Nicholas Leonardo and Brian A. Colle
School of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY
Despite overall improvements in medium-range (day 3-5) track forecasts of North Atlantic tropical cyclones (TC’s), some cases, such as Joaquin (2015) and Sandy (2012), can occasionally have track errors much larger than climatology. Meanwhile, results by Buckingham et al. (2010) showed that TC’s that underwent extratropical transition (ET) tend to have a significantly more negative along-track bias than non-transitioning cases. Thus, one question is whether large track error cases, particularly those with errors in the along-track direction, are associated with similar extratropical steering patterns that are prone to systematic model biases. How sensitive are TC tracks to these patterns and how early can their significance be traced back in the forecast?
This study verifies the 2008-2015 TC track forecasts of global ensembles, focusing on the ECMWF (51 members) ensemble for the North Atlantic. The NHC’s best-track data is used as the verifying analysis. The largest ensemble mean along-track errors from each day 3-5 forecast are analyzed. The forecasts are defined as “north” (“south”) cases if the verifying best-track (never) crosses north of 30N, thereby separating TC’s potentially undergoing ET at higher latitudes. For north and south cases separately, the top 20% most negative and most positive of these along-track errors are considered “slow” and “fast”, respectively. The model fields of each case are analyzed with standardized differences of the 10 “slowest” ensemble members and the 10 “fastest” ensemble members. In addition, the role of ET on the tracks of north TC's by using the cyclone phase space (CPS) diagnostic. Model fields are also evaluated against the CFSR reanalysis.
The along-track bias of the ECMWF is significantly more negative for north TC’s than for south. For north TC’s, the slow cases on average are associated with inherently stronger TC’s than the fast, though the intensity errors of slow and fast cases are comparable. From the CPS diagnostic of the slow north cases, the 10 slowest members tend to undergo ET later than the 10 fastest members 36 hours prior to the large mean track error. From vortex-relative composites, the slow north cases are associated with more amplified flow patterns than the fast north cases. Composited standardized differences imply that the slow north cases are under-amplifying upstream troughs and downstream ridges. The under-amplification of the ridge in the slow cases is driven by a significant under-prediction of the divergent outflow of the TC and its interaction with ambient PV gradients.