Analysis of spaced-antenna measurement precision using an X-band active phased-array weather radar

Since the first use of meteorological radars (Probert-Jones (1960)), research on radar algorithms and technology have improved their performance in several ways. The recent advent of phased-array weather radars have enabled new capabilities such as beam multiplexing and assimilation/prediction over short time-scales (Yu et al. (2006), Zrnic et al. (2007),Yussouf and Stensrud (2010)). The focus of this work is the application of phased-array radar technology to probe the wind-eld of convective storms with high spatial-resolution using the spaced-antenna method.

The spaced-antenna (SA) method has been routinely employed to probe the convective boundary layer and the Ionosphere, but interest in applying this method to prole precipitation is relatively recent. Existing literature on spaced-antenna weather radars have been mostly theoretical and/or based on numerical simulations (Zhang and Doviak (2007)), Venkatesh and Frasier (2012)). Hardwick et al. (2005) have documented spaced-antenna observations earlier, but do do not attempt analysis and subsequent interpretation of results. To date, there have not been any empirical studies evaluating the precision of SA retrievals on precipitation echoes. Herein, we present a case study based on real data comparing spaced-antenna retrievals to fine-resolution Doppler Beam Swinging (DBS) and coarse-resolution Variational retrievals.

The comparison between SA and DBS retrievals is based on the hypothesis that they both measure the same quantity, vx + r0 sx , where vx is the wind- speed along the spaced- antenna baseline, r0 is the range to resolution volume and sx is the shear of radial velocity along the spaced-antenna baseline. This hypothesis is derived using simple mathematics and is tested on close-range observations of a winter storm at 20 degrees elevation. We show that the X-band spaced-antenna results are in qualitative agreement with Doppler beam swinging results. Quantitatively, the correlation between the spaced-antenna and fine-resolution Doppler beam swinging retrievals is 0.57. This resultant correlation coefficient is, in part, due to the high standard deviation of the fine-resolution Doppler beam swinging retrievals. A Monte-Carlo based phased-array spaced-antenna simulator developed in Venkatesh and Frasier (2012) is used to interpret relative quanititive biases between SA and DBS retrievals.

The comparison of spaced-antenna and variational retrievals revealed a correlation of 0.4. The drop in correlation coecient (as compared to the DBS comparison) may be attributed to the differing retrieval resolutions and radial velocity shear. We also find that the highest wind-speeds (reported by the Variational method) depart significantly from corresponding spaced-antenna retrievals. However, the lack of ground truth independent of the radar of interest complicated further analysis of precision of either of these algorithms. An experiment to evaluate spaced-antenna and variational retrievals using dual-Doppler analysis from two mechanically scanned radars is planned to address this concern.