大渡河流域逐日降雨数据融合的误差分析

Daily Rainfall Data Fusion Method and its Error Analysis in Dadu River Basin

为有效结合不同来源降雨资料的优势,发展多源降雨融合技术,采用数据融合方法,生成融合逐日降雨数据(以下称为Grid数据),再与气象观测站点资料、ERA5和NCEP2格点资料进行误差对比,探讨Grid数据在大渡河流域的适用性。结果表明:(1)Grid数据基本能完整展示大渡河流域的日均降雨分布,总体的、各区域内的降雨分布与观测数据展示出的分布规律基本吻合;(2)从插值站点来看,Grid数据的降雨量量级与观测数据差距小,总体的、各区域内的最大降雨量和最小降雨量与观测数据接近,表现优于ERA5和NCEP2数据;(3)从标准差、均方根误差来看,Grid数据的误差要明显小于EAR5和NCEP2数据;(4)从2008-2018年全流域、1~4分区的面雨量时间序列来看,Grid数据与观测数据较ERA5和NCEP2数据更接近。综上,Grid融合数据能较好地反映大渡河流域实际降雨情况。

In order to effectively combine the advantages of rainfall data from different sources,and develop multi-source rainfall fusion technology,this paper uses data fusion method to generate fused daily rainfall data(hereinafter referred to as Grid data).Then,compare and analyze the error with meteorological station data,ERA5 data and NCEP2 data,in order to explore the applicability of the Grid data in Dadu River Basin.Conclusions:(1)The daily average precipitation distribution of Dadu River basin can be completely displayed by the Grid data.On the overall and within regions,and the precipitation distribution is basically consistent with the precipitation displayed by the meteorological observation data.(2)From the perspective of interpolation to stations,the difference between the rainfall magnitude of the Grid data and the observation data is small,and the overall maximum and minimum rainfall in each region are close to the observation data,which is better than the ERA5 and NCEP2 data;(3)From the standard deviation and the root mean square error,the Grid data is better than ERA5 data and NCEP2 data.(4)From the area rainfall time series of the whole basin and1-4 divisions,Grid data is closer to observational data than ERA5 and NCEP2 data.In a word,the daily rainfall Grid data can better reflect the actual precipitation of Dadu River Basin.