A Dynamical Approach to Seasonal Prediction of Atlantic Tropical Cyclone Activity
Abstract Analysis of ECMWF reanalyses and operational analyses covering the period between 1979?98 has confirmed that seasonal Atlantic tropical cyclone activity is strongly and negatively correlated with the observed vertical wind shear present in the main development region (MDR) between July and September. In 1983 and 1995, the least active and most active tropical cyclone years, respectively, anomalous shear was shown to be present in spring and to persist throughout each of the tropical cyclone seasons. While monitoring of MDR shear is recommended for highlighting the risk of such extreme events, the springtime MDR shear is not generally a good indicator of shear in the summer months. Seasonal forecasts of MDR shear made with the U.K. Met Office (UKMO) atmospheric GCM (AGCM) and observed SSTs for the years 1979?97 have been analyzed. The model possesses potential skill for predicting the MDR shear as determined by a consideration of the ensemble mean shear variability and an evaluation of the relative operating characteristics (ROC). The ROC analysis indicates high probabilistic skill, in particular for anomalously low shear events. Analysis of seasonal forecasts of MDR shear made with the UKMO AGCM with persisted SST anomalies for the years 1979?97 was also performed. Skill in predicting MDR shear is reduced but still significant. ROC analysis indicates probabilistic skill for the anomalously low shear events, which may be useful for some applications. Based on this work, the authors conclude that a dynamical approach to the seasonal forecasting of Atlantic tropical cyclone activity, which combines predicted MDR shear with a statistical model should be developed.