基于随机补片和DeepLabV3+的建筑物遥感图像变化检测

为有效解决传统遥感图像变化检测预处理复杂的问题,提出一种基于随机补片和DeepLabV3+的建筑物遥感图像变化检测方法。以ResNet50特征提取网络为基础,创建DeepLabV3+语义分割网络,并在图像和标签中创建大小为224像素×224像素的随机补片作为网络输入,训练建筑物提取网络;修改建筑物提取网络输入层为6通道,通过矩阵运算将两期遥感图像转换为一幅6通道非RGB图像,利用转换后的非RGB图像进行网络训练并验证变化检测精度。实验1利用ENVI5.3软件,采用马氏距离法进行变化检测;实验2采用改进的U-Net网络和随机补片,完成网络训练和精度验证;实验3使用实验2的训练数据和验证数据,采用随机补片和DeepLabV3+网络进行变化检测网络训练及精度验证。实验结果表明,该方法实验1、实验2、实验3建筑物变化检测平均交并比分别为24.43%、83.14%、89.90%,边界轮廓匹配分数分别为61.47%,80.24%、96.51%。     

Division of Gas Accumulation System and Laws Controlling Distribution of Natural Gas in Typical Petroliferous Basins of China

Considering the existing problems of the petroleum system, this paper brings forward the concept of natural gas accumulation system and presents the dividing principles. Then detailed statistics on the accumulation factors of the 32 typical natural gas accumulation systems in China and studies on the laws controlling distribution of gas are collected. The research shows that the petroleum accumulation system is the basic unit controlling petroleum generation, migration and accumulation. Generating intensity, generating amount, accumulating efficiency and migration distance plays an important role in the distribution of natural gas. Through analysis on results of resources evaluation, discovered reserves and residual reserves, potential areas in middle-scaled petroliferous basins in China are forecasted in this paper. Ordos, Sichuan, Tarim and Qaidam basins are found out to be the main basins developing and enriching gas accumulation systems.     

冲、爆、钻复合性钻进工艺在大滚石、孤石桩孔中的应用

在大口径工程施工中，采用以往一般方法对付某些特殊复杂地层效果欠使。本文所介绍的采用钢丝绳冲击钻迸、爆破与硬质合金回转钻进（即冲、爆、钻）相结合复合性钻进工艺，基本解决了在大滚石、孤石桩孔中难以施工的问题。     

A CASE STUDY ON CONCENTRATION AND DECENTRALIZATION：BEHAVIOR AND DYNAMIC MECHANISM OF SPATIAL EVOLUTION IN METROPOLITAN AREA，NANJING，CHINA

In rapid socio-economic development,the process of concentration and dispersal of various elements tends to be more dramatic,tremendously influencing the shaping and transformation of the space in metropolitan area.Survey of spatial concentration and decentralization has thus become a basic method in examining metropolitan spatial evolution.In this research,three elements were selected as the essential indicators of the process:demographic density distribu-tion,employment density distribution and business office location.Performance of these elements in Nanjing City was exam-ined historically.As Nanjing City could be regarded as a representative of metropolitan areas in China,its situation large-ly suggestes the general characteristics in similar areas of China.Hence based on the investigation of Nanjing City,four general implications were highlighted.First ,metropolitan areas in China are in a violent process and shift of spatial concentra-tion and decentralization.Second,from now to at least the near future,concentration will continue to be the central fea-ture.Third,the landscape of metropolitan areas basically exhibits a dual structure character.The gap in environmental and ecological qualities among different districts will continue for a long time.Fourth,Central Business District (CBD) is playing an important role in helping to convert the traditionally single-centered city structure into a polycentric one.     

Demeter观测的电离层非震变化特征研究

利用法国电磁卫星Demeter装载的ICE(电场探针)及IAP(等离子体分析仪)探测数据,分析研究2006年不同季节、不同地方时顶部电离层电场及各等离子参量的全球空间分布特征,结果发现,电场x分量及H+离子浓度、O+离子浓度、He+离子浓度、等离子体温度具有显著的季节变化和地方时变化,而电场y分量及z分量全球分布的季节...     

Joint PP and PS AVO inversion based on Zoeppritz equations

Considering Zoeppritz equations, reflections of PP and PS are only the function of ratios of density and velocity. So the inversion results will be the same if the ratios are the same but values of density, velocities of P-wave and S-wave are different without strict constraint. This paper makes efforts to explore nonlinear simultaneous PP and PS inversion with expectation to reduce the ambiguity of AVO analysis by utilizing the redundancy of multi-component AVO measurements. Accurate estimation of ratio parameters depends on independence of input data. There are only two independent AVO attributes for PP reflectivity (i.e. intercept and gradient) and two for PS reflectivity (i.e. pseudo-intercept and pseudo-gradient or extreme amplitude), respectively. For individual PP and PS inversion, the values of least-squares objective function do not converge around a large neighborhood of chosen true model parameters. Fortunately for joint PP and PS inversion the values of the least-squares objective function show closed contours with single minima. Finally the power function fitting is used to provide a higher precision AVO attributes than traditional polynomial fitting. By using the four independent fitting attributes (two independent attributes for PP and PS respectively), the inversion of four ratio parameters (velocities and densities) would be estimated with less errors than that in traditional method.     

三维倾斜界面P波共反射点道集NMO速度

为了实现三维倾斜界面真正的共反射点叠加,获得更加精确的P波速度模型,进行可靠的AVO/AVA分析,形成与三维多分量PS转换波共转换点道集配套的技术,在常速介质共反射点精确走时的基础上,推导并给出三维倾斜界面P波共反射点道集近似双曲时距关系及NMO速度;认识到近似时距在偏移距与深度比小于2时具有较高的拟合精度;指出P波共反射点道集NMO速度是P波速度、界面倾角、界面倾向和测线方位的函数,且随测线方位变化而呈现椭圆特征.与P波共中心点道集NMO速度不同.P波共反射点道集NMO速度小于介质速度.且随界面倾角增大而减小;椭圆长轴方向为倾斜界面走向方位,短轴方向为倾斜界面倾向方位.     

川滇地区电离层多参量异常监测系统的设计与实现

介绍了川滇地区电离层多参量异常监测系统的设计思路与功能实现,并将该系统应用于2019年6月17日四川长宁MS6.0地震的监测试验.结果表明:该系统实现了全球和中国区域垂直总电子含量VTEC、站点VTEC和F2层临界频率foF2异常变化的实时监测,有助于开展示范性地震电离层异常监测工作,其图形及数据结果可为地震-电离层异...     

Landslide susceptibility mapping using LiDAR and DMC data: a case study in the Three Gorges area, China

The objective of this study is to map landslide susceptibility in Zigui segment of the Yangtze Three Gorges area that is known as one of the most landslide-prone areas in China by using data from light detection and ranging (LiDAR) and digital mapping camera (DMC). The likelihood ratio (LR) and logistic regression model (LRM) were used in this study. The work is divided into three phases. The first phase consists of data processing and analysis. In this phase, LiDAR and DMC data and geological maps were processed, and the landslide-controlling factors were derived such as landslide density, digital elevation model (DEM), slope angle, aspect, lithology, land use and distance from drainage. Among these, the landslide inventories, land use and drainage were constructed with both LiDAR and DMC data; DEM, slope angle and aspect were constructed with LiDAR data; lithology was taken from the 1:250,000 scale geological maps. The second phase is the logistic regression analysis. In this phase, the LR was applied to find the correlation between the landslide locations and the landslide-controlling factors, whereas the LRM was used to predict the occurrence of landslides based on six factors. To calculate the coefficients of LRM, 13,290,553 pixels was used, 29.5 % of the total pixels. The logical regression coefficients of landslide-controlling factors were obtained by logical regression analysis with SPSS 17.0 software. The accuracy of the LRM was 88.8 % on the whole. The third phase is landslide susceptibility mapping and verification. The mapping result was verified using the landslide location data, and 64.4 % landslide pixels distributed in “extremely high” zone and “high” zone; in addition, verification was performed using a success rate curve. The verification result show clearly that landslide susceptibility zones were in close agreement with actual landslide areas in the field. It is also shown that the factors that were applied in this study are appropriate; lithology, elevation and distance from drainage are primary factors for the landslide susceptibility mapping in the area, while slope angle, aspect and land use are secondary.