基于EOF的高时空分辨率自动站温度观测资料质量控制

随着我国气象自动化观测技术的发展,全国已经建成了70 000多个自动观测站点,全面实现了气象观测自动化。自动化观测技术使得气象常规观测资料量得到了飞速增长,这也使得通过质量控制提高自动观测站资料的利用率尤为重要。利用江苏省气象局提供的2019年12月1日00时—7日23时共168个时次的地面自动站温度观测资料,及ECWMF (European Centre for Medium-Range Weather Forecasts)的ERA5(ECMWF Reanalysis V5)再分析资料中的温度格点资料作为背景场,结合常规质量控制方法及EOF (Empirical Orthogonal Function)质量控制方法,建立了适用于高时空密度的地面温度资料质量控制方法,并对我国中东部地面自动站温度观测资料进行质量控制试验。结果表明,在利用常规质量控制方法剔除观测资料中明显异常的资料后,针对自动站高密度的特点,通过选择合适的分析区域,EOF分析方法可以很好提取有组织的观测系统信息,从而保证剩余信息更好地满足随机分布特点,利用随机概率分布特点就可以很好剔除异常观测资料,并且可以避免天气变化的影响。

Quality control based on EOF for surface temperature observations from high temporal-spatial resolution automatic weather stations

With the development of automatic meteorological observation technology in China,approximately 70,000 automatic observation stations have been constructed across the country,and the automation of meteorological observation has been fully realized.Automatic observation technology causes the amount of meteorological observation data to increase rapidly,but how to improve the utilization rate of automatic observation data through quality control is of particular importance.This study uses a total of 168 temperature observation data of automatic ground stations,beginning from 00:00 to 23:00 BST on December 1,2019,provided by Jiangsu meteorological bureau,along with the temperature grid data in the ECMWF Reanalysis V5 (ERA5) Reanalysis data of European Centre for Medium-range Weather Forecasts (ECMWF).Next,by combining the general quality control method and Empirical Orthogonal Function (EOF) quality control method,this paper establishes a quality control method for surface temperature data with high spatial and temporal resolution.Next,real data quality control experiments on surface temperature observation of automatic stations in central and eastern China are conducted to verify the effectiveness of the new method.The results show that,according to the characteristics of the automatic station high density,by choosing the appropriate analysis area,the EOF analysis method can effectively extract organized observation system information,so as to ensure that the remaining information better meet the random distribution characteristics.After this,abnormal observations can be effectively removed according to the probability distributions,which in turn can help avoid the impact of changes in the weather.