Spatio-temporal Model Research Articles

Exposure to air pollution and non-neoplastic digestive system diseases: findings from the China health and retirement longitudinal study

ObjectivesIncreasing concern about air pollution’s impact on public health underscores the need to understand its effects on non-neoplastic digestive system diseases (NNDSD). This study explores the link between air pollution and NNDSD in China.MethodsWe conducted a national cross-sectional study using 2015 data from the China Health and Retirement Longitudinal Study (CHARLS), involving 13,046 Chinese adults aged 45 and above from 28 provinces. Satellite-based spatiotemporal models estimated participants’ exposure to ambient particulate matter (3-year average). An analysis of logistic regression models was conducted to estimate the association between air pollutants [particulate matter with a diameter ≤ 2.5 μm (PM2.5) or ≤10 μm (PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2), ozone (O3), and carbon monoxide (CO)] and NNDSD. Interaction analyses were conducted to examine potential modifiers of these associations.ResultsThe prevalence of NNDSD among participants was 26.29%. After adjusted for multivariate factors, we observed a 6% [odd ratio (OR) = 1.06, 95% confidence interval (CI): 0.94, 1.19], 23% (OR = 1.23, 95% CI: 1.09, 1.38), 26% (OR = 1.26, 95% CI: 1.12, 1.41), 30% (OR = 1.30, 95% CI: 1.16, 1.46), 13% (OR = 1.13, 95% CI: 1.01, 1.27) and 27% (OR = 1.27, 95% CI: 1.13, 1.43) increase in NNDSD risk with an interquartile range increase in PM2.5 (23.36 μg/m3), PM10 (50.33 μg/m3), SO2 (17.27 μg/m3), NO2 (14.75 μg/m3), O3 (10.80 μg/m3), and CO (0.42 mg/m3), respectively. Interaction analyses showed that PM2.5, SO2, and O3 had stronger effects on NNDSD risk among older adults, highly educated individuals, smokers, and married people, respectively.ConclusionThis study demonstrates that long-term exposure to PM2.5, PM10, SO2, NO2, O3, and CO is positively associated with NNDSD risk in Chinese adults aged 45 and above. Implementing intervention strategies to enhance air quality is essential for reducing the burden of NNDSD.

Physics-guided spatiotemporal approach for melt pool size prediction in laser powder bed fusion

ABSTRACT This study first proposes a physics-guided spatiotemporal graph. The melt pool temporal feature nodes and spatial feature nodes are extracted from melt pool images based on spatiotemporal correlations. These feature points are treated as graph nodes, and the edge weights in the graph are calculated based on an analytical heat transfer model. It supplements the physical information of accumulated heat. Based on this, the graph convolutional network (GCN) is used to extract spatiotemporal information of heat transfer, and the long short-term memory (LSTM) is used to capture the evolution patterns of the melt pool. Thus, a melt pool spatiotemporal feature extraction model is constructed for melt pool size prediction. The model achieves prediction errors within 12% for the melt pool area and 8.5% for the melt pool perimeter. Subsequently, interpretability analysis indicates that the physics-guided approach enhances the model's learning ability for the cumulative contribution of heat from neighboring tracks.

Dynamic inferences of crash risks in freeway merging zones: a spatio-temporal deep learning model

Freeway merging zones are critical for freeway operations and management due to potential crashes arising from complex vehicle merging behaviours. This paper investigates the application of a spatio-temporal deep learning model to infer crash risks in the zones. We first introduce a crash risk index based on Time-To-Collision and vehicle merging patterns. An innovation is that the developed spatio-temporal transformer model can analyze the evolving risk index. This model effectively captures dynamic risk features through a multi-head attention mechanism within its spatio-temporal learning components. Numerical experiments on nine inference tasks with varying spatial resolutions show improved performance of the model with lower resolution. Moreover, the ST-Transformer model is benchmarked against three advanced deep learning models, which consistently demonstrates its superiority in capturing spatio-temporal dependence in risk sequences. This investigation significantly contributes to a richer understanding of proactive traffic safety, providing valuable insights for advanced freeway management and driver assistance systems.

A Bayesian spatio-temporal model for cluster detection: identifying HPV suboptimal vaccine coverage

AbstractHuman papillomavirus (HPV) is a common sexually transmitted virus responsible for several types of cancer. HPV vaccines have been included in the systematic vaccination of the Valencia Region since 2008. Despite clinical agreement on the safety and effectiveness of the vaccines, vaccination coverage remains suboptimal in many areas. In order to facilitate the implementation of targeted strategies to enhance vaccination coverage, we develop here an easy-to-implement spatio-temporal clustering model that identifies groups of health districts that share similar behaviour. Namely, health districts are clustered twice. First, they are grouped into spatial clusters based on their underlying vaccination coverage that remains constant over time. Second, they are clustered according to their evolution of annual vaccination rates. A feature of our model is that it does not impose that geographically neighboring areas are assigned to the same spatial cluster or the same temporal trend. This flexibility allows us to explore different spatial and temporal structures. Suboptimal HPV vaccination coverage was found in some health districts. The results reveal the presence of three spatial clusters, with estimated coverage ranging between 82 and 97%, for 14-year-old girls and five clusters for 12-year-old girls, with coverage ranging between 61 and 96%. However, most health districts in this age group show an increasing vaccination trend. The need for future efforts to identify factors causing suboptimal vaccination coverage so that they can be acted upon specifically is highlighted.

Convolutional long short-term memory neural network for groundwater change prediction

Forecasting groundwater changes is a crucial step towards effective water resource planning and sustainable management. Conventional models still demonstrated insufficient performance when aquifers have high spatio-temporal heterogeneity or inadequate availability of data in simulating groundwater behavior. In this regard, a spatio-temporal groundwater deep learning model is proposed to be applied for monthly groundwater prediction over the entire Choushui River Alluvial Fan in Central Taiwan. The combination of the Convolution Neural Network (CNN) and Long Short-Term Memory (LSTM) known as Convolutional Long Short-Term Memory (CLSTM) Neural Network is proposed and investigated. Result showed that the monthly groundwater simulations from the proposed neural model were better reflective of the original observation data while producing significant improvements in comparison to only the CNN, LSTM as well as classical neural models. The study also explored the performance of the Masked CLSTM model which is designed to handle missing data by reconstructing incomplete spatio-temporal input images, enhancing groundwater forecasting through image inpainting. The findings indicated that the neural architecture can efficiently extract the relevant spatial features from the past incomplete information of hydraulic head observations under various masking scenarios while simultaneously handling the varying temporal dependencies over the entire study region. The proposed model showed strong reliability in reconstructing and simulating the spatial distribution of hydraulic heads for the following month, as evidenced by low RMSE values and high correlation coefficients when compared to observed data.

A comparison of statistical and machine learning models for spatio-temporal prediction of ambient air pollutant concentrations in Scotland

AbstractThe spatio-temporal prediction of air pollutant concentrations is vital for assessing regulatory compliance and for producing exposure estimates in epidemiological studies. Numerous approaches have been utilised for making such predictions, including land use regression models, additive models, spatio-temporal smoothing models and machine learning prediction algorithms. However, relatively few studies have compared the predictive performance of these models thoroughly, which is one of the novel contributions of this paper. For the specific challenge of predicting monthly average concentrations of NO$$_{2}$$ 2 , PM$$_{10}$$ 10 and $$\hbox {PM}_{2.5}$$ PM 2.5 in Scotland, we find that random forests typically outperform (or are as good as) more traditional statistical prediction approaches. Additionally, we utilise the best performing model to provide a new data resource, namely, predictions of monthly average concentrations (with uncertainty quantification) of the above pollutants on a regular 1 km$$^{2}$$ 2 grid for all of Scotland between 2016 and 2020.

EDH-STNet: An Evaporation Duct Height Spatiotemporal Prediction Model Based on Swin-Unet Integrating Multiple Environmental Information Sources

Given the significant spatial non-uniformity of marine evaporation ducts, accurately predicting the regional distribution of evaporation duct height (EDH) is crucial for ensuring the stable operation of radio systems. While machine-learning-based EDH prediction models have been extensively developed, they fail to provide the EDH distribution over large-scale regions in practical applications. To address this limitation, we have developed a novel spatiotemporal prediction model for EDH that integrates multiple environmental information sources, termed the EDH Spatiotemporal Network (EDH-STNet). This model is based on the Swin-Unet architecture, employing an Encoder–Decoder framework that utilizes consecutive Swin-Transformers. This design effectively captures complex spatial correlations and temporal characteristics. The EDH-STNet model also incorporates nonlinear relationships between various hydrometeorological parameters (HMPs) and EDH. In contrast to existing models, it introduces multiple HMPs to enhance these relationships. By adopting a data-driven approach that integrates these HMPs as prior information, the accuracy and reliability of spatiotemporal predictions are significantly improved. Comprehensive testing and evaluation demonstrate that the EDH-STNet model, which merges an advanced deep learning algorithm with multiple HMPs, yields accurate predictions of EDH for both immediate and future timeframes. This development offers a novel solution to ensure the stable operation of radio systems.

Traffic Awareness-Based Target Wake Time Scheduling in 802.11ax WLANs

In recent years, the rapid growth of mobile communication has led to the continuous expansion of network scale, and the energy consumption of wireless communication has also increased rapidly. Under the high-density networking conditions, frequent communication activities greatly increase the energy consumption. In order to reduce channel competition, optimize resource scheduling and reduce equipment power consumption, the target wake-up time (TWT) mechanism in IEEE 802.11ax system plays an important role through timely scheduling and transmission. To improve channel efficiency for multi-user (MU) transmission and tackle with complicated traffic conditions, this paper presents a traffic awareness-based TWT scheduling scheme (TAT). The traffic is predicted and classified using a spatio-temporal series model for traffic-based TWT parameter generation. Subsequently, the time–frequency scheduling table is developed to adaptively assign Resource Units (RUs) and time slices to active STAs. Simulation results show that TAT can guarantee the Quality of Service (QoS) under dynamic changes in power service and delay requirements while reducing power consumption, effectively meeting the energy-saving demands in the intelligent access scenarios for power terminals.

SAM-Net: Spatio-Temporal Sequence Typhoon Cloud Image Prediction Net with Self-Attention Memory

Cloud image prediction is a spatio-temporal sequence prediction task, similar to video prediction. Spatio-temporal sequence prediction involves learning from historical data and using the learned features to generate future images. In this process, the changes in time and space are crucial for spatio-temporal sequence prediction models. However, most models now rely on stacking convolutional layers to obtain local spatial features. In response to the complex changes in cloud position and shape in cloud images, the prediction module of the model needs to be able to extract both global and local spatial features from the cloud images. In addition, for irregular cloud motion, more attention should be paid to the spatio-temporal sequence features between input cloud image frames in the temporal sequence prediction module, considering the extraction of temporal features with long temporal dependencies, so that the spatio-temporal sequence prediction network can learn cloud motion trends more accurately. To address these issues, we have introduced an innovative model called SAM-Net. The self-attention module of this model aims to extract an inner image frame’s spatial features of global and local dependencies. In addition, a memory mechanism has been added to the self-attention module to extract interframe features with long temporal and spatial dependencies. Our method shows better performance than the PredRNN-v2 model on publicly available datasets such as MovingMNIST and KTH. We achieved the best performance in both the 4-time-step and 10-time-step typhoon cloud image predictions. On a cloud dataset consisting of 10 time steps, we observed a decrease in MSE of 180.58, a decrease in LPIPS of 0.064, an increase in SSIM of 0.351, and a significant improvement in PSNR of 5.56 compared to PredRNN-v2.

Long-Term Air Pollution Exposure and Severity of Idiopathic Pulmonary Fibrosis: Data from the IPF-PRO Registry.

While exposure to air pollution is a known risk factor for adverse pulmonary outcomes, its impact in individuals with idiopathic pulmonary fibrosis (IPF) is less well understood. To investigate the effects of long-term exposure to air pollution on disease severity and progression in patients with IPF and to determine whether genomic factors, such as MUC5B promoter polymorphism or telomere length, modify these associations. We performed analyses at enrollment and after one year of follow-up in the IPF-PRO Registry, a prospective observational registry that enrolled individuals with IPF at 46 US sites from June 2014 to October 2018. Five-year average pollution exposures (PM2.5, NO2, O3) prior to enrollment date were estimated at participants' residential addresses with validated national spatio-temporal models. Multivariable regression models estimated associations between pollution exposure and physiologic measurements (FVC, DLCO, supplemental oxygen use at rest) and quality of life measurements (St. George's Respiratory Questionnaire [SGRQ], EuroQoL, Cough and Sputum Assessment Questionnaire) at enrollment. Cox proportional hazard models estimated associations between pollutants and a composite outcome of death, lung transplant, or >10% absolute decline in FVC % predicted in the year after enrollment. Models were adjusted for individual-level and spatial confounders, including proxies for disease onset. Gene-environment interactions with MUC5B and telomere length were assessed. Of 835 participants, 94% were non-Hispanic Whites, 76% were male, mean (SD) age was 70 (7.7) years. In fully adjusted analyses, higher PM2.5 exposure was associated with worse quality of life per SGRQ activity score (3.48 [95% confidence interval (CI) 0.64, 6.32] per 2µg/m3 PM2.5) and EuroQoL scores (-0.04 [95%CI -0.06, -0.01] per 2µg/m3 PM2.5), and lower FVC % predicted and lower DLCO% predicted at enrollment. Each 3 parts per billion difference in O3 exposure was associated with a 1.57% [95% CI 0.15, 2.98] higher FVC % predicted at enrollment, although this effect was attenuated in multi-pollutant models. There was no association between NO2 and enrollment measures, between pollution exposure and one-year outcomes, or evidence for gene-environment interactions. In the IPF-PRO Registry, long-term exposure to PM2.5 was associated with worse quality of life and lung function at enrollment, but not with short-term disease progression or mortality. There was no evidence of effect modification by interaction of genomic factors with pollution. The reason for the unexpected relationship between O3 exposure and higher FVC is unclear.

Abstract 4129036: Air pollution and Cardiovascular Disease Incidence in a Pooled Analysis of 6 U.S. Cohorts

Introduction/Background: Long term air pollution is associated with cardiovascular disease (CVD) incidence, however the effects at low levels in the US are unknown. Research Questions/Hypothesis: To assess the effect of long-term exposure to ambient air pollution CVD incidence (myocardial infarction [MI], stroke, and all CVD) in a pooled analysis of epidemiological cohorts. Materials and Methods: We harmonized data from six cohorts (Cardiovascular Health Study, Nurses’ Health Study II, Multi-Ethnic Study of Atherosclerosis, and Reasons for Geographic And Racial Differences in Stroke Study, Sister Study, and Women’s Health Initiative) with both CVD incidence data and extensive covariate information. We predicted time-varying two-year average concentrations of particulate matter <2.5 microns (PM 2.5 ) and nitrogen dioxide (NO 2 ) from 1990-2019 at participant residential addresses using validated regionalized spatio-temporal models, combining spatio-temporal universal kriging and partial least squares regression with inputs from monitoring data from regulatory and researcher-deployed networks, satellite remote-sensing data and chemical transport modeling. Cox proportional hazards regression models adjusted for individual and area-level information, including smoking status, socio-economic factors, anthropometry, blood pressure, cholesterol, and other clinical variables. Results: Our analysis included over 300,000 participants and 10,229 incident CVD events. The median residential concentration of PM 2.5 was 10.1 µg/m 3 and NO 2 was 9.3 ppb. A 5 µg/m 3 increase in PM 2.5 was not significantly associated with a higher hazard ratio for all outcomes: 1.03 (95%CI: 0.96, 1.10) for CVD incidence, 1.03 (95%CI 0.93, 1.15) for MI, and 1.00 (95%CI 0.90, 1.10) for stroke. We found that a 10ppb increase in NO 2 was associated with a 1.06 (95%CI: 1.01, 1.11) higher hazard ratio for CVD, 1.07 (95%CI: 1.01, 1.11) for MI, and 1.02 (95%CI 0.96, 1.09) for stroke. Conclusions: In this analysis of a large, pooled cohort with excellent exposure, outcome, and covariate information, we found an association of NO 2 exposure—considered a surrogate for traffic-related air pollution—with CVD incidence and MI. Associations with particulate matter were not statistically significant, but could not exclude a small increased risk. We are extending this analysis to include 6 additional cohorts and an additional 700,000 participants to allow for generation of a highly resolved concentration-response function.

Collaborative control framework at isolated signalized intersections under the mixed connected automated vehicles environment

AbstractThis study proposes a collaborative control framework under the mixed traffic environment of connected and automated vehicles and connected human‐driven vehicles, which can simultaneously optimize the signal timing, lane settings, and vehicle trajectories at isolated intersections. Initially, considering the dynamics of traffic demand and incompatible signals, we analyze the vehicle delay of each lane. Based on the delay analysis, the spatiotemporal resource collaborative optimization model of lane setting and signal timing is established to minimize the average delay. Subsequently, in the buffer zone, a graph‐theoretic‐based sorting and platooning model provides a clear and concise representation of the transformation process from the initial state to the target state of vehicles, enabling the platoon formation. Additionally, trajectory optimization is integrated into the collaborative control framework by the optimal control model and car‐following model in the passing zone. Simulation experiments and sensitivity analyses demonstrate the effectiveness of the proposed framework in reducing average vehicle delay, improving fuel consumption, and coping with changing traffic demand at intersections.

A Deep Learning Model for Accurate Maize Disease Detection Based on State-Space Attention and Feature Fusion

In improving agricultural yields and ensuring food security, precise detection of maize leaf diseases is of great importance. Traditional disease detection methods show limited performance in complex environments, making it challenging to meet the demands for precise detection in modern agriculture. This paper proposes a maize leaf disease detection model based on a state-space attention mechanism, aiming to effectively utilize the spatiotemporal characteristics of maize leaf diseases to achieve efficient and accurate detection. The model introduces a state-space attention mechanism combined with a multi-scale feature fusion module to capture the spatial distribution and dynamic development of maize diseases. In experimental comparisons, the proposed model demonstrates superior performance in the task of maize disease detection, achieving a precision, recall, accuracy, and F1 score of 0.94. Compared with baseline models such as AlexNet, GoogLeNet, ResNet, EfficientNet, and ViT, the proposed method achieves a precision of 0.95, with the other metrics also reaching 0.94, showing significant improvement. Additionally, ablation experiments verify the impact of different attention mechanisms and loss functions on model performance. The standard self-attention model achieved a precision, recall, accuracy, and F1 score of 0.74, 0.70, 0.72, and 0.72, respectively. The Convolutional Block Attention Module (CBAM) showed a precision of 0.87, recall of 0.83, accuracy of 0.85, and F1 score of 0.85, while the state-space attention module achieved a precision of 0.95, with the other metrics also at 0.94. In terms of loss functions, cross-entropy loss showed a precision, recall, accuracy, and F1 score of 0.69, 0.65, 0.67, and 0.67, respectively. Focal loss showed a precision of 0.83, recall of 0.80, accuracy of 0.81, and F1 score of 0.81. State-space loss demonstrated the best performance in these experiments, achieving a precision of 0.95, with recall, accuracy, and F1 score all at 0.94. These results indicate that the model based on the state-space attention mechanism achieves higher detection accuracy and better generalization ability in the task of maize leaf disease detection, effectively improving the accuracy and efficiency of disease recognition and providing strong technical support for the early diagnosis and management of maize diseases. Future work will focus on further optimizing the model’s spatiotemporal feature modeling capabilities and exploring multi-modal data fusion to enhance the model’s application in real agricultural scenarios.

Research on Radial Vibration Model and Low-Frequency Vibration Suppression Method in PMSM by Injecting Multiple Symmetric Harmonic Currents

Driven by frequency conversion, the windings of a three-phase permanent magnet synchronous motor (PMSM) contain both odd and even harmonic currents. Due to the motor’s pole–slot conductance modulation, the interaction between the magnetic fields generated by these harmonic currents and the permanent magnet field results in harmonic radial vibrations of the motor. This paper analyzes the three-phase currents of the prototype and derives the radial magnetomotive force (MMF) spatiotemporal models for symmetric harmonic currents. By integrating Maxwell’s magnetic force formula and vibration response formula, the radial vibration models for symmetric harmonic currents are developed. The characteristics of vibrations caused by odd and even harmonic currents, as well as positive sequence and negative sequence harmonic currents, are analyzed separately. A cyclic sequence, low-frequency vibration suppression control method incorporating multiple harmonic current injections was designed. Experimental results of this method are compared with those obtained using an ideal sinusoidal current. Except for the second harmonic vibration, all other vibrations are significantly suppressed, with a maximum suppression rate of 92.28%. The total vibration level is reduced by 12.7619 dB, and the average torque is reduced by 0.67% with the total harmonic distortion of the current at 2.89%. The experimental results show that the vibration method in this paper has little influence on the average torque of the motor, the current distortion rate is small, and the vibration suppression effect is good.

Towards Integrating Federated Learning with Split Learning via Spatio-temporal Graph Framework for Brain Disease Prediction.

Functional Magnetic Resonance Imaging (fMRI) is used for extracting blood oxygen signals from brain regions to map brain functional connectivity for brain disease prediction. Despite its effectiveness, fMRI has not been widely used: on the one hand, collecting and labeling the data is time-consuming and costly, which limits the amount of valid data collected at a single healthcare site; on the other hand, integrating data from multiple sites is challenging due to data privacy restrictions. To address these issues, we propose a novel, integrated Federated learning and Split learning Spatio-temporal Graph framework (FS2G). Specifically, we introduce federated learning and split learning techniques to split a spatio-temporal model into a client temporal model and a server spatial model. In the client temporal model, we propose a time-aware mechanism to focus on changes in brain functional states and use an InceptionTime model to extract information about changes in the brain states of each subject. In the server spatial model, we propose a united graph convolutional network to integrate multiple graph convolutional networks. Integrating federated learning and split learning, FS2G can utilize multi-site fMRI data without violating data privacy protection and reduce the risk of overfitting as it is capable of learning from limited training data sets. Moreover, it boosts the extraction of spatio-temporal features of fMRI using spatio-temporal graph networks. Experiments on ABIDE and ADHD200 datasets demonstrate that our proposed method outperforms state-of-the-art methods. In addition, we explore biomarkers associated with brain disease prediction using community discovery algorithms using intermediate results of FS2G. The source code is available at https://github.com/yutian0315/FS2G.

Population Distribution Forecasting Based on the Fusion of Spatiotemporal Basic and External Features: A Case Study of Lujiazui Financial District

Predicting the distribution of people in the time window approaching a disaster is crucial for post-disaster assistance activities and can be useful for evacuation route selection and shelter planning. However, two major limitations have not yet been addressed: (1) Most spatiotemporal prediction models incorporate spatiotemporal features either directly or indirectly, which results in high information redundancy in the parameters of the prediction model and low computational efficiency. (2) These models usually incorporate certain basic and external features, and they can neither change spatiotemporal addressed features according to spatiotemporal features nor change them in real-time according to spatiotemporal features. The spatiotemporal feature embedding methods for these models are inflexible and difficult to interpret. To overcome these problems, a lightweight population density distribution prediction framework that considers both basic and external spatiotemporal features is proposed. In the study, an autoencoder is used to extract spatiotemporal coded information to form a spatiotemporal attention mechanism, and basic and external spatiotemporal feature attention is fused by a fusion framework with learnable weights. The fused spatiotemporal attention is fused with Resnet as the prediction backbone network to predict the people distribution. Comparison and ablation experimental results show that the computational efficiency and interpretability of the prediction framework are improved by maximizing the scalability of the spatiotemporal features of the model by unleashing the scalability of the spatiotemporal features of the model while enhancing the interpretability of the spatiotemporal information as compared to the classical and popular spatiotemporal prediction frameworks. This study has a multiplier effect and provides a reference solution for predicting population distributions in similar regions around the globe.

ConvLSTM-based spatiotemporal and temporal processing models for chaotic vibration prediction of a microbeam

The current work proposes a hybrid data-driven model, consisting of convolutional neural networks (CNN) and convolutional long short-term memory (ConvLSTM), to enable an innovative application in prediction for microbeam's chaotic vibrations. In the proposed CNNConvLSTM model, CNN is used as feature extractor while ConvLSTM retains long-term connectivity across the entire sequence of frames. The flexibility and effectiveness of the model are demonstrated by its capability to perform both spatiotemporal and temporal processing. Specifically, the spatiotemporal model predicts the chaotic vibrations at 19 selected locations within the microbeam at a specific instant, while the temporal model predicts the microbeam's chaotic vibrations at a fixed location over time. Additionally, two conventional spatiotemporal models and three conventional temporal models are built for comparison, demonstrating the superior performance of CNNConvLSTM in terms of shorter training time and lower training and testing loss. The CNNConvLSTM model can provide useful guidance when investigating chaotic vibrations for performance enhancement and design optimization of microbeam-related devices.

A groundwater level spatiotemporal prediction model based on graph convolutional networks with a long short-term memory

ABSTRACT The performance of regional groundwater level (GWL) prediction model hinges on understanding intricate spatiotemporal correlations among monitoring wells. In this study, a graph convolutional network (GCN) with a long short-term memory (LSTM) (GCN–LSTM) model is introduced for GWL prediction utilizing data from 16 wells located in the northeastern Xiangtan City, China. This model is designed to account for both the hybrid temporal dependencies and spatial autocorrelations among wells. It consists of two parts: the spatial part employs GCNs to extract spatial characteristics from a spatial self-similarity weight matrix and an attribute self-similarity wight matrix among wells; the temporal part utilizes a LSTM module to capture the temporal patterns of GWL sequences, along with monthly precipitation and temperature data. This model dynamically predicts changes in groundwater levels, achieving higher accuracy on average compared to single-well predictions using LSTM. By incorporating both temporal dependencies and spatial autocorrelations, the GCN–LSTM model demonstrated an average improvement in goodness-of-fit of approximately 11.21% over the LSTM-based model for individual wells. Its application holds significant reference value for the sustainable utilization and development of groundwater resources in Xiangtan City.

Enhancing Soil Salinity Evaluation Accuracy in Arid Regions: An Integrated Spatiotemporal Data Fusion and AI Model Approach for Arable Lands

Soil salinization is one of the primary factors contributing to land degradation in arid areas, severely restricting the sustainable development of agriculture and the economy. Satellite remote sensing is essential for real-time, large-scale soil salinity content (SSC) evaluation. However, some satellite images have low temporal resolution and are affected by weather conditions, leading to the absence of satellite images synchronized with ground observations. Additionally, some high-temporal-resolution satellite images have overly coarse spatial resolution compared to ground features. Therefore, the limitations of these spatiotemporal features may affect the accuracy of SSC evaluation. This study focuses on the arable land in the Manas River Basin, located in the arid areas of northwest China, to explore the potential of integrated spatiotemporal data fusion and deep learning algorithms for evaluating SSC. We used the flexible spatiotemporal data fusion (FSDAF) model to merge Landsat and MODIS images, obtaining satellite fused images synchronized with ground sampling times. Using support vector regression (SVR), random forest (RF), and convolutional neural network (CNN) models, we evaluated the differences in SSC evaluation results between synchronized and unsynchronized satellite images with ground sampling times. The results showed that the FSDAF model’s fused image was highly similar to the original image in spectral reflectance, with a coefficient of determination (R2) exceeding 0.8 and a root mean square error (RMSE) below 0.029. This model effectively compensates for the missing fine-resolution satellite images synchronized with ground sampling times. The optimal salinity indices for evaluating the SSC of arable land in arid areas are S3, S5, SI, SI1, SI3, SI4, and Int1. These indices show a high correlation with SSC based on both synchronized and unsynchronized satellite images with ground sampling times. SSC evaluation models based on synchronized satellite images with ground sampling times were more accurate than those based on unsynchronized images. This indicates that synchronizing satellite images with ground sampling times significantly impacts SSC evaluation accuracy. Among the three models, the CNN model demonstrates the highest predictive accuracy in SSC evaluation based on synchronized and unsynchronized satellite images with ground sampling times, indicating its significant potential in image prediction. The optimal evaluation scheme is the CNN model based on satellite image synchronized with ground sampling times, with an R2 of 0.767 and an RMSE of 1.677 g·kg−1. Therefore, we proposed a framework for integrated spatiotemporal data fusion and CNN algorithms for evaluating soil salinity, which improves the accuracy of soil salinity evaluation. The results provide a valuable reference for the real-time, rapid, and accurate evaluation of soil salinity of arable land in arid areas.

Nationwide analysis of air pollution hotspots across India: A spatiotemporal PM2.5 trend analysis (2008–2019)

IntroductionIndia experiences high levels of air pollution as measured by fine particulate matter <2.5 μm (PM2.5) across the country. With limited resources, it is imperative to identify the most impacted areas. We aimed to identify air pollution hotspots in India and analyze temporal trends. MethodsWe conducted a geospatial analysis using Getis-Ord Gi statistics on gridded-annual levels of PM2.5 disaggregated for every state/UT and three largest cities [Delhi, Kolkata & Mumbai] of India from 2008 to 2019. The annual average PM2.5 was derived from a validated and robust nationwide spatiotemporal model(1kmx1km). Hotspots were identified annually using Gi∗ score and p-value and temporal trends across 2 periods [T1:2008–2013 & T2:2014–2019] for each spatial unit. We classified temporal trends based on the number of occurrences of hotspots in T1 and in T2 as consistent (similar in T1 & T2), declining (decreasing in T2) and emerging (increasing in T2) hotspots. ResultsWe identified consistent hotspots in 9.9% followed by emerging hotspots in 2.6% of the country where 16% and 4.9% people live. In addition, we identified declining hotspots in 2.6% area with 3.4% of the population. Rajasthan had largest share of area identified as consistent hotspots while Uttar Pradesh had densely populated consistent hotspots. Maharashtra had both higher number of areas identified as emerging and declining hotspots. Among the largest cities, Kolkata had highest proportion of consistent hotspots. We identified 170 additional cities with either consistent or emerging hotspots beyond the non-attainment cities as defined by the National Clean Air Programme. ConclusionIndia continues to have large areas of consistent and emerging hotspots of air pollution where close to a fifth of India's population live. Identifying hotspots can inform strategic approach for targeted action in air quality management, appropriate resource allocation and a baseline for assessing intervention effectiveness and future programs and policies, including health.