河北省年均降水量插值方法比较  

Comparison of interpolation methods on annual mean precipitation in Hebei Province

作　　者：刘劲松[1,2] 陈辉[1,2] 杨彬云[1] 王卫[2] 相云[1] 赵超[1] 

机构地区：[1]河北省气象与生态环境实验室,石家庄050021 [2]河北师范大学资源与环境科学学院,河北省环境演变与生态建设实验室,石家庄050016

出　　处：《生态学报》2009年第7期3493-3500,共8页Acta Ecologica Sinica

基　　金：河北省气象与生态环境实验室开放基金资助项目(SYSKF0603);河北省自然科学基金资助项目(D2007000272);国家自然科学基金资助项目(40871073);国家“十一五”科技支撑计划重点资助项目(2006BAB15B02);河北省教育厅软科学资助项目(2006206)

摘　　要：以河北省及临近区域120个气象观测站点1971～2000年均降水量数据为基础,选择其中的40个作为检验站点,其余站点分别取80、40、20个作为插值站点,采用局部插值、整体插值、多元线性回归、综合模拟等多种插值模型讨论了降水空间插值问题,主要结论如下:插值站点数、模型类型、模型参数都会影响插值精度。局部插值模型相对误差最小值出现在Spline、IDW模型中,其次为Kridging模型,而整体模型Trend、多元线性回归模型误差均较大,但综合了局部插值模型和统计模型的综合模型一定程度上能改善插值精度及误差分布。河北省80和40个站点的最优插值模型为综合模型,20个站点的最优插值模型为IDW2。Methods, such as part interpolation, whole interpolation, multiple-linear regression and integrative model were used to interpolate the precipitation based on the annual mean precipitation from 120 climate stations in Hebei province and its adjacent areas （from 1971 to 2000）. Results showed that numbers of interpolation climate stations, type of interpolation model and parameters of model influenced the interpolating precision. The smallest mean relative error （MRE） appeared in Spline or IDW models, followed by Kridging models, and bigger for the error of whole model Trend and Linear Regression Models. The integrative model, such as part model and linear regression model, could improve the interpolating precision and error distribution. In conclusion, integrative model is the suitable interpolating model for 80 and 40 climate stations in Hebei province, and IDW2 is best for 20 climate stations.

关 键 词：河北省 年均降水量 插值 

分 类 号：P426.613[天文地球—大气科学及气象学]