【摘要】 利用地理信息系统软件ARC/INFO,在考虑海拔高度对气温影响和没有考虑海拔高度对气温影响的两种情况下,分别运用样条插值法、普通克立格法和逆距离权重法对中国623个气象站1961—2000年40 a的逐月平均气温进行了空间插值,并利用交叉检验方法对插值精度进行了评估,结果表明:考虑了海拔高度影响的3种插值方法的精度都有比较明显的提高,对于普通克立格法,平均绝对误差(MAE)从1.44℃降到0.82℃,均方根误差(RMSE)从2.31℃降到1.28℃;对于逆距离权重法,MAE从1.52℃降到0.88℃,RMSE从2.38℃降到1.34℃;对于样条插值法,MAE从1.54℃降到0.87℃,RMSE从2.49℃降到1.35℃。在6种插值方法中,考虑了海拔高度影响的普通克立格法是最优空间插值方法。更多还原

【Abstract】 Monthly mean temperature data from 623 meteorological stations in China from 1961 to 2000 were interpolated by using inverse distance weighing(IDW),ordinary kriging(OK) and spline with ARC/INFO,when altitude was taken into consideration or otherwise.Cross-validation were applied to evaluate the three interpolation method,from which the mean absolute error(MAE) and root mean squared error(RMSE) were calculated.The results showed that better simulation would be achieved if altitude were taken into consideration.MAE and RMSE by IDW decrease from 1.52 ℃ to 0.88 ℃ and from 2.38 ℃ to 1.34 ℃,respectively;for OK,from 1.44 ℃ to 0.82 ℃ and from 2.31 ℃ to 1.28 ℃,respectively;for SPLINE,from 1.54 ℃ to 0.87 ℃ and from 2.49 ℃ to 1.35 ℃,respectively.It revealed OK gave the best simulation,followed by SPLINE and then by OK.更多还原