论文部分内容阅读
新一代天气雷达VAD（Velocity Azimuth Display）风作为多层次水平风矢量的垂直廓线,具有时间频次高、易与业务数值预报模式分辨率相匹配、影响质量因素复杂等特点。在数值预报中有效同化该资料,可改善低空风场资料不足等状况。本文根据变分同化中对资料质量的特殊要求,集成现有VAD质量控制方案后,设计了逐步拟合质量控制方案。再根据VAD资料不同误差来源,综合集成了现有的VAD质量控制方案,形成逐级综合质量控制方案。通过分析逐级质量控制方案得到VAD资料在同化中的适应性,发现该资料相对背景场存在偏差。进一步分析表明,出现这种偏差的原因在于选择用以反演VAD径向原始数据中存在许多不符合VAD廓线的数据。据此设计了逐步拟合质量控制方案,该方法可逐步筛选出用以反演VAD的径向速度数据,提取出数据中的简谐波特征。通过北京站雷达获取的VAD廓线与分钟级探空实测对比,发现通过逐步拟合质量控制得到的风速更接近实况。相对背景场偏差分析也表明,通过逐步拟合质量控制,观测相对于背景场的偏差得到有效修正。一个月的连续试验结果显示,质量控制后获取的VAD资料可基本满足同化中对资料的无偏要求。“,”Developing the multiple iterative fitting quality control method for Velocity Azimuth Display(VAD)in order to meet the variational assimilation requirements.The radar VAD wind is mean horizontal wind at different heights with the characters of high time-frequency,suitable for applications in Numerical Weather Prediction model(NWP)and complex error source.Made good use of the VAD data in NWP,it could compensate the less wind information in the low level.However,whether it will play apositive impact or not in data assimilation,the key problem is the VAD data quality control.In this paper,the VAD data adaptability for assimilation using the traditional quality control method is analyzed firstly.A series of comprehensive quality control is carried out for different error sources.The result shows that the VAD data has large negative bias comparing with the background.Then by diagnosing the problem,it is found that many data do not meet the hypothesis among the raw radial velocity data used to retrieve VAD wind.In order to resolve this problem,the multiple iterative fitting quality control scheme is developed.The results show that the method could purify the data gradually to satisfy the harmonic characteristics.Compared with the sounding data,it is more accurate than without the multiple iterative fitting quality control.Furthermore,the negative bias compared with the background is also corrected.Case and monthly continuous test reveals that the retrieved VAD data could meet the requirements of quality control for data assimilation.