Depth Data Research Articles

Spatial coal seam modeling and reserve estimation using geostatistical methodologies by GIS technique in a selected coalfield in Ermenek Basin, Turkey

The accurate and reliable calculation of tonnage and quality of coal is vital for the success of mining projects. The Ermenek coal basin is located in the Western Taurus Mountains in the southern part of Turkey, which covers an area of approximately 620 km2 and is a Neogene coal basin. This paper generates spatial modeling of Pamuklu-Tepebaşı coal seams in the Ermenek Basin and the prediction of coal reserve by using geostatistical methods with Ordinary Kriging (OK) based on the GIS technique. The data used in the study were obtained from 82 boreholes belonging to a private company that has actively produced coal in the Pamuklu-Tepebaşı coalfield. In this study, the OK method was applied to spatial coal seam modeling for comparison of the performance of 11 distinct models. Using the rational quadratic models better explained the spatial structure of the coal seam depth data. The nugget-sill ratio indicated high spatial dependency with 0.01 and 0.02 for coal seam upper-lower surfaces, respectively. The kriged coal depth maps for upper and lower coal seam surfaces have been created using ArcGIS 10.2 software. As a result of these analyses, it was calculated that there is a coal reserve of 3.734.017 m3 using the determined surfaces with OK in the study area. Furthermore, developed spatial distribution maps in the study area could be helpful to plan further exploration activities.

Aerosol influence on cloud macrophysical and microphysical properties over the Tibetan Plateau and its adjacent regions.

This study uses aerosol optical depth (AOD) and cloud properties data to investigate the influence of aerosol on the cloud properties over the Tibetan Plateau and its adjacent regions. The study regions are divided as the western part of the Tibetan Plateau (WTP), the Indo-Gangetic Plain (IGP), and the Sichuan Basin (SCB). All three regions show significant cloud effects under low aerosol loading conditions. In WTP, under low aerosol loading conditions, the effective radius of liquid cloud particles (LREF) decreases with the increase of aerosol loading, while the effective radius of ice cloud particles (IREF) and cloud top height (CTH) increase during the cold season. Increased aerosol loading might inhibit the development of warm rain processes, transporting more cloud droplets above the freezing level and promoting ice cloud development. During the warm season, under low aerosol loading conditions, both the cloud microphysical (LREF and IREF) and macrophysical (cloud top height and cloud fraction) properties increase with the increase of aerosol loading, likely due to higher dust aerosol concentration in this region. In IGP, both LREF and IREF increase with the increase in aerosol loading during the cold season. In SCB, LREF increases with the increase in aerosol loading, while IREF decreases, possibly due to the higher hygroscopic aerosol concentration in the SCB during the cold season. Meteorological conditions also modulate the aerosol-cloud interaction. Under different convective available potential energy (CAPE) and relative humidity (RH) conditions, the influence of aerosol on clouds varies in the three regions. Under low CAPE and RH conditions, the relationship between LREF and aerosol in both the cold and warm seasons is opposite in the WTP: LREF decreases with the increase of aerosol in the cold season, while it increases in the warm season. This discrepancy may be attributed to a difference in the moisture condition between the cold and warm seasons in this region. In general, the influence of aerosols on cloud properties in TP and its adjacent regions is characterized by significant nonlinearity and spatial variability, which is likely related to the differences in aerosol types and meteorological conditions between different regions.

An Ensemble Learning Model Reveals Accelerated Reductions in Snow Depth Over Arctic Sea Ice Under High‐Emission Scenarios

AbstractThere are significant differences in snow depth predictions among different earth system models, and many models underestimate snow depth, restricting their application. Here, major factors influencing snow depth changes in the Coupled Model Intercomparison Project Phase 6 (CMIP6) were identified and evaluated. Based on satellite‐derived snow depth and CMIP6 data, an ensemble learning model based on multiple deep learning methods (hereafter referred to as the Multi‐DL model) was developed to predict future snow depth. According to satellite observations and two Operation IceBridge products, the Multi‐DL model yielded root mean square errors of 7.48, 6.20, and 6.17 cm. A continuous decrease in snow depth was observed from 2002 to 2100, and the rate of decrease accelerated with increasing emissions. Under the highest emission scenario, the first snow‐free year occurred in 2047, within the same decade as the first ice‐free year (2056). The predicted warm season snow depth was sensitive to sea ice velocity, sea ice concentration (siconc), precipitation, sea surface temperature (tos) and albedo, while the predicted cold season snow depth was sensitive to tos, air temperature, and siconc. The above parameters introduce some snow depth uncertainty. This method provides new ideas for predicting snow depth, and the generated snow depth records can provide data support for formulating Arctic‐related policies.

Improved flood depth estimation with SAR image, digital elevation model, and machine learning schemes

Study regionNhat Le river basin, the floodplains of Central coast of Vietnam. Study focusFlood disasters have a significant impact on population and economies worldwide. To accurately assess flood damage, it is crucial to estimate the water depth and predict the potential spread of damage. However, conventional methods of estimating flood depth can be time-consuming and costly due to the large amount of data required from hydraulic and rain-runoff models. To overcome these limitations, we propose a new approach that integrates a machine learning (ML) algorithm with a floodwater depth estimation tool (FwDET) based on SAR imagery and digital elevation models (DEMs). To evaluate our approach, we trained and tested six ML regression algorithms using 804 sets of inventory flood depth data from the historic flood on the Nhat Le River in Vietnam in October 2020 as the dependent variable. This approach significantly improves flood depth prediction accuracy, enabling a faster and more cost-effective estimation of flood damage. New hydrological insights for the regionOur research shows that by combining ML with FwDET, we were able to significantly enhance water depth prediction accuracy. We observed a 25.05% increase in the R2 coefficient and a 13.46% increase in the R coefficient. We recommend integrating satellite imagery, DEM, and ML for flood depth predictions, especially for floodplains with similar topographical conditions as the Nhat Le basin. This approach offers a practical, cost-effective, and near real-time solution for flood response. Implementing this approach can help improve flood damage assessment and mitigate the potential impacts of flood disasters on communities and economies.

Genetic Algorithm Based Path Planning for Seawater Depth Data Measurement in Real Scenarios

In this paper, firstly, the movement direction of the measuring ship is clarified, and according to the requirements of factors such as the range of overlap rate is controlled at 10% to 20% and the sea area is completely covered, the optimization equation system is derived, and the objective function is derived through the requirements, and the optimal solution can be derived from the two associations as the 34 measuring lines and the total length of 68 nautical miles of the shortest measuring lines. Using genetic algorithm to examine the real scene of seawater depth data measurement path planning, the length of the measurement line and coverage as an assessment index of the algorithm's subsequent optimization of the object of choice. Through the simulation of genetic mutation, crossover and selection operations, the total length of the shortest survey line is 21.5261 nautical miles, the percentage of missed sea area is 1.6%, and the overlap rate is 13.12% after the iterative solution of the genetic algorithm.

Design, Placement, and Laboratory and Field Evaluation of Rejuvenated Cold Recycled Asphalt Mixtures

Cold recycled (CR) pavements, the combination of reclaimed asphalt pavement and a recycling agent (foamed or emulsified asphalt) at ambient temperatures, can provide a sustainable method to rehabilitate, reconstruct, and construct roadways. However, CR pavement is not a one-for-one replacement of typical asphalt base mixtures. To utilize CR pavements as an alternative to a typical asphalt base mixture or surface course, additional materials such as rejuvenating agents may be needed. This research focuses on the formulation, plant production, and placement on the National Center for Asphalt Technology pavement test track off-ramp, and laboratory evaluation of three rejuvenated CR mixtures, one foamed asphalt CR mixture, and an engineered emulsion CR mixture. These mixtures were subjected to strength and moisture susceptibility testing, along with the indirect tensile asphalt cracking test (IDEAL-CT) and high-temperature indirect tensile test (HT-IDT). The influence of laboratory production versus plant production and the impact of rejuvenators on mixture performance and density were investigated. International roughness index and rut depth data for the off-ramp sections indicate adequate performance, except for the anionic emulsion with bio-based rejuvenator section. Laboratory test results indicate that the addition of rejuvenators into two of the three CR mixtures increased density and decreased moisture susceptibility and increased the CTIndex by at least 35% compared to non-rejuvenated CR mixtures. HT-IDT strengths revealed that two of the three rejuvenated mixtures performed slightly below the non-rejuvenated mixtures.

Attribution Analysis of Discharge Variations in the Inner Mongolia Basin of the Yellow River

Climate change and human activities are the main causes of discharge variation. Therefore, calculating the contribution of climate change and human activities to discharge variation in the Inner Mongolia Basin of the Yellow River can provide a reference basis for alleviating water resources stress in the Inner Mongolia Basin of the Yellow River. Firstly, based on the water balance theory and the Budyko hypothesis-based evapotranspiration formula, this study constructed a coupled water and heat balance equation and calibrated and validated the model using observed annual discharge depth data from the Toudao Guai hydrological station. The results show that the water-heat coupling model has good simulation accuracy at the annual scale, with NSE=0.76, RMSE=10.45, and RRMSE=19.5%, and overall, the simulated discharge results are close to the observed values. Secondly, this study also used the Mann-Kendall test method to determine the abrupt change points of discharge and analyzed the contribution rates of various meteorological factors. The results indicate that precipitation accounts for 88.79% of the discharge variation in the Inner Mongolia segment of the Yellow River basin, followed by land surface changes at 13.6%, while potential evaporation has the smallest contribution to discharge variation.

TransPose: 6D object pose estimation with geometry-aware Transformer

Efficient and accurate estimation of objects’ pose is essential in numerous practical applications. Due to the depth data contains abundant geometric information, some existing methods devote to extract features from 3D point cloud. However, these depth-based methods focus on extracting the point cloud local features and consider less about the global information. How to extract and utilize the local and global geometry features in depth information is crucial to achieve accurate predictions. To this end, we propose TransPose, a novel 6D pose framework that exploits Transformer Encoder with geometry-aware module to develop better learning of point cloud feature representations. To better extract local geometry features, we finely design the graph convolution network-based feature extractor that first uniformly sample point cloud and extract point pair features of point cloud. To further improve robustness to occlusion, we adopt Transformer to perform the propagation of global information, making each local feature obtains global information. Moreover, we introduce geometry-aware module in Transformer Encoder, which to form an effective constrain for point cloud feature learning and makes the global information exchange more tightly coupled with point cloud tasks. Extensive experiments indicate the effectiveness of TransPose, our pose estimation pipeline achieves competitive results on three benchmark datasets.

Exploring Ancillary Parameters for Quantifying Interpolation Uncertainty in Digital Bathymetric Models

The oceans remain one of Earth’s most enigmatic frontiers, with approximately 75% of the world’s oceans still unmapped. To create a seamless digital bathymetric model (DBM) from sparse bathymetric datasets, interpolation is employed, but this introduces uncertainties of unknown nature. This study aims to estimate and characterize these uncertainties, which is important in many fields, particularly nautical charting, and navigational safety. Complete seafloor coverage sonar depth data for five testbeds that varied in slope and roughness are sampled at a range of densities (1% to 50%) and interpolated across an entire area using Spline, Inverse Distance Weighting (IDW), and Linear interpolation. The resulting uncertainties are evaluated from both scientific and operational perspectives. Employing linear regression and machine learning techniques, the relationships between these uncertainties and ancillary parameters (distance to the nearest measurement, seabed roughness, and slope) are examined for quantifying and characterizing the interpolation uncertainty. Results indicate insignificant operational differences among the three interpolators in depth estimation, as well as the statistical significance of the examined uncertainty predictors. Additionally, findings suggest the potential presence of unaccounted-for factors shaping uncertainty, yet this work lays a foundational understanding for improving the estimate of uncertainty in DBMs.

Short-term load forecasting based on CEEMDAN and dendritic deep learning

Short-term load forecasting (STLF) plays a key role in improving the operational efficiency of real-time electricity markets by reducing load uncertainty, optimising the allocation of generation resources and increasing the reliability of electricity supply. Despite significant advances in the field, achieving high accuracy in STLF remains a challenge due to the inherent complexity and volatility of load data. This study proposes a novel STLF method that synergistically combines the strengths of a dendritic neuron model (DNM), long short-term memory, and complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) to address these challenges. First, historical load data is decomposed into components of different frequencies using CEEMDAN. Then, LSTM is used to extract the temporal characteristics of these load subsequence components in combination with various influencing factors such as temperature and humidity. Uniquely, this study innovatively integrates a dendritic neuron model instead of the traditional fully connected layer, a structural innovation that enhances the model’s ability to explore the intrinsic correlations of the time series data in depth. This allows the model to process and predict each component individually, and then later reconstruct and sum these independent prediction results to obtain a final prediction value with higher accuracy. Experiments are conducted on the Panama electricity load dataset to compare the proposed dendritic learning model with several state-of-the-art models, including CrossFormer and Transformer. The results confirm that the proposed model has significantly better prediction accuracy.

Effects of ferrite and graphite phases on scratch characteristics of nodular cast iron

In this work, we conduct micro-scratch tests of nodular cast iron with Vickers indenter under constant normal load and investigate the effects of individual phases (i.e., ferrite and graphite phases) on the scratch characteristics of nodular cast iron. Using a bimodal Gaussian model to fit the experimental data of penetration depth, residual depth, elastic recovery rate, and scratch hardness, we calculate the contribution percentages of individual phases to scratch variables. The scratch variables are independent of the scratch speed used in this work, while the penetration depth increases linearly with increasing the normal load. There exists a large difference between the contribution percentages and the volume fractions calculated from SEM images and XRD pattern. Such a large difference suggests that the bimodal Gaussian distribution analysis does not include the effect of the interaction between two phases on the scratch characteristics of dual-phase materials with significantly distinct mechanical properties and geometrical dimensions. A more refined method taking into account the interaction between two phases needs to be developed for the analysis of the scratch characteristics of dual-phase materials.

Learning Complementary Correlations for Depth Super-Resolution With Incomplete Data in Real World.

Depth information is a significant ingredient to visually perceive the physical world. However, mainstream depth sensors, e.g., time-of-flight (ToF) cameras, often measure incomplete and low-resolution depth data, resulting in low-quality visual perception. In this article, we try to address a potentially valuable task, i.e., depth super-resolution (DSR) with incomplete data, which recovers dense and high-resolution depth map from incomplete and low-resolution one. To tackle this task, we introduce a novel incomplete DSR (IDSR) framework, including a primary branch for DSR to recover high-frequency details, and an auxiliary branch for depth completion (DC) to fill missing pixels. More importantly, we propose two modules, joint correlation learning (JCL) and iterative-cross (IC), to enhance the learning of complementary information flows between the two branches. The former module aims to learn the correlative relationships of the two branches, whilst the latter module adequately fuses higher level representations for more precise predictions. Extensive experiments show that our framework is effective and achieves the state-of-the-art performance on the real-world RGB-D-D and the synthetic NYUv2 datasets.

Processes explaining increased ocean dynamic sea level in the North Sea in CMIP6

Ocean dynamic sea level (ODSL) is expected to be one of the major contributors to sea level rise in the North Sea during the 21st century. This component is defined as the spatial sea level anomaly due to ocean currents, wind stresses and local thermosteric and halosteric effects. Climate models from CMIP5 and CMIP6 show a large spread, as well as an increase between CMIP5 and CMIP6 North Sea ODSL projections. In this study, we apply linear regression models on CMIP5 and CMIP6 data to get a better understanding of the processes that influence ODSL change in the North Sea. We find that neither global surface air temperature nor global mean thermosteric sea level can reproduce ODSL projections based on a linear relation in CMIP6, whereas this was the case for CMIP5. Including the strength of the Atlantic meridional overturning circulation (AMOC) as an additional predictor enables us to reproduce long-term changes in ODSL for both ensembles. The sensitivity to the AMOC increased in CMIP6, which points to a difference in model dynamics between CMIP5 and CMIP6, and a more important role of the deep ocean. To investigate this further, we analyse mixed layer depth data in the North Atlantic. We find that models with a relatively deep mixed layer in the Greenland Sea over the period 1985–2004, project larger rise in ODSL in the North Sea for both CMIP5 and CMIP6. This implies that the location of deep water formation in the North Atlantic potentially influences ODSL in the North Sea. The number of these models increased from CMIP5 to CMIP6, again pointing to a different sensitivity to larger scale processes, potentially explaining the difference between the two ensembles.

A Review Work: Human Action Recognition in Video Surveillance Using Deep Learning Techniques

Despite being extensively used in numerous uses, precise and effective human activity identification continues to be an interesting research issue in the area of vision for computers. Currently, a lot of investigation is being done on themes like pedestrian activity recognition and ways to recognize people's movements employing depth data, 3D skeletal data, still picture data, or strategies that utilize spatiotemporal interest points. This study aims to investigate and evaluate DL approaches for detecting human activity in video. The focus has been on multiple structures for detecting human activities that use DL as their primary strategy. Based on the application, including identifying faces, emotion identification, action identification, and anomaly identification, the human occurrence forecasts are divided into four different subcategories. The literature has been carried several research based on these recognitions for predicting human behavior and activity for video surveillance applications. The state of the art of four different applications' DL techniques is contrasted. This paper also presents the application areas, scientific issues, and potential goals in the field of DL-based human behavior and activity recognition/detection.

Keep DRÆMing: Discriminative 3D anomaly detection through anomaly simulation

Recent surface anomaly detection methods rely on pretrained backbone networks for efficient anomaly detection. On standard RGB anomaly detection benchmarks these methods achieve excellent results but fail on 3D anomaly detection due to a lack of pretrained backbones that suit this domain. Additionally, there is a lack of industrial depth data that would enable the backbone network training that could be used in 3D anomaly detection models. Discriminative anomaly detection methods do not require pretrained networks and are trained using simulated anomalies. The process of simulating anomalies that fit the domain of industrial depth data is not trivial and is necessary for training discriminative methods. We propose a novel 3D anomaly simulation process that follows the natural characteristics of industrial depth data and generates diverse deformations, making it suitable for training discriminative anomaly detection methods. We demonstrate its effectiveness by adapting the DRÆM method to work on 3D anomaly detection, thus obtaining 3DRÆM, a strong discriminative 3D anomaly detection model. The proposed approach achieves excellent results on the MVTec3D anomaly detection benchmark where it achieves state-of-the-art results on both 3D and RGB+3D problem setups, significantly outperforming competing methods.

GNSS-R snow depth retrieval algorithm based on PSO-LSTM

The global navigation satellite system (GNSS)-interferometric reflectometry technique has been applied to retrieve snow depth, which has a high potential for application. The GNSS reflectometry classical algorithm retrieves the snow depth by extracting the frequency of the multipath signal and substituting it into an empirical formula. However, the retrieval errors of high and low snow depths are large due to the influence of factors such as surface vegetation and terrain environment. In this paper, we propose a snow depth retrieval algorithm based on a particle swarm optimized long short-term memory (PSO-LSTM) neural network. The algorithm extracted three characteristic parameters (frequency, amplitude, and phase) from the signal-to-noise ratio (SNR) data as inputs, and optimized the LSTM hyperparameters by the PSO algorithm to improve the retrieval accuracy for low snow depths and snow depths close to the antenna. The snow depth retrieval results of global positioning system L1 band SNR data collected from the P351 station in 2022 and AB33 station in 2017 were evaluated in this paper. The snow depth retrieval results of the PSO-LSTM algorithm for P351 station were in high agreement with the snow depth data provided by the snowpack telemetry network; the coefficient of determination ( R2 ) reached 0.986, and the root mean square error (RMSE) and mean absolute error (MAE) were 7.30 cm and 4.94 cm, respectively. Compared with the classical algorithm, the PSO-LSTM algorithm decreased the RMSE and MAE by 53.0% and 30.4% for the retrieval results of snow depths below 15 cm at the P351 station, and by 76.8% and 84.4% for the retrieval results of snow depths above 117 cm from the 1st day to the 137th day, respectively. Similarly, the RMSE, MAE, and R2 for the 2017 retrieval results at AB33 station were 5.90 cm, 4.25 cm, and 0.965, respectively. Compared with the classical algorithm, the PSO-LSTM algorithm decreased the RMSE and MAE by 47.9% and 33.0% for the retrieval results of snow depths below 15 cm at the AB33 station, and by 75.4% and 82.3% for the retrieval results of snow depths above 56 cm from the day 46 to day 120. In addition, the snow depth retrieval algorithm was proposed in this paper does not require antenna height and empirical formulas to realize snow depth retrieval, and at the same time, the algorithm effectively improved the retrieval accuracy for both high and low snow depths with strong robustness.

Keypoint Fusion for RGB-D Based 3D Hand Pose Estimation

Previous 3D hand pose estimation methods primarily rely on a single modality, either RGB or depth, and the comprehensive utilization of the dual modalities has not been extensively explored. RGB and depth data provide complementary information and thus can be fused to enhance the robustness of 3D hand pose estimation. However, there exist two problems for applying existing fusion methods in 3D hand pose estimation: redundancy of dense feature fusion and ambiguity of visual features. First, pixel-wise feature interactions introduce high computational costs and ineffective calculations of invalid pixels. Second, visual features suffer from ambiguity due to color and texture similarities, as well as depth holes and noise caused by frequent hand movements, which interferes with modeling cross-modal correlations. In this paper, we propose Keypoint-Fusion for RGB-D based 3D hand pose estimation, which leverages the unique advantages of dual modalities to mutually eliminate the feature ambiguity, and performs cross-modal feature fusion in a more efficient way. Specifically, we focus cross-modal fusion on sparse yet informative spatial regions (i.e. keypoints). Meanwhile, by explicitly extracting relatively more reliable information as disambiguation evidence, depth modality provides 3D geometric information for RGB feature pixels, and RGB modality complements the precise edge information lost due to the depth noise. Keypoint-Fusion achieves state-of-the-art performance on two challenging hand datasets, significantly decreasing the error compared with previous single-modal methods.

Data Fusion of RGB and Depth Data with Image Enhancement.

Since 3D sensors became popular, imaged depth data are easier to obtain in the consumer sector. In applications such as defect localization on industrial objects or mass/volume estimation, precise depth data is important and, thus, benefits from the usage of multiple information sources. However, a combination of RGB images and depth images can not only improve our understanding of objects, capacitating one to gain more information about objects but also enhance data quality. Combining different camera systems using data fusion can enable higher quality data since disadvantages can be compensated. Data fusion itself consists of data preparation and data registration. A challenge in data fusion is the different resolutions of sensors. Therefore, up- and downsampling algorithms are needed. This paper compares multiple up- and downsampling methods, such as different direct interpolation methods, joint bilateral upsampling (JBU), and Markov random fields (MRFs), in terms of their potential to create RGB-D images and improve the quality of depth information. In contrast to the literature in which imaging systems are adjusted to acquire the data of the same section simultaneously, the laboratory setup in this study was based on conveyor-based optical sorting processes, and therefore, the data were acquired at different time periods and different spatial locations. Data assignment and data cropping were necessary. In order to evaluate the results, root mean square error (RMSE), signal-to-noise ratio (SNR), correlation (CORR), universal quality index (UQI), and the contour offset are monitored. With JBU outperforming the other upsampling methods, achieving a meanRMSE = 25.22, mean SNR = 32.80, mean CORR = 0.99, and mean UQI = 0.97.

Enhancing automated lower limb rehabilitation exercise task recognition through multi-sensor data fusion in tele-rehabilitation

BackgroundTele-rehabilitation is the provision of physiotherapy services to individuals in their own homes. Activity recognition plays a crucial role in the realm of automatic tele-rehabilitation. By assessing patient movements, identifying exercises, and providing feedback, these platforms can offer insightful information to clinicians, thereby facilitating an improved plan of care. This study introduces a novel deep learning approach aimed at identifying lower limb rehabilitation exercises. This is achieved through the integration of depth data and pressure heatmaps. We hypothesized that combining pressure heatmaps and depth data could improve the model’s overall performance.MethodsIn this study, depth videos and body pressure data from an accessible online dataset were used. This dataset comprises data from 30 healthy individuals performing 7 lower limb rehabilitation exercises. To accomplish the classification task, three deep learning models were developed, all based on an established 3D-CNN architecture. The models were designed to classify the depth videos, sequences of pressure data frames, and combination of depth videos and pressure frames. The models’ performance was assessed through leave-one-subject-out and leave-multiple-subjects-out cross-validation methods. Performance metrics, including accuracy, precision, recall, and F1 score, were reported for each model.ResultsOur findings indicated that the model trained on the fusion of depth and pressure data showed the highest and most stable performance when compared with models using individual modality inputs. This model could effectively identify the exercises with an accuracy of 95.71%, precision of 95.83%, recall of 95.71%, and an F1 score of 95.74%.ConclusionOur results highlight the impact of data fusion for accurately classifying lower limb rehabilitation exercises. We showed that our model could capture different aspects of exercise movements using the visual and weight distribution data from the depth camera and pressure mat, respectively. This integration of data provides a better representation of exercise patterns, leading to higher classification performance. Notably, our results indicate the potential application of this model in automatic tele-rehabilitation platforms.

The Distribution and Evolution of Groundwater Level Depths and Groundwater Sustainability in the Hexi Corridor over the Last Five Years

Groundwater overexploitation for agricultural irrigation is prone to lead to numerous ecological concerns. This study delved into the present distribution and recent trend of groundwater levels in the plain areas of the Hexi Corridor in Northwest China according to the groundwater level depth (GWD) data from 264 monitoring wells in the Shiyang River Basin (SYB) and 107 in the Shule River Basin (SLB), recorded annually in April from 2019 to 2023. The key findings include the following: (1) Over the five-year span, the SYB’s GWD experienced change rates (CRs) ranging from −12.17 to 9.11 m/a (average: −0.13 m/a), with the number of monitoring wells showing increased and decreased GWDs accounting for 50% and 50%, respectively. By contrast, the SLB’s GWD exhibited CRs ranging from −1.87 to 2.06 m/a (average: 0.01 m/a), with the number of monitoring wells showing increased and decreased GWDs accounting for 52% and 48%, respectively; (2) the Wuwei (CR = 0.09 m/a) and Changning (0.58 m/a) basins in the SYB and the Yumen (0.06 m/a), Guazhou (0.05 m/a), and Huahai (0.03 m/a) basins in the SLB, witnessed rising groundwater levels. In contrast, the Minqin Basin (0.09 m/a) in the SYB and the southern Dunhuang Basin (0.04 m/a) in the SLB witnessed declines in the groundwater levels; (3) The groundwater sustainability assessment showed that the groundwater is still extremely unsustainable. This study’s insights are instrumental in targeted treatment, as well as the preparation and adjustment of sustainable groundwater protection strategies.