Extreme Analysis Differences between the ECMWF and GFS Models

At the National Centers for Environmental Prediction (NCEP) we have been studying the analyses differences between the Global Data Assimilation System (GDAS) and the European Center for Medium Range Weather Prediction (ECMWF) in order to track source regions of errors that grow throughout the forecast period and produce low skill 5-day forecasts. In order for researchers to understand the source regions that could lead to large forecast errors we developed a system that monitors the extreme analysis differences between the GFS and ECMWF. This study examines the extreme analysis differences for temperatures, winds, and heights from 00Z 14 May 2012 through 18Z 18 June 2012. The goal of this study is to locate the predominant regions and layers of the atmosphere that produce the highest number of extreme analysis differences. Results show that the southern hemisphere had the most extreme differences for winds, temperatures, and heights. The jet stream level (< 400 hPa) of the atmosphere had the most extreme analysis differences for height and winds, while the lower atmosphere had a significant number of extreme analysis differences for temperature. Overall this study is useful to identify potential analysis biases so that NCEP can research these areas and eventually lead to improvements of the GSI analysis and GFS model.