中文摘要：

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。

英文摘要：

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.