山西高原气温空间插值分析

Spatial Interpolation Analysis of Annual Average Air Temperature in Shanxi Plateau

基于山西高原及其周边区域共27个国家标准气象站1981—2005年的日气温资料,结合数字高程模型,利用反距离加权、全局多项式、局部多项式、径向基函数、样条函数、趋势面、普通克里金与协同克里金共8种空间内插方法对山西高原年均气温进行插值,然后用留一交叉验证法对各种插值方法进行精度对比,最后分析了山西高原多年平均气温的空间分布特征.结果表明:1)以高程为协同因子的协同克里金法精度最高,是适合山西高原气温插值的最优方法;2)基于地统计学的协同克里金和普通克里金比常规的插值方法精度要高;3)常规的插值方法按精度由高到低排序为径向基函数、局部多项式、反距离加权、全局多项式、趋势面、样条函数;4)山西高原气温在空间分布上具有明显的地域性特征,表现为从低纬度向高纬度递减的规律.

Based on daily air temperature data from 27 meteorological stations in Shanxi plateau and its surrounding areas,in combination with digital elevation model,the annual average temperature of Shanxi plateau from 1981 to 2005was interpolated using spatial interpolation methods including inverse distance weight,global polynomial interpolation,local polynomial interpolation,radial basis function,spline,trend surface,ordinary Kriging,and Co-Kriging.The precision of these methods was evaluated through leave-oneout cross-validation.The spatial distribution of temperature in Shanxi plateau was explored by the optimal method.The results showed that：1）DEM-adjusted Co-Kriging method is optimal for interpolating the temperature values of Shanxi plateau;2）The accuracy of Co-Kriging and ordinary Kriging based on geostatistics was higher than that of general interpolation methods;3）The accuracy of general interpolation methods in descending order was radial basis function method,local polynomial method,inverse distance weighting method,global polynomial method,trend surface method and spline function method;4）The spatial distribution of air temperature in Shanxi plateau presented regional characters,which showed a decrease trend from low latitudes to high latitudes.