Three-dimensional Network Model Research Articles

Correlations between an Urban Three-Dimensional Pedestrian Network and Service Industry Layouts Based on Graph Convolutional Neural Networks: A Case Study of Xinjiekou, Nanjing

Urban high-density development has led to the emergence of complex three-dimensional pedestrian networks. As a crucial component of city centers, these networks significantly influence the spatial distribution of service industries. Understanding the correlation between pedestrian networks and service industry layouts is vital for effective planning and development. This study proposes a technical framework for analyzing the relationship between three-dimensional pedestrian networks and service industry layouts. Using the Xinjiekou central area in Nanjing as a case study, we constructed a three-dimensional pedestrian network model using the sDNA method. Focusing on catering formats, we introduced a method to study the spatial distribution characteristics of service industries in three-dimensional spaces and employed a graph convolutional network model to systematically analyze the correlation between pedestrian network closeness and betweenness with catering formats. The results indicate that pedestrian network closeness is significantly positively correlated with the number and average spending of catering formats, while betweenness shows almost no correlation. High-closeness areas, due to their traffic convenience and walkability, are more conducive to the concentration of catering formats and higher spending levels. Our findings provide valuable insights for catering format location decisions and the optimization of three-dimensional pedestrian networks, contributing to sustainable urban development.

Heterogeneity properties and permeability of shale matrix at nano-scale and micron-scale

Heterogeneity of shale pores at nano-scale and micrometer-scale is of great significance to gas transport properties. In this study, the pore structure of shale samples from lower Silurian Longmaxi Formation in the Sichuan basin is investigated by field emission-scanning electron microscopy (FE-SEM) and x-ray micro-computed tomography (Xμ-CT) technology. Based on fractal theory, the lacunarity is introduced to describe the clustering degree of pores in shale matrix, which can compensate for the limitations of fractal dimension. Combining lacunarity with fractal dimension allows for quantification of subtle differences in pore spatial distribution. For FE-SEM images at nano-scales, the fractal dimension changes in a “U” shape, while lacunarity changes in a “∩” shape. For Xμ-CT images at micrometer-scale, both the fractal dimension and lacunarity change in a logarithmic function. Lacunarity at both nano-scale and micrometer-scale linearly decreases with the increase in fractal dimension. By three-dimensional (3D) pore network modeling analysis, the structure properties of the connected pores, such as the number of pores and throats, pore diameter, pore volume, pore surface, throat length, and coordination number, are quantitatively calculated, and these structure parameters show strong heterogeneity. The average coordination number of the connected pores ranges in 2.92–4.36. This indicates that these pores in shale matrix have poor connectivity. The permeability varies from 0.06 to 0.17 μm2 in two-dimensional (2D) Xμ-CT images but from 3.20 to 34.99 μm2 in a 3D structure. The permeability in the 3D structure is about two order higher in magnitude than that in the 2D Xμ-CT images.

A model of cement grout flow in a fracture network system

Understanding the mechanism of cement grout flow in fractured rocks is essential for controlling seepage in the construction of hydraulic projects. However, simulating grouting in fractured rocks is complex due to random fracture distribution and non-Newtonian grout flow. This study proposes a numerical model for simulating cement grout flow, considering the hydro-mechanical coupled process in a three-dimensional fracture network system based on the unified pipe-network method. A pipe-network equivalence method is used to derive equivalent pipes for the process of grouting displacement, and the flow field can be solved in unified pipe networks. The hydro-mechanical coupled process is achieved by incorporating an analytical solution for the stress distribution of the center borehole model on an infinitely large plane and a constitutive model for the estimation of the fracture aperture. The time-varying property of the cement grout flow is also considered in the simulation of grouting. The reliability of the numerical model is verified by comparing with analytical solutions for grout flow in a water-saturated fracture. Grouting experiments in fracture networks are also performed to demonstrate the applicability of the proposed method. The influence of various factors on the propagation of cement grout is further investigated in three-dimensional fracture network models.

Multiphase flow simulation of fractured karst oil reservoirs applying three-dimensional network models

The characteristics of karst reservoirs are extremely varied and anisotropic, exhibiting notable differences in porosity, permeability, and corresponding fluid flow pathways. Fractured karst petroleum reservoirs, such as distinct caverns and fractures, are an example of a typical discrete media type. The traditional reservoir modeling approach and discrete fracture-like local refinement models are unsuitable for field application of fractured karst oil reservoirs due to the needs of high fidelity geological description and huge computing efforts. Based directly on the spatial characteristics of seismic surveys, a numerical simulation model in three dimensions, akin to a node-like network, is presented here for cracked karst oil reserves. First, the watershed image processing technique and the automatic connection identification procedure are used to extract the three-dimensional node-network model. After that, automatic differentiation is used to build the numerical finite volume scheme, and the proper gradient-based adjoint approach is used to conduct the related historical matching rapidly. After validation by a synthetic model in a commercial simulator, this proposed three-dimensional network numerical model is used for a field reservoir block of deep formation in the Tarim basin to demonstrate its computational efficiency and viability for enormously comparable karst oil reservoirs.

Identification of Calculous Pyonephrosis by CT-Based Radiomics and Deep Learning.

Urgent detection of calculous pyonephrosis is crucial for surgical planning and preventing severe outcomes. This study aims to evaluate the performance of computed tomography (CT)-based radiomics and a three-dimensional convolutional neural network (3D-CNN) model, integrated with independent clinical factors, to identify patients with calculous pyonephrosis. We recruited 182 patients receiving either percutaneous nephrostomy tube placement or percutaneous nephrolithotomy for calculous hydronephrosis or pyonephrosis. The regions of interest were manually delineated on plain CT images and the CT attenuation value (HU) was measured. Radiomics analysis was performed using least absolute shrinkage and selection operator (LASSO). A 3D-CNN model was also developed. The better-performing machine-learning model was combined with independent clinical factors to build a comprehensive clinical machine-learning model. The performance of these models was assessed using receiver operating characteristic analysis and decision curve analysis. Fever, blood neutrophils, and urine leukocytes were independent risk factors for pyonephrosis. The radiomics model showed higher area under the curve (AUC) than the 3D-CNN model and HU (0.876 vs. 0.599, 0.578; p = 0.003, 0.002) in the testing cohort. The clinical machine-learning model surpassed the clinical model in both the training (0.975 vs. 0.904, p = 0.019) and testing (0.967 vs. 0.889, p = 0.045) cohorts.

Method for measuring blood flow and depth of blood vessel based on laser speckle contrast imaging using 3D convolutional neural network: A preliminary study

Laser speckle contrast imaging (LSCI) is a technology that can acquire real-time two-dimensional perfusion images non-invasively. However, measuring perfusion velocity under a tissue is difficult due to its optical properties. Dynamic scattered light occurs by perfusion is diffused by the surrounding tissue and it depends on the depth of tissue. We aimed to develop a three-dimensional convolutional neural network (3D-CNN) model based on laser speckle contrast images to measure the flow velocity and the depth of tissue using a flow phantom. The training data was selected as a region containing both static speckles and dynamic speckles, and the model was trained on the laser speckle contrast image data. The trained model showed a measurement accuracy of more than 90 % for velocity and 99 % for depth measurement. This study has the potential of a deep-learning model based on LSCI to analyze the blood flow and depth of blood vessels in bio-applications.

A 3D neuronal network read-out interface with high recording performance using a neuronal cluster patterning on a microelectrode array

In recent years, in vitro three-dimensional (3D) neuronal network models utilizing extracellular matrices have been advancing. To understand the network activity from these models, attempts have been made to measure activity in multiple regions simultaneously using a microelectrode array (MEA). Although there hve been many attempts to measure the activity of 3D networks using 2-dimensional (2D) MEAs, the physical coupling between the 3D network and the microelectrodes was not stable and needed to be improved. In this study, we proposed a neuronal cluster interface that improves the active channel ratio of commercial 2D MEAs, enabling reliable measurement of 3D network activity. To achieve this, neuronal clusters, which consist of a small number of neurons, were patterned on microelectrodes and used as mediators to transmit the signal between the 3D network and the microelectrodes. We confirmed that the patterned neuronal clusters enhanced the active channel ratio and SNR(signal-to-noise-ratio) about 3D network recording and stimulation for a month. Our interface was able to functionally connect with 3D networks and measure the 3D network activity without significant alternation of activity characteristics. Finally, we demonstrated that our interface can be used to analyze the differences in the dynamics of 3D and 2D networks and to construct the 3D clustered network. This method is expected to be useful for studying the functional activity of various 3D neuronal network models, offering broad applications for the use of these models.

Analysis and Characterization of Micro–Nano Pores in Coal Reservoirs of Different Coal Ranks

Coalbed methane represents a promising source of clean and efficient unconventional energy. The intricate network of micro–nano pores within coal serves as the primary adsorption space for gas, contributing to the complexity of gas migration channels. In this study, based on the box-counting method, three coal samples representing low, medium, and high ranks were subjected to high-precision micro-CT scanning and nano-CT scanning to generate three-dimensional (3D) pore network models using Avizo visualization software. This facilitated the accurate and quantitative characterization of the micro–nano pore structures within coal reservoirs. The results indicated that the face rate distribution range of each sample was large, indicating relatively strong heterogeneity in each sample. The volume fractal dimension of each sample, determined through micro–nano-CT scanning, was around 2.5, while the surface fractal dimension exhibited oscillatory characteristics with moderate uniformity. The pore equivalent radius and throat equivalent radius distributions were unimodal across all the samples, with the micro-CT scanning revealing a concentration primarily within the range of 100–400 μm for the pore equivalent radius and within 200 μm for the throat equivalent radius. Conversely, the nano-CT scanning exhibited concentrations primarily within the range of 500–2500 nm for the pore equivalent radius and within 2000 nm for the throat equivalent radius. The analysis of the 3D reconstruction structures indicated that the middle-rank coal exhibited more developed large–medium pores compared with the low-rank and high-rank coal, while the low-rank and high-rank coal exhibited relatively more micro–small pores. Furthermore, the low-rank coal exhibited the fewest number of pores but the largest average pore equivalent radius and throat radius. Additionally, the middle–high-rank coal exhibited a relatively larger number of pores. Despite the complex topological structures observed in each sample, a significant proportion indicated a coordination number of 0–20, indicating excellent connectivity within the coal samples. This study is conducive to the optimization of coalbed methane surface development blocks and the formulation of reasonable development plans.

Joint Optimization of Inventory and Schedule for Coal Heavy Rail Considering Production–Transportation–Sales Collaboration: A Spatio-Temporal-Mode Network Approach

China’s coal industry has long faced challenges due to imbalance between supply and demand divisions and the discrepancies between supply and demand prices. Research on optimizing coal transfer amid fluctuating coal prices is crucial for enhancing coal resource allocation, ensuring a balanced and stable supply of coal resources, and promoting the sustainable development of the coal industry. This paper focuses on optimizing coal transfer schemes within a comprehensive transportation system, considering fixed transportation routes and opening times of railroad transportation and waterway transportation, with road transportation serving as a supplementary network. A three-dimensional spatio-temporal network model is established to represent the spatial and temporal paths and transportation modes of coal logistics. The study uses the coal transportation network from Inner Mongolia and Shaanxi to East China as a case study, analyzing factors such as coal price fluctuation, transportation rates, and the ratio of self-produced to purchased coal at the production end via sensitivity analysis. The findings indicate that the proposed model can effectively guide transportation practices under the integrated production, transportation, and sales framework in the energy industry. Additionally, it shows that enterprises can achieve higher profits and maintain revenue and supply stability despite coal price fluctuations, aligning with the coal industry’s transformation and development in the current context.

Combining three-dimensional acoustic coring and a convolutional neural network to quantify species contributions to benthic ecosystems.

The seafloor is inhabited by a large number of benthic invertebrates, and their importance in mediating carbon mineralization and biogeochemical cycles is recognized. However, the majority of fauna live below the sediment surface, so most means of survey rely on destructive sampling methods that are limited to documenting species presence rather than event driven activity and functionally important aspects of species behaviour. We have developed and tested a laboratory-based three-dimensional acoustic coring system that is capable of non-invasively visualizing the presence and activity of invertebrates within the sediment matrix. Here, we present reconstructed three-dimensional acoustic images of the sediment profile, with strong backscatter revealing the presence and position of individual benthic organisms. These data were used to train a three-dimensional convolutional neural network model and, using a combination of data augmentation and data correction techniques, we were able to identify individual species with an 88% accuracy. Combining three-dimensional acoustic coring with deep learning forms an effective and non-invasive means of providing detailed mechanistic information of in situ species-sediment interactions, opening new opportunities to quantify species-specific contributions to ecosystems.

Accuracy of hygro-expansive curl predictions for paper sheets based on homogenised 2D and 3D network representations

Paper, a porous-fibrous network, is made up of hydrophilic fibres, which are notably susceptible to deformations due to variations in moisture content. The response of paper sheets to full or partial wetting, based on the water-induced fibre swelling, has mostly been predicted in the past using highly idealised two-dimensional network geometries. The purpose of the present paper is to establish how precisely the two-stage process of determining effective hygro-expansivity under uniform wetting by a two-dimensional model followed by bending analysis using a continuum model predicts the curl of a fully three-dimensional network. In the present multi-scale modelling work, the moisture-induced homogenised sheet scale curl, derived from hygro-elastic deformations using non-uniform moisture profiles, is analysed based on an idealised non-woven three-dimensional fibrous network finite element model. First, the effective paper sheet swelling properties, derived computationally for the uniform wetting condition, are compared with analytically homogenised hygro-elastic properties. Whereas the analytically predicted elastic modulus has an inaccuracy of only ∼5%, the hygro-expansivity coefficient is over-predicted by roughly a third for a network with straight fibres. The difference is smaller if waviness of the fibres and wrap-around in the fibre bonds are taken into account. Fibre alignment turns out to have little influence. Subsequently, paper sheet curl predictions made by the analytical model are compared with that computed from the three-dimensional network model. The inaccuracy of the fully analytical model, based on the two-dimensional network description, relative to the fully detailed three-dimensional computational model with straight fibres is on the order of 40%–50%. The fidelity of the simple model is primarily constrained by the inaccuracy of the effective hygro-expansion coefficient. If the expansivity extracted from the three-dimensional simulations under uniform wetting is inserted, the inaccuracy drops to less than 10%. The continuum bending description hence performs adequately, provided it is based on a reliable estimation of the uniform expansion behaviour of the material.

Dynamic simulation of immiscible displacement in fractured porous media

Investigating immiscible displacement in fractured porous media is essential for understanding the two-phase flow behavior within pores and fractures. In this work, a three-dimensional pore-fracture network model was developed to address the influence of fracture on flow patterns and to characterize fracture-matrix crossflow under different flow conditions. Sensitivity studies at a wide range of viscosity ratios and capillary numbers underscored that fracture significantly influenced flow patterns in the capillary fingering zone. Fracture with an advantageous path effect in the displacement direction caused a shift in the boundary of capillary fingering zone toward an increase in capillary numbers. As fracture aperture decreased and aspect ratio increased, there was a discernible decline in the crossflow rate. When fracture aperture equaled average matrix throat diameter, fracture lose advantageous path effect in compact displacement zone but retained it in viscous fingering and capillary fingering zones. Distinct matrix-fracture crossflow development processes were observed in different zones: in cross zone, following displacement breakthrough, the crossflow underwent a “long-term” process to attain stability. Viscous fingering zone promptly achieved stability post-breakthrough, whereas both capillary fingering and compact displacement zones had already reached a stable state before breakthrough. Nonlinear variations in breakthrough saturation were observed in the cross zone between compact displacement and capillary fingering zones. The control process of immiscible displacement exhibited variability under different flow conditions: compact displacement zone was characterized by matrix dominance, viscous fingering zone was jointly controlled by matrix displacement and fracture-matrix crossflow, and capillary fingering zone was primarily governed by fracture-matrix crossflow. These findings enhance scholarly comprehension of immiscible displacement behavior in fractured porous media.

Defects recognition of pine nuts using hyperspectral imaging and deep learning approaches

Pine nuts, as a highly nutritious and medicinally valuable food, are susceptible to various defects during their cultivation, harvesting, and transportation, which can reduce their quality. Therefore, rapid and accurate identification of pine nut defect types is of utmost importance to ensure the overall quality of the pine nuts. In this study, hyperspectral imaging (HSI) systems covering two spectral ranges (400–1000 nm and 900–1700 nm) were employed to capture hyperspectral images of healthy pine nuts and pine nuts with six types of defects. One-dimensional (1D) and three-dimensional (3D) Convolutional Neural Network (CNN) models with multi-head attention mechanisms were constructed using 1D spectra and 3D hyperspectral images, respectively. To validate the effectiveness of the proposed models, Support Vector Classifier (SVC) models were built using 1D spectra and used as a comparison. Overall, the proposed CNN models outperform traditional machine learning methods in two spectral ranges (400–1000 nm and 900–1700 nm). 1D CNN model in the near-infrared spectral range (900–1700 nm) achieved an accuracy of 90.23 % on the training set and 81.32 % on the validation set. Additionally, the Generalized Gradient-Weighted Class Activation Mapping (Grad-CAM++) visualization method was applied to conduct visual analysis on the 1D CNN and 3D CNN models, enabling the identification of important wavelength ranges and pixel regions in the models, thereby enhancing the interpretability of the decision-making process of the models. Overall, the results of this study demonstrated the feasibility of using a combination of hyperspectral imaging and convolutional neural networks for pine nut defects classification, and the visual analysis of the models provided new insights and understanding for pine nut defects identification.

Study on pore-crack evolution and connectivity of coal subjected to controlled electrical pulse based on CT scanning technology

Controlled electric pulse (CEP) technology is a novel method for enhancing coal seam permeability. To study the pore-crack evolution and connectivity of the coal body by CEP, a self-built CEP cracking and enhancing permeability of coal rock test system was used to test five groups of coal samples under different breakdown voltages (BVs). We established a three-dimensional crack structure and equivalent pore network model for before and after penetration using high-precision micro-CT technology to quantitatively study the evolution and connectivity of the pore and crack structures of the coal body. The results showed that with rising BV, coal body crushing intensifies, leading to an increased cracking ratio and fractal dimension. This summarised the three stages of coal sample crushing by CEP. In addition, the proportion of larger equivalent pores and throat radii in the breakdown coal body was significantly increased compared to that of raw coal, and the connectivity structure was improved to different degrees compared to that of raw coal. Eventually, a three-dimensional crack network structure formed inside the coal samples, which was broken at both ends and complex in the middle, providing a transport channel for coalbed methane mining.

Study on the characterization method and evolution law of the three-dimensional pore structure in frozen loess under loading

In this study, a functional relationship for frozen soil at different temperatures, confining pressures, and triaxial compressive strengths is established through macroscopic and mesoscopic comparative tests. Additionally, a three-dimensional pore network model at the mesoscopic scale is constructed. Morphological characterization parameters are introduced to quantify the pore structure, and the evolution of the pore structure in frozen soil during the stress process is analyzed, as well as the influence of temperature and confining pressure on the pore characteristics and failure morphologies. The results reveal that at the macroscale, frozen soil exhibits a power function relationship with temperature, confining pressure, and triaxial compressive strength. During mechanical loading, frozen soil undergoes compaction, pore development, and pore expansion stages, leading to changes in pore size and connectivity. Additionally, temperature and confining pressure significantly impact the pore characteristics and failure morphologies of frozen soil. At lower temperatures, frozen soil experiences severe bulging and brittle failure, accompanied by increased pore size, enhanced connectivity, and complex morphology. Increasing the confining pressure reduces the degree of bulging and damage, decreases the porosity and connectivity, enhances the complexity of the pore morphology, and results in a denser and more stable internal structure in frozen soil. Through this study, a better understanding of the damage behavior of frozen soil under different temperatures and confining pressures is achieved. Furthermore, this research provides a theoretical basis and reference for addressing related engineering problems.

Pore structure characterization of sandstone under different water invasion cycles using micro-CT

The meso-structural changes of rocks during repeated cycles of water invasion are the fundamental cause of macroscopic physical property damage. In this paper, based on the computed tomography scan images of rock samples under different numbers of water invasion cycles, a three-dimensional pore network model was constructed to analyze the changes in pore structure under the action of water invasion. The damage variable was introduced to quantitatively characterize the parameter damage of each pore and reveal the evolution of rock meso-damage. The results show that 81% of the pore radius is less than 10 μm under 0 water invasion cycles and that 76% of the pore radiuses are less than 10 μm after 10 water invasion cycles. After 10 water invasion cycles, the peak range of the pore radius distribution enlarged from the initial range of 2–4 μm to that of 4–6 μm and the proportion of pore throats with a radius less than 10 μm decreased from an initial 82–72%. With an increase of water invasion cycles, the proportion of large pores increased and the connectivity among pores enhances gradually. The damage variable of each pore parameter changed the most during 2–5 water invasion cycles. After 10 water invasion cycles, the maximum degree of damage that the pore volume reached was up to 41.44% and the minimum degree of damage of the pore coordination number was 5.80%. The test results helped to reveal the pore structure changes and the damage of rock samples during water invasion cycles.

Characterization and prediction of tensile properties of carbon fiber-reinforced thermoplastics composed of hybrid short carbon fiber/PA6 fiber nonwoven mats

Hybrid carbon fiber/resin fiber nonwoven mats are considered as a highly promising application of recycled carbon fibers due to their low cost and simple synthesis process. However, the mechanical properties of hybrid nonwoven mats are subject to uncertainties due to nonuniformity of fiber orientations during production procedures, which restricts their use in structural components. This study investigates the characteristics of various types of hybrid nonwoven mats with different volume fractions and fiber orientation distributions. Notably, the investigation yields noteworthy tensile properties, exemplified by a tensile strength of 367.20 MPa and a stiffness of 32.85 GPa, when the volume fraction is 28.9%. Tensile properties of the hybrid nonwoven mats and their fluctuations are then analyzed through statistical analyses. A three-dimensional microstructure-based fiber network model is established using the Monte Carlo method to represent the statistical and stochastic properties of the material. Furthermore, a prediction method of tensile properties based on microstructural properties in the model is proposed, which shows good agreement with all experimental results for adjustments of the volume fraction of carbon fibers, distribution of fiber lengths, and anisotropy of the materials, significantly reducing the experimental burden for optimization research. The tensile characteristics of composites are reliably anticipated with a nominal error margin below 10%. Additionally, this prediction method enables the prediction of scatter and stable standard deviation, making it possible to apply hybrid nonwoven mats in recycling and industrial processes.

Rock Mass Structure Classification of Caves Based on the 3D Rock Block Index

In large-scale water conservancy and hydropower projects, complex rock structures are considered to be the main factor controlling the stability of hydraulic structures. The classification of rock mass structure plays an important role in the safety of all kinds of large buildings, especially underground engineering buildings. As a quantitative classification index of rock mass, the rock block index is very common in the classification of borehole and dam foundation rock mass structures. However, there are few studies on the classification of underground engineering rock masses. Moreover, their classification criteria have disadvantages in spatial dimension. Therefore, this paper takes the long exploratory cave CPD1 in the water transmission and power generation system of the Qingtian pumped storage power station in Zhejiang Province as the research object and launches a study on the structural classification of the rock mass of a flat cave based on the 3D rock block index. According to the group distribution of joints, the sections are statistically homogeneous. Additionally, the Monte Carlo method is used to carry out random simulations to generate a three-dimensional joint network model. The virtual survey lines are arranged along the center of the shape of the three different orthogonal planes of the 3D joint network model to represent the boreholes, and the RBI values of the virtual survey lines on each orthogonal plane are counted to classify the rock mass structure of the flat cave in a refined manner using the rock block index of the rock mass in 3D. The above method realizes the application of the 3D rock block index in underground engineering and overcomes the limitations of traditional rock mass classification methods in terms of classification index and dimension. The results show that: (1) Three-dimensional joint network simulations built on statistical and probabilistic foundations can visualize the structure of the rock mass and more accurately reflect the structural characteristics of the actual rock mass. (2) Based on the 3D rock block index, the rock mass structure of the long-tunnel CPD1 is classified, from that of a continuous structure to a blocky structure, corresponding to the integrity of the rock mass from complete to relatively complete. The classification results are consistent with the evaluation results of horizontal tunnel seismic wave geophysical exploration. (3) Based on the 3D joint network model, it is reasonable and feasible to use the 3D rock block index as a quantitative evaluation index to determine the structure type of flat cave rock masses. The above method is helpful and significant in the classification of underground engineering rock mass structures.

Using time-series imagery and 3DLSTM model to classify individual tree species

ABSTRACT Classification of individual tree species (ITS) is critical for fine-scale forest surveys. However, it is difficult to obtain the complete and high-precision data needed for ITS classification in large areas. Lower spatial resolution time-series imagery is more accessible than other types of imagery and contains rich phenological information. In this study, after delineating individual tree crowns using a 0.2-m unmanned aerial vehicle (UAV) image, we used 3-m time-series imagery and a new 3DLSTM model to identify ITS at the Gaofeng Forest Farm in Guangxi Province, China. The 3DLSTM ITS classification model combines three-dimensional convolutional neural network (3D CNN) and long short-term memory (LSTM) models; thus, spatial, multiband, and time-series information can be extracted simultaneously to identify ITS more accurately. In this study, when only 3-m Planet time-series imagery was used for classification, the 3DLSTM model offered an ITS classification accuracy of 92.68%, outperforming two ITS classifiers (DenseNet or AlexNet model) based on one individual image. Moreover, the 3DLSTM model was better at identifying broad-leaved tree species than other deep-learning models. The experimental results proved that ITS classification could be significantly improved using only 3DLSTM and time-series images, offering the possibility of classifying large-scale ITS at a low cost.

Exploring nonlinear responses of lake nutrients and algal blooms to restoration measures: A three-dimensional flux network modelling approach

Excessive anthropogenic nitrogen (N) and phosphorus (P) inputs accelerate harmful algal blooms (HABs) and pose a significant global threat to lake ecosystems. Despite the adoption of extensive restoration measures to improve water quality, their effects on nutrient cycles and the underlying pathways remain unclear. In this study, a three-dimensional nutrient flux modelling approach was developed based on a validated Environmental Fluid Dynamics Code (EFDC) model and the Spatiotemporal Integral for Nutrient Networks (SINN) to compute nutrient storages and fluxes in lakes. We then conducted a scenario analysis, composed of 10 schemes, to assess the eight restoration measures in Lake Dianchi, China. The results showed that our model exhibited a high level of agreement with observed data and fluxes reported in previous studies. Of all the measures implemented, pollution interception by lakeshore wastewater treatment plants (WWTPs) accounted for the largest historical improvements in water quality, with reductions of total nitrogen (TN) by 9.3%, total phosphorus (TP) by 36.4%, and chlorophyll a (Chla) by 18.2%. Furthermore, combined implementation of transboundary water transfer and wastewater diversion contributed to a reduction of 13.3% in TN, 11.5% in TP, and 6.2% in Chla. To achieve greater improvements in water quality, efforts must be made to improve the emission standard of WWTPs and ensure compliance with water quality standard of inflow rivers. The cyanobacteria biomass catch can only be used as a supplementary approach to restrict the HAB peaks. The nonlinear responses of lakes were characterized by large loading abatement but slight changes in water quality. Through the SINN approach, we revealed that weakened internal cycling was the primary driver of water quality improvements in all restoration measures. Further analysis demonstrated that measures targeting reducing external loadings significantly weakened internal cycling processes. In contrast, measures targeting the internal loadings, such as sediment dredging, exhibited limited effectiveness in reducing the intensity of internal cycling and Chla in the lake. For eutrophic lakes with substantial legacy nutrients, the abatement of external loadings remains the principal strategy. The approach presented in this study can be used to assess and optimize lake restoration measures.