Spatial Prediction Models Research Articles

Spatial variation in housing construction material in low- and middle-income countries: A Bayesian spatial prediction model of a key infectious diseases risk factor and social determinant of health.

Housing infrastructure and quality is a major determinant of infectious disease risk and other health outcomes in regions where vector borne, waterborne and neglected tropical diseases are endemic. It is important to quantify the geographical distribution of improvements to dwelling components to identify and target resources towards populations at risk. This study aimed to model the sub-national spatial variation in housing materials using covariates with quasi-global coverage and use the resulting estimates to map predicted coverage across the world's low- and middle-income countries. Data on materials used in dwelling construction were sourced from nationally representative household surveys conducted since 2005. Materials used for construction of flooring, walls, and roofs were reclassified as improved or unimproved. Households lacking location information were georeferenced using a novel methodology. Environmental and demographic spatial covariates were extracted at those locations for use as model predictors. Integrated nested Laplace approximation models were fitted to obtain, and map predicted probabilities for each dwelling component. The dataset compiled included information from households in 283,000 clusters from 350 surveys. Low coverage of improved housing was predicted across the Sahel and southern Sahara regions of Africa, much of inland Amazonia, and areas of the Tibetan plateau. Coverage of improved roofs and walls was high in the Central Asia, East Asia and Pacific and Latin America and the Caribbean regions. Improvements in all three components, but most notably floors, was low in Sub-Saharan Africa. The strongest determinants of dwelling component quality related to urbanization and economic development, suggesting that programs should focus on supply-side interventions, providing resources for housing improvements directly to the populations that need them. These findings are made available to researchers as files that can be imported into a GIS for integration into relevant analyses to derive improved estimates of preventable health burdens attributed to housing.

Spatial Prediction of Soil Total Phosphorus in a Karst Area: Comparing GWR and Residual-Centered Kriging

Accurate soil total phosphorus (TP) prediction is essential to support sustainable agricultural practices and formulate ecological conservation protection policies, particularly in complex karst landscapes with high spatial variability and high phosphorus and cadmium content and interactions, complicating nutrient management. This study uses GIS and geostatistical methods to analyze the spatial distribution, influencing factors, and predictive modeling of soil TP in the karst region of northern Mashan County, Guangxi, China. Using 427 surface soil samples, we developed five predictive models: ordinary kriging (OK), regression kriging (RK) and geographically weighted regression kriging (GWRK) combined with environmental variables such as land uses, soil types, and topographic factors; residual mean-centered kriging (MM_OK), and residual median-centered kriging (MC_OK). Our results indicate that higher TP levels were observed in agricultural lands (paddy fields and dry land, at 766 and 913 mg·kg−1, respectively) may due to fertilization, while forests and shrublands showed lower TP levels (383 and 686 mg·kg−1, respectively), reflecting natural phosphorus cycling. The high-value areas of soil TP concentration are in the karst areas in the west and east of the study area, and the low-value area is in the Hongshui River valley in the north of Mashan. The spatial distribution of soil TP is affected by land use, soil type, and topography. The GWRK model exhibited superior accuracy (80.6%), with predicted concentration of TP closely aligning with observed TP values, effectively capturing fine spatial variations, and showing the lowest mean standardized error, average standard error, and mean absolute error. GWRK also achieved the highest R2 (0.67), demonstrating robust predictive capability. MM_OK and MC_OK models performed well and showed smoother spatial transitions, while the OK model displayed the lowest predictive accuracy (62%). By utilizing spatially adaptive weighting, GWRK and its residual-centered kriging method improve soil TP’s prediction accuracy and smoothness in karst areas, providing a reference for targeted soil conservation and sustainable agricultural practices in spatially complex karst environments.

Predicting origin-destination flows by considering heterogeneous mobility patterns

The accurate prediction of origin-destination (OD) flows is essential for advancing sustainable urban mobility and supporting resilient urban planning. However, the inherent heterogeneity of mobility patterns results in complex geographic unit relations, diverse spatial organizational structures, and the long-tailed effect on OD flow distribution. This study proposes a novel OD flow prediction method based on graph-based deep learning (named as HMCG-LGBM). Specifically, 1) a modularity-based graph reconstruction strategy is presented for geographic unit relation augmentation by eliminating weak connections; 2) the heterogeneous spatial organization of OD flows is captured by combining the community detection approach and graph attention mechanism with the introduction of socio-economic and spatial features; and 3) a weighted loss function with distribution smoothing paradigm is developed to enhance the prediction for low-probability mobility events, addressing the challenges posed by long-tailed distributions. Extensive experiments conducted on real-world datasets show that the predictive performance of the proposed method is significantly improved, with the RMSE and MAE reduced from the baselines by 11.1%–33.3% and 14.1%–22.2%, respectively. The results also demonstrate the robustness of the proposed method for dealing with imbalanced OD flow distributions, providing valuable insights for spatial interaction predictive modeling in the context of sustainable urban systems.

Simulating multi-scale optimization and variable selection in species distribution modeling

Species distribution modeling (SDM) is a fundamental tool in theoretical and applied ecology. However, relatively little is known about the performance of different approaches for scale optimization, model selection, and algorithmic prediction in the context of nonlinear, multiscale and interactive relationships between environmental variables and species occurrence. Modelers often struggle to optimize a tradeoff between ecological relevance, model robustness, complexity, and overfitting. In this paper, we investigated several methods designed to optimize spatial scale and variable selection in SDMs, in each case evaluating model fitness, parsimony and predictive performance. We used a simulation approach to produce a large pool of alternative underlying habitat relationships that reflect a broad range of realistic habitat associations. We also compared several different modeling algorithms, including logistic regression with a generalized linear model (GLM), Lasso and Elastic-Net Regularized GLMs (GLMNet), and random forest (RF), as well as alternative variable and scale selection methods. We found that GLM methods employing all-subsets dredge routines for variable selection were consistently the best predictors based on all criteria of our model performance assessment and across all attributes of the simulated underlying relationship, including nonlinearity and interaction. We had expected machine learning approaches, such as random forest, to perform better in these more complex forms of species-environment relationships. GLM using dredge variable selection was also the method that included the fewest spurious covariates and included the most correct predictors as a proportion of all predictors. We found that univariate scaling was the most robust method of variable and scale selection, along with Minimal Redundancy Maximal Relevancy (MRMR) which performed equivalently. The simulation experiment presented here provides a robust assessment of simulated multi-species distribution model performance, complexity and fidelity. By simulating a large range of potential habitat relationships with varying spatial scale, effect sizes, linearity, and interactions, we comprehensively evaluated model performance across gradients of complexity of the underlying relationships and violations of classical statistical assumptions. This study provides a valuable assessment and a broader example of the power and utility of controlled simulation experiments in habitat relationships and other ecological spatial predictive modeling.

Spatial correlation assessment of multiple earthquake intensity measures using physics-based simulated ground motions

Predictive models for spatial correlation play an effective role in the assessment of seismic risk associated with distributed infrastructure and building portfolios. However, existing models often rely on simplified approaches, assuming isotropy and stationarity. This paper verifies these assumptions by presenting a comprehensive study using a database of 3D physics-based simulated broadband ground motions for Istanbul, generated by the SPEED software. The results reveal significant event-to-event variability and nonstationary and anisotropic characteristics of spatial correlation influenced by source, path, and site effects. The development of nonstationary correlation models requires exploring influential metrics beyond spatial proximity and gaining a deep understanding of their impact, which is the focus of this study. Analysis of the spatial correlations of peak ground displacement, peak ground velocity, peak ground acceleration, and response spectral accelerations at different periods, employing both stationary and nonstationary correlation modelling methods and considering the finite fault model, indicates that the slip distribution pattern, direction and distance of station pairs relative to earthquake rupture, soil softness, and homogeneity of soil properties significantly influence the spatial correlations of near-field earthquake ground motions. Implementation of the introduced parameters in predictive spatial correlation models enhances the precision of regional seismic hazard assessments.

Spatial Mapping and Prediction of Groundwater Quality Using Ensemble Learning Models and SHapley Additive exPlanations with Spatial Uncertainty Analysis

The spatial mapping and prediction of groundwater quality (GWQ) is important for sustainable groundwater management, but several research gaps remain unexplored, including the inaccuracy of spatial interpolation, limited consideration of the geological environment and human activity effects, limitation to specific pollutants, and unsystematic indicator selection. This study utilized the entropy-weighted water quality index (EWQI), the LightGBM model, the pressure-state-response (PSR) framework and SHapley Additive exPlanations (SHAP) analysis to address the above research gaps. The normalized importance (NI) shows that NO3− (0.208), Mg2+ (0.143), SO42− (0.110), Cr6+ (0.109) and Na+ (0.095) should be prioritized as parameters for remediation, and the skewness EWQI distribution indicates that although most sampled locations have acceptable GWQ, a few areas suffer from severely poor GWQ. The PSR framework identifies 13 indicators from geological environments and human activities for the SMP of GWQ. Despite high AUROCs (0.9074, 0.8981, 0.8885, 0.9043) across four random training and testing sets, it was surprising that significant spatial uncertainty was observed, with Pearson correlation coefficients (PCCs) from 0.5365 to 0.8066. We addressed this issue by using the spatial-grid average probabilities of four maps. Additionally, population and nighttime light are key indicators, while net recharge, land use and cover (LULC), and the degree of urbanization have the lowest importance. SHAP analysis highlights both positive and negative impacts of human activities on GWQ, identifying point-source pollution as the main cause of the poor GWQ in the study area. Due to the limited research on this field, future studies should focus on six key aspects: multi-method GWQ assessment, quantitative relationships between indicators and GWQ, comparisons of various spatial mapping and prediction models, the application of the PSR framework for indicator selection, the development of methods to reduce spatial uncertainty, and the use of explainable machine learning techniques in groundwater management.

A two-stage spatial prediction modeling approach based on graph neural networks and neural processes

Spatial prediction models hold significant application value in fields such as environmental science, economic development, and geological exploration. With advancements in deep learning technology, graph neural networks (GNNs) offer a powerful and scalable solution for spatial data modeling. In these models, each spatial location is represented as a vertex, with the explanatory variables at each point converted into vertex features, and the vertex label corresponding to the target value at that point. GNNs utilize this representation to predict vertex labels, thereby performing spatial prediction. However, the residuals of GNN predictions are found to still exhibit spatial autocorrelation, indicating that traditional GNNs only achieve suboptimal results in terms of capturing complex spatial relationships. In this paper, we propose a two-stage spatial prediction method called Location Embedded Graph Neural Networks-Residual Neural Processes (LEGNN-RNP) to address this challenge. In the first stage, we employ LEGNNs, a new framework that integrates location features into GNNs through an attention mechanism, enhancing the learning of complex spatial models by considering both spatial context and attribute features. In the second stage, we model the residuals from LEGNN predictions to further extract spatial patterns from the data. Specifically, we introduce RNP, a neural process-based approach to model the Gaussian Process (GP) distribution of residuals. The distribution parameters (i.e., mean and covariance) are parameterized by a neural network, where the mean estimates the residual and the variance quantifies the prediction uncertainty. Experiments on four datasets demonstrate that our proposed two-stage method achieves state-of-the-art results by effectively extracting spatial relationships and significantly improving spatial prediction accuracy.

Spatial computational modelling illuminates the role of the tumour microenvironment for treating glioblastoma with immunotherapies

Glioblastoma is the most common and deadliest brain tumour in adults, with a median survival of 15 months under the current standard of care. Immunotherapies like immune checkpoint inhibitors and oncolytic viruses have been extensively studied to improve this endpoint. However, most thus far have failed. To improve the efficacy of immunotherapies to treat glioblastoma, new single-cell imaging modalities like imaging mass cytometry can be leveraged and integrated with computational models. This enables a better understanding of the tumour microenvironment and its role in treatment success or failure in this hard-to-treat tumour. Here, we implemented an agent-based model that allows for spatial predictions of combination chemotherapy, oncolytic virus, and immune checkpoint inhibitors against glioblastoma. We initialised our model with patient imaging mass cytometry data to predict patient-specific responses and found that oncolytic viruses drive combination treatment responses determined by intratumoral cell density. We found that tumours with higher tumour cell density responded better to treatment. When fixing the number of cancer cells, treatment efficacy was shown to be a function of CD4 + T cell and, to a lesser extent, of macrophage counts. Critically, our simulations show that care must be put into the integration of spatial data and agent-based models to effectively capture intratumoral dynamics. Together, this study emphasizes the use of predictive spatial modelling to better understand cancer immunotherapy treatment dynamics, while highlighting key factors to consider during model design and implementation.

KNNDM CV: k-fold nearest-neighbour distance matching cross-validation for map accuracy estimation

Abstract. Random and spatial cross-validation (CV) methods are commonly used to evaluate machine-learning-based spatial prediction models, and the performance values obtained are often interpreted as map accuracy estimates. However, the appropriateness of such approaches is currently the subject of controversy. For the common case where no probability sample for validation purposes is available, in Milà et al. (2022) we proposed the nearest-neighbour distance matching (NNDM) leave-one-out (LOO) CV method. This method produces a distribution of geographical nearest-neighbour distances (NNDs) between test and training locations during CV that matches the distribution of NNDs between prediction and training locations. Hence, it creates predictive conditions during CV that are comparable to what is required when predicting a defined area. Although NNDM LOO CV produced largely reliable map accuracy estimates in our analysis, as a LOO-based method, it cannot be applied to the large datasets found in many studies. Here, we propose a novel k-fold CV strategy for map accuracy estimation inspired by the concepts of NNDM LOO CV: the k-fold NNDM (kNNDM) CV. The kNNDM algorithm tries to find a k-fold configuration such that the empirical cumulative distribution function (ECDF) of NNDs between test and training locations during CV is matched to the ECDF of NNDs between prediction and training locations. We tested kNNDM CV in a simulation study with different sampling distributions and compared it to other CV methods including NNDM LOO CV. We found that kNNDM CV performed similarly to NNDM LOO CV and produced reasonably reliable map accuracy estimates across sampling patterns. However, compared to NNDM LOO CV, kNNDM resulted in significantly reduced computation times. In an experiment using 4000 strongly clustered training points, kNNDM CV reduced the time spent on fold assignment and model training from 4.8 d to 1.2 min. Furthermore, we found a positive association between the quality of the match of the two ECDFs in kNNDM and the reliability of the map accuracy estimates. kNNDM provided the advantages of our original NNDM LOO CV strategy while bypassing its sample size limitations.

Near real-time spatial prediction of earthquake-triggered landslides based on global inventories from 2008 to 2022

Near real-time prediction of earthquake-triggered landslides can rapidly forecast the spatial distribution of coseismic landslides just after a great earthquake, and provide effective support for emergency response. However, the prediction of earthquake-triggered landslides has always been a great challenge because of low accuracy and high false alarms. This work proposes a novel fuzzy deep learning (FuDL) model for near real-time earthquake-triggered landslide spatial prediction. Fuzzy learning theory is for the first time employed in earthquake-triggered landslide prediction. The FuDL has high generalization and robustness, effectively improving the accuracy of earthquake-triggered landslide prediction. Eighteen earthquake-triggered landslide inventories worldwide from 2008 to 2022 are employed to conduct ETL prediction. According to the chronological order, 15 earthquake-triggered landslides from 2008 to 2018 are adopted to train the FuDL model, and 3 earthquake-triggered landslides from 2019 to 2022 are utilized for near real-time earthquake-triggered landslide prediction. Furthermore, this work reveals that ground movement, relatively steep and high topography, and strong seismic intensity are critical factors affecting the spatial distribution of earthquake-triggered landslides. In addition, this work conducted a detailed analysis of the distribution patterns of earthquake-triggered landslides on a global scale.

Landslide hazard assessment for the Batxat area of Vietnam using GIS-based spatial prediction models

Located in the northwest of Laocai province, Batxat district has been frequently affected by natural disasters, including landslides and debris flows. Therefore, landslide hazard assessment (LHA) has been a significant task for planning, economic development, and minimizing human and property damage. For this purpose, landslide hazard maps were established in this study using the Analytic Hierarchy Process (AHP) and the combined Analytic Hierarchy Process - Frequency Ratio (AHP&FR) models. Ten landslide-related factors were selected, including elevation, slope, distance to road, distance to drainage, land use and land cover (LULC), average monthly rainfall, lithology, aspect, distance to fault, and relative relief. Afterwards, the weighted value of landslide-related factors and the landslide susceptibility index (LSI) were determined using the Analytic Hierarchy Process. The Frequency Ratio method was used to calculate the weighted value of factor classes. Two landslide hazard maps were established, and the study area was divided into five hazard zones: very low, low, moderate, high, and very high. The performance of the models was determined using the area under the curve (AUC) of the receiver operating characteristic (ROC), the seed cell area index (SCAI), and the precision of the predicted results (P). The AUC values for the success rate of these models were 0.72 and 0.75, and for the prediction rate were 0.67 and 0.70, respectively. The evaluation results of the models showed that, although both the AHP and combined AHP&FR models have good performance for landslide hazard mapping, the AHP&FR model produces more accurate outcomes than the AHP model.

Mapping human‐carnivore coexistence: approaches to integrating anthropogenic influences on carnivore distribution and connectivity modelling

AbstractInterdisciplinary approaches to modelling human‐wildlife coexistence have gained much attention in recent years, especially regarding carnivores which are subject to historical human persecution and dynamic attitudes and cultural values. Accordingly, there have been many attempts to gain insights into anthropogenic influences on carnivores through the use of species distribution models and connectivity analyses. However, there have been no global reviews on how anthropogenic influences are incorporated into such models. We conducted a literature review of global terrestrial carnivore species distribution and connectivity studies from 1995 to 2021 to evaluate the type of measure, frequency of use and spatial scale of the predictor variables used to describe environmental conditions and assess anthropogenic influences in such predictive models. We evaluated 2495 variables from 263 carnivore distribution and connectivity studies and found that variables used to describe environmental conditions are four times more commonly used than those used to assess anthropogenic influences on carnivores. However, the number of anthropogenic covariates used in carnivore distribution and connectivity studies has increased by 4.7% per year, and the spatial resolution of those covariates has simultaneously decreased by 7.6% per year. We observed that anthropogenic covariates were used more in studies occurring in the global north and for larger‐bodied carnivores (>15 kg), suggesting the need for better integration of anthropogenic metrics into predictive models more widely and for smaller‐bodied carnivores. To help guide and advance the further use of sociological data into predictive spatial models for carnivores, we propose a framework that emphasizes integration of sociological methodologies and data sources ranging from exploratory expert elicitation to targeted focus group interviews at multiple phases in the conservation planning process.

Spatial mapping and driving factor Identification for salt-affected soils at continental scale using Machine learning methods

Soil salinization is one of the most serious land-degrading threats globally, particularly in Asia, where there are 21 countries, e.g., Kazakhstan, China, Iran, and Indonesia, lack of accurate spatial information of salt-affected soils currently. Mapping the distribution of salt-affected soils and identifying their main driving factors is critical for sustainable development. Current studies for mapping the spatial distribution of salt-affected soils focused on recognizing the soil salinity by using the indicator of electrical conductivity of saturated soil paste extract (ECe), while cannot distinguish the soil sodicity. Meanwhile, the absence of uncertainty assessment is another problem for large-scale studies. To address these gaps, we developed an effective spatial prediction model that considered both soil salinity and sodicity using machine learning (ML) models and soil salinization data. This study used ECe, pH, and exchangeable sodium percent (ESP) as indicators, with sample sizes of 22,803, 30,821, and 12,961, respectively, to produce salt-affected soil maps with uncertainty assessments in Asia, and conducted the analysis of the drivers of soil salinization. The ML models involved cubist, random forest, quantile regression forest, quantile regression neural network, and a simple multiple linear regression model for comparison. The results showed that among these models, quantile regression forest exhibited the best performance, with mean absolute percentage error values of 2.54 %∼52.06 %, coefficient of determination values of 0.40 ∼ 0.74, and Nash-Sutcliffe efficiency values of 0.36 ∼ 0.71. The predicted maps, with 1 km resolution, demonstrated minimal uncertainty across over 60.4 % of the study area. Approximately 466.36 million hectares of topsoil (0–30 cm) and 377.68 million hectares of subsoil (30–100 cm) in Asia are salt-affected, mainly concentrated in arid, semi-arid, and coastal regions. Climatic factors (the annual precipitation and maximum temperature), topography (digital elevation model and slope), and salinity index V (SI5) were identified as the main drivers of soil salinization. Overall, this study provides important guidelines for soil salinization management and achieving ecological sustainability.

Unified Spatial Clustering of Territory Risk to Uncover Impact of COVID-19 Pandemic on Major Coverages of Auto Insurance

This research delves into the fusion of spatial clustering and predictive modeling within auto insurance data analytics. The primary focus of this research is on addressing challenges stemming from the dynamic nature of spatial patterns in multiple accident year claim data, by using spatially constrained clustering. The spatially constrained clustering is implemented under hierarchical clustering with a soft contiguity constraint. It is highly desirable for insurance companies and insurance regulators to be able to make meaningful comparisons of loss patterns obtained from multiple reporting years that summarize multiple accident year loss metrics. By integrating spatial clustering techniques, the study not only improves the credibility of predictive models but also introduces a strategic dimension reduction method that concurrently enhances the interpretability of predictive models used. The evolving nature of spatial patterns over time poses a significant barrier to a better understanding of complex insurance systems as these patterns transform due to various factors. While spatial clustering effectively identifies regions with similar loss data characteristics, maintaining up-to-date clusters is an ongoing challenge. This research underscores the importance of studying spatial patterns of auto insurance claim data across major insurance coverage types, including Accident Benefits (AB), Collision (CL), and Third-Party Liability (TPL). The research offers regulators valuable insights into distinct risk profiles associated with different coverage categories and territories. By leveraging spatial loss data from pre-pandemic and pandemic periods, this study also aims to uncover the impact of the COVID-19 pandemic on auto insurance claims of major coverage types. From this perspective, we observe a statistically significant increase in insurance premiums for CL coverage after the pandemic. The proposed unified spatial clustering method incorporates a relabeling strategy to standardize comparisons across different accident years, contributing to a more robust understanding of the pandemic effects on auto insurance claims. This innovative approach has the potential to significantly influence data visualization and pattern recognition, thereby improving the reliability and interpretability of clustering methods.

A SHAP-enhanced XGBoost model for interpretable prediction of coseismic landslides

In the coseismic landslide hazard assessment (CLHA), advanced machine learning (ML) models have garnered significant attention due to their effectiveness in handling the complex relationships between landslides and various influencing factors. However, explaining the decision-making mechanisms of machine learning models that predict landslide spatial distribution based on these influencing factors remains challenging. This study compares the predictive performance of four models—XGBoost, RF, LR, and SVM—optimized using the TPE algorithm, and introduces the SHAP algorithm into the XGBoost model to achieve both global and local interpretations of CLHA. In various tests, the optimized XGBoost model demonstrated the best predictive performance, achieving an accuracy of 0.864 and an AUC value of 0.886. Global interpretations indicate that the occurrence of coseismic landslides is primarily influenced by triggering factors such as hypocentral distance and distance from the seismogenic fault. Terrain roughness and elevation, on the other hand, make significant contributions among the conditioning factors. Single-factor dependence plots indicate that the contribution of individual factors to landslides varies across different ranges of their feature values. Analysis of two-factor dependence plots reveals that interactions between factors are also crucial in influencing the occurrence of landslides. Combining field surveys with local interpretations confirms significant variations in the contributions of influencing factors within local ranges. The main innovation of this study lies in the integration of the SHAP algorithm into the CLHA model, revealing the decision-making mechanisms of the model for spatial prediction of coseismic landslides.

Results of a Nationally Representative Seroprevalence Survey of Chikungunya Virus in Bangladesh.

There is an increasing global burden from chikungunya virus (CHIKV). Bangladesh reported a major epidemic in 2017, but it was unclear whether there had been prior widespread transmission. We conducted a nationally representative seroprevalence survey in 70 randomly selected communities immediately before the epidemic. We found that 69 of 2938 sampled individuals (2.4%) were seropositive to CHIKV. Seropositivity to dengue virus (adjusted odds ratio, 3.13 [95% confidence interval, 1.86-5.27]), male sex (0.59 [.36-.99]), and community presence of Aedes aegypti mosquitoes (1.80 [1.05-3.0]7) were significantly associated with CHIKV seropositivity. Using a spatial prediction model, we estimated that across the country, 4.99 (95% confidence interval, 4.89-5.08) million people had been previously infected. These findings highlight high population susceptibility before the major outbreak and that previous outbreaks must have been spatially isolated.

Characteristics and prediction of urban interaction networks from the perspective of traffic flow and text information flow

The degree of urban development depends on the degree of closeness between cities, which is reflected in the strength of interactions between cities, such as traffic and information flows. In this research, we compare and analyze the characteristics of urban interaction networks (UINs) at two spatial scales in Shandong Province and the whole country from two different perspectives of traffic flow and information flow and validate the spatial interaction characteristics reflected in the material space traffic flow from the perspective of textual spatial information flow. The UIN is constructed based on Tencent migration big data, and the assortative coefficient method is introduced to explore the assortative and interaction characteristics between core cities and edge cities in the traffic flow network. Introducing deep learning methods on a larger scale, the GCN_CD model is proposed for semi-supervised classification of nodes to realize community discovery for both traffic flow and information flow networks. The spatial interaction intensity prediction model [Formula: see text] is constructed taking into account geographic features, which improves the prediction accuracy. The results show that from the perspectives of traffic flow and text information flow, the urban interactive network in Shandong Province has the characteristics of Scale-Free and Small-World, showing certain homogeneity and strong spatial interaction. Shandong Province’s UIN formed Jinan, Qingdao, and other cities as the core, satellite cities around, the east and west ends of the remote echo structure. From the perspective of traffic flow, the national urban interactive network presents a jumping distribution, with a single core city as the dominant distribution structure. From the perspective of information flow, the dividing line between the eastern and western communities is obvious, and the internal aggregation of the community is strong. Distance attenuation effects have an impact on the strength of spatial interactions.

MGAtt-LSTM: A multi-scale spatial correlation prediction model of PM2.5 concentration based on multi-graph attention

The increase in air pollution has posed numerous new challenges for human society, making the exploration of an effective method for predicting air pollutant concentrations highly significant. The current research faces several primary challenges: the neglect of non-Euclidean characteristics of site distribution on data and the strong spatiotemporal dependencies in the dispersion process of pollutants. To address these issues, this paper constructs a spatiotemporal hybrid prediction model – the MGAtt-LSTM method – for predicting PM2.5 concentrations, which employs the dynamic multi-graph attention module (MGAtt) to tackle spatial dependencies and Long Short-Term Memory networks (LSTM) to address temporal dependencies. Additionally, extensive experiments are conducted by using historical air pollutant monitoring data and meteorological data from the Beijing-Tianjin-Hebei region. The results demonstrate that the proposed MGAtt-LSTM model achieved superior performance in concentration prediction compared to existing benchmark models.

Spatial Data Prediction Model Integrated with K-Nearest Neighbor Mechanism in Neural Networks

Traditional methods of spatial interpolation, such as inverse distance weighting (IDW) and ordinary Kriging (OK), utilize geographic distance and specific assumptions to simplify the computation of geospatial data complexity. Nevertheless, these conventional approaches are not as practical in obtaining high-precision estimation because of the intricate nonlinear relationship between geographic distance and correlation weights. In this study, a novel spatial interpolation technique, named WFNNKM, is introduced, which integrates the K-nearest neighbor (KNN) mechanism with a neural network to address this challenge. Firstly, the Lebesgue integral is used for clustering, and KNN tuples are obtained by clustering. Secondly, the KNN training task is constructed for interpolation points, and the bias parameters of each point are obtained through training. Finally, the pre-training parameters of a neural network are modified through bias parameters of nearest neighbors to obtain accurate prediction attribute values. In comparison with two conventional methods and three neural network approaches across three soil sample datasets, the results demonstrate a notably superior performance of the suggested approach compared to the five interpolation methods.

Geospatial artificial intelligence for estimating daytime and nighttime nitrogen dioxide concentration variations in Taiwan: A spatial prediction model

Nitrogen dioxide (NO2) is a major air pollutant primarily emitted from traffic and industrial activities, posing health risks. However, current air pollution models often underestimate exposure risks by neglecting the bimodal pattern of NO2 levels throughout the day. This study aimed to address this gap by developing ensemble mixed spatial models (EMSM) using geo-artificial intelligence (Geo-AI) to examine the spatial and temporal variations of NO2 concentrations at a high resolution of 50m. These EMSMs integrated spatial modelling methods, including kriging, land use regression, machine learning, and ensemble learning. The models utilized 26 years of observed NO2 measurements, meteorological parameters, geospatial layers, and social and season-dependent variables as representative of emission sources. Separate models were developed for daytime and nighttime periods, which achieved high reliability with adjusted R2 values of 0.92 and 0.93, respectively. The study revealed that mean NO2 concentrations were significantly higher at nighttime (9.60 ppb) compared to daytime (5.61 ppb). Additionally, winter exhibited the highest NO2 levels regardless of time period. The developed EMSMs were utilized to generate maps illustrating NO2 levels pre and during COVID restrictions in Taiwan. These findings could aid epidemiological research on exposure risks and support policy-making and environmental planning initiatives.