| description abstract | Analyses and predictions of explosive cyclogenesis over the western North Atlantic Ocean during the 1987/88 cold season were compared. The analyses were the manual and automated series produced at the National Meteorological Center (NMC). The forecasts were those produced by the nested grid model (NGM) and the ?aviation run? of the global spectral model (AVN) at NMC, and also by a simple checklist employed by the National Weather Service (NWS) Forecast Office, Boston. Skill of the forecast has evidently improved since the preceding year. Probability of detection of an event in a specified 24-h period, with the manual analyses used as verification, approached 72% for the NGM in the range of 0?24 h with a false alarm rate of 17%. In the range of 36?60 h, the values for the AVN forecasts were 42% and 30%. When the automated analyses were used for verification, forecast performance was somewhat better. The accuracy of the checklist forecasts was comparable to that of the AVN forecasts but not as good as that of the NGM predictions, in the small sample available for comparison. Deepening in the NGM forecasts over the range of 12?24 h was 2 mb less than in the manual analyses, with a correlation of 0.55. The accuracy was limited mainly by errors in timing, with the model failing to well represent the initial analyzed deepening but catching up later. The automated analyses displayed a similar failure, with a correlation of 0.49 between analyses. Uncertainty in initial analysis is a major factor limiting present accuracy, especially at short range. | |
