Atmospheric Radar Research Center

The WSR-88D radar has proven to be a vital instrument for meteorological and climatological applications. It has been shown that the warning lead time for severe weather has been significantly improved after the installation of a national network of WSR-88Ds [Polger et al., 1994]. The current WSR-88D system completes a volume coverage pattern (VCP) with a minimum time of 5 min, in which 14 elevation steps with a maximum elevation angle of 19.5 ° are employed [Crum and Alberty, 1993]. However, a faster scan with more complete elevation angle coverage is often required for rapidly evolving convective storms [e.g., Carbone et al., 1985].

Recent results from mobile radar has shown that the dynamics and structure of tornadoes can vary significantly over a few minutes [e.g., Alexander and Wurman, 2005; Bluestein et al., 2003]. In addition, Shapiro et al. [2003] have shown that single Doppler wind retrieval can be improved using data with rapid update on the order of 1 min. Therefore, rapid scan is needed not only to increase the warning lead time but also to advance the understanding of quickly evolving weather systems. An intuitive way to decrease the scan time for a given VCP is to increase the antenna rotation rate. However, the statistical error of the three spectral moments will increase due to the decrease in the number of available samples for processing.

A phased array radar with agile beam steering is an ideal candidate to address the fundamental limitation to achieve the goal of rapid scan and high data accuracy through the collection of independent sample pairs [Zrnic, 1977]. During the revisit time, the radar will scan other regions to maximize the usage of the radar resources [Smith et al., 1974] and for multi-function tasks, such as aircraft tracking. As a result, the radar beam will be multiplexed over a designated region to provide an equivalent number of samples such that relatively short scan time is needed to achieve the required data accuracy. This approach is termed beam multiplexing (BMX) [Curtis, 2002; Orescanin et al., 2005; Orescanin, 2005]. Preliminary statistical results of BMX implemented at the NWRT are presented in Figure 5.

The standard deviation (SD) of the reflectivity and velocity estimates at one radial from both BMX and step scan (SS) are denoted by red and blue lines, respectively. Additionally, the mean SNR over 20 realizations are superimposed and denoted by a black line. The theoretical SD for both BMX and SS was also fitted to the observational data and are denoted by green and cyan lines, respectively.
The statistical errors obtained by BMX are smaller than those by SS. The result indicates that rapid update can be achieved using BMX with the same quality current required in the WSR-88D.

It is evident that BMX can provide more accurate reflectivity and velocity estimations up to approximately 90 km, past where the SNR decreases below 10 dB. Note that the experiment was designed such that the scan time of a 28° sector is the same (1.792 s) for both BMX and SS. In other words, rapid scan can be achieved without compromising the accuracy of the estimates using BMX. In this experiment, it was found that the update rate could be increased by a factor of approximately two, on average.

The framework of BMX was first established at NSSL by Curtis [2002]. Through a collaboration between NSSL and OU, implementation and preliminary verification of BMX at NWRT has recently been reported by Orescanin et al. [2005] and Orescanin [2005]. The team proposes to leverage on the work being accomplished collaboratively by OU and NSSL to thoroughly study BMX and to integrate this concept with the whitening based technique (WBT) [Torres and Zrnic, 2003; Torres et al., 2004] to provide an optimal solution for rapid scan. Moreover, the revisit time in BMX can be managed for other tasks such as aircraft tracking, wind profiling, etc., which can be used to demonstrate the multi-function capability of PAR.

To provide meteorologically convincing evidence of the PAR’s capability for rapid scan and high data quality, the case of a simulated tornadic storm from the ARPS numerical model [Xue et al., 2003], will be used for the radar simulator. Realistic Level-I data from both BMX and conventional sampling scheme will be incorporated into the tornado detection algorithm [e.g., Mitchell et al., 1998; Wang et al., 2005] to demonstrate and quantify the increase of warning lead time using BMX.

It has been shown that BMX can provide an increase of scan rate with a factor of two for small spectrum width and medium to high SNR. Orescanin [2005] has shown preliminary results that further improvement in update time is feasible when BMX and WBT are implemented simultaneously. The BMX and WBT are complementary and were designed to increase the equivalent number of independent samples in sample time and range time, respectively. Simultaneous implementation of both techniques can provide near-optimal reduction of statistical error in the estimation.

Clutter filtering in the current BMX configuration is severely limited since only two correlated samples are used to estimate the spectral moments. In order to achieve more effective clutter suppression, more than two uniformly spaced pulses could be transmitted. The reduction in variance, however, will be limited through this process. Therefore, a generalized statistical investigation of BMX is warranted. The trade off between scan rate and data quality will be studied theoretically and verified using numerical simulations. It is also worth noting the potential of the application of model-based spectral processing for moment estimation given a limited number of samples provided by BMX.