风廓线雷达数据精度评价与质量控制方法

Accuracy Evaluation and Quality Control of Wind Profile Radar Data

为了更好地研究风廓线雷达数据的质量控制方法,选用了2018年6月地面无降水时段北京7个站点的风廓线雷达数据,将其与FNL再分析资料进行径向风和纬向风的对比,发现7个站点的数据与再分析资料均有较强的相关性,南郊站的数据观测误差相对较低。随后针对南郊站的风廓线雷达数据,先进行极值检验和垂直一致性检验,然后分别利用EOF分析重构法和高斯滤波法对其进行质量控制,通过质控前后的对比,并结合再分析资料,探寻这2种方法在时空尺度和误差分析上所产生的不同影响。结果表明:在空间尺度上,EOF分析重构法的数据平滑效果更好;但在时间尺度上,高斯滤波法的平滑效果更优;通过质量控制前后与FNL再分析资料的误差分析可以得出,高斯滤波法能较为明显地降低原始数据的观测误差。

In order to study the wind profile radar data quality control methods better,the data of wind profile radars at seven stations in Beijing during the period of no precipitation on the ground in June 2018 are selected and compared with FNL reanalysis data in terms of radial and zonal wind components.It is found that the data of seven stations are highly correlated with reanalysis data,and the data observation error of southern suburb station is relatively low.Firstly,for the wind profile radar data of the south suburb station,the extreme value test and the vertical consistency test are performed.And then,the EOF analysis reconstruction method and the Gaussian filtering method are used for quality control.Through the comparison before and after the quality control combined with the reanalysis data,the different effects of these two methods on space-time scale and error analysis are explored.The results show that the data smoothing effect of EOF is better in spatial scale,while the effect of Gauss filter is better in time scale.Through the error analysis of the data before and after quality control and FNL reanalysis data,it can be concluded that the Gaussian filtering method can reduce the observation error of the original data more significantly.