克里金日降水插值的不同变异函数比较分析

Comparison and Analysis of Different Variogram Functions Models in Kriging Interpolation of Daily Rainfall

摘要: 普通克里金法为一种广泛使用的地理统计的插值方法，但一般都只依据经验使用一个变异函数来计算插值结果。同时，现有研究着重研究年、月尺度下普通克里金法不同变异函数的比较，对日降水尺度研究较少。为了得到普通克里金法在日降水插值适应性较好的变异函数，比较和分析普通克里金法四种常用不同变异函数(指数、球型、高斯、线性)。选取了湘江洣水所在子流域及子流域周边一共43个雨量站1990~2005年逐日降水资料插值，从相关系数、检验指标、不同降水级别准确率这三方面比较插值结果，交叉验证的结果表明：1) 指数和球型函数对普通克里金法的日降水插值适应性较好；2) 普通克里金法的插值结果在降水量大的日子普遍偏小；3) 随着降水量级的增大，日降水插值的误差明显增大。

Abstract: Ordinary Kriging is a wide-used method of geostatistical interpolation. However, it is only one variogram function that is often used to calculate the estimated value depending on one’s experience. Meanwhile, there is lesser study about daily rainfall comparison of different variogram functions in ordinary Kriging because most of the recent researches emphasize yearly and monthly rainfall. To obtain the better fitted variogram function in ordinary Kriging for daily rainfall interpolation within four different variogram functions (exponential, spherical, gauss, linear). The study leans on 16-year (1990-2005) daily rainfall data at 43 raingages in and around Mishui Basin. According to the comparison of three aspects including the correlation coefficients, examine index, accuracy of different precipitation grade, the result of cross- validation shows that: 1) exponential and spherical functions can fit the daily rainfall interpolation better in ordinary Kriging; 2) the estimated value in ordinary Kriging is generally smaller than observed value on days of heavy rainfall; 3) the error of daily rainfall interpolation increases with the precipitation grade.