站点密度对复杂地形PRISM月降雨空间插值精度的影响

Effects of spatial station density on accuracy of spatial interpolation of monthly rainfall over complex terrain base on PRISM

【目的】利用降雨和高程与坡向等地形因子之间关系,分析站点密度对于坡面回归方程模型(PRISM)插值精度的影响,探究该模型的适用范围。【方法】以北京西北山区为例,基于研究区数字高程模型(DEM)、山地自动气象站点数据和降雨数据计算插值结果,采用反距离加权法(IDW)、克里金法(Kriging)和样条函数法(Spline)等插值方法,以及交叉验证和实测数据验证等方法进行数据对比,分析站点密度对插值结果的影响。【结果】当站点密度从0.55×10^(-2)个/km^2降低到0.18×10^(-2)个/km^2时,各种插值方法的插值精度均随站点密度的减少而降低,PRISM模型的变化程度最大,Spline的变化程度最小;同时当站点密度逐渐降低至0.18×10^(-2)个/km^2时,PRISM模型的插值误差超越Kriging和IDW,但仍在Spline之上。【结论】当站点密度较低时,PRISM模型优势不明显,建议使用IDW和Kriging。

[ Objective] Investigate the influence of station density on the accuracy of PRISM interpolation, in order to elucidatethe model＇s scope of application. [ Method] Based on the DEM of the northwest mountains of Beijing and rainfall data from the mountain automatic weather stations to calculate the result of interpolation. These results were then compared with the IDW, Kriging and Spline interpolation models using cross validation and test data validation. [ Result ] The precision of interpolation was reduced when site density was reduced from 0.55× 10-2 to 0.18× 10-2 sites/km2 but was reduced most when using the PRISM model and least when using the Spline model. Furthermore, when the site density was reduced to 0. 18× 10-2 sites/km2, the error of the PRISM model was less than that of the Spline model, but still greater than the Kriging and IDW models. [ Conclusion ] The present study suggests that the IDW and Kriging models should be used when stations arc sparse.