Cooperative Institute for Research in the Atmosphere

Preliminary Investigation of Ensemble Forecast Sensitivity to Observation Impact with GFS and RRFS

Forecast Sensitivity to Observation Impact (FSOI) helps to quantify the impact of assimilated observations on forecast skills in numerical weather prediction systems. At NOAA, an ensemble FSOI (EFSOI) package is available within the Gridpoint Statistical Interpolation (GSI) and the ensemble Kalman filter (EnKF) data assimilation system (i.e., GSI-EnKF).  This tool has been upgraded and implemented into the Global Forecast System (GFS) v.16 (operational since March 2021) and the Rapid Refresh Forecast System (RRFS) developing version (the successor of the current operational Rapid Refresh and High-Resolution Rapid Refresh systems).  There are differences in the practices of EFSOI between the GFS and RRFS.  One of the obvious differences is the cycling frequency.  The global model cycles and performs the EFSOI calculation every six hours, while we run regional ensemble analyses and EFSOI calculations using shorter frequencies (i.e., 1 or 3 hours).  The EFSOI uses the 24-hour global (and 3- or 6-hour regional) ensemble forecasts as inputs to estimate the observation impact at the analysis time.  We quantify the EFSOI of various types of data (e.g., in-situ, radiosonde, and satellite data), and a preliminary investigation of global and regional cycling ensemble data assimilation experiments will be presented.