基于DEM的江苏气温空间插值研究

Spatial interpolation of mean temperature of Jiangsu province based on DEM

以江苏省及周边39个常规气象站点1957—2001年的月平均气温数据和90 m空间分辨率的DEM数据为基础,采用基于DEM的多元线性回归插值方法,分析多年平均气温与海拔、坡度和坡向等地形因子的相关关系,建立适合该区域的多元回归空间插值模型.同时与反距离权重法（IDW）和克里格（Kriging）插值法等传统方法的计算结果进行对比,并用交叉验证方法比较5种插值方法的精度.结果表明：该研究区各月气温递减率在0.5-0.9℃/（100 m）左右;基于DEM的多元线性回归空间插值方法（MLR）无论从插值效果还是误差精度上,均优于其他传统插值方法.插值结果客观地表达了气温与各地形要素的相关性,反映了气温的空间变异性.

Based on the monthly mean temperature data of 39 meteorological stations in Jiangsu Province from 1957 to 2001 and Digital Elevation Model（ DEM） data with resolution of 90 m,a geographical knowledge aided spatial interpolation model was introduced to improve the spatial interpolation result. Comparative analysis was conducted between the knowledge based spatial interpolation and traditional methods of Inverse Distance Weighting（ IDW）,Co Kriging（ CK）,Ordinary Kriging（ OK）,Spline function（ SPLINE） using the cross validation results,it was concluded that the range of the monthly temperature lapse rate was about 0. 5-0. 9 ℃ /（ 100 m） in the research area,and the proposed interpolation method based on DEM was superior to other methods in both interpolation effect and accuracy. The monthly mean temperature distribution map with resolution of 90 m was established,which expressed the main trend of temperature changing with terrain and reflected the spatial variability of temperature.