Thursday, 2 July 2015: 2:30 PM
Salon A-2 (Hilton Chicago)
This paper studies the kinetic energy error growth in the National Centers for Environmental Prediction (NCEP) Global Forecast System (GFS) 0~192h forecasts over a region 0ºN~60ºN and 0ºW~180ºW, during a 92-day period from August to October in 2012. The 92-day and 10-day mean statistics are calculated for the analysis kinetic energy (KA), the forecast kinetic energy (KF), the error kinetic energy (KE), and the anomaly kinetic energy (KD, with anomaly defined as departure from the 92-day mean). Case-by-case statistics are also performed for Hurricanes Isaac (2012), Leslie (2012), and Sandy (2012) to examine the relationship between the hurricane track forecast skills and the accompanying kinetic energy error growth between different scales. The results show a systematic loss of kinetic energy (with less anomaly kinetic energy) in GFS forecasts and a systematic upscale growth of error kinetic energy with time. The 92-day mean errors of wind vector from day five to day seven are found to be flow-dependent and out-of-phase with synoptic scale circulations; wind vector error fields are cyclonic circulations over anticyclones and anticyclonic circulations over cyclones. This implies rapid error growth and error saturation on the synoptic scales. Following the study by Fang and Kuo (2015) on the scale predictability of the fractal atmosphere, we use the ratio of KE to KA as an indicator to differentiate good versus poor forecasts, and examine the relationship between GFS forecast kinetic energy error growth and the track forecasts of the three hurricanes on individual cases. We found that when GFS starts producing poor synoptic forecast (based on the kinetic energy error growth) from day five to day seven, hurricane track errors also increase sharply.
![](images/flag.png)
![](images/blue_ribbon.png)