投影寻踪动态聚类模型及其在天然草地分类中的应用

投影寻踪聚类模型在多因素聚类分析中被广泛应用并取得了满意的效果,然而,该模型还存在如密度窗宽参数取值由经验确定等不足,有待改进提高.本文针对投影寻踪聚类模型的不足,首次把投影寻踪的思想和动态聚类方法结合起来,构造投影指标,提出了投影寻踪动态聚类新模型.新模型在整个运算过程中不需人为给定参数,聚类结果客观、明确,而且它还具有稳定性好、操作简便等特点.天然草地分类的实际应用表明,投影寻踪动态聚类模型切实可行,取得了很好的效果,在多因素聚类分析领域具有广阔的应用前景.     

Simulating potential vegetation distribution in a mountainous region of the Eastern Tibetan Plateau:A case study of Barkam County

The main objective of this study is to simulate the potential vegetation types on the basis of environmental parameters.The paper took Barkam County in a mountainous region of the Eastern Tibetan Plateau as the study area.The vegetation distribution was mapped in 1994 and 2007 based on TM remote sensing images by object-oriented interpretation method.We overlaid the two maps to find out the vegetation patches which have not changed,and took them as stable types.Fifty per cent of the stable patches were randomly sampled to operate the logistic regression with related environmental parameters;others were used as test data of simulated results.Seven environmental parameters were mapped,including elevation,slope,aspect,surface curvature,solar radiation,temperature and precipitation,based on DEM data and meteorological site data by GIS technology.The relationship between the spatial distribution of vegetation and environmental variables were quantified by logistic regression.The distribution probabilities of each vegetation type were calculated.Finally,the spatial distribution of potential vegetation was simulated.This research can provide a scientific basis for vegetation restoration and ecological construction in this area.     

PCA-SVR在煤层瓦斯含量预测中的应用

针对煤层瓦斯含量与其影响因素之间存在着复杂的非线性关系,建立了基于主成分分析和支持向量回归机的煤层瓦斯含量预测模型。该模型有效地解决了小样本、非线性预测的问题,并发挥了主成分分析法消除输入变量间相关性的优点,减少了输入变量个数,提高了预测精度和收敛速度。通过实证分析,该模型的预测精度高,能够直接用于煤矿现场预测煤层瓦斯含量。     

区域农业市场化发展对水土流失的影响--以江西省为例

为了检验区域农业市场化发展对于水土流失的影响,以江西省为例,结合对水土流失驱动力因素的分析,构建了区域水土流失驱动力模型,并据此分析了1988~2000年区域农业市场化发展对水土流失治理的贡献。结果表明,农业市场化程度的提高,有利于提高水土流失治理的效率。1988~2000年,江西省的农业市场化水平提高了15.2％,由此通过农户行为作用,水土流失面积减少61.4471万hm2,相当于2000年水土流失面积的18.36％。因此,区域农业市场化程度每提高1％,水土流失面积将减少40.43万hm2。最后,就区域农业市场化进程中如何进一步推进水土流失治理,提高水土保持效果,提出了相关建议。     

Issues with using police citations to assign responsibility in quasi-induced exposure

Crash fault determination is one of the most critical issues in applications of quasi-induced exposure. Traditionally, the driver citation issued by the investigating police officer is the primary source to assign responsibility for motor vehicle crashes. Such citations are based on the “evidence” or observation of a moving violation (such as engaged hazardous actions) in combination with non-moving violations (such as suspended driver license) prior to the crash. The objective here is to identify the contributing factors that may lead to driver citations in two-vehicle crashes in addition to the hazardous action. Multivariate binary logistic regression modeling is employed to explore the behavior of the investigating police officer in terms of issuing citation at the crash scene. A series of explanatory parameters including roadway characteristics, environmental factors, and driver and vehicle attributes is assessed. The results show that whether the crash type was a hit-and-run, alcohol and illegal drug use, driver gender, driver age, and injury severity all appear to have significant impacts on the investigating officer’s decision-making. Specific examples are given to demonstrate how two factors hit-and-run and drinking status can skew the exposure estimates in the context of quasi-induced exposure. The findings will help to serve as a basis to select appropriate parameters in assigning crash responsibility in quasi-induced exposure applications; and we make recommendations to modify existing crash database for better safety research in the future.     

消防服用织物材料热湿舒适性综合评价

为提高消防服的热湿舒适性能,减少消防员在灭火救援过程中的热应激反应,基于现有消防服用织物材料的物性参数和单项热湿舒适性指标,采用多元回归分析的方法综合评价消防服各层材料的热湿舒适性能,研究各外层材料的单项热湿舒适性指标与物性参数之间的关系。结果表明:以黑色芳砜伦为外层材料、Goretex为防水透气层材料、Nomex针刺毡为隔热层材料、Nomex/FR-VISCOSE(50%Nomex,50%阻燃黏胶)为舒适层材料的消防服热湿舒适性最好,并得出织物的吸湿速率常数、透湿率和干燥率与物性参数之间的显著多元回归模型。     

基于潜变量回归模型的城市火灾风险评估

为评估城市火灾风险等级,针对城市火灾发生的可能性与严重性2个基本问题,建立评估指标体系,提出一种基于潜变量回归的评估模型,并给出无须迭代的确定性算法。该模型和算法无需事先指定风险等级评估分数的训练样本,不必事先指定指标汇总的加权系数,利用合理的配方约束,完成潜变量回归,实现城市火灾风险按地区或者按行业的等级评估,利用现成的软件计算,同样适用于其他具有类似数据结构的评估问题。     

矿区疏干诱发岩溶塌陷特征分析及预测

针对某矿山疏干诱发岩溶塌陷的现状,分析其塌陷分布特征。提出6个影响塌陷的控制参数。采用相关分析方法量化各控制参数对塌陷的影响。相关性分析结果表明,塌陷的主要影响因素是第4系土层厚度、水位变幅及岩溶含水层渗透系数,其中岩溶含水层渗透系数对塌陷的影响最显著。应用Logistic回归方法,建立岩溶塌陷预测模型。划分出矿区塌陷危险区的具体空间位置,共有3个区域:高危区(渗透系数大于5 m/d);危险区(渗透系数为2.5～5 m/d);活跃区(渗透系数小于2.5 m/d)。     

People’s dependency on forest resources and contributions of forests to the livelihoods: a case study in Chittagong Hill Tracts (CHT) of Bangladesh

The study was conducted to investigate people’s forest dependency and contributions of forests to the livelihoods in the three hill districts of Chittagong Hill Tracts (CHT), Bangladesh. The study revealed that forest resources contribute significantly to rural community’s livelihoods and economic well-being. Using generalized logistic regressions, this study has also investigated the socioeconomic factors that affect people’s dependency on forest resources of CHT. The study revealed that people’s income from forest and forest-related occupation was positively aligned with their forest dependency. However, their education level significantly reduced their dependency on forest resources. Thus, educating the forest-dependent people and supporting alternative income generation activities should be appreciated for effective forest management in CHT. This study will be helpful to forest and environmental policy makers for CHT.     

中国产区苹果越冬冻害的风险评估

越冬期冻害是影响苹果种植分布和质量形成的主要气象灾害之一。论文利用1961-2014年2 084个气象站点资料和冻害调查数据,基于二分类Logistic回归分别建立了越冬期初冬冻害和极端低温冻害发生的概率预测模型,并按照风险极低、低、中度、高、极高等5个等级进行了空间划分。结果表明：建立的初冬冻害和极端低温冻害风险概率预测模型均通过了Hosmer-Losmer检验,独立样本的预测准确率分别达到了83.6%和91.4%。中国产区苹果的越冬冻害主要以初冬冻害为主,覆盖了除黄河故道和云南产区外的大部分果区,而极端低温冻害主要分布在纬度或海拔较高的环渤海湾北部产区、黄土高原西北部和北疆。两种冻害的高风险区域面积基本相当,中度以上风险面积较大的省份依次为甘肃、辽宁、河北和山西,但苹果种植面积最大的陕西和山东遭遇两种冻害特别是极端低温冻害的风险概率总体较低。