结合软数据的同位置协同克里金估值新方法

摘要

协同克里金法结合软数据可以提高地质体建模中估值的准确性.通过分析传统的结合软数据同位置协同克里金估值过程中存在的问题及原因,提出并推导了一种新的结合软数据的同位置协同克里金估值方法.该方法采用软数据的期望值以保证外推的可靠性,利用主变量观测值滤去协同变量的期望值以估计变异函数参数.该方法的特点是:当待估点位于主变量相关变程之内时,估值的结果主要反映主变量变化特征;当待估点落在相关变程之外时,估值结果的差别主要来源于各自点上协同变量(软数据)的期望值.最后,用实例验证了新估值方法的优点.%CoKriging combined with soft data can improve the accuracy of building geological model. By analyzing the reasons for the problems of the conventional CoKriging and collocated CoKriging, this paper deduces a new collocated CoKriging with soft data. The method uses the expected value of soft data in order to ensure the reliability of extrapolation, the use of the main variables filtered the expected value of co-observations to estimate parameters of variogram. The advantage of the new collocated CoKriging is that the interpolation reflects feature of main variable where the points are within the interrelated range of main variable, and the difference of interpolation comes from expect of each co-variable where the points are beyond the interrelated range of main variable. Finally,examples show the advantages of the new valuation method.