不同插值方法对降水量空间不确定性的影响

鉴于点状数据的降水插值方法在时间和空间两个方面具有不确定性,从空间角度出发,以山东省多年平均降水量为例,采用交叉验证方法,对反比距离权重法、克里金方法、径向基函数法、全局多项式法和局部多项式法5种常用插值方法的整体插值精度和分区后各分区的插值精度分别进行分析验证。研究认为:山东省整体插值的最优方法为反比距离权重法;分区插值的最优方法中,平原地区为克里金方法、丘陵地区为全局多项式法、鲁中山地为反比距离权重法。并在此基础上,对整体插值精度和分区插值精度作了比较。研究结果可为区域降水插值模型的选取提供指导,同时分区插值的思路可为后续相关研究提供借鉴。

Influence of Different Interpolation Methods on Spatial Uncertainty of Rainfall

The uncertainty of precipitation using different forms of interpolation comes from two aspects: time and space.We,taking Shandong’s annual average precipitation as an example,make a comparation of the five common methods,that is,inverse distance weighting,kriging,radical basis function,global polynomial and local polynomial,in integral interpolator and zonal interpolator.By using cross-validation,the methods’precision was given.The conclusion is that the best integral interpolator is the inverse distance weighting method,and the zonal methods are kriging in plain,global polynomial in hilly ground,inverse distance weighting in mountain area.Then the exactness of integral interpolator and zonal interpolator was analyzed and the precipitation distribution of Shandong was given.The research on integral interpolation and zonal interpolation can be a guide for the choice of spatial interpolation,meanwhile,the mentality of zonal interpolator can be used as a reference for following researches.