Demonstration of Super-resolution Measurements with Phased Array Radar

Conventional weather radar with a dish antenna has an effective beamwidth in the scanning direction larger than the intrinsic beamwidth because during the dwell time the beam position changes, thus illuminating more of the volume in the lateral dimension. The dwell time is typically set so that the radar variables are estimated from data obtained within a beamwidth in azimuth and this yields a degraded resolution with an effective beamwidth about 40% larger than the intrinsic beamwidth. Because high resolution measurements are desirable, meteorologists have devised scanning strategies to produce a smaller effective beamwidth than the one corresponding to sampling in azimuth at a beamwidth apart. On the WSR-88D such finer spacing of radials is termed “super-resolution”. It uses a heavily tapered window for the time series data over one beamwidth but overlaps the results at one half the beamwidth. Although a similar approach can, in principle, be applied to the multifunction phased array radar (MPAR), it is impractical and not matched to the preferred scanning mode of phased array radars. Therefore a method is proposed in a companion paper (Zhang et al. 2015) that is suitable for phased array weather radars for collecting data with a stationary beam at a discrete position in azimuth. In this paper, the method is demonstrated with data collected by phased array radar: the national weather radar testbed (NWRT).

Time series data were collected at azimuthal spacing of half a beamwidth apart in a sector that covers 90o to test this concept. The data from three adjacent radials are combined to obtain the spectral moments. Because the pulse repetition time (PRT) was 800 μs and the number of pulses per radial was 40, it is possible to mimic a typical WSR-88D Doppler scan. Using every fourth sample, a surveillance scan (for reflectivity at the lowest elevation) of the super-resolution mode was emulated. Examined are weighted averages of moment estimates from adjacent radials as well as joint signal processing from these radials. The dwell time per radial of processed data is chosen to match the current super-resolution mode on the WSR-88D. Furthermore, a slightly modified CLEAN-AP ground clutter filter (Torres and Warde 2014) is applied and evaluated with the data. A data set from a supercell tornadic storm is also examined. A radial of this data consist of two batches of pulses; one batch has 12 samples at a PRT of 3.0 ms and the other has 25 samples at a PRT of 0.8 ms. A combination of time series as well as a combination of second order moments from adjacent radials is used to estimate the reflectivity and mean Doppler velocity. It is demonstrated that acceptable ground clutter canceling can be achieved on data from either of the two PRTs. Comparisons of results obtained with various window functions and total dwell times are made.

References: Torres, S., and D. Warde, 2014: Ground clutter mitigation for weather redar using the autocorrelation spectral density. J. Atmos. Oceanic Technol., in press. Zhang, G., D.Zrnic, L. Borowska, and Y. Al-Rashid, 2015: Hybrid Scan and Joint Signal Processing for a High Efficiency MPAR. 31st Conference on Environmental Information Processing Technologies, American Meteorological Society Annual Meeting, 2015, Phoenix, Az.