changing environmental conditions are assumed to lead to dynamic groundwater flow paths, saltwater-freshwater mixing zones, and a spatio-temporally variable groundwater biogeochemistry. Previous studies mainly focused on the upper meters of subterranean estuaries (STE), while the deep subsurface remained a black box. This study presents spatial (cross-shore) and temporal (∼ six-weekly, over 1.5 years) dynamics of the groundwater biogeochemistry that were observed down to 24 m below the ground surface (mbgs) of a sandy high-energy beach on Spiekeroog Island (Germany).In addition to redox conditions along a cross-shore transect ranging from oxic to Fe oxide reducing/slightly sulfidic, we found a previously unknown, distinct vertical redox zonation as well. Temporal variations of the biogeochemistry within low salinity groundwater at the most landward station close to the dune base were mainly driven by storm flood related seawater infiltration. Around the high water line, the extent of the upper saline plume (USP) varied over time. Furthermore, temporal dynamics of the O2 saturation at 6 mbgs indicated a seasonally shifting depth of the oxycline at this location. In the lower intertidal zone, groundwater solute concentrations displayed a temporally variable zone of deep freshwater discharge.Regarding the impact of the deep STE on the groundwater biogeochemistry of the discharge zone, our data revealed that nutrient, Mn, and Fe release along the deep flow paths through the USP towards the discharge zone was limited, likely due decreasing availability of labile organic matter and subsequent slowing down of metabolic processes with depth. High concentrations of metabolites in the upper ∼ 2 mbgs of the discharge zone were, therefore, rather attributed to the incorporation of labile organic matter during continuous and storm flood related sediment relocation and/or the contribution of older waters, e.g., the subtidal saltwater wedge.

Natural Hydrogen, the Ultimate Green Energy: Current Status and Prospects

Natural hydrogen is green hydrogen that is naturally produced on Earth, that is, an energy source that does not emit carbon without using additional energy for production. Since carbon is emitted from the existing gray hydrogen and blue hydrogen, green hydrogen with perfect eco-friendliness has been required. Natural hydrogen is produced through four mechanisms: serpentinization, radioactive decomposition, magma gas removal, and rock destruction and can exist in the dissolved ․ content of fluid mammals. Representative sites discovered to date include Mali, Africa, Brazil, and Australia, and each topography has different characteristics. Although reserves are estimated using geochemical and geophysical exploration technologies, the reliability is not high because the deep hydrogen flow in various geological environments is not considered. Based on this, this study drew the following conclusions. 1) Hydrogen can realize carbon neutrality with high energy content and independence, but 2) economic feasibility cannot be evaluated due to the uncertainty of the reserves and current reserves of natural hydrogen. Therefore, 3) for commercial use, leakage detection technology is required along with an understanding of the mechanism for generating natural hydrogen.