Kriging插值中条件数据点个数的选择

Kriging插值时使用的数据量越大,插值结果越可靠,但数据量对耗时的影响则呈指数增加。为了保证尽量多的数据参与Kriging插值,又能维持合理的计算耗时,必须正确地选择条件数据点个数。通过对地质统计学中Kriging插值算法的深入解析,利用Kriging插值中产生的平均权系数分析插值过程中的条件数据点个数的选择,当平均权系数最小时,找到这个平衡点,让插值过程中的计算在保证计算精度的同时,尽量减少计算耗时。并改变Kriging插值的维数、网格大小和变差函数的变程,分析所选择的条件数据点个数对此3种因素的敏感度,得到条件数据点个数选择40～60最为合适,主要的影响因素为计算维数。

Selection of number of conditional data in Kriging interpolation

When using Kriging interpolation to the unknown location,the larger the data volume is,the more reliable the result is,but the influence degree of data volume on time consumption increases exponentially.In order to ensure the more data to participate Kriging interpolation and maintain the proper calculation of time consumption,the number of conditional data must be selected reasonably.This paper analyzes the selection of number of conditional data in the process of interpolation by deeply analyzing Kriging interpolation algorithm in geostatistics and using average weights of Kriging.When the average weight is the smallest,a balance point,which can improve the accuracy in interpolation calculating and also can reduce the time consumption,can be found out.Through changing the dimensionality of Kriging interpolation,the size of cells and range of variogram,the sensitivity in choosing the number of conditional data is analyzed.The selected range of conditional data from 40 to 60 is the best and the main influence factor is calculation of dimensionality.