Signal Processing for Performance Improvement of Phased Array Weather Radar

In this presentation, we will show a new signal processing methods, which is a minimum mean square error (MMSE) beamformer method, for a phased array weather radar system. And also, the MMSE beamforming method is applied to the observation data from the X-band Phased Array Weather Radar (PAWR) at Osaka University. The PAWR has achieved three-dimensional precipitation observations in less than 30 second and within the range of 60 km. Fourier beamforming (FR) method is used in the elevation angles, and achieves a further elevation beamwidth reduction to 1.2° by all 128 antenna elements for receiver beamforming. The fan beam, which is used as transmitted waves and has the beam width of 5°, is useful to support rapid scanning. On the other hand, the received signals with the PAWR tend to be affected by the influence of ground clutter. The PAWR has been operating in urban areas, and covers the central part of the City of Osaka. Clutter echoes have much influence on the observations in high elevation angles. Consequently, there is a possibility of the overestimation of the precipitation echoes by the clutter echoes.

Some of the DBF methods (FR, and MMSE) were applied to the PAWR at Osaka University, to evaluate their effect in ground clutter suppression. The DBF methods were applied to the IQ signals recorded by the PAWR. The IQ signal was obtained by each antenna element, pulse, azimuth sector, and range bin information. In addition to the use of the mentioned DBF methods, a two-step correction method was proposed, to correct missing samples or multichannel signals with spikes caused by mechanical problems. The correction method was used as a pre-processing procedure before applying MMSE. The resulting scheme was termed “MMSE with pre-processing.”

The proposed correction procedure was shown to be capable of effectively handling those spikes in both domains, which are a phase and amplitude. From the comparison between the results of the different DBF methods, FR overestimated the received power at low elevation angles less than where strong clutter exists, because of the high sidelobe level. The MMSE approach clearly detected the clutter and suppressed its effect at low elevation angles (<15) much more effectively than FR. However, the clutter influence remained at elevation angles higher than , a range of angles where the results obtained with MMSE were equivalent to the ones obtained with FR. This remaining clutter was caused by the mentioned large errors in the recorded phase and amplitude measurement for some channels. The sidelobe suppression in the clutter direction is not enough to effectively eliminate the clutter at high elevation angles. Finally, the MMSE with pre-processing was shown to be very effective in mitigating the clutter at all elevation angles, because of its better suppression of the sidelobes at low elevation angles; it is therefore superior to the other analyzed DBF methods as a scheme for clutter reduction. The pre-processing worked to calibrate the phased array antenna using clutter echoes.

We also applied the DBFs to the precipitation data with the PAWR. As a results, we found that MMSE with pre-processing is also effective for the precipitation data even if the strong clutter echoes exist. The obtained results indicate that the pre-processing method significantly improved the clutter performance of MMSE. This presentation should be useful for improvement of the performance of the PAWR observations.