喀斯特地区降水量空间插值方法对比——以贵州省为例

(贵州师范大学 地理与环境科学学院, 贵阳 550025)

喀斯特山区; 降水量; 空间插值; 贵州省

Comparison of Spatial Interpolation Methods of Precipitation-A Case of Karst Area in Guizhou Province
MO Yueshuang, SUO Huiying, JIAO Shulin, ZHAO Zongquan,ZHANG Jie, ZHAO Meng, LIU Wei, LI Yinjiu

(School of Geography and Environmental Science, Guizhou Normal University, Guiyang 550025, China)

karst mountains area; spatial interpolation; precipitation; Guizhou Province

备注

利用贵州省17个气象站1960—2018年降水资料,采用反距离权重法(IDW)、普通克里金法(OK)、样条函数插值法(Spline)、趋势面插值法(Trend)4种方法,按多年平均、月平均、最大3个月及最小3个月不同降水量指标进行分析,对贵州省降水量空间插值模拟结果进行了交叉检验。结果 表明:4种方法的年均降水插值精度和最大3个月降水插值精度为Spline>OK>Trend>IDW; 最小3个月降水量插值结果显示OK法的插值结果最好,IDW法效果最差; 月降水量插值结果的MAE,MRE和RMSE的变化都呈现出明显的季节性,平均绝对误差(MAE)和均方根误差(RMSE)表现为秋冬两季误差低于春夏两季,平均相对误差(MRE)的值在春季和秋季较高,4种插值方法误差值排序为IDW>Trend>Spline>OK,OK法的插值结果较好,随着降水量的增加,模拟的降水量插值情况变差。(3)贵州省年降水量在空间上变化较大,总体上由南向北减少,年降雨量最高区域主要位于西南部,毕节地区降水最少; 采用Spline法对年均降水量进行插值,出现渐变平滑条带现象,其插值结果较为均匀准确,在空间分布上符合贵州省的实际情况,空间插值较好。

Based on precipitation data from 17 meteorological stations located at Guizhou, China, we used four interpolation methods such as reverse distance weight(IDW), Ordinary Kriging(OK), Spline, and Trend to simulate and cross-validate the average monthly precipitation from 1960 to 2018, in which the application involved yearly, monthly, the highest 3-month, and the lowest 3-month time scale. The results showed that the interpolation accuracies on yearly, monthly, the highest 3-month, and the lowest 3-month decreased in the order: Spline>OK>Trend>IDW, the OK was the best approach since it had the highest accuracy on the lowest 3-month time step, however, the result of the IDW method was unsatisfactory; the values of MAE(Mean Absolute Error), MRE(Mean Absolute Error), and RMSE(Root Mean Square Error)on the monthly precipitation interpolation changed over seasons; RMSE and MAE in the autumn and winter were lower than those in the spring and summer, and MRE in the spring and autumn was higher than ones in other seasons; the error accuracies decreased in the order: IDW>Trend>Spline>OK; the result of OK method was good, but the effect of simulated precipitation became worse over the change of raising rainfall; the spatial distribution of the average yearly rainfall was of significant difference in Guizhou Province; overall, the spatial change of annual precipitation was decreasing from the south to the north, and the area of the annual highest rainfall took place in the southwest; the lowest rainfall happened in the Bijie region. The Spline method was used to interpolate the average annual precipitation and appeared a gradual smooth strip. The interpolation result of the Spline was accurate, and its spatial distribution was consistent with the actual rainfall situation through comparing with observations.