A Hybrid Spatial Dependence Model Based on Radial Basis Function Neural Networks (RBFNN) and Random Forest (RF)

The majority of spatial data reveal some degree of spatial dependence. The term “spatial dependence” refers to the tendency for phenomena to be more similar when they occur close together than when they occur far apart in space. This property is ignored in machine learning (ML) for spatial domains of application. Most classical machine learning algorithms are generally inappropriate unless modified in some way to account for it. In this study, we proposed an approach that aimed to improve a ML model to detect the dependence without incorporating any spatial features in the learning process. To detect this dependence while also improving performance, a hybrid model was used based on two representative algorithms. In addition, cross-validation method was used to make the model stable. Furthermore, global moran’s I and local moran were used to capture the spatial dependence in the residuals. The results show that the HM has significant with a R2 of 99.91% performance compared to RBFNN and RF that have 74.22% and 82.26% as R2 respectively. With lower errors, the HM was able to achieve an average test error of 0.033% and a positive global moran’s of 0.12. We concluded that as the R2 value increases, the models become weaker in terms of capturing the dependence.

Wind power forecasting errors modelling approach considering temporal and spatial dependence

The uncertainty of wind power forecasting significantly influences power systems with high percentage of wind power generation. Despite the wind power forecasting error causation, the temporal and spatial dependence of prediction errors has done great influence in specific applications, such as multistage scheduling and aggregated wind power integration. In this paper, Pair-Copula theory has been introduced to construct a multivariate model which can fully considers the margin distribution and stochastic dependence characteristics of wind power forecasting errors. The characteristics of temporal and spatial dependence have been modelled, and their influences on wind power integrations have been analyzed.Model comparisons indicate that the proposed model can reveal the essential relationships of wind power forecasting uncertainty, and describe the various dependences more accurately.

Geospatial Coronavirus Vulnerability Regression Modelling for Malawi Based on Cumulative Spatial Data from April 2020 to May 2021

In the past two to three years, the world has been heavily affected by the infectious coronavirus disease and Malawi has not been spared due to its interconnection with neighboring countries. There is no management tool to identify and model the vulnerabilities of Malawi’s districts in prioritizing health services as far as coronavirus prevalence and other infectious diseases are concerned. The aim of this study was to model coronavirus vulnerability in all districts in Malawi using Geographic Information System (GIS) to monitor the disease’s cumulative prevalence over the severely affected period between 2020 and 2021. To achieve this, four parameters associated with coronavirus prevalence, including population density, percentage of older people, temperature, and humidity, were prepared in a GIS environment and used in the modelling process. A multiscale geographically weighted regression (MGWR) model was used to model and determine the vulnerability of coronavirus in Malawi. In the MGWR modelling, the Fixed Spatial Kernel was used following a Gaussian distribution model type. The Results indicated that population density and older people (age greater than 60 years) have a more significant impact on coronavirus prevalence in Malawi. The modelling further shows that Malawi, between April 2020 and May 2021, Lilongwe, Blantyre and Thyolo were more vulnerable to coronavirus than other districts. This research has shown that spatial variability of Covid-19 cases using MGWR has the potential of providing useful insights to policymakers for targeted interventions that could otherwise not be possible to detect using non-geovisualization techniques.

Dynamic Spatio-Temporal Modeling in Disease Mapping

Spatio-temporal models are valuable tools for disease mapping and understanding the geographical distribution of diseases and temporal dynamics. Spatio-temporal models have been proven empirically to be very complex and this complexity has led many to oversimply and model the spatial and temporal dependencies independently. Unlike common practice, this study formulated a new spatio-temporal model in a Bayesian hierarchical framework that accounts for spatial and temporal dependencies jointly. The spatial and temporal dependencies were dynamically modelled via the matern exponential covariance function. The temporal aspect was captured by the parameters of the exponential with a first-order autoregressive structure. Inferences about the parameters were obtained via Markov Chain Monte Carlo (MCMC) techniques and the spatio-temporal maps were obtained by mapping stable posterior means from the specific location and time from the best model that includes the significant risk factors. The model formulated was fitted to both simulation data and Kenya meningitis incidence data from 2013 to 2019 along with two covariates;Gross County Product (GCP) and average rainfall. The study found that both average rainfall and GCP had a significant positive association with meningitis occurrence. Also, regarding geographical distribution, the spatio-temporal maps showed that meningitis is not evenly distributed across the country as some counties reported a high number of cases compared with other counties.

GACNet: A Generative Adversarial Capsule Network for Regional Epitaxial Traffic Flow Prediction

With continuous urbanization,cities are undergoing a sharp expansion within the regional space.Due to the high cost,the prediction of regional traffic flow is more difficult to extend to entire urban areas.To address this challenging problem,we present a new deep learning architecture for regional epitaxial traffic flow prediction called GACNet,which predicts traffic flow of surrounding areas based on inflow and outflow information in central area.The method is data-driven,and the spatial relationship of traffic flow is characterized by dynamically transforming traffic information into images through a two-dimensional matrix.We introduce adversarial training to improve performance of prediction and enhance the robustness.The generator mainly consists of two parts:abstract traffic feature extraction in the central region and traffic prediction in the extended region.In particular,the feature extraction part captures nonlinear spatial dependence using gated convolution,and replaces the maximum pooling operation with dynamic routing,finally aggregates multidimensional information in capsule form.The effectiveness of the method is evaluated using traffic flow datasets for two real traffic networks:Beijing and New York.Experiments on highly challenging datasets show that our method performs well for this task.

Hierarchical Geographically Weighted Regression Model

In spatial analysis, two problems of the scale effect and the spatial dependencehave been plagued scholars, the first law of geography presented to solve the spatialdependence has played a good role in the guidelines, forming the Geographical WeightedRegression (GWR). Based on classic statistical techniques, GWR model has ascertainsignificance in solving spatial dependence and spatial non-uniform problems, but it hasno impact on the integration of the scale effect. It does not consider the interactionbetween the various factors of the sampling scale observations and the numerous factorsof possible scale effects, so there is a loss of information. Crossing a two-stage analysisof “return of regression” to establish the model of Hierarchical Geographically WeightedRegression (HGWR), the first layer of regression analysis reflects the spatial dependenceof space samples and the second layer of the regression reflects the spatial relationshipsscaling. The combination of both solves the spatial scale effect analysis, spatialdependence and spatial heterogeneity of the combined effects.

Geostatistics as a Methodology for Studying the Spatiotemporal Dynamics of Ramularia areola in Cotton Crops

Geostatistics as a methodology for studying the spatiotemporal dynamics of Ramularia areola in cotton crops. Geostatistics is a tool that has been used to study plant pathology, by modeling the spatiotemporal pattern of diseases, generating hypotheses about their epidemiological aspects in order to use tactics and strategies of rational control. The objective of this study was to use geostatistics to study the spatiotemporal dynamics of Ramularia areola in cotton crops. The experiment was conducted at the experimental area of Mato Grosso State University-Tangará da Serra campus, and arranged in a 2 × 3 factorial design, with randomized blocks, with two spaicngs (0.45 and 0.90 cm) and three conditions of soil coverage (no cover, P. glaucum and C. spectabilis). Geostatistical analysis of data was performed using data from temporal and spatial progress of R. areola, obtained through assessments of the incidence and severity of the disease in plants, and spatial dependence, and analyzed using semivariogram fittings. Through the isotropic exponential semivariogram model, it was possible to check the distribution pattern and spatial dependence of Ramularia leaf spot. Spatial dependence was observed for the disease—moderate to strong for most data evaluated. The pathogen spread from the primary source of inoculum, from the center portion towards the edges, forming foci originating from a source of secondary inoculum.

Spatial-temporal distribution of fatal yellowing in different oil palm genetic materials in eastern amazon

Oil palm (Elaeis guineensis Jacq.) is one of the agricultural crops with the greatest potentialfor vegetable oil production in Brazil. However, a disease of unknown etiology popularlyknown as Fatal Yellowin (FY) has caused damage to Brazilian farmers particularly in theeastern region of the Amazon. So, the objective of this study was to evaluate the spatialdependence of FY on three oil palm genotypes, grown for many years in an organic produc-tion system in the Amazon region. The study area had 4104 ha, divided into 139 plots. Ineach plot, the monthly incidence of disease was monitored forming a database. The num-ber of diseased plants per year, number of accumulated diseased plants, number of dis-eased plants per hectare, growth rate of diseased plants and incidence of accumulateddisease were evaluated. The results indicated spatial distribution of the variables adjustedto the gaussian, spherical and exponential models, with predominance of the first model.This increases the purpose that FY is caused by biotic factors. The highest range achievedin the study was 2929 m indicating the susceptibility of the tested genotypes. Some plotsclose to the river had the highest incidence of the disease on the study, probably associatedwith higher soil moisture.

Absorbing Boundary Conditions for Solving N-Dimensional Stationary Schr¨odinger Equations with Unbounded Potentials and Nonlinearities

We propose a hierarchy of novel absorbing boundary conditions for the onedimensional stationary Schr¨odinger equation with general(linear and nonlinear)potential.The accuracy of the new absorbing boundary conditions is investigated numerically for the computation of energies and ground-states for linear and nonlinear Schr¨odinger equations.It turns out that these absorbing boundary conditions and their variants lead to a higher accuracy than the usual Dirichlet boundary condition.Finally,we give the extension of these ABCs to N-dimensional stationary Schr¨odinger equations.