Initial evaluation of the National Weather Service's Dual Polarization, Legacy, and NSSL's Q2 QPE products

A major objective of the recent dual polarization (DP) technology upgrade to the Weather Surveillance Radar – 1988 Doppler (WSR-88D) system is to improve quantitative precipitation estimates (QPE). Using DP variables derived from vertical and horizontal reflectivity (ZV and ZH), different rain rates (R) are invoked. The traditional “Legacy” Precipitation Processing System (PPS) algorithm uses the single variable ZH.

An assessment was conducted of the performance of the WSR-88D DP and Legacy QPE algorithms as well as the experimental “Q2” QPE product from the National Severe Storm Laboratory's Multi-Radar, Multi-Sensor (MRMS) system. The Legacy algorithm has five R relations that can be changed at each Weather Forecast Office while Q2 uses a different precipitation process identification scheme to switch between four R relations at each grid point and volume scan.

The analysis period spans April through September 2012, limited primarily by the deployment schedule. The analysis includes storms primarily east of the Rocky Mountains with the exception of several events in southern Arizona. In total, 135 radar-storm events were examined within a 150 km radius of the radar in order to eliminate QPE artifacts from the DP Melting Layer Detection Algorithm. Ground verification includes 24 hour rain gauge data from the Hydrometeorological Automated Data System (HADS) gauge network, CoCoRaHS observers, COOP observers, and local mesonets such as the Oklahoma Mesonet.

Overall, DP QPE demonstrated a 19% improvement in RMS error over the Legacy QPE for all gauge amounts and a 23% improvement for gauge amounts greater than 2”. Q2 QPE was statistically equal to DP. Further, storms were grouped by imputed microphysical characteristics in order to provide insight to algorithm functionality. Again, DP outperformed Legacy for all storm types except for quiescent “pulse” storms with no hail. In addition, select cases of superior and inferior performance are described in detail.

While these results are encouraging, further analysis is needed that includes investigating performance using “continental” vs. “tropical” R relations within the DP algorithm, increasing reliance on the use of specific differential phase (KDP), and introducing a R relation based on specific attenuation, especially for areas of partial beam blockage. Additional efforts include monitoring and correcting for differential reflectivity (ZDR) bias errors, evaluating methods for mitigating the melting layer discontinuity, and conducting precipitation verification analysis on an hourly and sub-hourly basis. Finally, the MRMS “Q3” algorithm suite, that utilizes DP concepts, is introduced as an optimum solution for operational QPE.