Extended Abstract (1.6M)P2.1
Improved forecast skill using pseudo-observations in the NCEP GFS
Jordan C. Alpert, NOAA/NWS/NCEP, Camp Springs, MD; and D. Carlis, B. Ballish, and K. Kumar
On approximately a monthly basis, poor forecasts or “Skill Score Dropouts” plague GFS performance. Other national center forecasts, for example European Centre for Medium-range Weather Forecasts (ECMWF), often do not exhibit this loss in skill. We explore initial condition differences when dropouts occur to define a climatology of these events. We attempt to quantify the differences between the GFS and ECMWF forecast models when there are dropouts, and define initial condition areas that have an impact on 5-d forecasts. ECMWF standard initial conditions are converted to simulated or “pseudo” RAOB observations and inserted into the Gridpoint Statistical Interpolation (GSI) to create a new analysis from which new forecasts are made and compared with operational forecasts. The results show improvement in 5-day skill scores in practically all Southern and most Northern hemispheric cases. Initial condition areas that influence the sensitivity of forecast skill can be found by creating a hybrid initial condition, selectively overlaying the ECMWF pseudo-analysis over the GFS initial condition. The regions used are “patches” over special areas, e.g., ambiguity in observation quality areas, or latitude/longitude bands to isolate problems that alter downstream 5-d forecasts. Statistics will be presented to show areas of meteorological action with an emphasis on the Southern Hemisphere. How these results fit with recent adjoint methods to derive the impact of observations on analysis will be discussed.
Poster Session 2, Observing Systems
Wednesday, 14 January 2009, 2:30 PM-4:00 PM, Hall 5
Next paper