Improving the Processing Algorithm of Beijing MST Radar Power Spectral Density Data

Abstract

Beijing MST radar is a unique large instrument for atmospheric dynamic structure detection in Chinese Meridian Project. It plays an essential role in the in-depth understanding of the vertical structure of atmospheric wind, waves and turbulence in the troposphere, lower stratosphere, mesosphere, and lower thermosphere in North China. Since the completion of Beijing MST radar in 2011, wind data have been well acquired. However, there is still a need for improving the extraction of some elements. To achieve this goal, the power spectral density data processing algorithms is improved mainly from two aspects including accurate noise level estimation and target signal recognition. The improved algorithm derived data, radar products, radiosonde data and ERA5 reanalysis data from 1 January to 31 December in 2012 are statistically analyzed and compared. A log-linear fitting scheme is put forward and applied to realize rapid implementation of objective determination of the noise level. The root mean square error(RMSE) of noise values between the log-linear fitting and the conventional scheme is about 0.43 dB and mean values are 168.6 dB and 168.5 dB, respectively. Results show that the noise level estimation can be fast and accurate using the log-linear fitting scheme. Based on the property that atmospheric signals have spatio-temporal consistency and diffident signals show different spectral characteristics, the target signal can be accurately identified and extracted by the improved algorithms. The RMSE of zonal wind speed between the improved algorithm derived data and radiosonde data at different height are in the range of 2-3 m·s-1 while the RMSE of zonal wind speed between radar products and radiosonde data at different height are 3-4 m·s-1. Moreover, the mean value of the spectral width derived by the improved algorithm is 2.5 m·s-1, which is less than the mean value of radar products. Under precipitation weather condition, the mean bias and RMSE of horizontal wind speed between the improved algorithm derived data and radiosonde data at different height are both less than values between radar products and radiosonde data. Results show that the improved algorithm can reduce non-atmospheric signals such as noise and intermittent clutter and effectively suppress signals caused by precipitation. Thus the effectiveness and reliability of the improved algorithm are verified, and it is relatively easy to implement.

Keywords

• power spectral density data processing
• log-linear fitting scheme
• accurate identify of the target signal