Enhanced Moment Estimation of OU-PRIME Through Advanced Signal Processing

The quality of moment data is essential to various radar algorithms used for weather detection, estimation, and forecasting. The improvement of moment estimation would enhance the performance of those algorithms. In addition to sampling error, clutter and noise are main factors that affect the data quality. Ground clutter can be removed by the conventional filtering techniques (e.g., IIR filter) if the weather signal has a non-zero radial velocity. However, weather signal could be corrupted as well if the radial velocity is close to zero (e.g., when the storm moves perpendicularly to the radar beam). Recent studies have shown that spectrum-based methods have great potential to improve clutter filtering [e.g., Gaussian model adaptive processing (GMAP) algorithm] but there remain issues to be resolved for polarimetric radar applications. Noise can significantly degrade data quality, especially for polarimetric radar data when SNR is low (e.g., <10 dB) Improving moment estimation by removing both clutter and noise is highly desirable.

Recently a state-of-art research weather radar, OU-PRIME (Polarimetric Radar for Innovations in Meteorology and Engineering), has been built on the research campus of University of Oklahoma (OU). Advanced signal-processing algorithms are currently under development for OU-PRIME. This paper presents the preliminary results of the algorithms. The statistical characteristics (i.e., phase and amplitude of dual-channel) of weather signal, clutter and noise are analyzed in both the frequency and time domains. Clutter is identified and removed in spectrum while noise is reduced by excluding the zero-lag of auto-correlation function. Preliminary results show the combined frequency-time processing has great potential to improve the moment estimation for polarimetric weather radar.