Statistical-dynamical seasonal prediction of tropical cyclones affecting New York State

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 7 January 2015: 1:30 PM
122BC (Phoenix Convention Center - West and North Buildings)
Edmund K. M. Chang, Stony Brook University, Stony Brook, NY; and H. M. Kim

Tropical cyclones bring about destructive weather such as high winds, heavy rain, and storm surge, and can cause significant losses of life and property. Recent storms such as Irene and Sandy remind us that the heavily populated northeastern United States is an area pronged to be affected by tropical cyclones. Thus more skillful prediction of tropical cyclone activity is important to allow emergency management to be better prepared for mitigating their impacts.

As part of the effort of the New York State Resiliency Institute for Storms and Emergencies (NYS RISE), a novel statistical-dynamical seasonal prediction model has been developed to forecast the tropical cyclone activity affecting New York State during the upcoming hurricane season. Physical basis of the model related to large scale anomalies of the atmosphere-ocean system will be discussed. The model makes use of observations and Climate Forecasting System version 2 (CFSv2) model predictions of the large scale conditions which are then statistically downscaled to forecast the seasonal tropical cyclone activity. The current model can provide a skillful pre-season prediction in March, and an update can be provided in June. Cross validation shows that the correlation between hindcasts of the seasonal number of tropical cyclones that cross New York State and the observed number is as high as 0.53 for the period 1979-2013 for the June forecasts, and forecasts of the probability of one or more tropical cyclones impacting New York State have a Brier Skill Score of 0.38 compared to climatology, and are skillful for 26 out of the 35 seasons. For the 2014 season, the model predicts that the probability of one or more tropical cyclones (in any stage of their lifecycle) crossing New York State is 0.30, which is below the climatological probability of 0.43.

One should note that this model is different from others that scale the climatological probability of tropical cyclones affecting local regions (such as a state or a county) by the seasonal prediction of basin wide tropical cyclone activity. Such a strategy is not expected to work well for New York State, since the correlation between the observed number of tropical cyclones crossing New York State each season and the basin-wide number of tropical cyclones is less than 0.1 during the period 1979-2013. Hence even a perfect seasonal forecast of basin-wide tropical cyclone activity will not be particularly useful for New York State.