基于支持向量机的城市道路降雨径流污染预测模型

为实现城市降雨径流污染有效预测,以文献中的实测数据作为样本,选取雨型、平均雨强、峰值雨强、降雨历时、雨前干期、大气降尘量、PM 10、车流量、路面材料及城市功能区等10项影响因子作为模型输入量,选取径流污染指标COD的场次降雨平均浓度EMC及初期冲刷指数FF30作为模型输出量,基于支持向量机(SVM)构建了城市道路径流污染预测模型。结果表明:EMC-SVM及FF30-SVM模型均具有较高的预测精度,EMC-SVM模型校验参数RMSE、MBE远小于数据集EMC均值,CE、CC达到0.815及0.933;FF30-SVM模型校验参数RMSE、MBE远小于数据集FF30均值,CE、CC分别为0.866及0.932;选用径向基函数(RBF)作为核函数,使用k折交叉验证法对模型参数进行寻优,对于EMC-SVM及FF30-SVM模型寻得的最优参数(c,g)分别为(64.0,0.001953125)、(2.0,0.0625)。     

多机器学习模型下逐小时PM2.5预测及对比分析

为了能及时、准确的估算出PM_2.5浓度及污染等级,分别构建了K最邻近模型（KNN）、BP神经网络模型（BPNN）、支持向量机回归模型（SVR）、高斯过程回归模型（GPR）、XGBoost模型和随机森林模型（RF）6个PM_2.5浓度预测模型,选取江西省赣州市为实验区域,采用2017~2018年逐小时气象站数据、PM_2.5浓度数据和Merra-2再分析数据开展PM_2.5预测实验.结果表明,缺少污染物观测数据时,利用能见度和气象因子等数据也能较好的预测PM_2.5浓度.在PM_2.5浓度预测精度方面,XGBoost模型最高,随机森林模型次之,高斯过程回归模型最差.6个模型的预测精度总体呈现冬季最高,秋季和春季次之,夏季最低.XGBoost模型的PM_2.5污染等级预测准确率高于其他模型,综合准确率达87.6%,并且XGBoost模型具有训练时间短,占用内存小等优点.XGBoost模型的变量重要性结果表明,能见度变量的重要性最高,相对湿度和时间变量次之.本研究可为环境部门准确预测、预报PM_2.5浓度提供参考.     

邻苯二甲酸二异癸酯联合甲醛致小鼠学习记忆障碍的氧化损伤机制研究

为了探究邻苯二甲酸二异癸酯(Diisodecyl Phthalate,DIDP)、甲醛(Formaldehyde,FA)二者联合暴露致小鼠学习记忆障碍的氧化损伤机制,以及维生素E(Vitamin E,Vit E)对二者所致氧化损伤的保护作用,将80只SPF级3周龄雄性KM小鼠随机平均分为8组:(A)阴性对照组;(B)0.15 mg·kg~(-1)·d~(-1)DIDP组;(C)1.5 mg·kg~(-1)·d~(-1)DIDP组;(D)15 mg·kg~(-1)·d~(-1)DIDP组;(E)150 mg·kg~(-1)·d~(-1)DIDP组;(F) 1 mg·m~(-3)FA组;(G)1 mg·m~(-3)FA+15 mg·kg~(-1)·d~(-1)DIDP组;(H)1 mg·m~(-3)FA+15 mg·kg~(-1)·d~(-1)DIDP+100 mg·kg~(-1)·d~(-1)Vit E组.连续染毒3周.于染毒第13 d开始进行水迷宫实验,末次给药后24 h内取小鼠脑组织制作切片和组织匀浆,检测脑组织匀浆中活性氧(ROS)、还原型谷胱甘肽(GSH)、丙二醛(MDA)、8-羟基脱氧鸟苷(8-OHdG)的含量.结果表明,(A)组～(E)组小鼠脑海马组织氧化损伤逐渐加重,小鼠学习记忆能力逐渐下降;(G)组较(F)组小鼠脑海马组织氧化损伤加重,小鼠学习记忆能力下降;(H)组较(G)组小鼠脑海马组织氧化损伤减轻,小鼠学习记忆能力有所提高.研究显示,DIDP可通过氧化应激作用,引起脑海马组织损伤,导致小鼠学习记忆能力下降.     

Complex dynamics in the uptake of new farming practices: a case study for organic waste application

Adverse environmental effects of intensive agriculture, together with scarcity in phosphates and water, urge farmers to find more sustainable practices. An example of such a sustainable practice is on-farm processing of organic waste. This paper explores three mechanisms that can lead to a widespread uptake of this technique: (1) economies of scale, (2) information sharing, and (3) adjustment of social norms. Although each of these mechanisms has been studied before, this paper provides new insights by considering the interactions that might exist between the different mechanisms when they are applied to real-life situations. Based on a pilot study, we developed a multi-criteria mathematical programming model at individual farm level. We used this model to simulate the uptake of on-farm processing of organic waste, as a result of the three mechanisms and their interactions. Our results show that each mechanism results in an increased uptake, but is not likely to cause a widespread uptake. Interaction between the mechanisms, will lead to a much higher uptake. This result suggests that simultaneous consideration of multiple mechanisms is essential to understand the behaviour of social–ecological systems.     

Reflexive governance: exploring the concept and assessing its critical potential for sustainable development. Introduction to the special issue

ABSTRACT

Elaborated in publications on transition management, sustainability governance and deliberative environmental governance, ‘reflexive governance’ addresses concerns about social-ecological vulnerabilities, flawed conceptualisations of human-nature relations fragmented governance regimes and conditions for a sustainability transition. Key barriers to reflexive government include unavoidable politics; the influence of broader discursive systems that shape actors’ strategic interests; and structural and deliberate limitations to the range of admitted epistemological understandings, normative perspectives and material practices. Against this background, the contributions to the special issue provide novel conceptual linkages between reflexive governance and boundary objects, intercultural dialogue, conflict management heuristics, discourse linguistics, theories of the policy cycle and reflexive law, network and learning theories, and Lasswell’s ‘developmental constructs’. Based on the contributions, we identify five inherent conceptual tensions of reflexive governance: between the openness of horizontal learning processes and the desired direction towards sustainable development; between reflexive governance as a normative or procedural concept; between expected learning orientations and other, strategic orientations; between governance as a precondition for reflexivity and reflexive learning as a precondition for reorganized governance structures; and between reflexivity as an open-ended, evolutionary process and the need to strategically defend the space for reflexivity against powerful groups with an interest in the status quo.     

Barriers and enablers to climate change adaptation in hierarchical governance systems: the case of Vietnam

Governments fulfil important roles in increasing the adaptive capacity of local communities to respond to climate change impacts, particularly in developing countries. Existing studies on how governments enable and constrain the ways in which local level communities learn and build their adaptive capacity, however, generally adopt network or market-oriented types of governance. However, the most vulnerable regions to climate change impact in the world are generally governed through hierarchical policy systems. This research aims to understand how the hierarchical policy system in Vietnam creates enables and/or constrains the policy capacity of policy actors to contribute to effective climate change adaptation. We conducted interviews (n?=?26) with key actors at multiple levels of government. Our findings show the importance of clear legal institutions, available financing for implementing policies, and the training of governmental staff, particularly at district and commune levels where the policy capacities are generally too low to deal with climate change impacts. We conclude that any efforts to support local actors (i.e. smallholder farmers) should include investments in policy capacity to ensure uptake and upscaling of adaptation actions more broadly.     

Mentoring with(in) care: A co‐constructed auto‐ethnography of mutual learning

Research into workplace mentoring is principally focussed on predictors and psychosocial and instrumental outcomes, while there is scarcely any in‐depth research into relational characteristics, outcomes and processes. This article aims to illustrate these relational aspects. It reports a co‐constructed auto‐ethnography of a dyadic mentoring relationship as experienced by mentor and protégé. The co‐constructed narrative illustrates that attentiveness towards each other and a caring attitude, alongside learning‐focussed values, promote a high‐quality mentoring relationship. This relationship is characterised, among other things, by person centredness, care, trust and mutual influence, thereby offering a situation in which mutual learning and growth can occur. Learning develops through and in relation and is enhanced when both planned and unplanned learning takes place. In addition, the narrative makes clear that learning and growth of both those involved are intertwined and interdependent and that mutual learning and growth enrich and strengthen the relationship. It is concluded that the narrative illustrates a number of complex relational processes that are difficult to elucidate in quantitative studies and theoretical constructs. It offers deeper insight into the initiation and improvement of high‐quality mentoring relationships and emphasises the importance of further research into relational processes in mentoring relationships. Copyright © 2015 John Wiley & Sons, Ltd.     

基于自然正交展开的神经网络长期预报模型

对月降水量的前期500hPa高度场、海温场相关预报因子进行E0F展开，并取其中与预报量相关程度较高的主成分，结合人工神经网络技术，建立了一种新的短期气候预测模型。将这种新的预报模型与同样根据这些预报因子建立的回归预报模型进行了对比分析。结果表明，这种新的短期气候预测模型由于集中了众多预报因子的预报信息，并有效地利用了神经网络方法的非线性映射能力，因此比传统预报方法的预报精度显著提高，并且稳定性好，具有很好的应用前景。     

事故调查报告中事故原因分析的改进*——基于企业事故案例学习视角

为进一步提高我国事故调查报告的质量,更好地满足企业对事故案例学习的需要,通过案例分析事故调查报告中事故原因部分存在的问题及其原因,并在此基础上提出采用事故致因理论来改进事故调查报告的思路;应用24Model改进天嘉宜化工有限公司“3·21”特别重大爆炸事故的事故原因分析,以此来验证本文思路的可行性和有效性。研究结果表明,使用事故致因理论能够更加科学地归纳事故原因,增强措施的针对性和可操作性,可有效改善事故调查报告的质量。     

Workers’ compensation claim counts and rates by injury event/exposure among state-insured private employers in Ohio, 2007–2017

Introduction: This study analyzed workers’ compensation (WC) claims among private employers insured by the Ohio state-based WC carrier to identify high-risk industries by detailed cause of injury. Methods: A machine learning algorithm was used to code each claim by U.S. Bureau of Labor Statistics (BLS) event/exposure. The codes assigned to lost-time (LT) claims with lower algorithm probabilities of accurate classification or those LT claims with high costs were manually reviewed. WC data were linked with the state’s unemployment insurance (UI) data to identify the employer’s industry and number of employees. BLS data on hours worked per employee were used to estimate full-time equivalents (FTE) and calculate rates of WC claims per 100 FTE. Results: 140,780 LT claims and 633,373 medical-only claims were analyzed. Although counts and rates of LT WC claims declined from 2007 to 2017, the shares of leading LT injury event/exposures remained largely unchanged. LT claims due to Overexertion and Bodily Reaction (33.0%) were most common, followed by Falls, Slips, and Trips (31.4%), Contact with Objects and Equipment (22.5%), Transportation Incidents (7.0%), Exposure to Harmful Substances or Environments (2.8%), Violence and Other Injuries by Persons or Animals (2.5%), and Fires and Explosions (0.4%). These findings are consistent with other reported data. The proportions of injury event/exposures varied by industry, and high-risk industries were identified. Conclusions: Injuries have been reduced, but prevention challenges remain in certain industries. Available evidence on intervention effectiveness was summarized and mapped to the analysis results to demonstrate how the results can guide prevention efforts. Practical Applications: Employers, safety/health practitioners, researchers, WC insurers, and bureaus can use these data and machine learning methods to understand industry differences in the level and mix of risks, as well as industry trends, and to tailor safety, health, and disability prevention services and research.