Optimal Temporal Frequency of the Next Generation Phased Array Radar Observations for Storm-Scale Data Assimilation

One goal of the cross-agency Spectrum Efficient National Surveillance Radar (SENSR) program is to evaluate the feasibility of the next generation multi-function surveillance radar network in order to replace the aging operational radar infrastructure currently in place. A potential replacement candidate to be employed is the phased array radar (PAR) system. Current operational weather surveillance Doppler radars (WSR-88Ds) take ~5 min or more to produce a full volumetric scan of the atmosphere, but PAR technology allows for full volumetric scanning of the atmosphere every ~1 min. However, the new PAR system will have to serve as a multi-function radar (i.e., MPAR), so full volumetric MPAR data may be temporally limited due to serving other purposes. How this temporal limitation might affect NSSL’s Warn-on-Forecast (WoF) system is mostly unclear. Radar data assimilation is critical for the WoF system, which is a rapid update-cycle storm-scale ensemble data assimilation and forecast system for high-impact severe weather. As part of the SENSR and WoF programs, this study’s goal is to determine the optimal temporal frequency of MPAR observations for storm-scale data assimilation.

On 31 May 2013, a deadly tornado and flash flood event occurred in parts of central and eastern Oklahoma. NSSL’s National Weather Radar Testbed (NWRT) MPAR in Norman, OK started scanning this event more than an hour before the first (and strongest) tornado developed and scanned most of the tornadic supercell’s evolution. Before assimilating these radar observations, a multi-scale (15- and 3-km) GSI-EnKF data assimilation system is completed using WRF-ARW to provide initial and boundary conditions for the MPAR data assimilation experiments, which are conducted on a 1-km grid. Only conventional observations are assimilated onto the multi-scale grids to create the mesoscale background. After a 45-min spin-up period on the 1-km grid starting at 2100 UTC, available MPAR data are assimilated from 2145 UTC to 2300 UTC every 5 min using CAPS’s 4D-EnSRF data assimilation system. Using the final ensemble analysis, short-term ensemble forecasts are produced. Data assimilation experiments are performed using every 1-, 2-, and 5-min volume of MPAR data. The latter experiment is equivalent to currently assimilating WSR-88D data. The results from subjectively evaluating and objectively verifying the ensemble analyses and forecasts against MRMS-derived products, such as reflectivity and azimuthal shear, will be presented.