Monitoring the recovery of a previously exploited surf-zone fish community in the St Lucia Marine Reserve,South Africa,using a no-take sanctuary area as a benchmark

Limiting accessibility to a fishing area can reduce fishing effort effectively and may therefore promote local recovery of depleted stocks. In January 2002, beach driving was banned in South Africa, thereby reducing angler access to large areas of the coastline, particularly in less-developed areas. In November 2001 a project had been established in the St Lucia Marine Reserve, on the east coast of South Africa, to compare surf-zone fish populations inside a no-take sanctuary zone with those in an adjacent exploited area. Subsequent to the ban, the aim of the project was adapted such that surf-zone fish populations were monitored for potential recovery in what had been the exploited area, located to the north of Cape Vidal, which anglers could no longer access easily, because of the prohibition on beach driving. Standardised research fishing was conducted at two sites in the previously exploited area and two sites in the no-take sanctuary. Conventional stock-status indicators showed evidence of recovery in the four most common species caught in the previously exploited area, in terms of both abundance and biomass. Generalised additive mixed models were used to account for the influence of targeting specific species; however, subtle differences in habitat between the sampling sites, improved angling skill over time, variability in recruitment, and differential species-specific responses complicated interpretation of results. The implications of the findings are discussed in terms of the future monitoring and management of marine protected areas both within the St Lucia Marine Reserve and farther afield.     

深海半潜浮式生产平台关键理论与技术问题

深海半潜浮式生产平台FPU(floating production unit)是国际海洋油气资源开发的关键基础性装备。针对半潜浮式生产平台的技术特点,系统阐述其各方面特性(包括:总布置、系泊和立管、水动力性能、结构设计与分析等)及相关研究现状和未来发展趋势。并结合我国南海地理、水文和资源环境等特点,对未来半潜浮式生产平台的开发提出了一些建议。     

Characterizing the urban temperature trend using seasonal unit root analysis: Hong Kong from 1970 to 2015

This paper explores urban temperature in Hong Kong using long-term time series. In particular, the characterization of the urban temperature trend was investigated using the seasonal unit root analysis of monthly mean air temperature data over the period January 1970 to December 2013. The seasonal unit root test makes it possible to determine the stochastic trend of monthly temperatures using an autoregressive model. The test results showed that mean air temperature has increased by0.169?C(10 yr)~(-1)over the past four decades. The model of monthly temperature obtained from the seasonal unit root analysis was able to explain 95.9% of the variance in the measured monthly data — much higher than the variance explained by the ordinary least-squares model using annual mean air temperature data and other studies alike. The model accurately predicted monthly mean air temperatures between January 2014 and December 2015 with a root-mean-square percentage error of 4.2%.The correlation between the predicted and the measured monthly mean air temperatures was 0.989. By analyzing the monthly air temperatures recorded at an urban site and a rural site, it was found that the urban heat island effect led to the urban site being on average 0.865?C warmer than the rural site over the past two decades. Besides, the results of correlation analysis showed that the increase in annual mean air temperature was significantly associated with the increase in population, gross domestic product, urban land use, and energy use, with the R~2 values ranging from 0.37 to 0.43.     

小波神经网络模型在高铁路基沉降预测中的应用研究

为了提高变形监测数据预测的精度与可靠性,提高神经网络预测方法的稳定性,尝试将小波分析与BP神经网络相结合的小波神经网络应用于高铁路基处的沉降监测数据处理中。综合小波分析与神经网络算法的优点,建立松散型及紧致型小波神经网络预测分析模型。通过实验数据对比分析,验证了采用紧致型小波神经网络预测模型能够较好地用来处理路基的动态变形监测数据,预测稳定性及预测精度较高。     

结合PCNN的红外与可见光图像多尺度融合

针对红外与可见光图像融合对比度不高,易丢失细节信息等问题,提出了一种非下采样Contourlet变换域内基于特征激励的自适应PCNN红外与可见光图像融合方法。PCNN模型采用平均梯度和赋时矩阵来自适应调节其链接强度和迭代次数等参数。对NSCT多尺度多方向高低频子带系数,分别采用特征激励PCNN,根据点火时间图的区域能量来选择融合系数。实验结果表明,该方法能够有效地融合红外和可见光图像信息,对比度高,细节保持好,无论在视觉效果上还是客观评价指标上,优于常用的图像融合方法。     

基于Markovian切换的时滞回归神经网络Lagrange全局均方指数稳定

对一类激励函数是Lurie型（包括有界和无界激励函数）的具有Markovian切换的时滞回归神经网络的Lagrange全局均方指数稳定性进行了研究,得到了回归神经网络在Markovian切换状态下的Lagrange全局均方指数稳定的充分判据,并通过数值例子验证了所得结论的正确性和有效性.     

利用卷积神经网络的高光谱图像分类

针对高光谱图像分类中对光谱信息利用不足的问题,提出一种基于卷积神经网络在光谱域开展的分类算法。该算法通过构建五层网络结构,逐像素对光谱信息开展分析,将全光谱段集合作为输入,利用神经网络展开代价函数值的计算,实现对光谱特征的提取与分类。实验中采用三组高光谱遥感影像数据进行对比分析,以India Pines数据集为例,提出的基于卷积神经网络的分类方法的分类正确率达到90.16%,比RBF-SVM方法高出2.56%,相比三种传统的深度学习方法高出1%~3%,训练速度也较为理想。实验结果表明,本文所提出的算法充分利用了高光谱图像中逐像素点的光谱域信息,能够有效提高分类正确率。与传统学习算法相比,在较少训练样本的情况下,更能发挥其良好的分类性能。     

基于集合预报系统的日最高和最低气温预报

根据欧洲中心集合预报系统2 m气温预报的集合统计值,提出了BP-SM方法,针对中国512个台站2016年3月的日最高（低）气温做预报分析。将集合预报系统的模式直接输出、BP和BP-SM方法得到的日最高（低）气温进行了比较,结果表明:预报时效越长,BP-SM方法较之BP方法的预报优势也更明显;在1至5 d的预报中,BP-SM方法显著降低了预报绝对误差,误差在2℃以内的准确率大部分在60%以上,部分站点达到90%;正技巧评分均值大多高于30%,在青藏高原东部和南部地区超过了60%。预报正技巧站点次数在绝对误差≤2℃（1℃）范围内有所提高,对日最高气温预报准确率的提高略好于日最低气温;BP-SM方法有效地降低了预报系统偏差,较大预报误差出现次数显著减少。      

Renaming and Rebranding within U.S. and Canadian Geography Departments, 1990–2014

Between 2000 and 2014, more than thirty geography departments adopted revised or new names, with some entirely dropping geography. Although renaming and rebranding efforts are not new to higher education, the rapid pace at which geography department names have changed raises questions about the discipline's identity and health. We examine the renaming trend within geography programs together with intended and unexpected factors as perceived by faculty. Specifically, we look at the renaming and rebranding trend within the context of four pillars offered by Pattison (1964 Pattison, W. D. 1964. The four traditions of geography. Journal of Geography 63 (5): 211–16.[Taylor &; Francis Online], [Web of Science ®] , [Google Scholar]) to define geography's principal academic domains—earth-science, man–land, area/regional studies, and spatial traditions.     

智慧旅游城市旅游竞争力评价

提升旅游竞争力是智慧旅游城市建设的核心目标。在借鉴前人研究的基础上,建立包括5个一级指标、14个二级指标和33个三级指标的智慧旅游城市旅游竞争力评价指标体系,选取北京、南京、武汉、成都、大连、厦门等12个首批国家智慧旅游试点城市,通过模拟仿真运算构建智慧旅游城市旅游竞争力评价BP神经网络模型,并运用该模型对上述智慧旅游城市旅游竞争力进行评价和分析。结果表明：① 旅游科技创新竞争力是影响智慧旅游城市旅游竞争力最关键的一级指标,其他一级指标按重要性排序依次为旅游经济发展竞争力、旅游发展保障竞争力、旅游发展潜力竞争力和旅游环境支撑竞争力;② 中国智慧旅游城市旅游竞争力整体水平不高且极不均衡,根据评价等级和竞争态势分为5类。第一类北京市是中国智慧旅游城市旅游竞争力的标杆,5项一级指标评价值均领先于其他城市,总评价值高达0.887,评价等级AA、竞争态势“优势”;第二类南京市旅游环境支撑竞争力评价值与北京市并列第1,旅游经济发展竞争力、旅游科技创新竞争力和旅游发展保障竞争力评价值均位列第2,总评价值0.536,明显高于除北京市外的其他智慧旅游城市,评价等级BB、竞争态势“较强”;第三类武汉、大连、成都、厦门、镇江、温州、烟台7市代表目前中国智慧旅游城市旅游竞争力的普遍水平,5项一级指标评价值偏低且差距不明显,总评价值在0.3~0.4之间,评价等级B、竞争态势“一般”;第四类福州、洛阳2市绝大多数一级指标评价值较低,总评价值在0.2~0.3之间,评价等级CC、竞争态势“较弱”;第五类黄山市总评价值仅0.176,评价等级C、竞争态势“弱势”;③ 构建的智慧旅游城市旅游竞争力评价BP神经网络模型具有较好的科学性、普适性和可操作性,根据评价值的大小判断智慧旅游城市旅游竞争力的优劣,比照指标体系寻找差距并加以完善,同时对评价指标进行动态监测,将是智慧旅游城市旅游竞争力研究的重要发展方向。