29th Conference on Hurricanes and Tropical Meteorology


Verification of the National Weather Service Tropical Cyclone Intensity Probabilities and Future Plans

Michael J. Brennan, NOAA/NWS/NHC, Miami, FL; and D. P. Brown, R. Knabb, and M. DeMaria

The National Hurricane Center (NHC) began issuing intensity probabilities for tropical cyclones (TCs) in the 1990s. The original product used the long-term mean NHC intensity forecast errors and the deterministic NHC intensity forecast to compute the likelihood that a tropical cyclone would fall within certain intensity categories (dissipated, tropical depression, tropical storm, hurricane) and the various categories of the Saffir-Simpson Hurricane Scale at specified forecast times during the next 72 hours. However, these probabilities did not take into account land interaction. In 2008, the NHC and CPHC began computing the intensity probabilities from a set of 1,000 realizations, or alternate tracks and intensities, that vary around the official forecast based on a Monte Carlo sampling of historical errors in the NHC and CPHC track and intensity forecasts. The Monte Carlo technique accounts for land interaction and should provide a more accurate assessment of the chances that the intensity of a tropical cyclone will fall within the various categories. This study provides a verification of Monte Carlo-based intensity probabilities issued by NHC and CPHC with each advisory in the central North Pacific, eastern North Pacific, and Atlantic basins during the 2008 and 2009 hurricane seasons. The probabilities will be verified using the NHC and CPHC best tracks.

The study will also discuss future enhancements to the product, including the computation of intensity probabilities valid at landfall. The current product provides probabilities at specific times during the forecast period. For example, if a storm is near land at a specific time, many of the 1,000 tracks may have already intersected land, which results in weaker storms in those realizations. Because of this, intensity probabilities tend to be spread nearly equally among the various intensity categories when a tropical cyclone is forecast to be near land. To obtain accurate landfall intensity probabilities, a future enhancement will be to produce conditional probabilities using the tracks and intensities of only those realizations that cross the coast between pre-defined points. The intensity of each realization when it crosses the coast between the pre-defined points can be used to create landfall intensity probabilities. This study will provide a summary of the methodology used to produce the landfall intensity probabilities, and will show examples from recent storms of how these probabilities could be used in NHC and CPHC operational products in the future, as well as potential uses for decision makers such as emergency managers.

extended abstract  Extended Abstract (616K)

wrf recordingRecorded presentation

Session 15D, Probabilistic Forecasting
Friday, 14 May 2010, 8:00 AM-9:45 AM, Tucson Salon A-C

Previous paper  Next paper

Browse or search entire meeting

AMS Home Page