面向大范围降水空间插值的普通克里金模型开发与实例分析

A Matlab Based Ordinary Kriging Model Developed for Rainfall Spatial Interpolation Over Macro Regions and A Case Application

基于Matlab平台,开发了面向大范围降水空间插值的普通克里金模型—Matlab based ordinary Kriging(MatOK).与已有模型相比,MatOK的主要特点是:(1)在降水空间变异函数计算环节,采用SCE-UA算法拟合理论变异函数;(2)在OK方程组估值环节,引入降水空间发生概率的计算,完善了日等短时间尺度降水空间估值方法;(3)通过对OK方程组的标准化处理,有效提高了模型数值计算的稳定性.同时,将MatOK初步应用于集水面积为83374km2的赣江流域年、月、日降水空间插值中,并重点讨论了MatOK的计算稳定性和计算效率,结果初步说明了采用MatOK进行大范围区域降水空间插值是可行的.

An automated ordinary Kriging model based on Matlab platform （MatOK）was specially developed for rainfall spatial interpolation over macro regions. The proposed model has three main traits compared with other ordinary Kriging （OK） models. Firstly, in the part of rainfall spatial variogram computation, the parameters of theoretical variogram are optimized using the Shuffled Complex Evolution method-University of Arizona （SCE-UA）. Secondly, in the part of OK equation estimation, daily and other short time step rainfall interpolation is improved by adding a process of precipitation occurrence probability（POP） calculation. Thirdly, the model＇s numerical stability is raised through OK equation＇ s standardization. Then, based on 325 rain gauges observations during 2003--2009 in the Ganjiang River Basin with a drainage area of 83374km2,a case application of yearly,monthly and daily rainfall spatial interpolation using MatOK was conducted. Results preliminarily show that MatOK is feasible for rainfall spatial interpolation over macro regions.