Adaptation and implementation of the Gridpoint Statistical Interpolation (GSI) for Rapid Refresh

The Global System Division (GSD) at NOAA/ESRL has developed a new high frequency data assimilation and forecast system known as the Rapid Refresh (RR). This system is scheduled to replace the NCEP operational Rapid Update Cycle (RUC) in 2010. The RR combines WRF-ARW model and the Gridpoint Statistical Interpolation (GSI) analysis system. The RR domain covers all of North America, a significant expansion compared to the CONUS domain coverage of the Rapid Update Cycle (RUC). As with the RUC, the RR will provide hourly updated situational awareness guidance for short-range forecast challenges, especially aviation (convection, icing, turbulence, ceiling and visibility), severe weather, and surface conditions. The domain expansion will provide a consistent set of hourly updated grids over the entire North American continent and adjacent regions (Puerto Rico and Caribbean).

The GSI was developed by NCEP in cooperation with other institutions and has been applied successfully in both global and regional operational forecast systems (GFS and NAM). Over the past few years, GSD has had a major effort to optimize GSI for application in the Rapid Refresh. This has included work to improve the compatibility with the WRF-ARW model and to incorporate specific analysis features from the RUC into GSI. An outcome of this work is enhanced capabilities for the GSI, including 1) a generalized Cloud analysis package to generate cloud and precipitation fields using METAR, satellite, radar observations; 2) a radar and lightning assimilation procedure to initialize ongoing convection based on reflectivity/lighting converted temperature tendency, and 3) use of RH as analysis variable for ARW background fields. Work is ongoing to incorporate special procedures to better use surface data (improved forward model and anisotropic error covariances). Finally, significant effort has been made to ensure that the GSI analysis works consistently with ARW model, in terms of physical options and background IO, resulting in an hourly cycled RR with performance similar to that of the RUC.

We will provide a description of these GSI developments for the RR application and illustrate their impact on the RR analyses and forecasts.