基于距离相关系数和支持向量机回归的PM2.5浓度滚动统计预报方案

针对目前空气质量统计预报方法存在的主要缺陷,本文提出了距离相关系数和支持向量机回归相结合的统计预报方案DC-SVR.利用淮安市2013年1—12月PM_(2.5)观测资料和常规气象观测资料,首先在选入预报当日气象要素的基础上,增加选取前期污染物和气象要素作为预报因子,再采用距离相关系数分季节从预报因子中筛选出重要预报因子,最后采用支持向量机回归对PM_(2.5)浓度值进行逐日滚动统计预报.研究发现,淮安地区气温和气压对PM_(2.5)的距离相关性要高于其他气象要素,夏秋季PM_(2.5)与气象要素的距离相关性较春冬季好.基于距离相关系数和支持向量机回归建立DC-SVR模型,PM_(2.5)的试预报值和实测值的全年相关系数高达0.76,平均偏差仅为1.13μg·m~(-3),平均绝对误差为23.47μg·m~(-3).通过与支持向量机回归、人工神经网络的统计预报效果对比,DC-SVR模型有效降低预报因子维数且能自适应选取最佳参数,预报精度显著优于其他3种统计预报方案,可为业务化预报提供参考.     

A measurement of community disaster resilience in Korea

Building a community that is resilient to disasters has become one of the main goals of disaster management. Communities that are more disaster resilient often experience less impact from the disaster and reduced recovery periods afterwards. This study develops a methodology for constructing a set of indicators measuring Community Disaster Resilience Index (CDRI) in terms of human, social, economic, environmental, and institutional factors. In this study, the degree of community resilience to natural disasters was measured for 229 local municipalities in Korea, followed by an examination of the relationship between the aggregated CDRI and disaster losses, using an ordinary least squares (OLS) regression method and a geographically weighted regression (GWR) method. Identifying the extent of community resilience to natural disasters would provide emergency managers and decision-makers with strategic directions for improving local communities' resilience to natural disasters while reducing the negative impacts of disasters.     

Trace metals distribution and their dependence on some physico-chemical parameters in creek sediment

The concentrations of the trace metals Mn, Zn, Cr, V, Co, Ni, Ba, Pb, Cu, Cd, As and Hg, their spatial distribution, and their correlation as a function of physico-chemical parameters were investigated for grab sediment samples collected from different locations across the Thane Creek, Mumbai, India. Hydride generation atomic absorption spectrometry, inductively coupled plasma atomic emission spectrometry, and differential pulse anodic stripping voltammetry were adopted for chemical analyses. Various geochemical indices were evaluated. The studied metal/metalloids were compared with sediment quality guideline values. Besides, the role of different physico-chemical parameters of the sediments such as pH, CaCO₃, cation exchange capacity, total carbon, and of the relative composition in respect to very coarse, coarse and medium sand, fine sand and very fine sand, silt and clay fraction were also critically investigated. Positive one-to-one correlations between fine sand and very fine sand, silt and clay, and total carbon with sediment metal concentrations have been noticed. Multiple regression analysis reveals strong positive correlations between sediment metal concentrations and total carbon content only.     

An analysis of assessment outcomes from eight years' operation of the Australian border weed risk assessment system

The majority of Australian weeds are exotic plant species that were intentionally introduced for a variety of horticultural and agricultural purposes. A border weed risk assessment system (WRA) was implemented in 1997 in order to reduce the high economic costs and massive environmental damage associated with introducing serious weeds. We review the behaviour of this system with regard to eight years of data collected from the assessment of species proposed for importation or held within genetic resource centres in Australia. From a taxonomic perspective, species from the Chenopodiaceae and Poaceae were most likely to be rejected and those from the Arecaceae and Flacourtiaceae were most likely to be accepted. Dendrogram analysis and classification and regression tree (TREE) models were also used to analyse the data. The latter revealed that a small subset of the 35 variables assessed was highly associated with the outcome of the original assessment. The TREE model examining all of the data contained just five variables: unintentional human dispersal, congeneric weed, weed elsewhere, tolerates or benefits from mutilation, cultivation or fire, and reproduction by vegetative propagation. It gave the same outcome as the full WRA model for 71% of species. Weed elsewhere was not the first splitting variable in this model, indicating that the WRA has a capacity for capturing species that have no history of weediness. A reduced TREE model (in which human-mediated variables had been removed) contained four variables: broad climate suitability, reproduction in less or than equal to 1year, self-fertilisation, and tolerates and benefits from mutilation, cultivation or fire. It yielded the same outcome as the full WRA model for 65% of species. Data inconsistencies and the relative importance of questions are discussed, with some recommendations made for improving the use of the system.     

Modelling plant species richness using functional groups

Conservation biologists increasingly rely on spatial predictive models of biodiversity to support decision-making. Therefore, highly accurate and ecologically meaningful models are required at relatively broad spatial scales. While statistical techniques have been optimized to improve model accuracy, less focus has been given to the question: How does the autecology of a single species affect model quality? We compare a direct modelling approach versus a cumulative modelling approach for predicting plant species richness, where the latter gives more weight to the ecology of functional species groups. In the direct modelling approach, species richness is predicted by a single model calibrated for all species. In the cumulative modelling approach, the species were partitioned into functional groups, with each group calibrated separately and species richness of each group was cumulated to predict total species richness. We hypothesized that model accuracy depends on the ecology of individual species and that the cumulative modelling approach would predict species richness more accurately. The predictors explained plant species richness by ca. 25%. However, depending on the functional group the deviance explained varied from 3 to 67%. While both modelling approaches performed equally well, the models of the different functional groups highly varied in their quality and their spatial richness pattern. This variability helps to improve our understanding on how plant functional groups respond to ecological gradients.     

Watershed Regressions for Pesticides (warp) Models for Predicting Atrazine Concentrations in Corn Belt Streams1

Stone, Wesley W. and Robert J. Gilliom, 2012. Watershed Regressions for Pesticides (WARP) Models for Predicting Atrazine Concentrations in Corn Belt Streams. Journal of the American Water Resources Association (JAWRA) 48(5): 970‐986. DOI: 10.1111/j.1752‐1688.2012.00661.x Abstract: Watershed Regressions for Pesticides (WARP) models, previously developed for atrazine at the national scale, are improved for application to the United States (U.S.) Corn Belt region by developing region‐specific models that include watershed characteristics that are influential in predicting atrazine concentration statistics within the Corn Belt. WARP models for the Corn Belt (WARP‐CB) were developed for annual maximum moving‐average (14‐, 21‐, 30‐, 60‐, and 90‐day durations) and annual 95th‐percentile atrazine concentrations in streams of the Corn Belt region. The WARP‐CB models accounted for 53 to 62% of the variability in the various concentration statistics among the model‐development sites. Model predictions were within a factor of 5 of the observed concentration statistic for over 90% of the model‐development sites. The WARP‐CB residuals and uncertainty are lower than those of the National WARP model for the same sites. Although atrazine‐use intensity is the most important explanatory variable in the National WARP models, it is not a significant variable in the WARP‐CB models. The WARP‐CB models provide improved predictions for Corn Belt streams draining watersheds with atrazine‐use intensities of 17 kg/km² of watershed area or greater.     

Survivor needs or logistical convenience? Factors shaping decisions to deliver relief to earthquake‐affected communities,Pakistan 2005–06

In Bureaucratizing the Good Samaritan , Waters (2001) argues that bureaucratic rationality distracts humanitarian agencies from the needs of the people they are supposed to assist, in favour of other values that their institutional frameworks dictate. We test his claim by investigating the response to the Pakistan 2005 earthquake. One of us (Dittemore) worked with the United Nations Joint Logistics Centre in the theatre, managing a relief cargo shipment database. The response, known as 'Operation Winter Race', was hampered by extreme logistical challenges, but ultimately succeeded in averting a second disaster resulting from cold and starvation. We use statistical models to probe whether survivor needs significantly guided decisions to deliver relief to affected communities. Needs assessments remained incomplete and incoherent. We measure needs through proxy indicators and integrate them, on a Geographic Information System (GIS) platform, with logistics and relief delivery data. We find that, despite strong logistics effects, needs orientations were significant. However, the strength of decision factors varies between commodity types (food versus clothing and shelter versus reconstruction materials) as well as over the different phases of the response. This study confirms Thomas's observation that logistics databases are rich 'repositories of data that can be analyzed to provide post-event learning' (Thomas, 2003, p. 4). This article is an invitation for others to engage in creative humanitarian data management.     

Modeling perceived collision risk in port water navigation

An increase in the likelihood of navigational collisions in port waters has put focus on the collision avoidance process in port traffic safety. The most widely used on-board collision-avoidance system is the automatic radar plotting aid which is a passive warning system that triggers an alert based on the pilot’s pre-defined indicators of distance and time proximities at the closest point of approaches in encounters with nearby vessels. To better help pilot in decision making in close quarter situations, collision risk should be considered as a continuous monotonic function of the proximities and risk perception should be considered probabilistically. This paper derives an ordered probit regression model to study perceived collision risks. To illustrate the procedure, the risks perceived by Singapore port pilots were obtained to calibrate the regression model. The results demonstrate that a framework based on the probabilistic risk assessment model can be used to give a better understanding of collision risk and to define a more appropriate level of evasive actions.     

Challenges in radioecology

Today, radioecology covers a broad scientific field; from the source to long term environmental impact from ionizing radiation. To summarize key challenges within radioecology, the present paper focuses upon knowledge gaps related to processes, mechanisms and variables contributing most to the overall uncertainties in environmental impact assessments. A series of sources related to the nuclear weapon cycle and the civil nuclear cycle has contributed, is still contributing or can potentially contribute to release of radionuclides to the environment in the future. The speciation of most radionuclides depends on the source and release conditions, and will influence ecosystem transport, biological uptake, doses and effects in flora and fauna. Radionuclides may also co-occur in contaminant mixtures (e.g., metals, organics), which potentially could lead to synergisms or antagonisms. Thus, challenges associated with the links between the source or release term – radionuclide speciation – ecosystem transfer – exposure – response relationships are highlighted.     

Assessment Tools for Urban Catchments: Defining Observable Biological Potential1

Abstract: Urbanization represents a strong and increasingly more prevalent impact on stream quality worldwide. One of the characteristic effects of increased urbanization is a consistent decline in biological stream condition. The characterization of this biological degradation with increasing urbanization presents a number of advantages for the study and management of urban streams and catchments. In this paper, the limitation of biological condition with urbanization, called observed biological potential, is characterized. Using an urban intensity index and a biological index developed specifically for urban systems in the Baltimore, Maryland; Cleveland, Ohio; and San Jose, California regions, two principal techniques were compared (quantile regression and bin regression) to define observed biological potential along urban gradients. Quantile regression was selected as the preferable tool for describing observed biological potential given the consistency with which it can be applied and its statistical efficiency, however, bin quantile regression performed similarly. Having identified a numeric approximation of observed biological potential, two methods for identifying factors related to distance from potential as a way of identifying critical environmental factors affecting biological condition in urban areas were explored. The results of this work can be used for identifying benchmarks for urban stream biological condition, identifying limiting catchment characteristics, and prioritizing urban stream management efforts.