P10 GSI background error sensitivity for AFWA pre-operational implementation testing

Newman, Kathryn, National Center for Atmospheric Research, Ming Hu, National Oceanic and Atmospheric Administration, and Hui Shao, National Center for Atmospheric Research

The Air Force Weather Agency (AFWA) is in the process of implementing the Gridpoint Statistical Interpolation (GSI) 3D-Var system into their operational configuration in 2013. The Developmental Testbed Center (DTC) is conducting tests within a functionally similar environment to the AFWA pre-operational GSI system to help determine the best operational configuration. Two parallel real-time systems, as well as multiple short retrospective tests were run at the DTC to uncover issues with the AFWA pre-operational set-up and test individual changes. A drop in forecast skill was identified due to a change in the Background Error (BE), which led to sensitivity studies of various operational BEs including the North American Mesoscale (NAM) model, the Global Forecast System (GFS), and the Rapid Refresh (RR). Further sensitivity tests were conducted on a domain-specific BE generated from 3 months of real-time forecasts, and tuned versions of the domain-specific BE. Results showed that use of the NAM BE resulted in the highest forecast skill within the AFWA pre-operational domain. The BE factors and analysis fit of the NAM BE were investigated to determine the characteristics that lead to better forecasts. Additionally, further regional BE studies were conducted to investigate the behavior of domain-specific regional BEs generated using the Ensemble Perturbation method compared to the National Meteorological Center (NMC) method.