支持向量机优化的克里金插值算法及其海洋资料对比试验

针对常规克里金插值方法中变异函数为有限确定函数,难以准确刻画实际数据分布（特别是复杂要素资料的空间结构）的不足,基于最小二乘支持向量机从实际资料场中拟合重构变异函数的研究思想,提出了一种改进的插值方法——支持向量机—克里金插值算法（SVM-Kriging）。采用常规的克里金方法和支持向量机—克里金插值方法（SVM-Kriging）进行插值试验和对比分析,结果表明：支持向量机—克里金方法（SVM-Kriging）的变异函数源自实际的数据场,它克服了常规克里金插值对变异函数选择的依赖性以及变异函数选择的主观性和人为性,表现出较好的针对性和客观性,较为有效地改善了插值效果。     

加入多时相纹理的遥感变化检测

基于遥感技术的变化检测是遥感应用的一个重要方面.传统基于遥感的变化检测方法一般是利用光谱信息,较少注意多时相图像间的光谱特征相关关系分析.本文运用地统计学的伪交叉变差函数(Pseudo Cross Variogram)计算多时相图像纹理,定量表达多时相图像间的空间相关关系,并将得到的多时相纹理信息与光谱信息一起用于多时相变化检测.实验结果表明,加入多时相纹理信息可以显著提高变化检测精度,是一种有效的方法.     

基于Kriging方法的空间数据插值研究

介绍了Kriging插值方法及其实质,提出一种变异函数理论模型参数估计的新方法,给出变异函数理论模型有效性评定的统计指标,并通过算例予以验证。最后,通过实例与反距离加权法相比较,证实Kriging插值的优越性。     

Use of indicator variograms for an enhanced spatial analysis

Flat variograms often are interpreted as representing a lack of spatial autocorrelation. Recent research in earthquake engineering shows that nearby field noise can substantially mask a prominent spatial autocorrelation and result in what appears to be a purely random spatial process. A careful selection of threshold in assigning an indicator function can yield an indicator variogram which reveals underlying spatial autocorrelation. Although this application involves use of seismic data, the results are relevant to geostatistical applications in general.     

Using geostatistics to elucidate temporal change in the spatial variation of aeolian sediment transport

Little is known about the spatial and temporal scales of variation in aeolian processes. Studies that aim to investigate surface erodibility often sample aeolian sediment transport at the nodes of a regular grid of arbitrary size. Few aeolian transport investigations have the resources to obtain sufficient samples to produce reliable models for mapping the spatial variation of transport. This study reports the use of an innovative nested strategy for sampling multiple spatial scales simultaneously using 40 sediment samplers. Reliable models of the spatial variation in aeolian sediment transport were produced and used for ordinary punctual kriging and stochastic simulated annealing to produce maps for several wind erosion events over a 25 km² playa in western Queensland, Australia. The results support the existence of a highly dynamic wind erosion system that was responding to possibly cyclic variation in the availability of material and fluctuations in wind energy. The spatial scale of transport was considerably larger than the small scale expected of the factors controlling surface erodibility. Thus, it appears that transport cannot be used as a surrogate of erodibility at the scale of this investigation. Simulation maps of transport provided considerably more information than those from kriging about the variability in aeolian sediment transport and its possible controlling factors. The proposed optimal sampling strategy involves a nested approach using ca 50 samplers. Copyright © 2003 John Wiley & Sons, Ltd.     

General polygonal variogram functions: Evaluation of estimation variance integrals

The application of regionalized variables requires the estimation of the variogram function and the evaluation of its integral. By representing the variogram by a general polygonal function the requisite integrals may be easily computed by a closed form representation of simple integrals. This paper provides the integration formulas for two-dimensional variogram functions whose domain is represented as a finite collection of rectangles. The integration formulas essential for a fully developed polygonal approach to an extensive statistical evaluation of geostatistical quantities are presented.     

Scale Effect on Principal Component Analysis for Vector Random Functions

Principal component analysis (PCA) is commonly applied without looking at the spatial support (size and shape, of the samples and the field), and the cross-covariance structure of the explored attributes. This paper shows that PCA can depend on such spatial features. If the spatial random functions for attributes correspond to largely dissimilar variograms and cross-variograms, the scale effect will increase as well. On the other hand, under conditions of proportional shape of the variograms and cross-variograms (i.e., intrinsic coregionalization), no scale effect may occur. The theoretical analysis leads to eigenvalue and eigenvector functions of the size of the domain and sample supports. We termed this analysis growing scale PCA, where spatial (or time) scale refers to the size and shape of the domain and samples. An example of silt, sand, and clay attributes for a second-order stationary vector random function shows the correlation matrix asymptotically approaches constants at two or three times the largest range of the spherical variogram used in the nested model. This is contrary to the common belief that the correlation structure between attributes become constant at the range value. Results of growing scale PCA illustrate the rotation of the orthogonal space of the eigenvectors as the size of the domain grows. PCA results are strongly controlled by the multivariate matrix variogram model. This approach is useful for exploratory data analysis of spatially autocorrelated vector random functions.