基于最小二乘配置法和奇异值截断法向下延拓航空重力数据的仿真研究

基于最小二乘配置法建立的航空重力异常与其相应地面重力异常的协方差函数,由于协方差矩阵严重病态,观测误差被放大从而影响向下延拓结果的稳定性和精度;引入基于 GCV 法选择正则化参数的奇异值截断法后,能抑制协方差矩阵的病态性对观测噪声的放大影响。 基于 EGM2008 重力场模型分别计算陆地和海域的航空重力异常作为基础数据,用本文方法向下延拓获取相应的地面重力异常,实验结果验证该方法的有效性。     

融合多源重力数据的Tikhonov正则化配置法

针对最小二乘配置法在融合处理多源重力数据过程中可能出现的病态性问题,特别引入Tikhonov正则化方法,对配置法计算模型进行正则化改造,建立了相应的正则化配置模型。使用EGM2008位模型模拟产生航空重力和海面船测重力数据进行了融合处理仿真实验,实际验证了正则化处理方法的有效性。     

基于奇异值分解的多通道带乘性噪声系统的最优滤波算法

提出 1种基于奇异值分解 (SVD)的多通道带乘性噪声系统的最优滤波方法。该方法基于多通道带乘性噪声系统的最优滤波理论[1] ,利用奇异值分解作为工具 ,将原算法中的协方差矩阵P进行奇异值分解 ,可以在一定程度上避免在递推过程中 ,由于计算误差和舍入误差的积累而引起的协方差矩阵P失去对称性 ,因而导致算法失效的问题。在保证算法在线性最小方差意义下为最优的同时 ,具有很好的数值稳定性和鲁棒性。仿真中对改进后算法和原算法估计效果做了对比 ,仿真结果证明了本文方法的有效性。     

多站激光扫描点云配准算法研究

随着Li DAR的迅速发展,三维激光扫描技术已经被广泛运用于各个领域。点云数据配准过程中,传统的配准算法比较依赖特征点匹配精度,粗差点的存在会较大程度地影响配准精度和配准效率。通过对罗德里格矩阵、整体最小二乘原理的分析,提出了一种基于整体最小二乘的罗德里格矩阵算法。在该算法实现的过程中,能够考虑到系数矩阵误差,降低算法实现过程中特征点坐标误差对参数求解的影响。实验结果表明,本文算法比参数算法和严密的罗德里格算法精度更高、稳定性更强,在初始对应点坐标误差较大的情况下仍能获得精度较为稳定的变换参数。     

深海热液温度场测量重建算法比较

介绍利用声学方法测量深海热液温度场的基本原理。对宽单峰、窄单峰和双峰温度场模型,采用最小二乘法和傅立叶正则化方法分别进行重建仿真,给出重建温度场的绝对误差、相对误差和均方根误差。比较最小二乘法与正则化方法对几种温度场模型的重建结果。仿真结果表明:最小二乘法对宽单峰模型温度场具有较高的重建精度,而正则化方法对高温区相对测量域偏小的窄单峰模型及双峰模型温度场比最小二乘法有更好的还原结果。     

最小二乘配置中两种局部协方差函数的比较

在分析Moritz和Tscherning/Rapp两种协方差函数模型技术特点基础上,利用EGM2008模型计算的局部重力异常数据,分别采用基于Moritz和Tscherning/Rapp协方差函数的最小二乘配置法计算了大地水准面高,以分析两种协方差函数模型的适用性和可靠性。实验结果表明:Tscherning/Rapp协方差函数模型要优于Moritz协方差函数。     

大地测量中法方程病态性问题的初探

在大地测量中,法方程病态问题大量存在。由于病态性的存在,致使按最小二乘法解算参数估值精度变差发生扭曲,最终使所得结果不能使用。重点研究奇异值分解法(SVD)和误差转移法解决病态问题的思想、途径、对关键问题的处理、适用范围等。最后通过GPS实测数据来验证这两种方法的有效性,其解算的结果表明:这两种方法能解决法方程病态性问题,且算法简便、易懂,有较强的应用价值。     

功率谱密度函数的构建及在重力场元计算中的应用

应用输入输出法逼近局部重力场的关键在于构建有效反映局部重力场特征的功率谱密度函数模型、利用"间接法",基于Moritz协方差模型、Tscherning/Rapp协方差模型和Moritz两分量模型分别构建对应的功率谱密度函数模型,并采用非线性最小二乘法在EGM2008模型的基础上拟合了模型系数。拟合结果表明:在2倍相关长度的相关距离内,Moritz两分量模型求得的理论协方差值与经验协方差值的符合度最好。设计了以EGM2008模型重力异常为基础数据,计算大地水准面高的仿真实验。结果表明:三种模型中,Moritz两分量模型的计算结果精度最高,同传统Stokes方法的计算结果精度相当,与拟合的理论协方差曲线图所反映的结果一致。     

利用卫星测高垂线偏差计算其他重力场元的特性分析

利用卫星测高技术确定海洋重力场,垂线偏差数据作为导出观测量在实际工作中被普遍采用。利用物理大地测量边值问题的定义以及扰动位在球面边界条件下的解,给出了由垂线偏差计算大地水准面高、重力异常和扰动重力的公式。分析了不同积分计算公式在重力场阶谱表达形式下对垂线偏差误差的抑制作用,也分析了不同积分核函数的变化特性,得出基本结论:在利用卫星测高数据求解海洋重力场时,当以格网化海面垂线偏差数据计算重力场参数时,求解的大地水准面高的有效性和稳定性优于重力异常和扰动重力。     

不规则重磁测线网误差模型的约束最小二乘平差

传统实用的海洋重力、磁力测线网平差方法主要为规则测线网设计,并只就线偏差进行调整,用于处理航次数据之间的拼接时效果不理想。为了解决重磁测线数据拼接和建库工作中碰到的误差难以调平的问题,设计、建立了不规则海洋重磁测线网的任意复杂误差模型的最小二乘平差方法,将线偏差纳入到任意复杂(如非线性和周期性)的误差模型中,归纳出最小二乘条件方程系数矩阵的直接构制法。针对可能出现的解不稳定,放弃不易实用的权矩阵,引入两种解决方法:先验确定部分测线误差分布的固定测线法和视误差大小实施系数矩阵对角线微扰动的泛约束法。     

Local Oceanic Vertical Deflection Determination with Gravity Data Along a Profile

Deflections of the vertical (DOVs) over oceans cannot be directly measured, which restricts their applications. A local covariance function of anomalous potential is put forward in this paper in conjunction with the least-squares collocation (LSC) method to compute the oceanic DOVs utilization of oceanic gravity data along a profile. The covariance functions of gravity field quantities have been derived directly as functions of x, y and z without the need to introduce coordinate transformations corresponding to along- or cross-profile components. In the proposed methodology, gravity data along a profile were used to calculate the residual gravity anomaly using the remove-compute-restore technique. The residual gravity anomaly was used to calculate the parameters of the proposed covariance function of the local anomalous gravity field, which was used in the LSC to compute the residual DOVs along the profile. The residual DOVs added model DOVs to recover the DOVs along the profile. The results of a simulation experiment prove that the proposed methodology is feasible and effective.     

Combined satellite altimetry and shipborne gravimetry data processing

The recovery of quantities related to the gravity field (i.e., geoid heights and gravity anomalies) is carried out in a test area of the central Mediterranean Sea using 5' × 5' marine gravity data and satellite altimeter data from the Geodetic Mission (GM) of ERS‐J. The optimal combination of the two heterogeneous data sources is performed using (1) the space‐domain least‐squares collocation (LSC) method, and (2) the frequency‐domain input‐output system theory (IOST). The results derived by these methods agree at the level of 2 cm in terms of standard deviation in the case of the geoid height prediction. The gravity anomaly prediction results by the same methods vary between 2.18 and 2.54 mGal in terms of standard deviation. In all cases, the spectral techniques have a much higher computational efficiency than the collocation procedure. In order to investigate the importance of satellite altimetry for gravity field modeling, a pure gravimetric geoid solution, carried out in a previous study for our lest area by the fast collocation approach (FCOL), is used in comparison with the combined geoid models. The combined solutions give more accurate results, at the level of about 15 cm in terms of standard deviation, than the gravimetric geoid solution, when the geoid heights derived by each method are compared with TOPEX altimeter sea surface heights (SSHs). Moreover, nonisotropic power spectral density functions (PSDs) can be easily used by IOST, while LSC requires isotropic covariance functions. The results show that higher prediction accuracies are always obtained when using a priori nonisotropic information instead of isotropic information.     

Time-Dependent Spatial Wavenumber Spectra Derived from Merged Sea Surface Temperature Data

Monthly wavenumber spectra of sea surface temperatures (SST) have been estimated in two regions near the Kuroshio, in the recirculation and the Kuroshio Extension regions, using the merged SST product to determine the statistical parameter (spatial decorrelation scale) required for optimal interpolation of a high-resolution SST dataset. The two-dimensional wavelet transform was used for analysis. Estimates were made of daily mean and daily minimum SSTs. These do not significantly differ, which suggests that the same covariance matrix can be used for the daily mean and minimum in the merging procedure. The seasonality of wavenumber spectra is significant. There are also large differences between those in the recirculation region and in the Kuroshio Extension region. Therefore, it is recommended that the covariance matrix in the merging process for high-resolution SST dataset be defined as a function of time and space. Improvements of the merging methodology are discussed. This revised version was published online in July 2006 with corrections to the Cover Date.     

Comparison of Satellite Altimetric Gravity and Global Geopotential Models with Shipborne Gravity in the Red Sea

The determination of high-resolution geoid for marine regions requires the integration of gravity data provided by different sources, e.g. global geopotential models, satellite altimetry, and shipborne gravimetric observations. Shipborne gravity data, acquired over a long time, comprises the short-wavelengths gravitation signal. This paper aims to produce a consistent gravity field over the Red Sea region to be used for geoid modelling. Both, the leave-one-out cross-validation and Kriging prediction techniques were chosen to ensure that the observed shipborne gravity data are consistent as well as free of gross-errors. A confidence level equivalent to 95.4% was decided to filter the observed shipborne data, while the cross-validation algorithm was repeatedly applied until the standard deviation of the residuals between the observed and estimated values are less than 1.5 mGal, which led to the elimination of about 17.7% of the shipborne gravity dataset. A comparison between the shipborne gravity data with DTU13 and SSv23.1 satellite altimetry-derived gravity models is done and reported. The corresponding results revealed that altimetry models almost have identical data content when compared one another, where the DTU13 gave better results with a mean and standard deviation of ?2.40 and 8.71 mGal, respectively. A statistical comparison has been made between different global geopotential models (GGMs) and shipborne gravity data. The Spectral Enhancement Method was applied to overcome the existing spectral gap between the GGMs and shipborne gravity data. EGM2008 manifested the best results with differences characterised with a mean of 1.35 mGal and a standard deviation of 11.11 mGal. Finally, the least-squares collocation (LSC) was implemented to combine the shipborne gravity data with DTU13 in order to create a unique and consistent gravity field over the Red Sea with no data voids. The combined data were independently tested using a total number of 95 randomly chosen shipborne gravity stations. The comparison between the extracted shipborne gravity data and DTU13 altimetry anomalies before and after applying the LSC revealed that a significant improvement is procurable from the combined dataset, in which the mean and standard deviation of the differences dropped from ?3.60 and 9.31 mGal to ?0.39 and 2.04 mGal, respectively.     

噪声相关时多通道带乘性噪声系统最优滤波

针对多通道乘性噪声系统问题的实际需要 ,推广 Rajasekaran滤波算法 ,利用线性最小方差的概念和投影公式 ,导出含有乘性噪声统计特性参数阵特殊乘法运算的新息协方差矩阵 ,并利用这个中间矩阵 ,在观测为多通道 ,且各个通道的乘性噪声不同 ,以及系统的动态噪声和观测噪声同时刻相关的情况下 ,导出状态递推滤波算法 ,该算法在线性最小方差意义下是最优的。并对该算法进行仿真研究 ,仿真结果表明了该算法的有效性     

解析四维集合变分参数优化方法研究

传统的四维变分数据同化方法在同化观测资料的同时可以对数值模式参数进行优化,然而传统的四维变分方法需要针对不同的数值模式编写特有的伴随模式,因此算法的可移植性差,同时计算时耗费大量资源。本文提出了一种新的基于解析四维集合变分的参数优化方法,该方法以迭代搜索得到的模式参数为基准展开扰动并构建样本集合,由此显式地计算协方差矩阵,并得到代价函数极小值的解析解,从而避免了伴随模式的使用。基于Lorenz-63模型对该方法进行单参数和多参数数值试验和优化效果检验,并在不同的同化时间窗口长度和观测采样间隔情况下,采用传统四维变分方法与之进行对比,结果显示,新方法表现出与传统四维变分相同的优化性能,都能有效收敛到真值,而新方法不需要计算伴随模式,可移植性好。本文还测试了不同的集合成员个数和模式参数真值的情况下新方法的同化效果,结果表明,新方法对集合样本个数及模型参数真值不敏感,采用较少的集合样本即可完成数据同化。     

Preconditioned Optimizing Utility for Large-dimensional analyses (POpULar)

I present the derivation of the Preconditioned Optimizing Utility for Large-dimensional analyses (POpULar), which is developed for adopting a non-diagonal background error covariance matrix in nonlinear variational analyses (i.e., analyses employing a non-quadratic cost function). POpULar is based on the idea of a linear preconditioned conjugate gradient method widely adopted in ocean data assimilation systems. POpULar uses the background error covariance matrix as a preconditioner without any decomposition of the matrix. This preconditioning accelerates the convergence. Moreover, the inverse of the matrix is not required. POpULar therefore allows us easily to handle the correlations among deviations of control variables (i.e., the variables which will be analyzed) from their background in nonlinear problems. In order to demonstrate the usefulness of POpULar, we illustrate two effects which are often neglected in studies of ocean data assimilation before. One is the effect of correlations among the deviations of control variables in an adjoint analysis. The other is the nonlinear effect of sea surface dynamic height calculation required when sea surface height observation is employed in a three-dimensional ocean analysis. As the results, these effects are not so small to neglect.     

THE BOUGUER ANOMALIES AND DEPTHS OF MOHOROVICIC DISCONTINUITY IN THE SOUTH CHINA SEA REGION

The South China Sea is situated at the continsntal margin of South China. In this region, there are both continental and oceanic crusts. The values of Bouguer gravity anomalies on the continental shelf are low positive or low negative. Because the depth of the Mohorovicic discontinuity in this region is about 26-32 km below sea level, the crust belongs to the continental type. The values of Bouguer gravity anomalies in the deep-sea region are more than 250 mgal and the depth of the Moho-surface is about 10-15 km below sea level, so the crust is of oceanic type. The values of gravity anomalies and depths of the Moho-surface, obtained over the continental (and island) slope, range between those regions mentioned above, so the crust belongs to the transitional type. The continental crust is inferred to be directly in contact with the oceanic crust as a result of a lithospheric fault.