Predictability of North Atlantic Tropical Cyclone Activity on Intraseasonal Time Scales
Abstract Recent work suggests that there may exist skill in forecasting tropical cyclones (TC) using dynamically based ensemble products, such as those obtained from the ECMWF Monthly Forecast System (ECMFS). The ECMFS features an ensemble of 51 coupled ocean–atmosphere simulations integrated to 32 days once per week. Predicted levels of TC activity in the North Atlantic Ocean with these monthly ensemble forecasts is compared with the observed variability during the months of June–October during 2008 and 2009. Results indicate that the forecast system can capture large-scale regions that have a higher or lower risk of TC activity and that it has skill above climatology for the Gulf of Mexico and the “Main Development Region” on intraseasonal time scales. Regional forecast skill is traced to the model’s ability to capture the large-scale evolution of deep-layer vertical shear, the frequency of easterly waves, and the variance in 850-hPa relative vorticity. The predictability of TC activity, along with the forecast utility of the ECMFS, is shown to be sensitive to the phase and intensity of the Madden–Julian oscillation at the time of model initialization.