A Bayesian approach to the detection of gross errors based on posterior probability

Existing methods for gross error detection, based on the mean shift model or the variance inflation model, have hardly considered or taken advantage of the potential prior information on the unknown parameters. This paper puts forward a Bayesian approach for gross error detection when prior information on the unknown parameters is available. Firstly, based on the basic principle of Bayesian statistical inference, the Bayesian method—posterior probability method—for the detection of gross errors is established. Secondly, considering either non-informative priors or normal-gamma priors on the unknown parameters, the computational formula of the posterior probability is given for both the mean shift model and the variance inflation model, respectively, under the condition of unequal weight and independent observations. Finally, as an example, a triangulation network is computed and analyzed, which shows that the method given here is feasible.     

Sensitivity equations for hyperbolic conservation law-based flow models

The present paper focuses on the governing equations for the sensitivity of the variables to the parameters in flow models that can be described by one-dimensional scalar, hyperbolic conservation laws. The sensitivity is shown to obey a hyperbolic, scalar conservation law. The sensitivity is a conserved scalar except in the case of discontinuous flow solutions, where an extra, point source term must be added to the equations in order to enforce conservation. The propagation speed of the sensitivity waves being identical to that of the conserved variable in the original conservation law, the system of conservation laws formed by the original hyperbolic equation and the equation satisfied by the sensitivity is linearly degenerate. A consequence on the solution of the Riemann problem is that rarefaction waves for the variable of the original equation result in vacuum regions for the sensitivity. The numerical solution of the hyperbolic conservation law for the sensitivity by finite volume methods requires the implementation of a specific shock detection procedure. A set of necessary conditions is defined for the discretisation of the source term in the sensitivity equation. An application to the one-dimensional kinematic wave equation shows that the proposed numerical technique allows analytical solutions to be reproduced correctly. The computational examples show that first-order numerical schemes do not yield satisfactory numerical solutions in the neighbourhood of moving shocks and that higher-order schemes, such as the MUSCL scheme, should be used for sharp transients.     

Effects of underground cavities on Rayleigh waves—Field and numerical experiments

This paper presents the results of a multi-channel analysis of surface waves (MASW) test conducted over near-surface mine workings, with the objective of delineating the underlying. Numerical studies were carried out to explain and extend the results. The displacement time histories at the surface show amplitude changes in the region over the void, and the Fourier spectra show significant energy concentration on and in the vicinity of the cavity. Different numerical models are constructed and the responses at the surface of the medium and around voids of different sizes and embedment depths are monitored. The numerical results show that part of the incident energy is trapped within the void. The trapped energy bounces back and forth inside the void, until it is attenuated by radiation. The effect of the trapped energy is seen as a concentration of energy over the void region in the frequency domain. The amount of trapped energy is a function of the size and embedment of the void, as well as of the frequency content of the source. The void absorbs part of the energy and radiates it as body waves. Therefore, the recorded responses at the surface carry valuable information about the void. The characteristics of the void can be extracted from the surface responses by analyzing the responses in time, frequency, and spatial domains.     

基于边缘特征匹配的遥感影像变化检测预处理方法

提出了一种基于边缘特征匹配的遥感影像变化检测预处理方法，在进行不同时相遥感影像配准时，一并解求两期影像的辐射校正系数，同时实现两期影像的配准和辐射校正。实验表明，经预处理后，两期影像的色调基本一致，地物没有明显的几何变形。     

一种运用纹理和光谱特征消除投影差影响的建筑物变化检测方法

针对不同时期高分辨率遥感影像变化检测中城区建筑物因投影差差异所产生的误检测现象,提出了一种综合应用光谱和纹理特征的建筑物变化检测方法。以变化和未发生变化地物影像的散度作为可分性依据,首先对光谱差分影像在混合高斯密度模型下建模,并采用马尔可夫最小错误概率准则提取初始变化区域,往往含有错判的建筑物。然后将误判建筑物影像类和真实变化影像类构成训练集,通过引入多通道Gabor滤波器,提取训练集的纹理差分特征,并采用分类别PCA变换实施纹理差分特征的选择。最后对选择出的纹理差分特征依据高斯混合密度模型建模,并用马尔可夫最小错误概率提取真变化区域,即可去除光谱信息检测所产生的伪变化。试验表明,本文方法能够较好地解决建筑物变化的错判问题,提高了影像变化检测的精度。     

基于正常轮廓更新的自适应异常检测方法

根据网络系统发生正常改变的基本特征,提出了确定网络系统正常改变的“三条件”计算方法,其计算结果可作为更新正常轮廓的依据。对正常轮廓的更新问题进行了深入探讨,提出了自适应异常检测的实现机制。并以网络流量分析为例,验证了在异常检测中应用这一方法的可行性。     

基于多结构元的形态学边缘检测方法研究

针对传统边缘检测算子的缺点,研究了基于数学形态学的边缘检测方法,给出了基于多结构元的抗噪型形态学边缘检测算法的具体流程,进行了实验对比分析,表明该方法具有较好的边缘提取能力和抗噪性。     

西藏地震台网检测能力与定位能力评估

根据西藏自治区地震台网的记录,对现有台站的检测能力和定位能力进行了评估,再把“十五”计划增加的台站加入西藏地震台网,估算了该区新台网的检测能力和定位能力。     

Detection of spheriodal free oscillation excited by Peru 8.2 <Emphasis Type="Italic">M</Emphasis><Subscript>s</Subscript> earthquake with five international superconducting gravimeter data

In this paper, authors obtain the spectral peaks of the earth free oscillation and check all normal modes from ₀S₀ to ₀S₄₈ accurately, with the Fourier analysis and the maximum entropy spectrum method dealing jointly with six groups of the observational residual data from five international superconducting gravimeter stations. By comparing the observational results in this paper with three former groups of observations or models, authors notice an extra discrepancy between two observational ₀S₂ modes excited separately by Peru earthquake and Alaska earthquake, which probably mirrors the anisotropy of the Earth’s inner core. The analysis on the splitting ₁S₂ mode shows that the asymmetric factor of rotationwise spectral splitting is possible to be different from that of anti-rotationwise spectral splitting.     

CHAMP卫星快速科学轨道数据的使用及精度评定

指出了德国GFZ数据中心提供的CHAMP卫星快速科学轨道数据存在的两个问题，即：1)轨道数据文件开始、结束段精度较低；2)轨道数据中存在粗差。针对这些问题，给出了CHAMP卫星快速科学轨道数据的拼接及粗差探测方法。利用德国慕尼黑技术大学提供的精密约化动力法轨道对CHAMP卫星快速轨道数据的精度进行了评估。结果表明：CHAMP卫星快速科学轨道的位置精度在10～20cm左右，速度精度为0．3～0．5mm／s。