Impact of Assimilating Multiple-radar Data Through The GSI System On Numerical Prediction of Tropical Storm Erin (2007)

Abstract

A rare event occurred over Oklahoma in August 2007 when Atlantic tropical storm Erin (2007) re-intensified over western Oklahoma three days after making a landfall. The storm re-developed an eye, an eye wall structure and spiral rain bands after weakening significantly over western Texas, producing strong winds and heavy flooding that claimed several lives and caused extensive property damage. Being over land and covered by the operational WSR-88D radar network, this event provides a good opportunity to test the assimilation of both radial velocity and reflectivity observations of operational radars into high-resolution NWP models, for the prediction of such storms of tropical nature, including the prediction of track, intensity, structure and precipitation. Although not planned for the initial operational implementation of the High-Resolution Rapid Refresh (HRRR) system (Benjamin et al. 2009, this meeting), we look further ahead to consider the use of the NCEP GSI (Grid-point Statistics Interpolation) 3DVAR system at storm-scale resolution. In our work, the GSI is enhanced and used to assimilate level-II radial velocity data, and mosaic reflectivity data produced by the National Severe Storms Lab (NSSL). The radial velocity data are variationally analyzed directly with GSI and the reflectivity data are assimilated within the GSI framework using a complex cloud analysis package based on that of the ARPS Data Analysis System (ADAS). Thirty-minute assimilation cycles of up to 6 hours are performed on a large 3-km grid, using the WRF-ARW model. An ARPS radar data processing package is adopted to prepare the radial velocity data, and an ARPS-based radar emulation package is adopted to allow direct comparison of the model forecast against radar observations. The operational 12-km NAM analyses provide the boundary conditions for the 3 km grid and the initial guess to start the radar data assimilation cycles. The forecasts are compared against that starting from the operational NAM analysis directly. Significant improvement is found with the radar data assimilation. The impacts of assimilating radar radial data or reflectivity data alone or in combination are examined. The scale of the background error covariance is tuned for optimal results when analyzing radial velocity data in GSI. The default scale used by conventional data as well as the reduced scale used in operational implementation of GSI for NAM analyses of radar data are found to be too large for the 3-km analysis. A 1-km grid is further nested within the 3-km grid during the forecast period. Detailed results will be reported in the extended abstract and at the conference.