测风激光雷达与风廓线雷达的探测性能评估及数据融合

测风激光雷达和风廓线雷达作为L波段探空测风的有效补充，均可以提供高时空分辨率的大气风场信息，然而由于工作原理和适用条件存在明显差异，在探测性能上各有优缺点，单一设备的探测数据已不能满足精细化预报的要求。本研究使用2020年1—5月北京南郊观象台的L波段探空资料对同址观测的测风激光雷达和风廓线雷达进行了数据质量评估，结果表明测风激光雷达与探空的一致性较高，U、V分量的相关系数分别为0.97和0.98，均方根误差分别为1.1和0.95 m·s-1，然而在2 km以上数据获取率较低且偏差较大；风廓线雷达与探空相比，U、V分量的相关系数分别为0.94和0.93，均方根误差分别为2.94和2.91 m·s-1，风廓线雷达的探测距离虽然更远，但在0.5 km以下和6 km以上的测量偏差较大。考虑到两种测风雷达在不同探测高度上的性能优缺点，提出分段曲面拟合法对两者的水平风资料进行融合处理，并选取个例对融合效果进行验证，结果表明，融合后的风廓线与融合前相比，风向和风速的一致性均得到明显提升。

Detection Performance Evaluation and Data Fusion of Doppler Wind Lidar and Wind Profiling Radar

As the effective supplements to L-band radiosonde, both the Doppler wind lidar and wind profiling radar can provide atmospheric wind field information with high spatial and temporal resolution. However, due to the obvious differences between Doppler wind lidar and wind profiling radar in terms of working principles and applicable conditions, the two wind measuring instruments have their own advantages and disadvantages in detection performance, and the detection data of a single instrument can no longer satisfy the requirements of refined forecasting. In this study, L-band radiosonde data from the Nanjiao observation site in Beijing from January to May 2020 are used to evaluate the observational data quality of the Doppler wind lidar and wind profiling radar at the same site. The results show that the Doppler wind lidar has a high consistency with the L-band radiosonde. The correlation coefficients of U and V components of Doppler wind lidar and L-band radiosonde are 0.97 and 0.98, respectively, and the root mean square errors of U and V components of Doppler wind lidar and L-band radiosonde are 1.1 and 0.95 m·s^-1, respectively. However, when the detection height is above 2 km, the data acquisition rate of Doppler wind lidar is low and the deviation is large. Compared to Doppler wind lidar, the consistency between the wind profiling radar and L-band radiosonde is slightly worse. The correlation coefficients of U and V components between the wind profiling radar and L-band radiosonde are 0.94 and 0.93, respectively, and the root mean square errors of U and V components between the wind profiling radar and L-band radiosonde are 2.94 and 2.91 m·s^-1, respectively. Compared to Doppler wind lidar, wind profiling radar has a longer detection range but a lower temporal resolution, in addition to a larger measurement bias below 0.5 km and above 6 km. Considering the advantages and disadvantages of the doppler wind lidar and wind profiling radar at different detection heights, the horizontal wind measurements from the two wind measuring instruments are first spliced and fused, and then the segmented surface fitting method is used to correct the outliers and compensate for the missing values of the fused wind profiles. Two cases are selected to validate and analyse the effect of fusion. The results show that the consistency between the wind direction and wind speed of the fused wind profile and the L-band radiosonde is significantly improved compared to the measurements of the individual devices before the fusion process.