Diagnosis of GFS Height, Wind and TC Motion Errors during 2012 North Atlantic Hurricane Season

Over the past five years, we have witnessed significant improvement in GFS hurricane track forecast skill for forecasts out to four days. The improvement in GFS hurricane track forecast from Day 5 and beyond is not as evident. In this study, we examine GFS systematic forecast errors in geopotential height and wind during the 2012 North Atlantic hurricane season that are relevant to hurricane track forecast errors. Evaluation of GFS Day 1 through Day 8 forecast averaged over the 92-day hurricane season (defined here as 1 August–31 October) shows a distinct distribution of height and wind errors. The GFS predicted troughs and ridges become progressively de-amplified with increasing forecast lead time. The weakening of subtropical high over the Atlantic with time is particularly evident early in the season. The GFS error field shows a tropospheric-deep anticyclone over the eastern U.S. by Day 5. This robust anticyclone in the GFS wind error field appears to be associated with a wave train emanating from the tropical east Pacific. Since this major anticyclonic circulation is a robust feature of GFS systematic error, the motion of storms in western Atlantic and eastern Gulf of Mexico are likely affected. To gain further insight on GFS hurricane track forecasts, we perform a detailed analysis of GFS forecast errors for Hurricane Isaac and Hurricane Sandy. These detailed analyses illustrate how GFS forecast errors in sub-synoptic scale and mesoscale circulations can impact the track forecast of individual storms. Our preliminary analysis suggests that errors in the treatment of convection might be responsible for the excitement of the wave train emanating from East Pacific as well as forecast errors of sub-synoptic and mesoscale circulations that affect the storm track.