【作者】 刘全明；

【导师】 陈亚新；

【作者基本信息】 内蒙古农业大学 ， 农业水土工程， 2002， 硕士

【摘要】 本论文根据区域水环境预测中水盐监测与评价的要求，进行了水盐空间变异性与稳健变异函数的研究。为了在水土科学中引入稳健地质统计学的理论和方法，以弥补传统地质统计学的不足，本文利用河套灌区内两个实验区的春季水盐采样数据，进行了采样数据的经典统计分析和采样尺度的分析，计算了传统地质统计学变异函数，分析了特异值的干扰性，为此，进行了剔除特异值后计算稳健变异函数的尝试；同时，直接计算了两种稳健变异函数，将传统地质统计学变异函数和三种稳健变异函数进行了对比分析，得出以下结论：（1）通过数据的经典统计分析发现水盐数据多数为有偏分布，特异值的存在和干扰是主要因素之一。（2）不同尺度下的实验变异函数受特异值等因素的影响而有不同，要尽可能选择适当的尺度来计算。（3）由于特异值的存在和干扰，影响了变异函数的稳健性和区域化变量的结构性，因此，在水盐空间变异函数的推求中，处理特异值是非常必要的。（4）稳健估计方法对提高变异函数的统计特性有明显的改进效果，文中涉及的3种稳健估计方法所得到的稳健变异函数明显的优于传统地质统计学的实验变异函数。这3种方法分别为影响系数法、位置的尤伯估计、克瑞西-豪金斯方法，其中，影响系数法的稳健变异函数优于其他两种方法。3种方法拟和的理论变异函数与原始数据拟和的理论变异函数比较后发现明显的规律，块金常数C₀明显降低，区域化变量的随机性减小；理论变异函数的Ⅰ指标值明显减小，表明拟和效果更好；基台值C₀+C、C减小，反映了变量在该方向上的变化幅度降低；稳健变异函数曲线不同程度地存在C／C₀减小的趋势，方差变小了，变程a变化不大，说明稳健估计方法拟和的模型更可靠。两个垂直方向的稳健变异函数套合后进行克立格估值的最优等值线图与模型模拟图反映了实际的水盐含量。该图为积累多年水盐动态预测，提供了基础性的评价依据。（5）稳健估计方法除结果具有可靠性的优点外，还有控制稳健性和保证信息量之间的平衡的优点。（6）通过计算与分析表明，稳健地质统计学的方法可以移植到水土科学中应用，这对于加强水盐空间变异性以及稳健变异函数的基础研究有着重要的应用价值与学术价值。更多还原

【Abstract】 On the basis of requirements on water-salt monitoring and evaluation in zone water environment forecasting, this paper studies the Time-Space Variability of Water-Salt and the Robust Variogram. in order to replant theory and methods of Robust Geostatistics into Water-Soil science, and remedy the lack of the traditional Geostatistics, utilizing water-salt spring sampling data from two experiment section in Hetao irrigation zone, analyzes sutra statistical identity of the sampling data and sampling scale, calculates traditional Geostatistics Variogram, analyzes disturbance of outliers, then stab calculating the Robust Variogram after eliminating the outliers, while directly calculates two kinds of Robust Variogram, analyzes the traditional Geostatistics and three kind of Robust Variogram by comparing them, conclusions as follows(1)the water-salt sampling data almost attribute departure distribution,disturbance of the outliers is a main factor.(2)effect on experimental Variogram in different scales is different,we should select suitable scale to calculate.(3)because of the disturbance of the outliers,it affects Robustness of the Variogram and spatial structures of regional variables,so it is very essential to eliminate the outliers for calculating water-salt spatial Variogram.(4)the method of robust estimating can improve effect of the Variogram statistic character,three kinds of robust estimating methodes in this paper are distinctly superior to the traditional Geostatistics experimental Variogram, consists of the effective modulus method, location Huber estimating and Cressie.Hawkins method, among them, the effective modulus method is obviously superior to the others. Comparing academic Variogram by simulating using the three methods and that from raw data, we can find obviously rules: gold nugget constant C0 debases obviously, randomiciry of the regional variables minishes, statistical value minishes greatly, showing that simulating effect become better;C0+C minshes, it show that changing extent of the variable on this aspect debases; curves of the Robust Variogram have trend that C/C0 minish disparately, variance become small, changing area "a" changes not too, all show that the former from the Robust estimating is more trusty. Kriging gaging former plots and classic isoline maps of two vertical direction combination structure model reflect actual water-salt content. The maps offer basal ’ estimating basis for accumulating many years of Water-salt dynamic forecasting. (5) the Robust estimating has advantages of credible result and that control balance between Robustness and ensuring information quantity. (6)via calculating and analyzing, we cansay that the Robust Geostatistics methods can be introduced into the Water-Soil science, this has very important appliance and learning merit for enhancing studying on the Water-Salt spatial variability along with basal studying on the Robust Variogram.更多还原