Spatial Weighting Scheme Research Articles

Graph convolutional networks with learnable spatial weightings for traffic forecasting applications

How to select a suitable spatial weighting scheme for convolutional graph neural networks (ConvGNNs) is challenging. In this study, we propose a ConvGNN, termed learnable graph convolutional (LGC) network, which learns spatial weightings between a road and its k-hop neighbours as learnable parameters in the spatial convolutional operator. A dynamic LGC (DLGC) network is further proposed to learn the dynamics of spatial weightings by explicitly considering the temporal correlations of spatial weightings at different times of the day. A multi-temporal DLGC (MTDLGC) network is developed for forecasting traffic variables in road networks. Results of case study suggest that the MT-DLGC network can achieve higher prediction accuracy than other state-of-the-art baselines. Both LGC and DLGC networks can be used as general spatial weighting schemes for baselines with better forecasting performance than existing spatial weighting schemes, e.g., graph attention. The source code of this study is available publicly at https://github.com/Mayaohong/MTDLGC.

Segment-Level Spatial Spillover Effects of Exogenous Characteristics of Arterials on Crash Frequency

Studies of spatial effects on road traffic safety are highly skewed in relation to aggregation units. As a result, the spatial effect of road traffic crashes at segment-level aggregation is rarely investigated. In light of that, this study investigates spatial spillover effects to determine the effects of independent variable changes initiated in one segment on the dependent variable of its neighboring segments. Additionally, the performance of the implemented models is evaluated under five different spatial weighting approaches. Spatial spillover effect of fatal and injury crashes aggregated at the segment level is estimated with the spatial lag of explanatory variables model under the general framework of Poisson and negative binomial models. The results indicate (i) spatial spillover effects are better modeled with spatial relation conceptualized by inverse distance spatial weighting schemes; (ii) spatial lag of explanatory variables under negative binomial and inverse distance spatial weighting provides the best model fitting; (iii) exogenous interactions among influence factors of fatal and injury crashes have a noteworthy effect in determining the crash frequency of neighboring segments. Convincingly, spatial spillover effect seems to significantly affect the results of conventional count modeling at the segment levels. To that effect, this study expects to provide empirical evidence and complement the existing literature on the impact of changes in explanatory variables initiated in a given segment on the safety of segments in proximity.

Spatial interdependence and spillovers of fiscal grants in Benin: Static and dynamic diffusions

This paper investigates the spatial diffusion of an intergovernmental grant in Benin. Using static and dynamic spatial models, we estimate the spillover effects of the Fonds d’Appui au Développement des Communes (FADeC) on per capita local government expenditure in the 77 municipalities from 2008 to 2015. Neighborliness – a measure of interdependence – is captured through geographic and distance-based spatial weighting schemes. In addition, we constructed a measurement of ethnic affinity as an alternative spatial weighting scheme to test for the existence of an ethno-spatial interdependence in local public finance in Benin. The empirical results suggest that a statistically significant share of the total effects of the FADeC stems from indirect elasticities or the diffusion process of grants received by neighboring jurisdictions, regardless of how we measure neighborliness. The results also confirm the existence of a robust ethno-spatial interdependence and complementarity in local government expenditure in Benin. The spillovers across ethnic neighbors are estimated to be 13.9% of the total effects in the short-run and 15.5% in the long-run. Put differently, the effects of the FADeC in a given municipality are influenced by the transfers received by its ethnic (and linguistic) neighbors. The findings point to the appeal of inter-governmental transfers for the decentralized financing of public services, especially in low-income countries where local bureaucratic capacity in raising own-source revenues might be limited. Supporting local governments with well-structured grants can not only be a channel to foster local public provision but also contribute to pushing geographic or ethnic neighboring localities to increase their own spending and generate positive spillovers that are crucial for wholistic regional development.

Reconstruction of a Global 9 km, 8-Day SMAP Surface Soil Moisture Dataset during 2015–2020 by Spatiotemporal Fusion

Soil moisture, a crucial property for Earth surface research, has been focused widely in various studies. The Soil Moisture Active Passive (SMAP) global products at 36 km and 9 km (called P36 and AP9 in this research) have been published from April 2015. However, the 9 km AP9 product was retrieved from the active radar and L-band passive radiometer and the active radar failed in July 2015. In this research, the virtual image pair-based spatiotemporal fusion model was coupled with a spatial weighting scheme (VIPSTF-SW) to simulate the 9 km AP9 data after failure of the active radar. The method makes full use of all the historical AP9 and P36 data available between April and July 2015. As a result, 8-day composited 9 km SMAP data at the global scale were produced from 2015 to 2020, by downscaling the corresponding 8-day composited P36 data. The available AP9 data and in situ reference data were used to validate the predicted 9 km data. Generally, the predicted 9 km SMAP data can provide more spatial details than P36 and are more accurate than the existing EP9 product. The VIPSTF-SW-predicted 9 km SMAP data are an accurate substitute for AP9 and will be made freely available to support research and applications in hydrology, climatology, ecology, and many other fields at the global scale.

A network approach to measuring state preferences

AbstractState preferences play an important role in international politics. Unfortunately, actually observing and measuring these preferences are impossible. In general, scholars have tried to infer preferences using either UN voting or alliance behavior. The two most notable measures of state preferences that have flowed from this research area are ideal points (Bailey et al., 2017) and S-scores (Signorino & Ritter, 1999). The basis of both these models is a spatial weighting scheme that has proven useful but discounts higher-order effects that might be present in relational data structures such as UN voting and alliances. We begin by arguing that both alliances and UN voting are simply examples of the multiple layers upon which states interact with one another. To estimate a measure of state preferences, we utilize a tensor decomposition model that provides a reduced-rank approximation of the main patterns across the layers. Our new measure of preferences plausibly describes important state relations and yields important insights on the relationship between preferences, democracy, and international conflict. Additionally, we show that a model of conflict using this measure of state preferences decisively outperforms models using extant measures when it comes to predicting conflict in an out-of-sample context.

MOANA: An Online Learned Adaptive Appearance Model for Robust Multiple Object Tracking in 3D

Multiple object tracking has been a challenging field, mainly due to noisy detection sets and identity switch caused by occlusion and similar appearance among nearby targets. Previous works rely on appearance models built on individual or several selected frames for the comparison of features, but they cannot encode long-term appearance changes caused by pose, viewing angle and lighting conditions. In this work, we propose an adaptive model that learns online a relatively long-term appearance change of each target. The proposed model is compatible with any feature of fixed dimension or their combination, whose learning rates are dynamically controlled by adaptive update and spatial weighting schemes. To handle occlusion and nearby objects sharing similar appearance, we also design cross-matching and re-identification schemes based on the application of the proposed adaptive appearance models. Additionally, the 3D geometry information is effectively incorporated in our formulation for data association. The proposed method outperforms all the state-of-the-art on the MOTChallenge 3D benchmark and achieves real-time computation with only a standard desktop CPU. It has also shown superior performance over the state-of-the-art on the 2D benchmark of MOTChallenge.

Multi-center convolutional descriptor aggregation for image retrieval

Recent works have demonstrated that the convolutional descriptor aggregation can provide state-of-the-art performance for image retrieval. In this paper, we propose a multi-center convolutional descriptor aggregation (MCDA) method to produce global image representation for image retrieval. We first present a feature map center selection method to eliminate the background information in the feature maps. We then propose the channel weighting and spatial weighting schemes based on the centers to boost the effect of the features on the object. Finally, the weighted convolutional descriptors are aggregated to represent images. Experiments demonstrate that MCDA can produce state-of-the-art retrieval performance, and the generated activation map is also effective for object localization.

Spatial Pyramid Pooling of Selective Convolutional Features for Vein Recognition

Deep neural network (DNN) has demonstrated astounding performance in large-scale image recognition task, and pre-trained DNN models trained for one task have also been applied to domains different from their original purposes. Following such an idea, a novel hand-dorsa vein recognition model is constructed by adopting DNN pre-trained on a large-scale database as a universal feature descriptor. Unlike most of these studies which adopt activations of the fully connected layer of DNN as the image representation, we adopt convolutional activations as the region representation. However, not local features of all regions are equally important for final classification. Thus, to solve this issue, a novel selective convolutional feature model based on spatial weighting is proposed to acquire more robust and discriminative feature representation. In specific, a spatial weighting scheme is applied on the convolutional activations to weigh the importance of local features at different regions for classification, which further enhances the discriminability of feature representation. Besides, to take full advantage of the spatial information of the convolutional activations, spatial pyramid pooling is introduced to obtain feature representation with rich spatial information. The final image representation is formed by concatenating the local features of different level of the spatial pyramid. Series rigorous experiments on the lab-made database are conducted to evidence the effectiveness and feasibility of the proposed model. What is more, an additional experiment with subset of PolyU database illustrates its generalization ability and robustness.

Weighted spatial based geometric scheme as an efficient algorithm for analyzing single-trial EEGS to improve cue-based BCI classification

There is a growing interest in analyzing the geometrical behavior of electroencephalogram (EEG) covariance matrix in the context of brain computer interface (BCI). The bottleneck of the current Riemannian framework is the bias of the mean vector of EEG signals to the noisy trials, which deteriorates the covariance matrix in the manifold space. This study presents a spatial weighting scheme to reduce the effect of noisy trials on the mean vector. To assess the proposed method, dataset IIa from BCI competition IV, containing the EEG trials of 9 subjects performing four mental tasks, was utilized. The performance of the proposed method is compared to the classical Riemannian method along with Common Spatial Pattern (CSP) on the dataset. The results show that when considering just two imagery classes, the proposed method performs on par with CSP method, whereas in the multi class scenario, the proposed algorithm outperforms the CSP approach on seven out of nine subjects. Incidentally, the proposed method obtains better accuracy for the majority of subjects compared to the classical Riemannian method.

A Review of Sampling Effects and Response Bias in Internet Participatory Mapping (PPGIS/PGIS/VGI)

AbstractGlobal interest in participatory mapping described as public participation GIS (PPGIS), participatory GIS (PGIS), and volunteered geographic information (VGI) continues to grow, but systematic study of spatial data quality and sampling effects is limited. This article provides a review and meta‐analysis of Internet‐based PPGIS studies conducted during the period 2006–2015 (n=26) to answer the following research questions: (1) How does mapping effort, as a proxy measure for spatial data quality, differ by sampling group? (2) Does the purpose and context of PPGIS influence mapping results? (3) What is the potential for mapping bias through sampling design? (4) Given the results, what should be the focus of future PPGIS research? Mapping effort was highest in sampling groups whose livelihoods were closely related to the purpose of the study, there was greater mapping effort in household sampling groups compared to volunteer groups, and participant domicile had strong effects on mapped results through spatial discounting. The use of online Internet panels provides higher response rates but lower spatial data quality. Future research should focus on increasing sampling response rates, assessing social trade‐offs using alternative spatial weighting schemes, and examining the capacity of the public to select land use alternatives as a complement to traditional expert‐driven planning systems.

Strategy of Spatial Panel Estimation: Spatial Spillovers Between Taxation and Economic Growth

Spatial panels are a powerful econometric tool for the estimation of space-dependent cross-sectional time-series models of economic phenomena. A plethora of parameters and possible specifications require a systematic approach to estimation. This paper presents a strategy of estimation to be considered in applied research on economic policy, including the concept of spatial spillovers and its local and global effects, direct and indirect impacts, as well as the role of different spatial weighting schemes. The paper presents fiscal factors affecting GDP between the years 2002–2015 in a number of European economies.

Identifying the Flypaper Effect in the Presence of Spatial Dependence: Evidence from Education inChina's Counties

AbstractIn the context ofChina without a median voter system, this study examines whether the “flypaper effect,” an unconditional lump‐sum grant from the upper governments to the county governments increases spending in a greater proportion than an equivalent rise in local income, holds true inChina. UsingChina's county‐level education data during 2007, the models have been estimated using a spatial econometric technique that accounts for spatial interaction behavior on public education expenditure across local governments. We find that, in the presence of spatial interdependence, there is no evidence of a “flypaper effect” when different spatial weighting schemes and the endogeneity problem of education grants are accounted for. Rather, the “anti‐flypaper effect” is found. Important policy implications are drawn forChina's fiscal decentralization reform.

A reactive transport benchmark on modeling biogenic uraninite re-oxidation by Fe(III)-(hydr)oxides

A reactive transport benchmark on uranium (U) bioreduction and concomitant reoxidation has been developed based on the multicomponent biogeochemical reaction network presented by Spycher et al. (Geochim Cosmochim Acta 75:4426–4440, 2011). The benchmark problem consists of a model inter-comparison starting with the numerical simulations of the original batch experiments of Sani et al. (Geochim Cosmochim Acta 68:2639–2648, 2004). The batch model is then extended to 1D and 2D reactive transport models, designed to evaluate the model results for the key biogeochemical reaction processes and their coupling with physical transport. Simulations are performed with four different reactive transport simulators: PHREEQC, PHT3D, MIN3P, and TOUGHREACT. All of the simulators are able to capture the complex biogeochemical reaction kinetics and the coupling between transport and kinetic reaction network successfully in the same manner. For the dispersion-free variant of the problem, a 1D-reference solution was obtained by PHREEQC, which is not affected by numerical dispersion. PHT3D using the sequential non-iterative approach (SNIA) with an explicit TVD scheme and MIN3P using the global implicit method (GIM) with an implicit van Leer flux limiter provided the closest approximation to the PHREEQC results. Since the spatial weighting schemes for the advection term and numerical dispersion played an important role for the accuracy of the results, the simulators were further compared using different solution schemes. When all codes used the same spatial weighting scheme with finite-difference approximation, the simulation results agreed very well among all four codes. The model intercomparison for the 2D-case demonstrated a high level of sensitivity to the mixing of different waters at the dispersive front. Therefore this benchmark problem is well-suited to assess code performance for mixing-controlled reactive transport models in conjunction with complex reaction kinetics.

Improving the predictive power of spatial statistical models of stream macroinvertebrates using weighted autocovariance functions

Spatial statistical stream-network models are useful for modelling physicochemical data, but to-date have not been fit to macroinvertebrate data. Spatial stream-network models were fit to three macroinvertebrate indices: percent pollution-tolerant taxa, taxa richness and the number of taxalacking out-of-network movement (in-stream dispersers). We explored patterns of spatial autocorrelation in the indices and found that the 1) relative strength of in-stream and Euclidean spatial autocorrelation varied between indices; 2) spatial models outperformed non-spatial models; and 3) the spatial-weighting scheme used to weight tributaries had a substantial impact on model performance for the in-stream dispersers; with weights based on percent stream slope, used as a surrogate for velocity because of its potential effect on dispersal and habitat heterogeneity, producing more accurate predictions than other spatial-weighting schemes. These results demonstrate the flexibility of the modelling approach and its ability to account for multi-scale patterns and processes within the aquatic and terrestrial landscape.

Detecting transient signals in geodetic time series using sparse estimation techniques

AbstractWe present a new method for automatically detecting transient deformation signals from geodetic time series. We cast the detection problem as a least squares procedure where the design matrix corresponds to a highly overcomplete, nonorthogonal dictionary of displacement functions in time that resemble transient signals of various timescales. The addition of a sparsity‐inducing regularization term to the cost function limits the total number of dictionary elements needed to reconstruct the signal. Sparsity‐inducing regularization enhances interpretability of the resultant time‐dependent model by localizing the dominant timescales and onset times of the transient signals. Transient detection can then be performed using convex optimization software where detection sensitivity is dependent on the strength of the applied sparsity‐inducing regularization. To assess uncertainties associated with estimation of the dictionary coefficients, we compare solutions with those found through a Bayesian inference approach to sample the full model space for each dictionary element. In addition to providing uncertainty bounds on the coefficients and confirming the optimization results, Bayesian sampling reveals trade‐offs between dictionary elements that have nearly equal probability in modeling a transient signal. Thus, we can rigorously assess the probabilities of the occurrence of transient signals and their characteristic temporal evolution. The detection algorithm is applied on several synthetic time series and real observed GPS time series for the Cascadia region. For the latter data set, we incorporate a spatial weighting scheme that self‐adjusts to the local network density and filters for spatially coherent signals. The weighting allows for the automatic detection of repeating slow slip events.

Spatial Pooling of Heterogeneous Features for Image Classification

In image classification tasks, one of the most successful algorithms is the bag-of-features (BoFs) model. Although the BoF model has many advantages, such as simplicity, generality, and scalability, it still suffers from several drawbacks, including the limited semantic description of local descriptors, lack of robust structures upon single visual words, and missing of efficient spatial weighting. To overcome these shortcomings, various techniques have been proposed, such as extracting multiple descriptors, spatial context modeling, and interest region detection. Though they have been proven to improve the BoF model to some extent, there still lacks a coherent scheme to integrate each individual module together. To address the problems above, we propose a novel framework with spatial pooling of complementary features. Our model expands the traditional BoF model on three aspects. First, we propose a new scheme for combining texture and edge-based local features together at the descriptor extraction level. Next, we build geometric visual phrases to model spatial context upon complementary features for midlevel image representation. Finally, based on a smoothed edgemap, a simple and effective spatial weighting scheme is performed to capture the image saliency. We test the proposed framework on several benchmark data sets for image classification. The extensive results show the superior performance of our algorithm over the state-of-the-art methods.

Does the spatial arrangement of disturbance within forested watersheds affect loadings of nitrogen to stream waters? A test using Landsat and synoptic stream water data

Remotely sensed maps of forest disturbance provide a powerful tool for predicting spatial and temporal variability in the loading of nitrogen to receiving waters, key data needed for effective watershed management of nutrient pollution. We hypothesize that the spatial arrangement of disturbances within small-forested watersheds can affect N loadings. To test this, we developed schemes for spatially weighting maps of yearly disturbance produced through change analysis of the Landsat Tasseled Cap Disturbance Index (DI), and evaluated the ability of each scheme to predict N concentrations, and subsequently estimated N loads, from forty low-order streams within the Savage River drainage of western Maryland, USA during the 2006–2010 water years, a period encompassing extensive defoliations by gypsy moths (Lymantria dispar). We generated a base scheme of unweighted, watershed averaged change in DI (ΔDI), and five other schemes that weighted ΔDI by either a pixel's flow accumulation value, the distance to the watershed outlet, or proximity to the stream. Over the five years, the flow accumulation scheme tended to perform better than other weighting schemes, and even explained slightly more variability than the base scheme during years of moderate N loads (R2=0.15 vs. 0.03 in 2007 and R2=0.30 vs. 0.18 in 2010). However, this best spatial weighting scheme explained comparable or less variability during the two post-defoliation years with larger N loads (R2=0.43 vs. 0.44 in 2008 and R2=0.31 vs. 0.48 in 2009). Thus, for the purposes of utilizing remote sensing information within watershed management of nutrient pollution, these results suggest that coarse-scale, high temporal frequency data such as MODIS could be well suited for characterizing forest disturbance and predicting the resultant episodic N loads.

A scoping review of spatial cluster analysis techniques for point-event data

Spatial cluster analysis is a uniquely interdisciplinary endeavour, and so it is important to communicate and disseminate ideas, innovations, best practices and challenges across practitioners, applied epidemiology researchers and spatial statisticians. In this research we conducted a scoping review to systematically search peer-reviewed journal databases for research that has employed spatial cluster analysis methods on individual-level, address location, or x and y coordinate derived data. To illustrate the thematic issues raised by our results, methods were tested using a dataset where known clusters existed. Point pattern methods, spatial clustering and cluster detection tests, and a locally weighted spatial regression model were most commonly used for individual-level, address location data (n = 29). The spatial scan statistic was the most popular method for address location data (n = 19). Six themes were identified relating to the application of spatial cluster analysis methods and subsequent analyses, which we recommend researchers to consider; exploratory analysis, visualization, spatial resolution, aetiology, scale and spatial weights. It is our intention that researchers seeking direction for using spatial cluster analysis methods, consider the caveats and strengths of each approach, but also explore the numerous other methods available for this type of analysis. Applied spatial epidemiology researchers and practitioners should give special consideration to applying multiple tests to a dataset. Future research should focus on developing frameworks for selecting appropriate methods and the corresponding spatial weighting schemes.

Spatially weighted principal component analysis (PCA) method for water quality analysis

Incorporating spatial information data into the principal component analysis is of importance. Some proposed methods of spatial weighting schemes to be applied to the ordinary PCA are reviewed and a new version of the method is proposed in the context of spatial analysis for geospatial multivariate analysis. In view of spatial variations in the hydrochemistry of rivers such combined version of the technique might be useful for reliable estimates.

Optimizing locations for the installation of automated external defibrillators (AEDs) in urban public streets through the use of spatial and temporal weighting schemes

ObjectiveOut-of-hospital cardiac arrest (OHCA) occurs when the heart is deprived of oxygen without immediate medical treatment. The use of publicly accessible automated external defibrillators (AEDs) is generally considered to be an effective pre-hospital measure. Although studies have been undertaken to determine the locations of AED installations, the spatial and temporal characteristics of OHCA occurrence have not been considered comprehensively. This study attempts to assess the feasibility of using the 7-Eleven chain of convenience stores as possible locations for the installation of AEDs to capture the spatial and temporal characteristics of OHCA patients. MethodsThe methodological framework was divided into two stages. The first stage involved the development of two weighting schemes, a temporally weighted model (TWM) and a spatially weighted model (SWM), to capture the temporal and spatial variations in selecting AED locations. In the second stage, we proposed a stirring genetic algorithm (SGA) to select the limited subset of 7-Elevens covering the most weighted OHCA occurrences from the first stage. ResultsWe considered two modes of conveyance, human running and vehicle transportation, by setting the service range of the 7-Elevens at 100 and 300 m. Among a total of 323 OHCA cases within 100 m of a 7-Eleven, our results showed that there were 177 cases (54.8%) to be covered in the TWM and 150 cases (46.44%) in the SWM. When the service range was increased to 300 m, the total number of OHCA cases increased to 1271, which included 659 cases (51.85%) to be covered in the TWM and 522 cases (41.07%) in the SWM. We also found that, in public streets in urban settings, the TWM selected more 7-Elevens in commercial areas, while the SWM selected more stores in residential areas. ConclusionsWe conclude that each 7-Eleven has a different role for allocating AEDs in an urban setting. The AEDs at 7-Elevens in commercial areas help to compensate for the temporal gap of emergency medical service (EMS) in nighttime occurrences and for a high incidence of OHCA patients. For convenience stores in residential areas, AEDs help to compensate for the spatial gap in areas that are far from fire stations.