不同插值方法模拟四川省逐月降水量的对比分析  

Comparison Analysis on Different Spatial Interpolation Methods to Simulate Monthly Precipitation in Sichuan Province

作　　者：李艳[1] 朱军[1,2] 胡亚[1,2] 张恒[1,2] 

机构地区：[1]西南交通大学地球科学与环境工程学院,成都611756 [2]西南交通大学高速铁路运营安全空间信息技术国家地方联合工程实验室,成都611756

出　　处：《水土保持研究》2017年第1期151-154,160,共5页Research of Soil and Water Conservation

基　　金：国家自然科学基金"基于基元组合和语义约束的虚拟高速铁路环境智能建模方法研究"(41271389);国家自然科学基金"面向高速列车耦合仿真的三维虚拟高速铁路精细化实体环境建模研究"(41401433)

摘　　要：我国西部地区气象观测点较少,致使一些地区的降水信息无法直接获取,采用空间插值方法来推测相邻地区的降水量是重要手段之一。针对四川地区观测点少的现状,结合空间分辨率为90m×90m的数字高程数据(DEM),分别采用常规反距离加权插值(IDW)、考虑各点高程的反距离加权插值(IDW)、局部多项式(LPI)、普通克里金(OK)、协同克里金(CK)对每个月降水量和1年平均每月降水量进行插值,并采用交叉检验法来验证插值结果,将平均误差(MAE)和均方根误差(RMS)作为评估插值效果的标准。结果表明:考虑高程的IDW插值误差小于常规IDW,可明显提高插值精度,克里金方法平均误差和均方根误差较小,优于反距离加权和局部多项式插值,在两种克里金方法中,与数字高程模型结合的CK方法的插值效果更好,更加适合四川省山区地形降水量数据的插值。The precipitation information canr t be obtained directly in some areas because the meteorological observation points are few in the western region of China. Using the spatial data interpolation methods to estimate the precipitation in the adjacent areas is one of the important means. For the less observation points in Sichuan Province, we combine with the digital elevation model （DEM） of the spatial resolution of 90 m× 90 m, and use the ordinary of inverse distance weighted interpolation （IDW）, the inverse distance weighted interpolation of considering each point elevation （IDW）, local polynomial interpolation （LPI）, the ordinary Kriging interpolation method （OK） and the collaborative Kriging interpolation method （CK） to interpolate the every monthly and annual average precipitation in Sichuan Province. The cross checking method was used to verify the results of interpolation, and the average error （MAE） and the root mean square error （RMS） were used as the criteria for evaluating the five interpolation methods. The results show that the IDW interpolation of considering the points elevation is more precise than the ordinary IDW, can significantly improve the accuracy of interpolation, Kriging average error and root mean square error is smaller than IDW and local polynomial interpolation, the collaborative Kriging has better interpolation results because of considering the influence of digital elevation model on rainfall. Therefore, collaborative Kriging is more suitable for spatial interpolation of rainfall data in mountain area.

关 键 词：空间插值 降水量模拟 交叉验证 误差分析 

分 类 号：P208[天文地球—地图制图学与地理信息工程] P332[天文地球—测绘科学与技术]