西太平洋马里亚纳海沟附近海山的金相柳珊瑚属(珊瑚虫纲:软珊瑚目:金柳珊瑚科)二新记录种

为探究热带西太平洋海山生物多样性,2016年利用"发现"号遥控无人潜水器(ROV)对西太平洋马里亚纳海沟附近的M2海山进行了采样调查,其中获得了3个深水金相柳珊瑚样本。通过光学显微镜和扫描电镜观察,鉴定其为黑发金相柳珊瑚Metallogorgia melanotrichos(WrightStuder,1889)和长刺金相柳珊瑚Metallogorgia macrospina Kükenthal,1919。这两种金相柳珊瑚均为该区域的新记录种,对其作了详细的形态描述,并依据特征性状对该属所有物种作了分类检索。     

浮力装置触发深海管道水平向整体屈曲效果的数值模拟研究

海洋油气资源的运输主要通过海底管道进行,管道在工作时受到较大的温度荷载,会产生整体屈曲变形。深海管道设计中常采用人为装置触发一定程度的水平向整体屈曲变形,来释放轴向的温度应力,浮力装置是常用的触发方式之一。本文通过数值模拟研究,分析了不同浮力大小和不同浮力施加范围下,管道水平向整体屈曲的临界屈曲力,得出临界屈曲力随浮力大小和施加范围变化的情况;并研究了不同土体阻力下,浮力装置触发整体屈曲的效果。研究表明,水平向土体阻力较大时,浮力装置触发水平向整体屈曲的效果较好。浮力装置的触发效果对轴向土体阻力不敏感。     

Bathymetry,substrate and fishing areas of Southeast Atlantic high-seas seamounts

Most of the Southeast Atlantic Ocean is abyssal, and global bathymetries suggest that only ～3.2% of the areas beyond national jurisdiction (ABNJ; also known as the high seas, as defined in the United Nations Convention on the Law of the Sea [UNCLOS]) are shallower than 2 500 m. This study mapped bathymetry and characterised substrates in selected seamount summit areas, including several that have been or may become fishing areas. The southernmost location, the Schmitt-Ott Seamount, has exposed volcanic bedrock with surrounding flats covered by thin biogenic sediments and/or coral rubble that appears ancient. At Wüst, Vema, Valdivia and Ewing seamounts the basaltic base appears to be overlain by coral caps and other coral substrates (sheets, rubble). Adjacent summit plains have biogenic sediments of varying thickness. Vema has a flat, roughly circular summit, <100 m deep, with the shallowest point being a 22-m-deep summit knoll; the upper slopes have ancient coral framework, but the summit has a mixture of coralline and volcanic rock and coarse sediments, including extensive areas with coralline algae and kelp forests. Valdivia Bank is a 230-m-deep, flat, rocky area (～11 × 5 km), protruding steeply from the extensive multi-summit Valdivia subarea of the Walvis Ridge. The distribution of past fisheries in the Convention Area of the South East Atlantic Fisheries Organisation (SEAFO) was considered in relation to the new information on bathymetry and substrate.     

高分辨率光学遥感场景分类的深度度量学习方法

针对高分辨率光学遥感影像场景具有同类型内部差异大、不同类型间相似度高导致部分场景识别困难的问题,本文提出了一种深度度量学习方法。首先在深度学习模型的特征输出层上为每类预设聚类中心,其次基于欧氏距离方法设计均值中心度量损失项,最后联合交叉熵损失项以及权重与偏置正则项构成模型的损失函数。该方法的目标是在特征空间上使同类型特征聚集并扩大类型间的距离以提高分类准确率。试验结果表明,本文方法有效地提升了分类准确率。在RSSCN7、UC Merced和NWPU-RESISC45数据集上,与现有方法相比,分类准确率分别提高了1.46%、1.09%和2.51%。     

结合深度学习和图割法的遥感影像建筑物检测

提出一种基于卷积神经网络和图割法的自动提取高分影像建筑物的方法。首先,通过卷积神经网络定位与检测建筑物的位置,逐一提取单个建筑物轮廓,利用检测结果分别建立建筑物和非建筑物的高斯混合模型（GMM）,然后结合最大流最小割的图像分割方式实现全局优化,完成建筑物初步提取,最后用形态学进行优化。通过试验证明了该方法的可行性。     

检索词优化与空间自适应的深网POI获取方法研究

针对检索词库构建困难、数据请求量受限制等相关问题,该文提出一种基于独立覆盖率排序和空间自适应剖分的深网POI信息搜索方法,通过候选检索词初步构建、贪婪式探测搜索、检索词优化降维、空间自适应剖分爬行等主要流程,利用逐步逼近POI搜索的最大覆盖最优解,实现对深网POI信息的全量获取,大幅度提高深网POI数据的召回率与采集效率,该方法对于丰富地理信息资源、提升空间信息服务与内容管理能力具有重要意义。     

Semantic segmentation of high spatial resolution images with deep neural networks

Availability of reliable delineation of urban lands is fundamental to applications such as infrastructure management and urban planning. An accurate semantic segmentation approach can assign each pixel of remotely sensed imagery a reliable ground object class. In this paper, we propose an end-to-end deep learning architecture to perform the pixel-level understanding of high spatial resolution remote sensing images. Both local and global contextual information are considered. The local contexts are learned by the deep residual net, and the multi-scale global contexts are extracted by a pyramid pooling module. These contextual features are concatenated to predict labels for each pixel. In addition, multiple additional losses are proposed to enhance our deep learning network to optimize multi-level features from different resolution images simultaneously. Two public datasets, including Vaihingen and Potsdam datasets, are used to assess the performance of the proposed deep neural network. Comparison with the results from the published state-of-the-art algorithms demonstrates the effectiveness of our approach.     

深度学习GoogleNet模型支持下的中分辨率遥感影像自动分类

提出了一种基于深度学习技术的遥感分类方法,它能有效解决中分辨率影像在分类过程中出现的像元混分问题。研究选用2016年5月12日武汉市Landsat 7 ETM+遥感影像,基于GoogleNet模型中的Inception V3网络结构,借助迁移学习方法,构建出遥感分类模型,实现了对武汉市主城区4类典型地物（不透水层、植被、水体和其他用地）的自动分类提取,并将分类结果与传统最大似然分类（ML）结果进行了对比分析。研究表明：基于深度学习方法的遥感影像总体分类精度高达88.33%,Kappa系数为0.834 2,明显优于传统ML方法总体分类精度83%和Kappa系数0.755 0,而且有效抑制了地物在分类过程中出现的像元混分现象。     

基于支持向量机的GNSS时间序列预测

在深度学习的理论框架下,针对预测全球卫星导航系统(GNSS)时间序列,传统的经验风险最小化预测模型误差大精度低,泛化性能差且对历史数据的经验依赖大的问题.提出一种采用结构风险最小化原则的基于支持向量机(SVM)的时间序列预测模型.通过和多层的BP神经网络预测模型预测效果比较,结果证明SVM预测模型拥有更好的时间序列预测效果.     

中国大陆强震孕育深处的变形模式

地震预测面临着地球内部的"不可入性"和大地震"非频发性"的问题,是极具挑战性且尚待解决的世界性科学难题之一(陈运泰,2009) .地震活动是断裂活动的直接体现,精定位的震源深度客观反映了断裂深部发生地震滑动变形的起始位置.断裂深部发生的慢滑动和重复地震等对揭示断裂深部行为具有重要的指示意义(Harris,2017).在...