长江流域逐月气温空间插值方法的探讨

长江流域地形地貌特征复杂。为了探索适合长江流域的逐月气温数据空间插值方法，在考虑海拔、经纬度、坡度、坡向对气温影响和没有考虑这些地形因子对气温影响的两种情况下，利用流域151个气象站2007年逐月气温与对应的站点地形因子进行回归分析，并与回归直线截距相加产生栅格化的回归气温值，同时对回归多项式的残差分别运用反距离权重法（IDW）、普通克立格法（OK）和样条函数法（SPLINE）气温进行了空间插值，然后将栅格化的回归气温值与残差的插值结果相加得到空间化的逐月气温数据，并利用交叉检验方法对插值精度进行了评估。结果表明：考虑了地形因子影响的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03℃降到060℃，均方根误差（RMSE）从238℃降到123℃；对于反距离权重法，MAE从110℃降到065℃，RMSE从248℃降到138℃；对于样条插值法，MAE从124℃降到074℃，RMSE从266℃降到151℃。考虑了地形因子影响的空间插值方法整体上要优于没有考虑地形因子的空间插值方法，其中，基于地形因子的普通克里格插值方法结果相对较好

DISCUSSION OF METHODS OF SPATIAL INTERPOLATION FOR MONTHLY TEMPERATURE DATA IN THE YANGTZE RIVER BASIN

There is complex terrain in the Yangtze River basin.In order to explore the best suitable method for monthly temperature data spatial interpolation,we used the monthly temperature data which was gained from stations of the Yangtze River basin in 2007 for spatial interpolation experiment,and considered the altitude,latitude and longitude,slope,aspect effects on the temperature.We used monthly temperature of 151 stations in the basin of 2007,and did regression analysis between the site topography and temperature.Intercept of the regression line with the sum of the return generated temperature value of the grid.Meanwhile,the residual of polynomial regression were used by anti-distance weighting(IDW),Ordinary Kriging(OK) and the spline function method(SPLINE) spatial interpolation.Then return to the grid of the interpolated temperature value and the results added to get the residuals of the monthly temperature.The results showed that better simulation would be achieved if topography were taken into consideration.MAE and RMSE by OK decreased from 1.03℃ to 0.60℃ and from 2.38℃ to 1.23℃,respectively;for IDW,from 1.10℃ to 0.65 ℃ and from 2.48℃ to 1.38℃,respectively;for SPLINE,from 1.24℃ to 0.7℃ and from 1.66℃ to 1.51℃,respectively.It revealed OK gave the best simulation which take topography into consideration.