智能网格气温降水实况插值方法优选研究

Optimal Selection of Interpolation Methods for Temperature and Precipitation in Intelligent Grid

为生成任意位置气象“实况”,在智能网格5 km空间格距整点气温和小时降水量“实况”的基础上,选取湖北省10个不同地形和气候分区的气象站为代表站,开展了双线性插值、最近邻插值、反距离权重插值等3种不同插值方法的对比试验。结果表明:(1)整点气温插值反距离权重法精度最高,平均绝对误差为0.53℃。插值效果随气温升高而降低、随海拔高度上升而降低。(2)小时降水量插值双线性插值法精度最高,平均绝对误差为0.17 mm。08时插值效果最佳,20时插值效果最差;插值效果随降水强度增大而降低。

Based on the temperature and one-hour accumulated precipitation datasets from the intelligent grid with resolution 5 km,the comparison upon the performance of bilinear interpolation,nearest neighbor interpolation and inverse distance weight interpolation in producing actual meteorological data at any location are conducted at 10 selected meteorological observation stations with different topography and climate conditions.It shows that,first,the inverse distance weighting method performs best in temperature interpolating,with average absolute error 0.53℃.The interpolation accuracy decreases with the increase of temperature and altitude.Second,the bilinear interpolation method of hourly precipitation interpolation performs better than the other two interpolation strategy,with average absolute the error 0.17 mm.The interpolation accuracy of hourly rainfall column with bilinear interpolation performs best at 08:00 UTC,while the worst at 20:00 UTC.