网平差在TAI计算中的应用

截至目前，用于TAI计算的GPS共视时间传递链路采用固定链路，以CRL、NIST和OP作为3个主要中心站，形成一个连接全球约50个时间实验室的时间比对链路，采用固定链路是为TAI计算方便，而人为规定的，具有简单、工作量较小等特点，但是整个TAI GPS共视时间比对对这3个中心站的依赖度过高，一旦这3个中心站不能正常观测，将影响TAI的计算，就GPS共视法这种技术本身而言，可支持任意链路的时间比对，并可形成网络化的比对结果，在TAI计算时，对GPS共视比对结果，采用网平差处理技术，可提高TAI计算的可靠性，避免因中心站观测故障而导致一个区无法参加比对的情况发生。     

Difference of urban development in China from the perspective of passenger transport around Spring Festival

Because cities have different attractions for laborers, people migrate from city to city, which may widen the gap between cities. Analyzing the fluxes and directions of migrant flows can help us clarify the regional difference of urban development. Using migration data inferred from passenger travels during 2016 Spring Festival from Tencent Location Big Data, this paper analyzed the unbalanced migration between cities and the spatial difference of urban development. Network analysis methods are employed to evaluate interactions among cities. A community detection method identifies 19 city communities, and the directions of migrant flows in the communities are explored. The PageRank algorithm is employed to evaluate the importance of cities on the migration network and divide the cities into 5 grades, and then the hierarchical structure of the migrant network is illustrated and analyzed. Indices based on migrant populations indicate that the most attractive cities for laborers are along the east coast and that cities in the central region export a significant number of laborers. PageRank and attractiveness values are compare with socio-economic data, and the results indicate that both PageRank and attractiveness are positively correlated with the economic and development level of cities, while PageRank works better. It suggests that Spring Festival travel data in China can be used as migration data, however, it should be facilitated with network methods to disclose the relationship between Spring Festival travel and urban development.     

基于小波变换和GRNN神经网络的黑河出山径流模型

对黑河山区流域月降水量和气温做Harr小波变换,并作为GRNN神经网络的输入,对黑河出山径流进行模拟和预测验证,效果较好。应用全球变化成果,在不同的气候情景下,对黑河出山径流进行预测。结果表明,黑河出山径流在未来一段时间内,径流量会有一定程度的增加,最终会减少。但模型对气温反应不敏感。去除气温重构的细节系数后,气温也成为一个敏感因素,但径流量却随气温的增加而增加。可推断,引进Haar小波变换的GRNN神经网络模型可应用于径流量对气温不敏感的流域。     

基于微博数据的西藏那曲比如县地震舆情分析

震后对网络舆情信息的监控与分析,对于相关部门开展震灾应急救援、掌握救灾动态、稳定民众情绪具有重要意义。本研究基于新浪微博数据,运用网络爬虫技术,获取西藏自治区那曲市比如县6.1级地震震后24小时及震后一周的相关微博及评论;利用Python中文分词组件“jieba”和ROST CM6软件,对数据进行分类、去重、分词等处理,得到结构化的分级、分类数据,并制成震后微博时间序列图、地理分布图、舆情热词词频表、情绪极性统计图等,实现微博舆情数据的可视化。研究结果表明,本次地震事件的微博舆情整体呈现积极情绪,微博活跃程度与当地经济发展程度密切相关。在本次地震舆情传播中,政府部门的舆情引导起到至关重要的作用,舆情传播中的防灾视频传播具有明确的正向引导作用。本研究对于中国西部欠发达少数民族地区的地震舆情分析及引导工作具有借鉴意义。     

滹沱河上游水系变迁问题的讨论

本文着重讨论了滹沱河上游水系变迁的三个争议较大的问题。研究表明：（１）阳武河袭夺问题是不存在的，其干、支流分别沿ＮＷ和ＮＥ两组性质不同的构造发育而成；（２）奇村宽谷在更新世期间曾长期作为滹沱河汊道与界河铺峡谷并存，后因西部山地河流的超补偿堆积而最终断流。（３）滹沱河是一条先成河，自上新世末形成以来，其总的流路并未改变。石岭关风口早于盆地而形成，可能是一条更古老的河流所造成。     

云南建水锰矿地质环境破坏评价及其修复技术

矿山开采对矿区地质环境造成严重的破坏.以建水锰矿地质环境修复为例,引入香根草生物修复技术,形成根系网络,提高矿区植被覆盖率,有效改良了土壤,水土保持效果显著.     

Road detection from high-resolution satellite images using artificial neural networks

This article treats the possibility of using artificial neural networks for road detection from high-resolution satellite images on a part of RGB Ikonos and Quick-Bird images from Kish Island and Bushehr Harbor, respectively. Attempts are also made to verify the impacts of different input parameters on network's ability to find out optimum input vector for the problem. A variety of network structures with different iteration times are used to determine the best network structure and termination condition in training stage.It was found that when the input parameters are made up of spectral information and distances of pixels to road mean vector in a 3 × 3 window, the network's ability in both road and background detection can be improved in comparison with simple networks that simply use spectral information of a single pixel in their input vector.     

The evolution of tidal creek networks,mary river,northern Australia

Tidal creek networks have in 50 years extended over 30 km inland across the coastal plains of the Mary River in northern Australia, invading freshwater wetlands and destroying the associated vegetation. The networks have grown at an exponential rate through a combination of main channel extension and tributary development, with concomitant widening of the creeks. A large tidal range, very small elevational differences over the plains, and the availability of preexisting channel lines (notably in the form of palaeochannels) have been major factors contributing to the rapid rate of expansion. Close parallels exist between these networks and terrestrial networks as regards modes of growth and planimetric properties. A channel is initiated when the diffuse flow of a seepage zone becomes concentrated through localized scour. Subsequent development is characterized by the rapid extension of long first-order channels, with most tributary addition occurring later. Model tests suggest that branching was more likely on exterior links in the early stages but that exterior and interior link branching became more equally likely through time. Although the headward limits of the main creeks seem to have been reached, tributary infilling will continue to progress upstream. Only in the most downstream parts is a stable drainage density being approached. The networks not only satisfy the laws of drainage network composition and the basic postulates of the random model but also depart from topologic randomness in similar ways to terrestrial networks. Both topologic and length properties have changed during evolution but largely at the link rather than network scale. The close correspondence with terrestrial networks may be due to the low relief and the relatively unconstrained nature of growth in which availability of space was the main determining factor.     

重庆地震台网双链路备份传输方案及成效

简要对比DDN、SDH光纤、卫星和无线CDMA等几种通讯方式的优缺点,介绍重庆区域测震台网所使用的传输方案,即以SDH光纤传输为主、无线CDMA为辅的双链路备份传输方案。重点介绍该传输方案的原理和实施方法,详细讲述台站及台网中心的路由配置和相关说明。通过在重庆市区域测震台网的运行,结果发现,传输效果较理想,测震台网台站运行率均达到98%以上,证明了该方案应用的可行性。     

地面与半航空瞬变电磁法三维联合反演

瞬变电磁一维反演方法存在耗时长、参数难以调控、过于依赖初始模型等缺陷。为此,提出一种基于Attention机制的卷积−双向长短时记忆神经网络(AC-BiLSTM)瞬变电磁实时反演方法,充分利用时间差,在非观测时间进行模型训练,在观测时间对当下采集数据进行实时反演。整个过程中,以实测数据加入一定比例正演数据作为数据集,以监督学习方式将采样时间−视电阻率作为输入特征,以测井约束的Occam反演结果作为学习目标,基于卷积神经网络和长短期记忆神经网络搭建编码器−解码器模型,并针对数据特性,在解码器部分加入Attention机制对隐藏层输出数据进行重点提取,最后经全连接层获得深度−电阻率数据。研究结果表明：AC-BiLSTM算法能充分挖掘数据时空特性,快速获得符合地层电性特征的电阻率成像结果,在瞬变电磁正演数据集上的预测值与正演模型拟合优度达0.898、均方根误差18.44、平均相对误差0.065,与单一长短期记忆神经网络及Occam方法相比,拟合优度分别提高了0.086、0.176,均方根误差分别减小了2.97、9.32,平均相对误差分别减小了0.012、0.068。通过对V8电法工作站实测瞬变电磁数据的AC-BiLSTM反演,快速实现了研究区地层的精准分层、圈定了煤矿采空区分布范围,获得成果与真实情况一致。研究成果突破了传统反演方法局限性,提高了瞬变电磁数据解释精度及效率。