Rapid Refresh: version 2 upgrade at NCEP and work toward version 3 and the North American Rapid Refresh Ensemble (NARRE)

An upgrade to version 2 of the NCEP operational Rapid Refresh (RAP) weather prediction system was made in February 2014. This upgrade to the RAP (RAPv2) included a number of significant enhancements to both the data assimilation and model prediction components of the RAP. This NCEP operational version of the RAP will support an operational version of the High Resolution Rapid Refresh (HRRR) slated for NCEP implementation in fall 2013 and described in a companion presentation within this conference by Alexander et al. Work is also progressing on further enhancements to the RAP system that have been incorporated into a real-time experimental version 3 of the RAP (RAPv3), being run by GSD and planned for operational implementation at NCEP in Spring 2015.

For RAPv2, the most significant improvement within the data assimilation module was the upgrade from a traditional 3DVAR technique to a hybrid ensemble technique, using covariance information from the 80-member global ensemble data assimilation system. This change, enabled by global assimilation developments and facilitated by use of the community-based Gridpoint Statistical Interpolation (GSI) analysis package, has resulted in significant forecast improvement for most upper-air variables and forecast lengths. Additional assimilation enhancements include improved cloud analysis and soil temperature and moisture adjustment procedures. Within the WRF-ARW model component, the most significant enhancements include switching from the MYJ planetary boundary layer (PBL) to a specially adapted MYNN PBL scheme, an upgrade from a 6-layer to a new 9-layer version of the Smirnova (RUC) land surface model formulation (LSM), and a modified roughness length specification. There are also smaller upgrades to other model components, including the Thompson microphysics and the radiation schemes.

A number of model and assimilation upgrades have been made for RAPv3 and additional refinements are being evaluated. These include upgrades to the new Grell-Freitas cumulus parameterization, upgrade to a new Thompson scheme that incorporates aerosols, enhancements to the satellite radiance assimilation (cycled bias correction and more satellite data types assimilated), and improvements to the hybrid-ensemble assimilation parameters. Development work is also underway toward the North American Rapid Refresh Ensemble (NARRE) forecasting system, planned for NCEP implementation in 2017. This 6-member system will include 3 members each from the ARW and 3 NMMB dynamic cores and a common ensemble-based data assimilation system.

In the presentation, RAPv2 and RAPv3 enhancements and the associated forecast improvements will be described, with a brief discussion of the preliminary work toward the NARRE.