风廓线雷达数据质量影响因子及处理算法

风廓线雷达系统误差和探测数据时空代表性影响风的数据质量。针对五波束探测风廓线雷达，提出雷达系统误差检测方法并分析风的空间不均匀分布和时间代表性对风数据质量的影响。在此基础上，通过比较4组三波束计算的两组水平风u，v分量离差进行风的空间均匀性判别，并比较了一致性平均和数学平均两种时间代表性处理算法间的测风精度差异。利用广东风廓线雷达站网2014年3—5月10部雷达数据进行方法应用和评估。结果表明:稳定大气条件下，3种型号雷达 (LC，PB，PA) 的有效数据高度分别达到3，6 km和10 km的雷达系统功能设计需求。经空间均匀性检验与时间一致性平均处理的风数据在降水期间质量优于业务雷达数据，3—5月10部雷达获取的两组u，v分量离差标准差约为1 m·s-1，表明经过空间一致性检验和时间一致性平均处理后的数据质量较好。

Quality Factors and Processing Algorithm for Wind Profiling Radar Data

In recent years, wind profiling radar (WPR) network in China is under rapid development. To take advantage of the network measurements in weather analysis and numerical prediction, it's of great significance to make full aware of quality factors and improve the current processing algorithm for WPR data.Many factors affect the quality of horizontal wind data detected by WPR, especially system error and meteorological background. According to five-beam WPR, a new method for examining system error from radar Doppler measurements is proposed. As for meteorological background, wind filed is assumed homogenous when it is detected by WPR, and the accuracy of horizontal wind data will decline when the assumption is not satisfied. During the period of precipitation, scattering caused by raindrops is much stronger than turbulence detected by WPR. And the assumption of homogenous wind breaks down easily for the cause that fall terminal velocity of precipitation particles changes rapidly in space when convective precipitation happens, which is a significant problem for WPR data quality control algorithm.However, two independent wind profiles can be measured with a five-beam WPR and differences between measured zonal winds and meridional winds can reflect errors caused by the inhomogeneity of wind field. In order to reduce such errors, all observations are examined to make sure data detected under circumstances where the wind filed is extremely inhomogenous are deleted. Besides, different averaging methods, such as consensus average and simple average, used to calculate hourly averaged winds also affect the accuracy of it and comparisons are conducted on two averaging methods.Combined with 10 radars of Guangdong WPR network, evaluation of the new methods for processing basic data is analyzed from March to May in 2014. Results show that 10 radars in Guangdong WPR network, including 8 boundary radars (LC), 1 troposphere radar Ⅰ(PA) and 1 troposphere radar Ⅱ(PB), meet the designed requirements respectively in terms of the maximum height of credible data in clear air, which is 3 km for LC radar, 6 km for PB radar and 10 km for PA radar. Furthermore, there are no large system errors in 10 radars except that the examining consequence is unsatisfactory during 1-2 km for PA radar. It is necessary to consider the atmospheric inhomogeneities that may cause great errors especially when it rains heavily, and consensus averaged wind is superior to simple averaged wind in median and high heights. Therefore, an improved algorithm according to examination of atmospheric inhomogeneities and consensus average is proposed to obtain hourly averaged winds. It is proved that winds obtained from the improved algorithm show better representation than the currently used data during precipitation, as the stand deviation of differences between two independent measured zonal wind values and meridional wind values are both close to 1 m·s-1.