Tropical cyclone position forecast errors have decreased by an average of 1% - 2% per year from 1970 – 1998 (McAdie and Lawrence 2000). However, forecast errors for landfalling storms have not significantly declined during the last 25 years (Aberson 2001). This lack of improvement can be attributed in part to large forecast errors that can occur for storms moving parallel to a nearby segment of coastline. Storms with tracks parallel to the coast were found to have between 15% and 50% greater errors within 30 hours of predicted landfall than those with more perpendicular tracks (Aberson and Powell 2001).

From the period 1976 – 2000, a significant number of Atlantic-basin storms have approached the southeast US coast, tracked parallel to the coastline (sometimes coming inland, sometimes remaining offshore), and then curved back out to sea. Of these storms, a small subset that moved through the region between Cape Hatteras, NC and Savannah, GA is of particular interest because of their potentially large forecast landfall position and time errors. This study investigates the impact on forecast landfall position and timing errors for these storms by running a series of adjoint sensitivity experiments using the ARPS (Advanced Regional Prediction System). Current work involves contrasting a very high (1 km or less) resolution of the coastal orography and other land surface parameters with the operational resolution, which presently is much coarser, ranging from about 10 to 80 km.

Future work involves expanding the data set to include a larger number of storms from the period and investigating whether the magnitude of the error reduction changes for this more inclusive set of storms.

To view the latest research, including climatological transition probabilities for storms that approach the SE US coastline, please visit: http://weather.ou.edu/~bbarrett/research.htm