Spatio-temporal Model Research Articles

Spatio-temporal Prediction of Air Quality Using Spatio-temporal Clustering and Hierarchical Bayesian Model

Air pollution can cause many negative impacts, so it is meaningful to establish an air quality prediction model with high accuracy for air pollution prevention and control. In previous studies, most models did not consider the spatio-temporal distribution characteristics of air quality. So this study proposes a novel spatio-temporal prediction approach STC-HBM, which combines spatio-temporal clustering (STC) and hierarchical Bayesian model (HBM) to build a spatio-temporal prediction model. In this study, we first perform spatio-temporal clustering on the Beijing-Tianjin-Hebei region in China and then apply hierarchical Bayesian models to different clusters separately. We consider three hierarchical Bayesian models: the separable spatio-temporal (SST) model, the Gaussian process (GP) model, and the autoregressive (AR) model. Since the prior distribution affects the prediction accuracy of the HBM, the resultant output of the Bayesian linear regression (BLR) model is used as the prior input of the HBM, which improves the flexibility of the model. The experimental results show that 13 cities in the BTH region are clustered into two clusters according to their spatio-temporal characteristics. Based on MAE, RMSE, MAPE, and coverage (CVG), cluster 1, and cluster 2 are better in both temporal and spatial prediction compared to the overall prediction model. In addition, for cluster 1, the models with the best prediction in time and space are AR and GP, respectively; for cluster 2, the models with the best prediction in time and space are SST and GP, respectively.

A spatio-temporal model for temporal evolution of spatial extremal dependence

Few spatio-temporal models allow temporal non-stationarity. When modeling environmental data recorded over the last decades of the 20th century until now, it seems not reasonable to assume temporal stationarity, since it would not capture climate change effects. In this paper, we propose a space–time max-stable model for modeling some temporal non-stationarity of the spatial extremal dependence. Our model consists of a mixture of max-stable spatial processes, with a rate of mixing depending on time. We use maximum composite likelihood for estimation, model selection, and a non-stationarity test. The assessment of its performance is done through wide simulation experiments. The proposed model is used to investigate how the rainfall in the south of France evolves with time. The results demonstrate that the spatial extremal dependence is significantly non-stationary over time, with a decrease in the strength of dependence.

Individual and joint exposure to air pollutants and patterns of multiple chronic conditions.

Existing research on the detrimental effects of air pollution and its mixture on multiple chronic conditions (MCC) is not yet fully recognized. Our objective was to examine if individual and joint exposure to air pollution is associated with the incidence and patterns of MCC. Totally 10,231 CHARLS 2015 participants aged over 45 years and 1,938 without MCC were followed up in 2018 and 2020. Residential-levelcumulative personal exposure concentrations of PM1, PM10, PM2.5, CO, O3, NO2, SO2, NO3-, Cl-, NH4+, and SO42- at the residential level were determined utilizing a spatio-temporal random forest model with a spatial resolution of 0.1° × 0.1°. In the cross-sectional and longitudinal research, logistic regression, cox regression analysis, and quantile g-computation were utilized to estimate the single and joint effect with MCC and its patterns, respectively. Interaction analyses and stratified analyses were also performed. A correlation was observed between the prevalence of cardiovascular illnesses and the presence of all 11 major air pollutants. PM2.5, PM10, NH4+, NO3-, CO, and SO42- are associated with an increased frequency of respiratory disorders. An increase of PM2.5, PM1, PM10, NO2, and SO2 (a 10µg/m3 rise), CO (a 0.1mg/m3 rise), and PMCs (Cl-, NH4+, NO3-, and SO42-) (a 1µg/m3 rise) corresponded to the HRs (95% CI) for developing MCC of 1.194 (95% CI: 1.043, 1.367), 1.362 (95% CI: 1.073, 1.728), 1.115 (95% CI: 1.026, 1.212), 1.443 (95% CI: 1.151, 1.808), 3.175 (95% CI: 2.291, 4.401), 1.272 (95% CI: 1.149,1.410), 1.382 (95% CI: 1.011, 1.888), 1.107 (95% CI: 1.003, 1.222), 1.035 (95% CI: 0.984, 1.088), and 1.122 (95% CI: 1.086, 1.160), respectively. SO2 was the predominant contributor to the combined effect (HR: 2.083, 95% CI: 1.659-2.508). Gender, age, drinking, and health status could modify the effects of air pollutants on MCC patterns. Long-term exposure to air pollution is correlated to the incidence and patterns of MCC in middle-aged and elderly Chinese individuals. Preventive methods are essential to safeguarding those susceptible to MCC.

Numerical investigation of lean methane flame response to NRP discharges actuation

This study investigates the response of a laminar methane-air flame to Nanosecond Repetitively Pulsed (NRP) discharges in a canonical wall-stabilized burner using a combined experimental and numerical approach. The flow and flame behaviors were modeled using Direct Numerical Simulation (DNS) with an Analytically Reduced Chemistry for a precise chemical description. A phenomenological model incorporating detailed plasma kinetics and experimental observations was developed to simulate plasma effects. Zero-dimensional plasma reactor simulations were used to build up a reduced-order model describing discharge energy distribution in the specific conditions studied. Experimental measurements of electrical profiles identified two discharge regimes: a low-energy Corona discharge and a higher-energy Glow discharge, characterized by distinct spatial energy distributions. Experimental flame response analysis revealed three major phases: marginal response up to 100 pulses, a downstream shift of the flame tip, and stabilization after 400 pulses. Numerical simulations indicated that the Corona regime is crucial for explaining initial flame responses, while the Glow regime influences later stages. Adjustments in the Vibrational-Translational (VT) energy relaxation time and energy deposition ratios between fresh and burnt gases were necessary to match experimental observations. Additionally, an accurate modeling of the transient and steady-state flame responses requires integrating both the specificity of the Corona and the Glow discharge regimes. Future work should focus on measuring or theoretically calculating N2(v) relaxation times in CH4-H2O-CO2 mixtures and analyzing the spatial energy distribution of discharges interacting with flames to enhance plasma-combustion coupled models.Novelty and significanceIn this work, a phenomenological plasma-assisted combustion model has been developed, to investigate a laminar premixed stagnation plate burner, focusing on VT energy relaxation time and spatio-temporal energy distribution modeling. For the first time, not only O2, N2 and O but also the fuel and combustion intermediates and products have been considered in the VT relaxation model. It revealed their strong influence on the overall flame response in a case where the discharge crosses a flame front, highlighting the strong beneficial effect of energy deposited in vibrational form. The study questions and investigates the energy distribution from fresh to burnt gases, challenging the conventional uniform energy distribution assumption. The experimental identification of two specific plasma regimes was necessary to predict the transient flame response. Additionally, energy deposited downstream of the flame, in fresh gases, was found to more efficient than in hot gases to enhance combustion.

Spatio-temporal modeling to identify factors associated with stunting in Indonesia using a Modified Generalized Lasso

This study investigates the factors associated with stunting prevalence in Indonesia, utilizing a generalized lasso framework with modified penalty matrices to accommodate spatio-temporal data structures. Novel approaches are introduced to construct the penalty matrices, with particular focus on defining neighborhood structures. The proposed method is applied to data from 34 Indonesian provinces, covering the years 2019 to 2023. The primary outcome is stunting prevalence, modeled against nine predictor variables: poverty, exclusive breastfeeding, low birth weight (LBW), high school completion, access to proper sanitation, unmet health service needs, Gross Domestic Product (GDP), calorie consumption, and protein consumption. A total of nine spatio-temporal models were compared, including a modified generalized lasso with three distinct penalty matrices for each two tuning selection methods and a generalized ridge regression with three penalty matrices. Results indicate that the generalized lasso model with a 3-nearest neighbor adjacency matrix outperformed the alternatives. Temporal variations were observed in the effects of exclusive breastfeeding, LBW, high school completion, and unmet health service needs. Positive associations with stunting prevalence were identified for poverty, exclusive breastfeeding, LBW, and unmet health service needs, while negative associations were found for high school completion rates, access to proper sanitation, GDP, calorie intake, and protein consumption. The strongest associations were observed in parts of Sumatra, Sulawesi, and Jakarta. These findings suggest that government interventions aimed at improving education, healthcare access, and poverty reduction may help alleviate stunting in Indonesia, particularly in regions with the greatest need.

Comparison of two biologically effective dose calculation models applied to single fraction stereotactic radiosurgery

BackgroundRecent studies suggest strong correlations between Biologically Effective Doses (BED) and single fraction stereotactic radiosurgery treatment outcomes, as demonstrated for vestibular schwannomas (VS), arterio-venous malformations and pituitary adenomas. The BEDs calculated in these studies consider an uniform dose delivery where the spatio-temporal aspects of dose delivery were neglected. PurposeThe aim of the study is to quantify the discrepancies between the BED values calculated with a simplified model of uniform dose delivery against the more complex model that incorporates the temporo-spatial incrementation of dose delivery and the bi-exponential effect of the sub-lethal damage repair. MethodsA software tool that computes the BED distributions based on individual isocenter dose matrices extracted from the GammaPlan (Elekta) treatment planning was developed. Two cohorts 5 VS and 5 jugular foramen schwannoma cases of various tumor volumes and isocenter number were utilized to benchmark the method. Their BEDs covering 98% of tumor volumes were compared against those determined with the uniform delivery model. ResultsThe BEDs covering 98% of the tumor volumes as calculated with both models show an approximately linear dependency with the treatment time. For all studied cases, the uniform delivery model overestimates the BEDs calculated with the full spatio-temporal delivery model. This discrepancy seems to accentuate with the tumor volume and treatment complexity. ConclusionsDespite their resemblance, the BED distributions provide a plethora of BED measures more suitable to characterize clinical outcomes than the unique peripheral BED value calculated with the simplified model of uniform dose delivery.

The effects of climate and population on human land use patterns in Europe from 22ka to 9ka ago

Between 22ka and 9ka ago, after the Last Glacial Maximum (LGM) and during the transition to the Holocene, mobile hunter-gatherer populations, differentiated by their stone tool assemblages, periodically dispersed and contracted across Europe. It is well understood that climate played an important role in human distributions and population sizes during the post-LGM Late Pleistocene and Early Holocene, but the question remains as to whether increasing population sizes drove early human dispersal. Here, using a spatiotemporal species distribution model to infer the fundamental and potential human climatic niche space of Late Upper Paleolithic, Final Paleolithic and Early Mesolithic Europe, and hypotheses derived from the Ideal Distribution Model, we test i) how changes in climate affected the size and extent of the projected potential human niche space, ii) for effects of changes in size of projected potential niche on regional human population sizes, iii) whether increasing human population sizes drove human dispersal into less climatically suitable habitats, and iv) whether populations associated with different high-order material culture groupings (macro-level technocomplexes) occupied different climatic spaces. We find that changes in climate correlate strongly with the size and extent of the projected potential human niche space, that increases in the projected human niche space correlate with increases in human population, that human population size is just beginning to become large enough to influence land use and dispersal patterns at the Pleistocene-Holocene transition, and that archaeological technocultural entities overlap in their fundamental climatic niche space. This overlap implies that changing tools forms cannot readily been seen as reflecting adaptations to changing climates.

A novel spatio-temporal attention-based bidirectional LSTM model for moisture content prediction in drying process

Accurate prediction of moisture content is significantly crucial for ensuring process stability and product quality in the cylinder drying process. However, the drying process exhibits complex spatio-temporal characteristics and strong interference, which make accurate prediction challenging for the deep learning approach. To address this issue, this article proposes a new spatio-temporal attention-based bidirectional long-short temporal memory network (STA-BiLSTM) model for accurate moisture content prediction. First, Maximum Relevance Minimum Redundancy (mRMR) is adopted to identify optimal features highly related to moisture content. Secondly, bidirectional long-short temporal memory (Bi-LSTM) network is utilized to extract temporal dependencies from the sequential data. Subsequently, spatio-temporal attention mechanisms are designed to adaptively focus on the most relevant features and timesteps, enhancing the model’s generalization ability. Finally, due to the harsh industrial environment, eXtreme Gradient Boosting (XGBoost) is adapted to improve generalizability and robustness. Extensive experiments on a real industrial dataset of the drying process demonstrate that the proposed STA-BiLSTM approach significantly outperforms alternative approaches for predicting moisture content, validating its effectiveness and superiority.

The Effect of Long-Term Particulate Matter Exposure on Respiratory Mortality: Cohort Study in China.

Particulate matter (PM), which affects respiratory health, has been well documented; however, substantial evidence from large cohorts is still limited, particularly in highly polluted countries and for PM1. Our objective was to examine the potential causal links between long-term exposure to PMs (PM2.5, PM10, and more importantly, PM1) and respiratory mortality. A total of 580,757 participants from the Guangzhou area, China, were recruited from 2009 to 2015 and followed up through 2020. The annual average concentrations of PMs at a 1-km spatial resolution around the residential addresses were estimated using validated spatiotemporal models. The marginal structural Cox model was used to estimate the associations of PM exposure with respiratory mortality, accounting for time-varying PM exposure. Results were stratified by demographics and lifestyle behaviors factors. Among the participants, the mean age was 48.33 (SD 17.55) years, and 275,676 (47.47%) of them were men. During the follow-up period, 7260 deaths occurred due to respiratory diseases. The annual average concentrations of PM1, PM2.5, and PM10 showed a declining trend during the follow-up period. After adjusting for confounders, a 6.6% (95% CI 5.6%-7.6%), 4.2% (95% CI 3.6%-4.7%), and 4.0% (95% CI 3.6%-4.5%) increase in the risk of respiratory mortality was observed following each 1-μg/m3 increase in concentrations of PM1, PM2.5, and PM10, respectively. In addition, older participants, nonsmokers, participants with higher exercise frequency, and those exposed to a lower normalized difference vegetation index tended to be more susceptible to the effects of PMs. Furthermore, participants in the low-exposure group tended to be at a 7.6% and 2.7% greater risk of respiratory mortality following PM1 and PM10 exposure, respectively, compared to the entire cohort. This cohort study provides causal clues of the respiratory impact of long-term ambient PM exposure, indicating that PM reduction efforts may continuously benefit the population's respiratory health.

DATA ANALYSIS AND MACHINE LEARNING APPLICATIONS IN ENVIRONMENTAL MANAGEMENT

The rapid expansion of data on air contaminants and climate change, particularly concerning public health, presents both opportunities and challenges for traditional epidemiological methods. This study aims to address these challenges by exploring advanced data collection, pattern identification, and predictive modeling techniques in the context of air pollution research. The focus is leveraging data mining and computational methods to enhance the understanding of air pollution's impact on public health, specifically ozone exposure. A comprehensive review of the scientific literature was conducted, utilizing databases such as Professor, Scholar, Embl, and Nih to identify relevant studies on air pollution epidemiology. The review highlights the integration of data mining, machine learning, and spatiotemporal modeling to improve the detection, analysis, and forecasting of air pollution-related health issues. The findings reveal a growing trend in applying data mining techniques within the field of air pollution epidemiology. Advanced methods, such as spatiotemporal analysis and geographic data mining, enable more precise tracking and forecasting of pollution-related health risks. Continuous advancements in artificial intelligence and the development of more sophisticated sensors and data storage technologies are enhancing the accuracy and reliability of air quality monitoring and public health predictions. This study highlights the transformative potential of integrating data mining and AI techniques into air pollution epidemiology. Exploring emerging technologies like spatiotemporal mining and next-generation sensors paves the way for more accurate, timely, and scalable solutions to monitor air quality and predict its impact on public health, opening new avenues for research and policy interventions.

Abstract A048: Association between ambient particulate matter and colorectal cancer incidence: The Multiethnic Cohort Study

Abstract Background: Growing evidence suggests that air pollution may be a risk factor for colorectal cancer (CRC). We examined the association between ambient particulate matter (PM) exposure and CRC risk within the Multiethnic Cohort Study (MEC), representing the first study to include a large sample of racial and ethnic minoritized populations. Methods: The MEC is a prospective study that recruited, from 1993 through 1996, men and women ages 45-75 years largely from five racial and ethnic groups (African American, Japanese American, Latino, Native Hawaiian, and Non-Hispanic White) living in Hawai‘i and California. This study included 98,675 California MEC participants, residing predominately in Los Angeles County. Satellite-based PM2.5 (PM with an aerodynamic diameter <2.5μm) exposures were estimated from a published spatiotemporal model. Kriging interpolation was used to estimate PM10 (PM <10μm) exposures using routine air monitoring data from the US Environmental Protection Agency. Time-varying PM exposures were assessed from time of recruitment through 12/31/2018. Incident invasive CRC cases were identified via linkage with the California Cancer Registry (n=3,217). Cox proportional hazards models were used to examine the association between PM and CRC risk using age as the time metric, with strata defined by age at cohort entry and race and ethnicity, and adjusting for demographic and lifestyle factors. Heterogeneity of the PM-CRC association was assessed by sex, race and ethnicity, and CRC subsite (right colon, left colon, rectum). Results: Risk of CRC increased with PM2.5 (per 10μg/m3 hazard ratio [HR] 1.13, 95% confidence interval [CI] 0.96-1.33) with a larger HR observed among females (HR 1.29, 95% CI 1.03-1.62) than males (HR 0.96, 95% CI 0.76-1.22) (Pheterogeneity=0.24). CRC risk was positively associated with PM2.5 across race and ethnicity (except for African American participants) with the largest HR observed among Latino participants (HR 1.47, 95% CI 1.03-2.08) (Pheterogeneity=0.20). There was no statistically significant heterogeneity of PM2.5 effect by tumor subsite (Pheterogeneity=0.07); although larger HRs were seen for left colon (HR 1.52, 95% CI 1.08- 2.14) and rectal cancers (HR 1.54, 95% CI 1.05-2.25) than right colon cancers (HR 1.09, 95% CI 0.88-1.35). PM10 was not associated with CRC risk (per 10 μg/m3; HR 1.02, 95% CI 0.93-1.22) with no evidence of heterogeneity in associations by sex or race and ethnicity (respective Pheterogeneity 0.75, 0.44). Although HRs for PM10 appeared to differ by CRC subsite (Pheterogeneity=0.03), the 95% CI of these HRs included the null (right HR 0.95, 95% CI 0.84-1.06; left HR 1.13, 95% CI 0.95-1.35; rectal HR 1.18, 95% CI 0.97-1.43). Conclusions: Preliminary findings suggest that PM2.5 exposure is associated with CRC risk with larger associations in females and, possibly, for left colon and rectal cancer. Additional studies are needed to confirm these findings and evaluate the potential biological pathways underlying these associations. Citation Format: Ugonna Ihenacho, Chiuchen Tseng, Jun Wu, Scott Fruin, Timothy V. Larson, Salma Shariff-Marco, Loïc Le Marchand, Daniel O. Stram, Lynne R. Wilkens, Christopher A. Haiman, Beate Ritz, Iona Cheng, Anna H. Wu. Association between ambient particulate matter and colorectal cancer incidence: The Multiethnic Cohort Study [abstract]. In: Proceedings of the 17th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2024 Sep 21-24; Los Angeles, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2024;33(9 Suppl):Abstract nr A048.

Improving the ERA5-Land Temperature Product through a Deep Spatiotemporal Model That Uses Fused Multi-Source Remote Sensing Data

Temperature is a crucial indicator for studying climate, as well as the social and economic changes in a region. Temperature reanalysis products, such as ERA5-Land, have been widely used in studying temperature change. However, global-scale temperature reanalysis products have errors because they overlook the influence of multiple factors on temperature, and this issue is more obvious in smaller areas. During the cold months (January, February, March, November, and December) in the Yellow River Basin, ERA5-Land products exhibit significant errors compared to temperatures observed by meteorological stations, typically underestimating the temperature. This study proposes improving temperature reanalysis products using deep learning and multi-source remote sensing and geographic data fusion. Specifically, convolutional neural networks (CNN) and bidirectional long short-term memory networks (BiLSTM) capture the spatial and temporal relationships between temperature, DEM, land cover, and population density. A deep spatiotemporal model is established to enhance temperature reanalysis products, resulting in higher resolution and more accurate temperature data. A comparison with the measured temperatures at meteorological stations indicates that the accuracy of the improved ERA5-Land product has been significantly enhanced, with the mean absolute error (MAE) reduced by 28.7% and the root mean square error (RMSE) reduced by 25.8%. This method obtained a high-precision daily temperature dataset with a 0.05° resolution for cold months in the Yellow River Basin from 2015 to 2019. Based on this dataset, the annual trend of average temperature changes during the cold months in the Yellow River Basin was analyzed. This study provides a scientific basis for improving ERA5-Land temperature reanalysis products in the Yellow River Basin and offers theoretical support for climate change research in the region.

Spatiotemporal modelling of airborne birch and grass pollen concentration across Switzerland: A comparison of statistical, machine learning and ensemble methods

BackgroundStatistical and machine learning models are commonly used to estimate spatial and temporal variability in exposure to environmental stressors, supporting epidemiological studies. We aimed to compare the performances, strengths and limitations of six different algorithms in the retrospective spatiotemporal modeling of daily birch and grass pollen concentrations at a spatial resolution of 1 km across Switzerland. MethodsDaily birch and grass pollen concentrations were available from 14 measurement sites in Switzerland for 2000–2019. To develop the spatiotemporal models, we considered spatiotemporal, spatial and temporal predictors including meteorological factors, land-use, elevation, species distribution and Normalized Difference Vegetation Index (NDVI). We used six statistical and machine learning algorithms: LASSO, Ridge, Elastic net, Random forest, XGBoost and ANNs. We optimized model structures through feature selection and grid search techniques to obtain the best predictive performance. We used train-test split and cross-validation to avoid overfitting and overoptimistic performance indicators. We then combined these six models through multiple linear regression to develop an ensemble hybrid model. ResultsThe 5th-95th percentiles of birch and grass pollen concentrations were 0–151 and 0–105 grains/m3, respectively. The hybrid ensemble model achieved the best RMSE on the test dataset for both birch and grass pollen with 94.4 and 19.7 grains/m3, respectively.Nonlinear models (Random forest, XGBoost and ANNs) achieved lower test RMSE's than linear models (LASSO, Ridge, Elastic net) for both pollen types, with RMSE's ranging from 105.9 to 140.5 grains/m3 for birch and from 20.0 to 25.4 grains/m3 for grass pollen. The Random forest algorithm yielded the best spatial and temporal performance among the six evaluated modelling methods. The ensemble hybrid model outperformed the six linear and nonlinear algorithms. Country-wide pollen concentration, land use, weather, and NDVI were important predictors. ConclusionNonlinear algorithms outperformed linear models and accurately explained complex, nonlinear relationships between environmental factors and measured concentrations.

Low-Rank Robust Subspace Tensor Clustering for Metro Passenger Flow Modeling

Tensor clustering has become an important topic, specifically in spatiotemporal modeling, because of its ability to cluster spatial modes (e.g., stations or road segments) and temporal modes (e.g., time of day or day of the week). Our motivating example is from subway passenger flow modeling, where similarities between stations are commonly found. However, the challenges lie in the innate high-dimensionality of tensors and also the potential existence of anomalies. This is because the three tasks, that is, dimension reduction, clustering, and anomaly decomposition, are intercorrelated with each other, and treating them in a separate manner will render a suboptimal performance. Thus, in this work, we design a tensor-based subspace clustering and anomaly decomposition technique for simultaneous outlier-robust dimension reduction and clustering for high-dimensional tensors. To achieve this, a novel low-rank robust subspace clustering decomposition model is proposed by combining Tucker decomposition, sparse anomaly decomposition, and subspace clustering. An effective algorithm based on Block Coordinate Descent is proposed to update the parameters. Prudent experiments prove the effectiveness of the proposed framework via the simulation study, with a gain of +25% clustering accuracy over benchmark methods in a hard case. The interrelations of the three tasks are also analyzed via ablation studies, validating the interrelation assumption. Moreover, a case study in station clustering based on real passenger flow data is conducted, with quite valuable insights discovered. History: Bianca Maria Colosimo served as the senior editor for this article. Funding: Dr. Yan is partially funded by DOE [DE-EE0009354] and NSF [CMMI 2316654]. Dr. Fugee is partially funded with the RGC [GRF 16201718 and 16216119]. The authors appreciate the help from the Hong Kong MTR Co. research, marketing, and customer service teams. Data Ethics & Reproducibility Note: The code capsule is available on Code Ocean at https://codeocean.com/capsule/6536164/tree and in the e-Companion to this article (available at https://doi.org/10.1287/ijds.2022.0028 ).

LLMDiff: Diffusion Model Using Frozen LLM Transformers for Precipitation Nowcasting.

Precipitation nowcasting, which involves the short-term, high-resolution prediction of rainfall, plays a crucial role in various real-world applications. In recent years, researchers have increasingly utilized deep learning-based methods in precipitation nowcasting. The exponential growth of spatiotemporal observation data has heightened interest in recent advancements such as denoising diffusion models, which offer appealing prospects due to their inherent probabilistic nature that aligns well with the complexities of weather forecasting. Successful application of diffusion models in rainfall prediction tasks requires relevant conditions and effective utilization to direct the forecasting process of the diffusion model. In this paper, we propose a probabilistic spatiotemporal model for precipitation nowcasting, named LLMDiff. The architecture of LLMDiff includes two networks: a conditional encoder-decoder network and a denoising network. The conditional network provides conditional information to guide the denoising network for high-quality predictions related to real-world earth systems. Additionally, we utilize a frozen transformer block from pre-trained large language models (LLMs) in the denoising network as a universal visual encoder layer, which enables the accurate estimation of motion trend by considering long-term temporal context information and capturing temporal dependencies within the frame sequence. Our experimental results demonstrate that LLMDiff outperforms state-of-the-art models on the SEVIR dataset.

Improvement of Spatio-Temporal Inconsistency of Time Series Land Cover Products

In recent years, time series land cover products have been developed rapidly. However, the traditional classification strategy rarely considers time continuity and spatial consistency, which leads to the existence of unreasonable changes among the multi-period products. In order to solve the existing problems, this paper proposes a matrix decomposition model and an optimized hidden Markov model (HMM) to improve the consistency of the time series land cover maps. It also compares the results with the spatio-temporal window filtering model. The spatial weight information is introduced into the singular value decomposition (SVD) model, and the regression model is constructed by combining the eigenvalues and eigenvectors of the image to predict the unreasonable variable pixels and complete the construction of the matrix decomposition model. To solve the two problems of reliance on expert experience and lack of spatial relationships, this paper optimizes the model and proposes the HMM Land Cover Transition (HMM_LCT) model. The overall accuracy of the matrix decomposition model and the HMM_LCT model is 90.74% and 89.87%, respectively. It is found that the matrix decomposition model has a better effect on consistency adjustment than the HMM_LCT model. The matrix decomposition model can also adjust the land cover trajectory to better express the changing trend of surface objects. After consistent adjustment by the matrix decomposition model, the cumulative proportion of the first 15 types of land cover trajectories reached 99.47%, of which 83.01% were stable land classes that had not changed for three years.

MeST-Former: Motion-enhanced Spatiotemporal Transformer for generalizable Deepfake detection

The rise of Deepfake technology has sparked significant concerns due to its potential for misuse and malicious manipulation of multimedia content. Various detection approaches aimed at detecting Deepfake videos have been proposed, mostly relying on the identification of spatial and temporal artifacts. However, due to the different contexts of source images and the variety of generation techniques, current Deepfake detection methods usually perform well on training datasets, and yet generalize poorly to those unseen identities in new datasets. This issue is widely known as the generalization challenge of Deepfake detection. To address this challenge, this paper proposes an advanced spatiotemporal Deepfake video detector, named Motion-enhanced Spatiotemporal Transformer (MeST-Former). MeST-Former is based on the spatiotemporal modeling capacity of the video Swin Transformer. The spatial and temporal features are obtained from the RGB and motion images, respectively. To enhance the generalization ability of MeST-Former to unseen identities in unseen datasets, the ID-related components in the spatial and temporal features are detached. Specifically, MeST-Former adopts the newly proposed Identity-Decoupling Attention (IDC-Att) module to disentangle the ID-related and ID-unrelated components. Only the ID-unrelated components are used to construct more generalizable spatiotemporal representations. This process makes the constructed spatiotemporal features identity-agnostic and more generalizable to unseen identities. We conducted extensive experiments to evaluate the performance of the MeST-Former. Our results indicate that MeST-Former achieves accurate and generalizable Deepfake detection performance. Notably, MeST-Former also demonstrates high efficacy in detecting AI-animated talking-head videos.

Multi-granularity spatiotemporal object modelling of waterborne traffic elements

The electronic navigational charts are crucial carriers for representing the multi-source heterogeneous data of Waterborne Traffic Elements (WTEs). However, their layer-based modelling method has some shortcomings in expressing the multi-granularity features, complex relationships, and dynamic evolution of elements. This paper proposes an objectification modelling method for WTEs based on the concept of multi-granularity spatiotemporal object modelling. A classification system for waterborne traffic objects is developed based on the relevance of behavior to elements; combining characteristics of waterborne traffic, a data model for waterborne traffic objects is constructed from eight aspects: spatiotemporal reference, spatiotemporal position, spatial form, basic information, attributes, behavioral ability, structure, and associative relationships. An object extraction function is also established, extracting object attributes and relationships between objects according to different element classes. Taking the Jiashan section of the Hangzhou-Shanghai Line in Zhejiang Province as the experimental subject, the multi-granularity spatiotemporal characteristics, dynamic evolution, and relationship expression of channel class objects are tested. The experimental results show that the proposed method provides the theoretical basis and data organization mode for the multi-granularity expression of WTEs.

A roadmap for generating annual bycatch estimates from sparse at-sea observer data

Abstract To support ecosystem-based fisheries management, monitoring data from at-sea observer (ASO) programs should be leveraged to understand the impact of fisheries on discarded species (bycatch). Available techniques to estimate fishery-scale quantities from observations range from simple mean estimators to more complex spatiotemporal models, each making assumptions with differing degrees of support. However, the resulting implementation and analytical trade-offs are rarely discussed when applying these techniques in practice. Using blue shark (Prionace glauca) bycatch in the Canadian pelagic longline fishery as a case study, we evaluated the performance of seven contrasting approaches to estimating total annual discard amounts and assessed their trade-offs in application. Results demonstrated that simple approaches such as mean estimator and nearest neighbors are feasible to implement and can be as efficient for prediction as complex models such as random forest and mixed-effects models. The traditionally used catch-ratio estimator consistently underperformed among all tested models, likely due to misspecified correlative relationships between target and bycatch species. Overall, efforts in model-based approaches were rewarded with very small gains in predictive ability, suggesting that such models relying on environmental, biological, spatial, and/or temporal patterns to improve prediction of bycatch may lack sufficient foundation in data-limited contexts.

Spatio-Temporal Heterogeneous Ensemble Learning Method for Predicting Land Subsidence

The prediction of land subsidence is of significant value for the early warning and prevention of geological disasters. Although numerous land subsidence prediction methods are currently available, two obstacles still exist: (i) spatio-temporal heterogeneity of land subsidence is not well considered, and (ii) the prediction performance of individual models is unsatisfactory when the data do not meet their assumptions. To address these issues, we developed a spatio-temporal heterogeneous ensemble learning method for predicting land subsidence. Firstly, a two-stage hybrid spatio-temporal clustering method was proposed to divide the dataset into internally homogeneous spatio-temporal clusters. Secondly, within each spatio-temporal cluster, an ensemble learning strategy was employed to combine one time series prediction model and three spatio-temporal prediction models to reduce the prediction uncertainty of an individual model. Experiments on a land subsidence dataset from Cangzhou, China, show that the prediction accuracy of the proposed method is significantly higher than that of four individual prediction models.