基于粒子群-支持向量机(PSO-SVM)的苯储罐泄漏事故风险预测

将支持向量机(SVM)模型运用于事故前苯储罐泄漏事故风险预测,为使模型性能最优, 用粒子群算法PSO优化SVM模型参数,建立了PSO-SVM风险预测模型。为验证模型风险预测性能,分别采用遗传算法(GA)和网格搜索法(GS)优化SVM参数,并比较测试集与PSO-SVM、GA-SVM、GS-SVM三种模型预测结果的均方误差及相关系数。然后进一步探讨模型中权重调整方式、种群规模对PSO-SVM模型预测性能的影响。研究发现,权重线性递减所建PSO-SVM预测值与测试集相关系数更高、均方误差更小、预测效果更好,种群规模没有影响PSOSVM模型预测值但会影响计算时间,这为危化品泄漏事故的风险预测提供了一种新的方法。     

A CRITICAL REVIEW OF VOLLENWEIDER'S NUTRIENT BUDGET MODEL AND OTHER RELATED MODELS1

ABSTRACT: Early attempts at nutrient budget modeling considered only the case where there was no loss of the material by sedimentation, i.e. the substance was assumed to be conservative. Nonstratified and stratified conditions have both been investigated under these terms. An elegant model, taking into account loss of a substance by sedimentation as well as flushing was presented by Vollenweider in 1969. Although this model has several shortcomings, it is particularly valuable because it can have immediate practical value in terms of water management policy development. These basic shortcomings in the model are analyzed and suggestions are made to alter the model to take these factors into account.     

GROUND WATER SIMULATION USING A ONE DIMENSIONAL FLOW MODEL1

ABSTRACT: In projects involving ground water problems, dependence on the mathematical modeling of the ground water flow phenomena is inescapable. At present, two dimensional flow models, which require tremendous amounts of computer time and storage, are generally used. When such bulky models are used for planning purposes, the two requirements (computer time and storage) can severely limit the number of alternatives that can be considered. A simple quantity and quality simulation model is developed here which requires considerably less computer time and storage and gives reasonably accurate results. The model was applied to simulate a ground water basin in San Luis Rey River in Southern California. The results were compared with those obtained by a USGS model. It was found that the simple model gave results which were consistentaly within five percent of the USGS model results, while the requirements on computer time and storage were drastically reduced.     

鄂尔多斯周缘地震带中长期地震趋势估计

本文基于一定的地震资料,分别应用最大熵原理、贝努里概型以及计算地震复发周期的经验关系式,对鄂尔多斯周缘地震带进行了中强地震的中长期地震趋势估计。     

Catastrophic Reproductive Failure, Terrestrial Survival, and Persistence of the Marbled Salamander

Abstract:  Wide variation in reproductive success is common among amphibians that breed in seasonal ponds, but persistence of adults can buffer against these fluctuations, particularly for long-lived species. We hypothesized that the frequent episodes of catastrophic failure of the marbled salamander ( Ambystoma opacum ) enhance the importance of high terrestrial survival. At Rainbow Bay in South Carolina reproductive success was poor (<1 metamorph/breeding female) in nearly half of the 22 years that the species bred. Complete failure occurred in 6 of 22 years. To study catastrophic failure, we adapted an age-structured, individual-based model with density-dependent growth and survival of larvae. The model was based on extensive data from local field studies and experiments. With consistently good survival in the pond stages, the simulated population required survival probabilities in the upland stages (juveniles and adults) near 0.5/year to persist and near 0.8/year to achieve the increases observed. Catastrophic failure, occurring randomly with probability 0.5/year, created additional fluctuations in the population, raised the thresholds of survival required for persistence, and caused extinction under conditions that were otherwise favorable. The marbled salamander at Rainbow Bay is not at great risk of extinction because of catastrophic failure, but the risk increases dramatically if life span is decreased or frequency of failure is increased. Any reduction in terrestrial survival will have deleterious consequences by reducing the breeding populations at equilibrium, even if it does not jeopardize persistence. Our model provides assessments of risk that can be applied to poorly studied species with similar life histories, such as the endangered flatwoods salamander ( A. cingulatum ).     

Assessing the Effects of Alternative Setback Channel Constraint Scenarios Employing a River Meander Migration Model

River channel migration and cutoff events within large river riparian corridors create heterogeneous and biologically diverse landscapes. However, channel stabilization (riprap and levees) impede the formation and maintenance of riparian areas. These impacts can be mitigated by setting channel constraints away from the channel. Using a meander migration model to measure land affected, we examined the relationship between setback distance and riparian and off-channel aquatic habitat formation on a 28-km reach of the Sacramento River, California, USA. We simulated 100 years of channel migration and cutoff events using 11 setback scenarios: 1 with existing riprap and 10 assuming setback constraints from about 0.5 to 4 bankfull channel widths (bankfull width: 235 m) from the channel. The percentage of land reworked by the river in 100 years relative to current (riprap) conditions ranged from 172% for the 100-m constraint setback scenario to 790% for the 800-m scenario. Three basic patterns occur as the setback distance increases due to different migration and cutoff dynamics: complete restriction of cutoffs, partial restriction of cutoffs, and no restriction of cutoffs. Complete cutoff restriction occurred at distances less than about one bankfull channel width (235 m), and no cutoff restriction occurred at distances greater than about three bankfull widths (∼700 m). Managing for point bars alone allows the setbacks to be narrower than managing for cutoffs and aquatic habitat. Results suggest that site-specific “restriction of cutoff” thresholds can be identified to optimize habitat benefits versus cost of acquired land along rivers affected by migration processes.     

Parameterizing,evaluating and comparing metapopulation models with data from individual-based simulations

Due to the lack of sufficient data and appropriate ecological information parameterizing predictive population dynamical models usually is a difficult task. The approach proposed in this study is meant to overcome this problem by using detailed individual-based simulations to generate artificial data. With short-term data samples, the models to be investigated can be parameterized and their predictions be compared. The flexibility of individual-based simulations as experimental tools also facilitates the evaluation and comparison of different (aggregated) model types. The presented approach is a step towards unifying models of different complexity. As an example we applied it to two metapopulation models of insect species in a highly fragmented landscape: the well-known incidence function model with a patch-based representation of space and a grid-based analogue. The models are tested with respect to their data requirement and recommendations for a better data sampling are derived.     

Ecological scenarios analyzed and evaluated by a shallow lake model

We applied the complex ecosystem model EMMO, which was adopted to the shallow lake Müggelsee (Germany), in order to evaluate a large set of ecological scenarios. By means of EMMO, 33 scenarios and 17 indicators were defined to characterize their effects on the lake ecosystem. The indicators were based on model outputs of EMMO and can be separated into biological indicators, such as chlorophyll-a and cyanobacteria, and hydro-chemical indicators, such as phosphorus. The question to be solved was, what is the ranking of the scenarios based on their characterization by these 17 indicators? And how can we handle high quantities of complex data within evaluation procedures? The scenario evaluation was performed by partial order theory which, however, did not provide a clear result. By subsequently applying the hierarchical cluster analysis (complete linkage) it was possible to reduce the data matrix to indicator and scenario representatives. Even though this step implies losses of information, it simplifies the application of partial order theory and the post processing by METEOR. METEOR is derived from partial order theory and allows the stepwise aggregation of indicators, which subsequently leads to a distinct and clear decision. In the final evaluation result the best scenario was the one which defines a minimum nutrient input and no phosphorus release from the sediment while the worst scenario is characterized by a maximum nutrient input and extensive phosphorus release from the sediment. The reasonable and comprehensive results show that the combination of partial order, cluster analysis and METEOR can handle big amounts of data in a very clear and transparent way, and therefore is ideal in the context of complex ecosystem models, like that we applied.     

基于CA-Markov模型的艾比湖流域平原区景观格局动态模拟预测

利用“3S”技术及CA-Markov模型,按照是否实施跨流域调水工程2种预案情况,以2005年为起始时刻,对新疆艾比湖流域平原区2020年景观格局进行模拟预测。结果显示：在调水工程未实施的情况下,2020年研究区的生态环境将进一步恶化,其突出表现为艾比湖湖泊水面将持续萎缩、裸露湖底盐漠面积进一步扩大,水资源短缺及生态环境恶化的结果将严重制约研究区社会经济的发展;在调水工程实施的情况下,2020年研究区内艾比湖湖泊水面将稳定增加至800 km^2以上、裸露湖底盐漠面积相应有所减少,生态环境恶化趋势得到改善,区域水资源短缺问题将有所缓解,可有效地促进研究区内社会经济的发展。     

结构风灾经济损失模型在GIS中的应用

在城市抗风防灾研究中，采用GIS技术已成为一种必要的手段，但现有结构风灾经济损失模型参数多，不易采集，造成建立GIS数据库具有一定的困难。因此，简化模型，降低数据采集难度的工作量是十分必要的，本文采用样本计算，制成表格供查询使用，在保证计算精度的前提下，使参数由12个减为2个，大大降低了数据采集和GIS数据库建立的难度。