基于GA-BP神经网络模型的登革热时空扩散模拟

目的 登革热是一个全球性公共卫生问题,从地理学时空数据分析的视角,探究登革热的时空特质、构建登革热时空过程模型,是有效预防、控制登革热的新方法、研究新热点。方法 基于时空数据挖掘、时空过程建模,综合环境、气象、地理、人口4大因素,分析登革热的空间相关性及登革热病例的空间自相关,挖掘登革热影响因子;针对BP(back propagation)神经网络模型易陷入局部最优的缺陷,引入遗传算法(GA)改进BP神经网络模型,用于登革热时空模拟。结果 登革热的时空扩散与温度、湿度、居民地、交通、人口密度呈显著相关;登革热病例之间呈显著自相关;登革热发生、扩散与环境、气象、地理、人口中的多种因子存在非线性关系;利用改进的GA-BP神经网络模型模拟登革热时空扩散,均方根误差达到0.081。结论 登革热发生、扩散是由多种因素综合影响的结果;GA-BP神经网络模型能够有效模拟登革热时空过程;此模型同样适用于其他伊蚊类传染病的模拟。     

Downscaling global information for regional benefit: coupling spatial models at varying space and time scales

Every environmental activity to a large extent is dependent on climate as natural processes are intrinsically linked with the waxing and waning of the seasons. The goal is to integrate global seasonal climate forecasts with local environmental decision support systems within an operational framework to deliver community benefits. This framework is designed to support the downscaling of coarse resolution seasonal forecasts to drive biological or hydrological applications at the regional level. Some of the challenges and complexities in coupling spatial simulations operating at varying spatial and temporal resolutions will be discussed from several viewpoints, illustrating the value of multidisciplinary collaboration in a virtual team and benefits from the globalisation of research. This project demonstrates how a state Government is evolving an existing service to enhance the use of seasonal climate forecasts for sustainable environmental and natural resource management.     

Bayesian inference for additive mixed quantile regression models

Quantile regression problems in practice may require flexible semiparametric forms of the predictor for modeling the dependence of responses on covariates. Furthermore, it is often necessary to add random effects accounting for overdispersion caused by unobserved heterogeneity or for correlation in longitudinal data. We present a unified approach for Bayesian quantile inference on continuous response via Markov chain Monte Carlo (MCMC) simulation and approximate inference using integrated nested Laplace approximations (INLA) in additive mixed models. Different types of covariate are all treated within the same general framework by assigning appropriate Gaussian Markov random field (GMRF) priors with different forms and degrees of smoothness. We applied the approach to extensive simulation studies and a Munich rental dataset, showing that the methods are also computationally efficient in problems with many covariates and large datasets.     

Modelling and forecasting tourism from East Asia to Thailand under temporal and spatial aggregation

Tourism is one of the key service industries in Thailand, with a 5.27% share of Gross Domestic Product in 2003. Since 2000, international tourist arrivals, particularly those from East Asia, to Thailand have been on a continuous upward trend. Tourism forecasts can be made based on previous observations, so that historical analysis of tourist arrivals can provide a useful understanding of inbound trips and the behaviour of trends in foreign tourist arrivals to Thailand. As tourism is seasonal, a good forecast is required for stakeholders in the industry to manage risk. Previous research on tourism forecasts has typically been based on annual and monthly data analysis, while few past empirical tourism studies using the Box–Jenkins approach have taken account of pre-testing for seasonal unit roots based on Franses [P.H. Franses, Seasonality, nonstationarity and the forecasting of monthly time series, International Journal of Forecasting 7 (1991) 199–208] and Beaulieu and Miron [J.J. Beaulieu, J.A. Miron, Seasonal unit roots in aggregate U.S. data, Journal of Econometrics 55 (1993) 305–328] framework. An analysis of the time series of tourism demand, specifically monthly tourist arrivals from six major countries in East Asia to Thailand, from January 1971 to December 2005 is examined. This paper analyses stationary and non-stationary tourist arrivals series by formally testing for the presence of unit roots and seasonal unit roots prior to estimation, model selection and forecasting. Various Box–Jenkins autoregressive integrated moving average (ARIMA) models and seasonal ARIMA models are estimated, with the tourist arrivals series showing seasonal patterns. The fitted ARIMA and seasonal ARIMA models forecast tourist arrivals from East Asia very well for the period 2006(1)–2008(1). Total monthly and annual forecasts can be obtained through temporal and spatial aggregation.     

Predicting daily streamflow using rainfall forecasts,a simple loss module and unit hydrographs: Two Brazilian catchments

The performance of a simple, spatially-lumped, rainfall–streamflow model is compared with that of a more complex, spatially-distributed model. In terms of two model-fit statistics it is shown that for two catchments in Brazil (about 30,000 km² and 34,000 km²) with different flow regimes, the simpler catchment models, which are unit hydrograph-based and require only rainfall, streamflow and air temperature data for calibration, perform about as well as more complex catchment models that require additional information from satellite images and digitized maps of elevation, land-use and soils. Simple catchment models are applied in forecasting mode, using daily rainfall forecasts from a regional weather forecasting model. The value of the rainfall forecasts, relative to the case where rainfall is known, is assessed for both catchments. The results are discussed in the context of on-going work to compare different modelling approaches for many other Brazilian catchments, and to apply improved forecasting algorithms based on the simple modelling approach to the same, and other, catchments.     

A Bayesian Joint Mixture Framework for the Integration of Anatomical Information in Functional Image Reconstruction

We present a Bayesian joint mixture framework for integrating anatomical image intensity and region segmentation information into emission tomographic reconstruction in medical imaging. The joint mixture framework is particularly well suited for this problem and allows us to integrate additional available information such as anatomical region segmentation information into the Bayesian model. Since this information is independently available as opposed to being estimated, it acts as a good constraint on the joint mixture model. After specifying the joint mixture model, we combine it with the standard emission tomographic likelihood. The Bayesian posterior is a combination of this likelihood and the joint mixture prior. Since well known EM algorithms separately exist for both the emission tomography (ET) likelihood and the joint mixture prior, we have designed a novel EM² algorithm that comprises two EM algorithms—one for the likelihood and one for the prior. Despite being dove-tailed in this manner, the resulting EM² algorithm is an alternating descent algorithm that is guaranteed to converge to a local minimum of the negative log Bayesian posterior. Results are shown on synthetic images with bias/variance plots used to gauge performance. The EM² algorithm resulting from the joint mixture framework has the best bias/variance performance when compared with six other closely related algorithms that incorporate anatomical information to varying degrees.     

Using a Bayesian framework and global sensitivity analysis to identify strengths and weaknesses of two process-based models differing in representation of autotrophic respiration

Process-based models are powerful tools for sustainable and adaptive forest management. Bayesian statistics and global sensitivity analysis allow to reduce uncertainties in parameters and outputs, and they provide better insight of model behaviour. In this work two versions of a process-based model that differed in the autotrophic respiration modelling were analysed. The original version (3PGN) was based on a constant ratio between net and gross primary production, while in a new version (3PGN¹) the autotrophic respiration was modelled as a function of temperature and biomass. A Bayesian framework, and a global sensitivity analysis (Morris method) were used to reduce parametric uncertainty, to highlight strengths and weaknesses of the models and to evaluate their performances. The Bayesian approach allowed also to identify the weaknesses and strengths of the dataset used for the analyses. The Morris method in combination with the Bayesian framework helped to identify key parameters and gave a deeper understanding of model behaviour. Both model versions reliably predicted average stand diameter at breast height, average stand height, stand volume and stem biomass. On the contrary, the models were not able to accurately predict net ecosystem production. Bayesian model comparison showed that 3PGN¹, with the new autotrophic respiration model, has a higher conditional probability of being correct than the original 3PGN model.     

Spatial-temporal association between fine particulate matter and daily mortality

Fine particulate matter (PM_2.5) is a mixture of pollutants that has been linked to serious health problems, including premature mortality. Since the chemical composition of PM_2.5 varies across space and time, the association between PM_2.5 and mortality could also change with space and season. A statistical multi-stage Bayesian framework is developed and implemented, which provides a very broad and flexible approach to studying the spatiotemporal associations between mortality and population exposure to daily PM_2.5 mass, while accounting for different sources of uncertainty. The first stage of the framework maps ambient PM_2.5 air concentrations using all available monitoring data (IMPROVE and FRM) and an air quality model (CMAQ) at different spatial and temporal scales. The second stage of the framework examines the spatial temporal relationships between the health end-points and the exposures to PM_2.5 by introducing a spatial-temporal generalized Poisson regression model. A method to adjust for time-varying confounders such as seasonal trends is proposed. A common seasonal trends model uses a fixed number of basis functions to account for these confounders, but the results can be sensitive to the number of basis functions. Thus, instead the number of the basis functions is treated as an unknown parameter in the Bayesian model, and a space-time stochastic search variable selection approach is used. The framework is illustrated using a data set in North Carolina for the year 2001.     

A Bayesian method for improving probabilistic wave forecasts by weighting ensemble members

New innovations are emerging which offer opportunities to improve forecasts of wave conditions. These include probabilistic modelling results, such as those based on an ensemble of multiple predictions which can provide a measure of the uncertainty, and new sources of observational data such as GNSS reflectometry and FerryBoxes, which can be combined with an increased availability of more traditional static sensors. This paper outlines an application of the Bayesian statistical methodology which combines these innovations. The method modifies the probabilities of ensemble wave forecasts based on recent past performance of individual members against a set of observations from various data source types. Each data source is harvested and mapped against a set of spatio-temporal feature types and then used to post-process ensemble model output. A prototype user interface is given with a set of experimental results testing the methodology for a use case covering the English Channel.     

Cover – volcanic ash and airspace closures

Trace gases are important components for climate change process, and Earth’s climate is sensitive to change in their atmospheric concentrations; therefore, proper assessment of trace gases is essential for ongoing global climate simulation. The spatio-temporal variations of four trace gases, namely carbon monoxide (CO), nitrogen dioxide (NO₂), ozone (O₃), and carbon dioxide (CO₂), over Bangladesh during the last decade are analysed using the remote-sensing data sets of the Atmospheric Infrared Sounder (AIRS) and Ozone Monitoring Instrument (OMI). Monthly, seasonal, and annual mean variations of trace gases were assessed. Higher CO, O₃, and CO₂ concentrations show west-to-east gradient, indicating the impact of both local meteorology and emissions on variations in trace gases. On the other hand, total NO₂ concentration increases over Dhaka because of large population density, high traffic emission, larger industrial activities, and highly polluted air. The inter-annual variations of trace gases are mainly due to large-scale climatic phenomena such as El Niño and La Niña conditions. All the trace gases show strong seasonality, with higher levels during pre-monsoon season and lower levels during monsoon season, which are caused by the seasonal variations in biomass burning (BB), long-range transportation, and rainfall in South and Southeast Asia (S–SE Asia). However, O₃ concentration reveals minimum loading during winter season, associated with the reduction of O₃ formation in cold days due to insufficient heat. These findings are important to estimate regional climate variability due to trace gases.     

Non-parametric regression for space–time forecasting under missing data

As more and more real time spatio-temporal datasets become available at increasing spatial and temporal resolutions, the provision of high quality, predictive information about spatio-temporal processes becomes an increasingly feasible goal. However, many sensor networks that collect spatio-temporal information are prone to failure, resulting in missing data. To complicate matters, the missing data is often not missing at random, and is characterised by long periods where no data is observed. The performance of traditional univariate forecasting methods such as ARIMA models decreases with the length of the missing data period because they do not have access to local temporal information. However, if spatio-temporal autocorrelation is present in a space–time series then spatio-temporal approaches have the potential to offer better forecasts. In this paper, a non-parametric spatio-temporal kernel regression model is developed to forecast the future unit journey time values of road links in central London, UK, under the assumption of sensor malfunction. Only the current traffic patterns of the upstream and downstream neighbouring links are used to inform the forecasts. The model performance is compared with another form of non-parametric regression, K-nearest neighbours, which is also effective in forecasting under missing data. The methods show promising forecasting performance, particularly in periods of high congestion.     

Spatio-temporal distribution of skipjack in relation to oceanographic conditions in the west-central Pacific Ocean

The skipjack tuna, Katsuwonus pelamis, is an economically important oceanic species widely distributed in the west-central Pacific Ocean (WCPO). The spatio-temporal distribution of Katsuwonus pelamis with respect to oceanographic and climatic variables during 1995–2010 in the west-central Pacific was examined in this study using purse seine fishery data from South Pacific Fisheries Commission (SPC). ‘Gravitational centre’ of two temporal scales (i.e. monthly and yearly) of catch per unit effort (CPUE) was calculated to represent the variability of local stock abundance on fishing grounds. Significant inter-annual and seasonal variabilities were observed. Monthly longitudinal ‘centres of gravity’ were correlated with sea surface temperature anomaly (SSTA) in Niño 3.4 region and monthly latitudinal ‘centres of gravity’ reflect a ‘South–North’ migration pattern of Katsuwonus pelamis. The distribution–habitat associations were quantitatively evaluated including SST between 28–30°C, sea surface height (SSH)in the range 90–100 cm, gradient SST between 0.1 and 0.7°C 10 km^?1,and chlorophyll-a(chl-a) between 0.1 and 0.6 mg m^?3 by an empirical cumulative distribution function (ECDF). Four clusters of yearly ‘gravitational centres’ were classified using the k-means method, which could be defined as warmpool fishing ground (WFG) and cold-tongue fishing ground (CFG) according to their oceanographic habitat. The integrated environmental distribution map combined with the developed model (R² = 0.28, p < 0.0001) provides an approach for predicting hotspots of Katsuwonus pelamis. This study improves our understanding of the spatio-temporal dynamics of skipjack tuna, which is critical for sustainable management of this important fisheries resources.     

A modeling and spatio-temporal analysis framework for monitoring environmental change using NPP as an ecosystem indicator

We present and describe a modeling and analysis framework for monitoring protected area (PA) ecosystems with net primary productivity (NPP) as an indicator of health. It brings together satellite data, an ecosystem simulation model (NASA-CASA), spatial linear models with autoregression, and a GIS to provide practitioners a low-cost, accessible ecosystem monitoring and analysis system (EMAS) at landscape resolutions. The EMAS is evaluated and assessed with an application example in Yellowstone National Park aimed at identifying the causes and consequences of drought. Utilizing five predictor covariates (solar radiation, burn severity, soil productivity, temperature, and precipitation), spatio-temporal analysis revealed how landscape controls and climate (summer vegetation moisture stress) affected patterns of NPP according to vegetation functional type, species cover type, and successional stage. These results supported regional and national trends of NPP in relation to carbon fluxes and lag effects of climate. Overall, the EMAS provides valuable decision support for PAs regarding informed land use planning, conservation programs, vital sign monitoring, control programs (fire fuels, invasives, etc.), and restoration efforts.     

Evaluation of soil moisture from satellite observations over South America

A study was performed to evaluate the surface soil moisture derived from Advanced Microwave Scanning Radiometer for the Earth Observing System (AMSR-E) sensor observations over South America. Other soil moisture and rainfall datasets were also used for the analysis. The information for the soil data came from the Eta regional climate model, and for the rainfall data from the Tropical Rainfall Microwave Mission (TRMM) satellite. Statistical analysis was used to evaluate the quality of the soil moisture and rainfall products, with estimates of the correlation coefficient (R), χ² and Cramer's phi (?_c). The results show high correlations (R > 0.8) of the AMSR-E soil moisture products with the Eta model for different regions of South America. Comparison of soil moisture products with rainfall datasets showed that the AMSR-E C-band soil moisture product was highly correlated with the TRMM satellite rainfall datasets, with the highest values of χ² and ?. The results show that the AMSR-E C-band soil moisture products contain important information that can be used for various purposes, such as monitoring floods or droughts in arid areas or as input within the framework of an assimilation scheme of numerical weather prediction models.     

On Bayesian estimation and model comparison of an integrated structural equation model

In this paper, we introduce a Bayesian approach to the estimation and model comparison of an integrated two-level nonlinear structural equation model with mixed continuous, dichotomous, and ordered categorical data that may be missing at random. This general model can accommodate nonlinearities of latent variables and the effects of fixed covariates on measurement and structural equations in within-groups and between-groups models. A sampling-based algorithm that combines the Gibbs sampler and the Metropolis-Hastings algorithm is proposed for posterior simulation. A procedure that utilizes path sampling is implemented to compute the Bayes factor for model comparison under the framework of the proposed integrated model. Empirical performances of Bayesian methodologies are illustrated via analysis of a real example.     

Identification of sparse FIR systems using a general quantisation scheme

This paper presents an identification scheme for sparse FIR systems with quantised data. We consider a general quantisation scheme, which includes the commonly deployed static quantiser as a special case. To tackle the sparsity issue, we utilise a Bayesian approach, where an ?₁ a priori distribution for the parameters is used as a mechanism to promote sparsity. The general framework used to solve the problem is maximum likelihood (ML). The ML problem is solved by using a generalised expectation maximisation algorithm.     

The EM algorithm in a distributed computing environment for modelling environmental space–time data

Statistical models for spatio-temporal data are increasingly used in environmetrics, climate change, epidemiology, remote sensing and dynamical risk mapping. Due to the complexity of the relationships among the involved variables and dimensionality of the parameter set to be estimated, techniques for model definition and estimation which can be worked out stepwise are welcome. In this context, hierarchical models are a suitable solution since they make it possible to define the joint dynamics and the full likelihood starting from simpler conditional submodels. Moreover, for a large class of hierarchical models, the maximum likelihood estimation procedure can be simplified using the Expectation–Maximization (EM) algorithm.In this paper, we define the EM algorithm for a rather general three-stage spatio-temporal hierarchical model, which includes also spatio-temporal covariates. In particular, we show that most of the parameters are updated using closed forms and this guarantees stability of the algorithm unlike the classical optimization techniques of the Newton–Raphson type for maximizing the full likelihood function. Moreover, we illustrate how the EM algorithm can be combined with a spatio-temporal parametric bootstrap for evaluating the parameter accuracy through standard errors and non-Gaussian confidence intervals.To do this a new software library in form of a standard R package has been developed. Moreover, realistic simulations on a distributed computing environment allow us to discuss the algorithm properties and performance also in terms of convergence iterations and computing times.     

Nowcasting model of low wind profile based on neural network using SODAR data at Guarulhos Airport,Brazil

A generalized regression neural network model was tested – as a nowcasting tool – to forecast the low wind profiles up to 45 min (i.e. at heights of 10, 100, 200, and 300 m) at the Guarulhos International Airport, São Paulo, Brazil. A data set representing over 4 years was generated from sonic detection and ranging and surface meteorological station, which registered vertical wind profiles with intervals from 10 m to approximately 500 m in height every 15 min, and surface meteorological variables were collected each minute, respectively. These data were simultaneously used to train, validate, and test the proposed model. The u and v forecasts generated at 300, 200, and 100 m were better than at 10 m, which could certainly be attributable to the surface roughness. In addition, the results also revealed that the performance of the model is time-dependent – decreasing over time – and that this may be correlated with the fact that the neural network is a statistical rather than physical model. The forecasts of wind components u and v are slightly biased (or closely matched to observations) at all heights, and forecast intervals with maximum values have median and average errors equal to 0.070 and ?0.017 ms^?1, respectively. The forecast model’s results were evaluated using the values of four categorical statistics: probability of detection; probability for non-events; bias; and false-alarm ratio, with respectable minimum and maximum values for u wind principal components equal to 0.841, 0.833, 0.159, 0.981 at 10 m for 45-min forecasts and 0.989, 0.987, 0.011, 0.999 at 300 m for 15-min forecasts, respectively.