Spatio-temporal Model Research Articles

Air quality forecasting using a spatiotemporal hybrid deep learning model based on VMD–GAT–BiLSTM

The precise forecasting of air quality is of great significance as an integral component of early warning systems. This remains a formidable challenge owing to the limited information of emission source and the considerable uncertainties inherent in dynamic processes. To improve the accuracy of air quality forecasting, this work proposes a new spatiotemporal hybrid deep learning model based on variational mode decomposition (VMD), graph attention networks (GAT) and bi-directional long short-term memory (BiLSTM), referred to as VMD–GAT–BiLSTM, for air quality forecasting. The proposed model initially employ a VMD to decompose original PM2.5 data into a series of relatively stable sub-sequences, thus reducing the influence of unknown factors on model prediction capabilities. For each sub-sequence, a GAT is then designed to explore deep spatial relationships among different monitoring stations. Next, a BiLSTM is utilized to learn the temporal features of each decomposed sub-sequence. Finally, the forecasting results of each decomposed sub-sequence are aggregated and summed as the final air quality prediction results. Experiment results on the collected Beijing air quality dataset show that the proposed model presents superior performance to other used methods on both short-term and long-term air quality forecasting tasks.

Association between air pollution, meteorological factors and hand, foot, and mouth disease incidence in Zhejiang Province: A Bayesian spatio-temporal modelling study

Temperature and humidity are known to influence seasonal patterns of hand, foot and mouth disease (HFMD) transmission. However, the effect of air pollution and climate-air pollution interaction on the intensity of HFMD epidemics is poorly understood. This study aims to quantify the exposure-response effects of air pollution on HFMD risk by level of meteorological indicators in Zhejiang Province. A spatiotemporal Bayesian hierarchical model was used to determine the exposure-response association between air pollution, meteorological factors and the relative risk (RR) of HFMD. The model was fitted to monthly HFMD case data for the 92 counties in Zhejiang Province between January 2013, and December 2021. We assessed both linear interaction effects between air pollutants and meteorological factors on HFMD, and the modification role of extreme weather conditions on the nonlinear air pollutant-HFMD association. A 10 μg/m³ increase in PM10 and NO2 concentration raised the HFMD risk by 6.6% (RR = 1.066, 95% CI: 1.042, 1.092) and 20.8% (RR = 1.208, 95% CI: 1.162, 1.256) respectively. HFMD was negatively associated with O3 (RR = 0.980, 95% CI: 0.968, 0.993), SO2 (RR = 0.820, 95% CI: 0.776, 0.868) and CO (RR = 0.770, 95% CI: 0.621, 0.956). Temperature, humidity, and precipitation positively interacted with most air pollutants on HFMD, while wind speed had a negative interaction effect. In higher temperature conditions, the HFMD risk associated with PM10 increased. Elevated humidity levels amplified the risk associated with PM10, as did CO concentrations within a certain range. Heavy rainfall intensified the risk of HFMD associated with PM10, SO2, and CO. Results of this study showed that air pollution was associated with HFMD incidence, with different climate factors playing various modification roles in the associations and primarily leading to increased risk under extreme conditions. These findings provide a theoretical basis for personal protection and government prevention and control measures.

A deep learning-based convolutional spatiotemporal network proxy model for reservoir production prediction

Accurate production prediction is crucial in the field of reservoir management and production optimization. Traditional models often struggle with the complexities of nonlinear relationships and high-dimensional data, which hinders their ability to capture the variability of the production process efficiently and results in time-consuming calculations. To overcome these limitations, this paper introduces an innovative proxy modeling technique employing a convolutional spatiotemporal neural network. This method utilizes convolutional neural networks to extract spatial features from high-dimensional data, while the Transformer is used to model and predict complex temporal dynamics in production. To validate the effectiveness of the proposed proxy model, two case studies involving four injection and nine production wells within two-dimensional (2D) and three-dimensional (3D) non-homogeneous reservoirs were conducted, with the R2 coefficient serving as the primary evaluation metric. As the number of training iterations and data volume increase, the proxy model demonstrates rapid convergence. In tests conducted on the 2D and 3D datasets, the average R2 value exceeded 0.96 and 0.94. These results confirm the accuracy and stability of the proxy model. It also shows that the proxy model can accurately describe the geological and fluid seepage characteristics of the reservoir, which in turn can achieve a highly accurate match with the real data. In addition, the computational time is reduced by two orders of magnitude compared to traditional models. Compared with the long short-term memory method, the accuracy of the prediction results is increased by 30%, which greatly enhances efficiency and accuracy. To some extent, the presented proxy model can provide some guidance for the efficient history match of production data.

Method for constructing typical operation scenarios in power systems considering spatiotemporal correlation of renewable energy generation

Abstract The proposed article focuses on developing a methodology for constructing typical operational scenarios in power systems considering the spatiotemporal characteristics of new energy output. The integration of new energy sources on a large scale into power systems can lead to issues such as line congestion and supply-demand imbalances. To support the planning and scheduling of future high-capacity power systems, this paper introduces a method that accounts for the spatiotemporal characteristics of new energy outputs. The method begins by proposing a spatiotemporal fusion-based multi-head attention model for generating new energy output scenarios. This model employs multi-head attention mechanisms to extract the spatiotemporal characteristics of new energy outputs, eliminating redundancy in the feature data. The processed spatiotemporal feature information is then transferred to a gated feature fusion model for integration, using an encoder structure based on Long Short-Term Memory (LSTM) to generate new energy output scenarios. The paper, by means of an advanced Fuzzy C-Means (FCM) clustering algorithm, identifies typical scenarios of net energy output. Subsequently, the proposed method’s efficacy and precision are confirmed through simulation analysis with real data from a certain area.

A new attention-based CNN_GRU model for spatial-temporal PM2.5 prediction.

Accurately predicting thespatial-temporal distribution of PM2.5 is challenging due to missing data and selecting an appropriate modeling method. Effective imputation of missing data must consider the relationships between variables while preserving their inherent variability and uncertainty. In this study, we employed machine learning techniques to impute missing data by analyzing the relationships between meteorological variables and other pollutants. Subsequently, we introduced an innovative spatiotemporal hybrid model, AC_GRU, which integrates a one-dimensional convolutional neural network (CNN), GRU, and an attention-based network to predict PM2.5 concentrations in urban areas. The AC_GRU model utilizes meteorological variables, PM2.5 concentrations from nearby air quality monitoring stations, and concentrations of other pollutants as inputs. This approach allows the model to learn spatiotemporal correlations within the time-series data, enhancing the accuracy of PM2.5 predictions. Additionally, the attention mechanism improves prediction accuracy by automatically weighting the past input variables based on their importance for future PM2.5 predictions. The experimental results demonstrate that our AC_GRU model outperforms state-of-the-art methods, making it a valuable tool for urban air quality management and public health protection.

O'Hare Airport roadway traffic prediction via data fusion and Gaussian process regression

This study proposes an approach of leveraging information gathered from multiple traffic data sources at different resolutions to obtain approximate inference on the traffic distribution of Chicago's O'Hare Airport area. Specifically, it proposes the ingestion of traffic datasets at different resolutions to build spatiotemporal models for predicting the distribution of traffic volume on the road network. Due to its good adaptability and flexibility for spatiotemporal data, the Gaussian process (GP) regression was employed to provide short-term forecasts using data collected by loop detectors (sensors) and supplemented by telematics data. The GP regression is used to make predictions of the distribution of the proportion of sensor data traffic volume represented by the telematics data for each location of the sensors. Consequently, the fitted GP model can be used to determine the approximate traffic distribution for a testing location outside of the training points. Policymakers in the transportation sector can find the results of this work helpful for making informed decisions relating to current and future transportation conditions in the area.

Pressure swing adsorption process modeling using physics-informed machine learning with transfer learning and labeled data

Pressure swing adsorption (PSA) modeling remains a challenging task since it exhibits strong dynamic and cyclic behavior. This study presents a systematic physics-informed machine learning method that integrates transfer learning and labeled data to construct a spatiotemporal model of the PSA process. To approximate the latent solutions of partial differential equations (PDEs) in the specific steps of pressurization, adsorption, heavy reflux, counter-current depressurization, and light reflux, the system's network representation is decomposed into five lightweight sub-networks. On this basis, we propose a parameter-based transfer learning (TL) combined with domain decomposition to address the long-term integration of periodic PDEs and expedite the network training process. Moreover, to tackle challenges related to sharp adsorption fronts, our method allows for the inclusion of a specified amount of labeled data at the boundaries and/or within the system in the loss function. The results show that the proposed method closely matches the outcomes achieved through the conventional numerical method, effectively simulating all steps and cyclic behavior within the PSA processes.

Spatiotemporal and weather effects on the reproductive success of piping plovers on Prince Edward Island, Canada.

Piping plovers (Charadrius melodus sp.) rank among North America's most endangered shorebird species, facing compounding environmental challenges that reduce habitat availability and suppress recruitment and survival rates. Despite these challenges, research on the direct effects of climate variability and extremes on their breeding ecology remains limited. Here, we employ a spatiotemporal modelling approach to investigate how location, nest timing and weather conditions influence reproductive success rates in a small breeding population of C. m. melodus in Prince Edward Island (PEI), Canada from 2011 to 2023. Analysis of 40 years of monitoring records from a subset of nesting sites revealed that flooding and predation have been persistent sources of reproductive failures in this population, with unexplained losses increasing in recent years. Contrary to our hypotheses, our modelled results did not support a negative impact of extreme high temperatures and strong precipitation events on reproductive outcomes. Instead, we identified a positive effect of T MAX and no effect of strong precipitation, perhaps due to limited exposure to extreme high temperatures (>32°C) and context-specific risks associated with precipitation-induced flooding. However, trends in regional climate change are likely to increase exposure to-and the influence of-such factors in the near future. Our models also identified spatiotemporal variability in apparent hatch success over the study period, as well as worse hatch outcomes across popular beachgoing regions and for delayed nesting attempts. While our results offer preliminary insights into factors affecting breeding success in this population, further research will be imperative to enhance understanding of constraints on recruitment. To this end, we encourage the collection and analysis of additional time-series data of prey populations, human activities, fine-scale weather data and predator/flood risks associated with each nest on PEI.

Spatiotemporal Characteristics and Factors Influencing the Cycling Behavior of Shared Electric Bike Use in Urban Plateau Regions

At present, most of the research on shared electric bikes mostly focuses on the scheduling, operation and maintenance of shared electric bikes, while insufficient attention has been paid to the behavioral characteristics and influencing factors of shared cycling in plateau cities. This paper takes Kunming as a research case. According to the user’s cycling behavior, the spatiotemporal cube model and emerging hotspot analysis are used to explore the spatiotemporal characteristics of the citizens’ cycling in the plateau city represented by Kunming, and the method of geographical detectors is used to study the specific factors affecting the shared travel of citizens in Kunming and conduct interactive detection. The findings are as follows: ① the use of shared electric bikes in Kunming varies greatly on weekdays, showing a bimodal feature. In space, the overall distribution of cycling presents a “multi-center” agglomeration feature with distance decay from the center of the main urban area. ② The geographic detector factor detection model quantitatively analyzes the interactive influence between factors, providing a good supplement to the independent influence results of each factor. Through the dual factor interactive detection model, we found that the overall spatiotemporal distribution of cycling during each time period is most significantly affected by the distribution of service facilities, followed by transportation accessibility, land use, and the natural environment. The research results can assist relevant departments in governance of urban shared transportation and provide a reference basis, and they also have certain reference value in urban pattern planning.

An experiential account with recommendations for the design, installation, operation and maintenance of a farm-scale soil moisture sensing and mapping system

Context The research explores the benefits of real time tracking of soil moisture for various land management contexts and the importance of spatio-temporal modelling and mapping to gain clear and visual understanding of soil moisture fluxes across a farm. Aims This research aims to outline the key processes required for building an operational on-farm soil moisture monitoring system where the product is highly granular daily soil moisture maps depicting variations temporally, spatially and vertically. Methods We describe processes of capacitance soil moisture probe installation, data collection infrastructure, sensor calibration, spatio-temporal modelling, and mapping. Key results An out-of-bag soil moisture evaluation modelling system was tested for nearly 2 years. We found a model accuracy (RMSE) estimate of 0.002 cm−3 cm−3 and concordance of 0.96 were found. This result is averaged over this period but fluctuated daily, and related to rainfall patterns across the target farm, which were not directly incorporated into the modelling framework. As expected, incorporating prior estimates of soil moisture into the modelling framework contributed to very accurate estimates of real time available soil moisture. Conclusions This research promotes the importance of iterative improvements to the soil moisture monitoring system, particularly in areas of sensor recalibration and spatio-temporal modelling. We stress the need for a longer-term view and plan for ongoing maintenance and improvement of such systems in the emerging digital farming ecosystem. Implications The results of this research will be useful for researchers and practitioners involved in the design and implementation of on-farm soil monitoring systems.

Settlement estimation during foundation excavation using pattern analysis and explainable AI modeling

With the rapid expansion of underground engineering, accurate settlement estimation during foundation excavation using monitoring data has gained prominence. Previous studies have typically overlooked data patterns, relying solely on time-series models, which yielded limited accuracy and short-term predictability. To address these issues, this paper performs a thorough pattern analysis covering both the temporal and spatial properties, and proposes a spatiotemporal modeling method grounded in explainable artificial intelligence technique. A real-world engineering case study is conducted to validate the effectiveness of the proposed method. Counterintuitively, the results reveal a weak temporal effect within the settlement data, with only the last step playing a dominant role, whereas the spatial correlation among measuring points is notably more significant. Compared to conventional models, the proposed method consistently outperforms, achieving significantly higher R2 scores and excelling in long-term estimation with reductions of at least 78% in RMSE, 80% in MAE, and 72% in MAPE.

Exposure to ambient ozone and sperm quality among adult men in China

BackgroundLimited evidence exists regarding the association between ozone exposure and adverse sperm quality. We aimed to assess the association between ozone exposure and sperm quality, and identify susceptible exposure windows. MethodsWe recruited 32,541 men aged between 22 and 65 years old attending an infertility clinic in Wuhan, Hubei Province, China from 2014 to 2020. Ozone data were obtained from a satellite-based spatiotemporal model. Generalized linear models were used to estimate the association between ozone exposure and sperm quality parameters, including sperm concentration, sperm count, sperm total motility, and sperm progressive motility during the entire stage of sperm development (0–90 days before ejaculation) and three crucial stages (0–9 days, 10–14 days and 70–90 days before ejaculation). Stratified analyses were performed to evaluate whether associations varied by age, body mass index, and education levels. ResultsThe final analysis included 27,854 adult men. A 10 μg/m3 increase in ozone concentrations during the entire stage of sperm development was associated with a −4.17 % (95 % CI: −4.78 %, −3.57 %) decrease in sperm concentration, −6.54 % (95 % CI: −8.03 %, −5.60 %) decrease in sperm count, −0.50 % (95 % CI: −0.66 %, −0.34 %) decrease in sperm total motility, and −0.07 % (95 % CI: −0.22 %, 0.09 %) decrease in sperm progressive motility. The associations were stronger during 70–90 days before ejaculation and among men with middle school and lower education for sperm concentration. ConclusionsOzone exposure was associated with decreased sperm quality among Chinese adult men attending an infertility clinic. These results suggest that ozone may be a risk factor contributing to decreased sperm quality in Chinese men.

Tropical cyclone ensemble forecast framework based on spatiotemporal model

Tropical cyclone ensemble forecast framework based on spatiotemporal model

High modularity, more flexible of brain networks in patients with mild to moderate motor impairments after stroke

Stroke is recognized as a network communication disorder. Advances in neuroimaging technologies have enhanced our comprehension of dynamic cerebral alterations. However, different levels of motor function impairment after stroke may have different patterns of brain reorganization. Abnormal and adaptive patterns of brain activity in mild-to-moderate motor function impairments after stroke remain still underexplored. We aim to identify dynamic patterns of network remodeling in stroke patients with mild-to-moderate impairment of motor function. fMRI data were obtained from 30 stroke patients and 31 healthy controls to establish a spatiotemporal multilayer modularity model. Then, graph-theoretic measures, including modularity, flexibility, cohesion, and disjointedness, were calculated to quantify dynamic reconfiguration. Our findings reveal that the post-stroke brain exhibited higher modular organization, as well as heightened disjointedness, compared to HCs. Moreover, analyzing from the network level, we found increased disjointedness and flexibility in the Default mode network (DMN), indicating that brain regions tend to switch more frequently and independently between communities and the dynamic changes were mainly driven by DMN. Notably, modified functional dynamics positively correlated with motor performance in patients with mild-to-moderate motor impairment. Collectively, our research uncovered patterns of dynamic community reconstruction in multilayer networks following stroke. Our findings may offer new insights into the complex reorganization of neural function in post-stroke brain.

An ensemble neural network approach for space–time landslide predictive modelling

There is an urgent need for accurate and effective Landslide Early Warning Systems (LEWS). Most LEWS are currently based on temporally-aggregated measures of rainfall derived from either in-situ measurements or satellite-based rainfall estimates. Relying on a summary metric of precipitation may not capture the complexity of the rainfall signal and its dynamics in space and time in triggering landslides. Here, we present a proof-of-concept for constructing a LEWS based on an integrated spatio-temporal modelling framework. Our proposed methodology builds upon a recent approach that uses a daily rainfall time series instead of the traditional scalar aggregation. Specifically, we partition the study area into slope units and use a Gated Recurrent Unit (GRU) to process satellite-derived rainfall time series and combine the output features with a second neural network (NN) tasked with capturing the effect of terrain characteristics. To assess if our approach enhances accuracy, we applied it in Vietnam and benchmarked it against a modelling counterpart where we replaced the rainfall time series with the corresponding scalar representative of the cumulated precipitation. The corresponding duration was set at 14 days as it proved to produce the best performance. Our results show that our protocol leads to better performance in hindcasting landslides when making use of the rainfall as a continuous signal over time. While not tested here, our approach can be extended to rainfall obtained from weather forecasts, potentially leading to actual landslide forecasts.

Developing a Research Network of Early Warning Systems for Infectious Diseases Transmission Between China and Australia.

This article offers a thorough review of current early warning systems (EWS) and advocates for establishing a unified research network for EWS in infectious diseases between China and Australia. We propose that future research should focus on improving infectious disease surveillance by integrating data from both countries to enhance predictive models and intervention strategies. The article highlights the need for standardized data formats and terminologies, improved surveillance capabilities, and the development of robust spatiotemporal predictive models. It concludes by examining the potential benefits and challenges of this collaborative approach and its implications for global infectious disease surveillance. This is particularly relevant to the ongoing project, early warning systems for Infectious Diseases between China and Australia (NetEWAC), which aims to use seasonal influenza as a case study to analyze influenza trends, peak activities, and potential inter-hemispheric transmission patterns. The project seeks to integrate data from both hemispheres to improve outbreak predictions and develop a spatiotemporal predictive modeling system for seasonal influenza transmission based on socio-environmental factors.

Novel STAttention GraphWaveNet model for residential household appliance prediction and energy structure optimization

The multifaceted challenge of carbon dioxide (CO2) emissions during building operations, construction, and material production perpetuates energy shortages and climate adversities. Urgent imperatives underscore the need for effective energy reduction and carbon mitigation, driving the building sector toward a paradigm shift in clean, low-carbon evolution. However, the complex multidimensional and nonlinear characteristics of influencing factors of building energy consumption make it difficult in precisely understanding the spatial and temporal relationships between the features in the predictive modeling of buildings. Therefore, a novel Graph Wavenet (GWN) based on the spatio-temporal attention fusion mechanism (STA-GWN) is proposed for spatio-temporal graph prediction modeling of structures, which can capture hidden spatial relationships using an adaptive dependency matrix. To overcome the drawbacks of standard predefined adjacency matrices and their inability to capture faraway dependencies, a spatio-temporal attention fusion mechanism is introduced to extract latent spatio-temporal features dynamically. By combining extracted features, building energy consumption can be predicted more precisely. Moreover, the proposed model becomes more interpretable, making it easier to reveal how spatio-temporal features relate to building energy consumption. Lastly, the proposed method is applied to energy conservation and emission reduction in the building industry. A predictive model for electrical energy consumption during two public datasets with an operational phase of buildings is established to analyze and optimize energy utilization in residential structures that rely on electricity. The experimental results demonstrate that the STA-GWN model surpasses other deep learning methods in terms of prediction accuracy and robustness. Moreover, the proposed model can offer optimal solutions for building energy processes, reducing approximately 201.44 kg (Kg) and 64.08 Kg of CO2 emissions in the appliances energy prediction dataset and the individual-household-electric-power-consumption (IHEPC) dataset, respectively.

A spatio-temporal predictive learning model for efficient sea surface temperature forecasting

A spatio-temporal predictive learning model for efficient sea surface temperature forecasting

Spatiotemporal modelling of northern shrimp Pandalus borealis distribution patterns throughout Canada’s subarctic and arctic regions

Northern shrimp Pandalus borealis occur throughout Canada’s Atlantic Ocean, where they are thought to form a single population spanning from Baffin Bay to the tail of the Grand Bank. Here, they play an important role in the ecosystem as prey for many taxa and have been targeted by a lucrative large-scale fishery since the 1970s. Yet, we still understand little about which (and how) ecosystem and environmental factors influence their distribution and abundance. We used survey data collected over 29 yr throughout 23 degrees of latitude to develop a spatiotemporal model predicting northern shrimp density. We confirmed that both top-down drivers (e.g. predation pressure), as well as bottom-up drivers (e.g. bottom temperature) play important roles in determining both the presence and abundance of northern shrimp. The model was used to predict the density of northern shrimp throughout the entire study area from 2005 to 2022. Our results highlight the importance of understanding ecosystem and environmental dynamics in relation to northern shrimp population patterns and trends within resource assessments.

Dynamics of a spatiotemporal SIS epidemic model with distinct mobility range

In this study, we consider an epidemic model for susceptible-infected-susceptible populations with nonlocal dffusion, described by convolution operators. Our main objective is to model different diffusion strategies for the susceptible and infected populations by using distinct kernel functions in the convolution operator to model the different movements of individuals for each class. This approach eliminates the simplification of reducing the stationary problem to a single equation, which makes the analysis more challenging. However, we have shown the existence of at least one positive endemic steady state when the basic reproduction number is greater than one, using the index degree theory of Leray-Schauder. Additionally, we investigated the asymptotic profile of the endemic equilibrium state for both, large and small diffusion rates, to demonstrate the persistence or extinction of the disease. Overall, our study indicates that restricting individual movements only partially cannot eliminate the disease unless the total population size is small.