INTEGRATING SERVICE‐LEARNING INTO WATERSHED MANAGEMENT PROGRAMS: OPPORTUNITIES AND CHALLENGES1

ABSTRACT: The objective of this article is to open a dialogue on integrating service‐learning into community based watershed management programs and to discuss opportunities and challenges that a service‐learning program presents to universities and communities. The article presents the concept and definition of service‐learning, and arguments concerning why institutions of higher education and university faculty and students should be involved with community based watershed management programs. The article describes a case study for developing a service‐learning program for watershed management at Virginia Tech and discusses lessons learned from the case study. The paper concluded that to make a service‐learning program sustainable, there should be a long term plan, regular and effective communication with the stakeholders, and some incentives for faculty and students for long term commitment to the community based watershed management programs.     

GROUND WATER NITRATE REDUCTION IN GIANT CANE AND FOREST RIPARIAN BUFFER ZONES1

ABSTRACT: Ground water contamination by excess nitrate leaching in row‐crop fields is an important issue in intensive agricultural areas of the United States and abroad. Giant cane and forest riparian buffer zones were monitored to determine each cover type's ability to reduce ground water nitrate concentrations. Ground water was sampled at varying distances from the field edge to determine an effective width for maximum nitrate attenuation. Ground water samples were analyzed for nitrate concentrations as well as chloride concentrations, which were used as a conservative ion to assess dilution or concentration effects within the riparian zone. Significant nitrate reductions occurred in both the cane and the forest riparian buffer zones within the first 3.3 m, a relatively narrow width. In this first 3.3 m, the cane and forest buffer reduced ground water nitrate levels by 90 percent and 61 percent, respectively. Approximately 40 percent of the observed 99 percent nitrate reduction over the 10 m cane buffer could be attributed to dilution by upwelling ground water. Neither ground water dilution nor concentration was observed in the forest buffer. The ground water nitrate attenuation capabilities of the cane and forest riparian zones were not statistically different. During the spring, both plant assimilation and denitrification were probably important nitrate loss mechanisms, while in the summer nitrate was more likely lost via denitrification since the water table dropped below the rooting zone.     

供水管线震害量化参数--渗漏面积的估算方法

为了分析管线破坏状态对震后供水管网功能的影响，首先需要对管线的震害状态进行量化。建议用管线渗漏面积标定管线的破坏状态。在管线震害分析和理论计算的基础上，提出了震后管线渗漏面积的计算方法，为进一步进行供水管网抗震分析和设计提供了研究基础。     

QUALITY OF HYDROMETEOROLOGICAL DATA IN COLD REGIONS1

ABSTRACT: Many users of hydrometeorological records are not aware of the number of inconsistencies and biases that occur in hydrometeorological records. Examples are presented illustrating how the exposures of sites for measurement of precipitation, wind, snow on the ground, and evaporation determine to a large extent how useful the records are for estimating areal conditions. For areas where precipitation in the form of snow produces a significant portion of the runoff, a smaller number of quality records may be more valuable for modeling than a much larger number of records of lower quality. Information is presented to show that the overall value of an operational hydrometeorological network is dependent upon how consistent and representative of average conditions the collected records are, especially for mountainous areas in cold regions.     

神经网络及其在水库调洪演算中的应用

结合洪水灾害管理的特点建立了水库调洪演算的人工神经网络模型，阐述了其基本原理及算法，并结合某水库调洪演算实例说明了其应用。     

基于神经网络的重大危险源动态分级研究

利用自组织神经网络对重大危险源进行动态分级研究，介绍了神经网络的模式聚类即分级法的自组织学习过程和算法，克服了以往危险源分级方法的某些局限性。在计算机上进行的模拟计算表明：基于自组织神经网络的重大危险源动态分级方法是合理可行的     

西江年最高水位的神经网络预报模型

对西江洪水发生的特征进行分析表明，洪水发生频率高，具有明显阶段性特征，并与流域面雨量密切相关。利用前期环流场、海表温度（SST）场及环流特征量资料选择初选预报因子，然后对初选预报因子作EOF分解构造综合预报因子，结合人工神经网络方法建立了西江年最高水位预报模型，并对预报模型进行独立样本试验。结果表明，该预报模型对历史样本拟合精度高，试报效果明显好于传统的逐步回归模型，可在汛期预测业务中应用。     

基于模糊评判人工神经网络的重大危险源辨识研究

国家已经制定了重大危险源辨识标准,辨识依据是危险物质的数量,主要适用于化工行业.本文对于国家重大危险源辨识标准不适合的行业,结合各行业发生事故的特点和工艺特征,综合利用模糊评判法和人工神经网络,辨识企业内存在的重大危险源.在某化工厂进行了检验,实践证明了方法的科学性和合理性.     

BP神经网络技术在交通工具火灾预警中的应用

通过对交通工具火灾成因机理以及现有典型交通工具火灾实例的分析研究,建立了预警评价指标体系。根据非线性理论和模式识别原理以及交通工具火灾的特点,采用基于BP神经网络的智能灾害诊断方法,对交通工具火灾发生的可能性和危险性进行评估和预测。研究表明:BP神经网络方法是解决非线性系统问题的一种有效方法,与传统的预警方法相比,该方法具有速度快、效率高、可信度好、自学能力强等特点。采用BP网络进行交通工具火灾预警时,只需输入影响交通工具火灾发生的相关指标因素,网络便可在较短的时间内得出可靠的预警结果。     

基于神经网络和遗传算法的城市火灾风险评价模型

以消防安全工程学与系统安全工程理论为基础,结合我国城市发展特征及消防安全管理状况,建立了城市区域火灾风险评价指标体系;针对神经网络易陷入局部极小而引起评价指标权值分布不合理的缺陷,提出了基于神经网络和遗传算法的城市火灾风险评价模型,该模型以火灾发生的可能性以及灾后的严重程度为输入单元,火灾风险等级为输出单元,采用误差反算法训练BP网络,最终得出火灾风险等级范围,有效地解决了城市火灾的动态性和非线性特征;研究实例证明了该模型的有效性,可为城市的消防安全管理提供确实可行的参考依据。