基于智能网格降水产品的降尺度释用技术研究

利用空间分辨率为5km的广西智能网格降雨量预报产品以及桂南区域自动气象站降雨量实况资料,采用最邻近点法、双线性插值、双三次插值和统计降尺度方法,研究和评估空间分辨率在1km的格点降雨量插值预报释用技术。结果表明:(1)对于分辨率较高的智能网格预报产品,最邻近点法、双线性插值和双三次插值的预报效果无明显差异,其中最邻近点法的ETS评分和预报偏差略好于其它两种方法;(2)采用一元线性回归的统计降尺度方法比其它3种方法的降水预报能力高,并可以提高强降水的预报评分,对构造更高分辨率的网格预报产品具有一定参考作用。

Research on downscaling interpretation technology based on intelligent grid precipitation products

Based on the 5 km spatial resolution of Guangxi intelligent grid rainfall forecast products and the rainfall data of automatic weather stations in southern Guangxi,the nearest neighbor method,bilinear interpolation,bicubic interpolation and statistical downscaling method were used to evaluate the grid point rainfall interpolation forecast interpretation technology with spatial resolution of 1 km.The results showed that(1)for smart grid forecast products with higher resolution,there was no significant difference in the forecast effects of the nearest neighbor method,bilinear interpolation,and bicubic interpolation.The ETS score and forecast deviation of the nearest neighbor method were slightly better than the other two methods.(2)The statistical downscaling method using univariate linear regression had higher precipitation forecasting ability than the other three methods,and could improve the forecast score of heavy precipitation,which has a certain reference effect for constructing higher-resolution grid forecast products.