基于BP人工神经网络的水体遥感测深方法研究

利用Landsat7ETM＋遥感图像反射率和实测水深值之间的相关性,建立了动量BP人工神经网络水深反演模型,并对长江口南港河段水深进行了反演.结果表明：具有较强非线性映射能力的动量BP神经网络模型能较好地反演出长江口南港河段的水深分布情况;由于受长江口水体高含沙量的影响,模型对小于5 m的水深值反演精度较高,而对大于10 m的水深值反演精度较低.     

基于MSA特征的遥感图像多目标关联算法

遥感图像中多目标关联存在以下两个问题:一是低时间分辨率观测使得目标状态信息无法准确估计,基于Kalman滤波的多目标关联算法不再适用;二是基于图像特征的目标关联算法又无法处理大场景观测中多个目标关联引起的模糊性.针对上述问题,提出一种基于多尺度自卷积特征匹配和关联代价矩阵最优化的多目标关联算法.实验表明该算法对遥感图像中多目标关联问题具有一定的适用性.     

对我国西南地区河谷深厚覆盖层成因机理的新认识

近年来,在我国水能资源开发过程中,发现各河流现代河床以下普遍堆积厚达数十米甚至上百米的松散堆积物。河谷深厚覆盖层的存在,不仅严重制约了工程坝址的选择,影响相关流域水电资源的开发利用,也给坝工设计带来巨大的困难。由于深厚覆盖层埋藏于现代河床之下,其形成年代一般先于一二级阶地,有悖于河流发育演化的常理,其成因一直令人费解。首次将河谷深切和深厚堆积事件与全球气候变化、海平面升降运动、地壳运动等有机地联系起来,并提出冰期、间冰期全球海平面大幅度升降,是导致河流深切成谷并形成深厚堆积的主要原因的新观点。在此基础上,引入层序地层学原理,从理论上较好地解释了全球气候变化导致海平面和河流侵蚀基准面大幅变化,并产生河谷深切和深厚堆积的原因和过程。最后,进一步将沿河大型古滑坡的孕育和发生与河谷深切事件相联系,提出沿河大型古滑坡是在河谷深切期因前缘临空较好而形成的新观点,从而对沿河古滑坡前缘剪出口高程往往低于现代河床数十米的原因给出了较合理的解释。     

藏南冈底斯带西段麦拉花岗岩锆石SHRIMP定年及地质意义

用SHRIMP测定了冈底斯岩带西段麦拉山口岩体同碰撞黑云母二长花岗岩锆石的U-Pb年龄,黑云母二长花岗岩锆石SHRIMP U-Pb年龄为47.1±1.1Ma。这就为前人提出的雅鲁藏布江地区洋盆闭合、板块开始碰撞的时间为50Ma左右提供了同位素年代学的证据。     

改进的边缘角度直方图在遥感图像检索中的应用

针对基于形状的遥感图像检索中区域分割困难这个问题,本文提出利用边缘角度直方图作为形状特征对遥感图像进行检索;针对遥感影像分辨率和颜色反差对检索性能的影响,本文提出采用多尺度分析的思想,提取图像在多个尺度上的边缘角度直方图特征进行检索。通过单一尺度边缘角度直方图特征和多尺度边缘角度直方图特征检索的对比实验,揭示了本文所提出的改进方案使遥感图像的检索精度提高了约31个百分点。     

信息时代地图集编制特征综述

随着计算机技术的发展,地图集的编制也发生了很大的变化.本文主要从地图集的表示内容、成图方法、资料情况、地图符号、地图整饰5个方面来阐述地图集编制的新特征.对近几年来国家、各省市地图集的编制情况作以概括.     

Developing a SPASE Query Language

The advent of the Virtual Observatory has begun an evolution in the space physics data environment. A number of nascent and discipline specific Virtual Observatories have started to emerge with an emphasis on data search and retrieval. As this new data environment takes shape an emphasis will be placed on interdisciplinary communication in attempts to address large scale and global problems. To this end we formulate the development of a query language to facilitate Virtual Observatory to Virtual Observatory communication. Furthermore, we outline the goals of such a language, how it would work and how existing community efforts can be leveraged to speed the development of this query language.

Lermontovskoe Tungsten Skarn Deposit: the Oldest Mineralization in the Sikhote-Alin Orogen, Far East Russia

Abstract. Lermontovskoe tungsten skarn deposit in central Sikhote-Alin is concluded to have formed at 132 Ma in the Early Cretaceous, based on K-Ar age data for muscovite concentrates from high-grade scheelite ore and greisenized granite. Late Paleozoic limestone in Jurassic - early Early Cretaceous accretionary complexes was replaced during hydrothermal activity related to the Lermontovskoe granodiorite stock of reduced type. The ores, characterized by Mo-poor scheelite and Fe^3+- poor mineral assemblages, indicate that this deposit is a reduced-type tungsten skarn (Sato, 1980, 1982), in accordance with the reduced nature of the granodiorite stock.
The Lermontovskoe deposit, the oldest mineralization so far known in the Sikhote-Alin orogen, formed in the initial stage of Early Cretaceous felsic magmatism. The magmatism began shortly after the accretionary tectonics ceased, suggesting an abrupt change of subduction system. Style of the Early Cretaceous magmatism and mineralization is significantly different between central Sikhote-Alin and Northeast Japan; reduced-type and oxidized-type, respectively. The different styles may reflect different tectonic environments; compressional and extensional, respectively. These two areas, which were closer together before the opening of the Japan Sea in the Miocene, may have been juxtaposed under a transpressional tectonic regime after the magmatism.     

Cretaceous isochron ages from K–Ar and Ar / Ar dating of eclogitic rocks in the Tso Morari Complex, western Himalaya, India

The Tso Morari Complex, which is thought to be originally the margin of the Indian continent, is composed of pelitic gneisses and schists including mafic rock lenses (eclogites and basic schists). Eclogites studied here have the mineral assemblage Grt + Omp + Ca-Amp + Zo + Phn + Pg + Qtz + Rt. They also have coesite pseudomorph in garnet and quartz rods in omphacite, suggesting a record of ultrahigh-pressure metamorphism. They occur only in the cores of meter-scale mafic rock lenses intercalated with the pelitic schists. Small mafic lenses and the rim parts of large lenses have been strongly deformed to form the foliation parallel to that of the pelitic schists and show the mineral assemblages of upper greenschist to amphibolite facies metamorphism. The garnet–omphacite thermometry and the univariant reaction relations for jadeite formation give 13–21 kbar at 600 °C and 16–18 kbar at 750 °C for the eclogite formation using the jadeite content of clinopyroxene (X_Jd = 0.48).

Phengites in pelitic schists show variable Si / Al and Na / K ratios among grains as well as within single grains, and give K–Ar ages of 50–87 Ma. The pelitic schist with paragonite and phengite yielded K–Ar ages of 83.5 Ma (K = 4.9 wt.%) for paragonite–phengite mixture and 85.3 Ma (K = 7.8 wt.%) for phengite and an isochron age of 91 ± 13 Ma from the two dataset. The eclogite gives a plateau age of 132 Ma in Ar/Ar step-heating analyses using single phengite grain and an inverse isochron age of 130 ± 39 Ma with an initial ⁴⁰Ar / ³⁶Ar ratio of 434 ± 90 in Ar/Ar spot analyses of phengites and paragonites. The Cretaceous isochron ages are interpreted to represent the timing of early stage of exhumation of the eclogitic rocks assuming revised high closure temperature (500 °C) for phengite K–Ar system. The phengites in pelitic schists have experienced retrograde reaction which modified their chemistry during intense deformation associated with the exhumation of these rocks with the release of significant radiogenic ⁴⁰Ar from the crystals. The argon release took place in the schists that experienced the retrogression to upper greenschist facies metamorphisms from the eclogite facies conditions.     

长江口水域多光谱遥感水深反演模型研究

利用Landsat-7 ETM 遥感影像反射率和实测水深值之间的相关性可以探测水深。该文介绍单波段、双波段比值和多波段3种线性回归模型以及动量BP人工神经网络水深反演模型。选择长江口北港河道上段作为研究区,利用上述模型,分两种情况进行水深反演:一是以河道全部历史样本建模;二是将河道按自然水深划分为浅水区和深水区分别建模。结果表明:神经网络模型预测精度高于线性回归模型;水深分区后线性回归和神经网络模型预测误差均有所减小。