基于数据增广训练的深度神经网络方法压制地震多次波

地震数据中存在的多次波影响偏移成像,误导地震资料的解释,因此通常视为相干噪声而被去除.为了对多次波进行智能化衰减,本文提出了一种基于数据增广训练的使用深度神经网络的多次波压制方法.设计的深度神经网络包括卷积编码和卷积解码过程,其中卷积编码过程学习全波场数据中的一次波特征,卷积解码过程利用这些特征来重构一次波并压制多次波和随机噪声.在训练阶段,旋转训练集并在输入数据中加入随机噪声构成增广训练数据集来提升神经网络的抗噪稳定性和泛化性,通过迁移学习让深度神经网络具备跨工区压制多次波的能力.简单模型与Sigsbee2B模型三套模拟数据的实例验证了本文方法在一次波重构和多次波压制中的有效性、稳定性和良好泛化性;一套崎岖海底模型地震物理模拟数据的应用实例表明本文方法具有应用于复杂条件下压制地震多次波的能力.

     

一种改进的基于非高斯性最大化的预测反褶积算法

The predictive deconvolution algorithm （PD）, which is based on second-order statistics, assumes that the primaries and the multiples are implicitly orthogonal. However, the seismic data usually do not satisfy this assumption in practice. Since the seismic data （primaries and multiples） have a non-Gaussian distribution, in this paper we present an improved predictive deconvolution algorithm （IPD） by maximizing the non-Gaussianity of the recovered primaries. Applications of the IPD method on synthetic and real seismic datasets show that the proposed method obtains promising results.     

多层螺旋CT肝脏三期增强扫描质量评价的探讨

多层螺旋CT和高压注射器的应用，使全肝动态多期增强扫描成为可能，但具体应用中，要想获得满意的肝脏三期增强效果，技术因素和患者因素也不容忽视。本文从科室影像服务器中提取在线的肝脏三期增强影像资料26例进行主观评价和CT值测量分析，结果表明，所有病例影像显示均无不满意评价，其中门脉期和实质期显示满意率均为100％，但动脉期显示满意率为65．4％，由此看出肝脏三期增强扫描的技术难点在动脉扫描时机的把握上，能否正确设置造影剂注射后的扫描延迟时间至关重要，文中对能够影响扫描延迟时间的因素进行说明并提出相应操作对策。     

Assessment of human health impact from exposure to multiple air pollutants in China based on satellite observations

Assessment of human health impact caused by air pollution is crucial for evaluating environmental hazards. In this paper, concentrations of six air pollutants (PM₁₀, PM_2.5, NO₂, SO₂, O₃, and CO) were first derived from satellite observations, and then the overall human health risks in China caused by multiple air pollutants were assessed using an aggregated health risks index. Unlike traditional approach for human health risks assessment, which relied on the in-situ air pollution measurements, the spatial distribution of aggregated human health risks in China were obtained using satellite observations in this research. It was indicated that the remote sensing data have advantages over in-situ data in accessing human health impact caused by air pollution.     

引入强跟踪滤波器的 IMM 算法在导航定位解算中的应用

针对单一模型对城市机动载体运动状态估计不足的问题,本文在标准交互式多模型算法的基础上引入强跟踪滤波器,并以“当前”模型和“CV”模型作为基础模型进行交互式模型滤波。利用信号源仿真动态数据分析对比了利用单一模型算法、标准交互式多模型算法和引入强跟踪滤波器的交互式多模型算法的定位解算性能。实验结果表明：本文提出的基于强跟踪滤波器的交互式多模型算法相比于“当前”模型滤波和标准IM M 算法径向误差方差分别从19.11和16.78降低到13.24．     

Application of Lithium Isotope Geochemistry to the Study of the Continental Geothermal System

The lithium-rich brine in salt lakes is the main raw material of the world’s lithium products, while the continental geothermal fluids with a high salinity often contain a high concentration of lithium. Continental geothermal system is the focus in the study of geothermal formation mechanism. However, less attention is paid to the system due to the complexity of lithology, and the application of lithium isotopes in this field has not been systematically recognized. The newest application and progress of lithium isotope geochemistry in continental geothermal research in recent years were discussed, the problems in this field were put forward, and future research methods and directions were expected. The study of continental geothermal fluids should attach great importance to the application of Li-B-Sr-U multi-isotopic method, and should also be combined with water-rock reaction experiments under different temperature conditions. Moreover, in the future, the research on continental geothermal system should pay more attention to the various sediment/rock lithium isotopic compositions and their spatio-temporal distribution characteristics in the regional or geothermal field’s scales, mineralogy of reservoir rock, and behaviors of lithium isotopes related to the formation of secondary minerals in the process of water-rock interaction, in order to reveal the complex process of fluid evolution in the geothermal system and provide scientific reference for the exploration, exploitation and utilization of lithium resources in the system.     

西部缺水地区多样化水资源开发模式探讨

最近我国政府提出了西部大开发的战略决策。发展西部地区首要任务是研究解决水资源缺乏问题。40多年来南京大学在地下水探寻和开发、水资源评价和水污染的模拟等方面取得了系统研究成果。特别是运用新构造理论和优势面理论 ,在地下水资源探寻的研究中取得显著成效 ,近几年来作者提出的缺水山区水资源多样化开发模式研究课题 ,在河北顺平严重缺水区取得初步突破。这对于解决西部的水资源匮乏问题 ,促进西部大开发会具有重要的实际意义。     

Mudbank regime off the Kerala coast during monsoon and non-monsoon seasons

Field experiments conducted in the nearshore ocean to understand the dynamics of mudbank off Kerala, south-west coast of India, are highlighted. Real time monitoring of the nearshore ocean off Purakkad, Kerala was accomplished using pressure transducers for nearshore surface wave measurements, and current sensors for nearshore velocity measurements. Comprehensive information on the spatial structure of mudbank was obtained from aerial surveys. Extensive data collected on surface waves and currents in the nearshore ocean, indicate that the infra-gravity (IG) waves (leaky modes and trapped edge wave modes), and far infra-gravity (FIG) waves coupled with strong shoreline reflections and undertow play an important role in the dynamics associated with the mudbanks off Kerala during the monsoon season. During the non-monsoon season evidence for progressive edge waves in the infragravity frequency band, an energetic gravity wave band and a strong undertow with weak reflections was observed.     

Seismic characterization of reservoirs with multiple fracture sets using velocity and attenuation anisotropy data

Knowledge about the spatial distribution of the fracture density and the azimuthal fracture orientation can greatly help in optimizing production from fractured reservoirs. Frequency-dependent seismic velocity and attenuation anisotropy data contain information about the fractures present in the reservoir. In this study, we use the measurements of velocity and attenuation anisotropy data corresponding to different seismic frequencies and azimuths to infer information about the multiple fracture sets present in the reservoir. We consider a reservoir model with two sets of vertical fractures characterized by unknown azimuthal fracture orientations and fracture densities. Frequency-dependent seismic velocity and attenuation anisotropy data is computed using the effective viscoelastic stiffness tensor and solving the Christoffel equation. A Bayesian inversion method is then applied to measurements of velocity and attenuation anisotropy data corresponding to different seismic frequencies and azimuth to estimate the azimuthal fracture orientations and the fracture densities, as well as their uncertainties. Our numerical examples suggest that velocity anisotropy data alone cannot recover the unknown fracture parameters. However, an improved estimation of the unknown fracture parameters can be obtained by joint inversion of velocity and attenuation anisotropy data.