用加权多项式回归进行球状模型变差图的最优拟合

本文提出了一种用加权多项式回归对球状模型和二级套合球状模型的变差函数进行最优拟合的方法。在已经较好地算出各实验变差函数值并选定球状模型或二级套合球状模型为理论模型的条件下,应用这种方法可以编出程序,并在计算机上快速、自动地算出最优拟合的变差图。其结果唯一确定,不因人而异,可避免人为的误差,又可为地质统计学计算的全盘自动化创造良好条件。经过几个实例验算表明,该方法简单,效果良好。

ON THE BEST FITTING OF THE SPHERICAL MODEL VARIOGRAM WITH THE WEIGHTED POLYNOMIAL REGRESSION

The paper suggests an approach, utilizing the weighted polynomial regression for the best fitting of the variogram about spherical model or two-order nested spherical model. Under the conditions that the values of experimental variogram were well calculated and the spherical model or the two-order nested spherical model was selected as the theoretical model of variogram, we are able to use the approach to write a program, and the computer will give the best fitted variogram rapidly and automatically. The result is uniquely determined, being independent of the person who works at it,so it ispossible to avoid the artificial errors and to create favorable conditions for the overall automatization of the calculation in geostatistics. After an application of this approach to some practical examples, it shows that this approach is quite simple and very effective.