Benefits from the Assimilation of Rapid-Scan Phased Array Radar Data for Storm-scale NWP: Optimal Configurations and Best Strategies

The past decade has seen the development of two major technologies that will help develop the capability for real-time prediction of severe convective storms. Since 2003, NSSL has been actively developing the Multifunction Phased Array Radar (MPAR), a potential next-generation operational weather radar system using phased array panels to enable complete volumes every 30-60 seconds (Heinselman and Torres, 2012). During the same period, data assimilation techniques have been extended from the synoptic and mesoscale to encompass nonhydrostatic scales, enabling more complete assimilation of radar observations.

In an observing system simulation experiment by Yussouf and Stensrud (2010), assimilating synthetic rapid-scan radar observations of a splitting supercell using an Ensemble Kalman Filter improved ensemble analyses and forecasts relative to assimilating radar data using emulated WSR-88D radar data. Demonstrating a similar impact using real phased-array radar data has proven more challenging. This requires separating forecast errors arising from the less frequent WSR-88D updates from those due to mesoscale background errors, model errors, and model imbalances generated with each new analysis. Experience has shown that these errors rapidly become large enough to inhibit differentiation of the impact of assimilating convectional versus rapid-scan radar data.

This presentation will focus on real-data analyses and forecasts of two large tornadic storms that occurred in central Oklahoma (May 2011 and 2013). Each dataset has a long time series (> 2 hours) of one-minute volume scans collected while the storm was within 100 km of the MPAR. A large number of storm-scale radar assimilation experiments have been conducted using both synchronous and asynchronous local ensemble transform Kalman filter (LETKF) data assimilation. We will present results from a broad cross-section of the assimilation parameter space (e.g., observation errors, “super-observation” resolution, covariance localization) as well as some unique data assimilation strategies that are enabled by the high temporal data density of the MPAR. With proper tuning of the assimilation system, rapid-scan radar observations do improve storm-scale analyses and forecasts relative to the current temporal density available from a optimally-tuned WSR-88D assimilation system.

References

Heinselman, P. L., and S. M. Torres, 2011: Hightemporal resolution capabilities of the National Weather Radar Testbed Phasedarray Radar. J. Appl. Meteor. Climatol., 50, 579–593.

Yussouf, N., D. J. Stensrud, 2010: Impact of Phased-Array Radar Observations over a Short Assimilation Period: Observing System Simulation Experiments Using an Ensemble Kalman Filter. Monthly Weather Review, 138, 517–538, doi:10.1175/2009MWR2925.1