92nd American Meteorological Society Annual Meeting (January 22-26, 2012)

Tuesday, 24 January 2012: 11:15 AM
Probabilistic Tropical Storm Position Forecasts
Room 238 (New Orleans Convention Center )
Zoltan Toth, NOAA/ESRL/GSD, Boulder, CO; and W. Smith and P. T. McCaslin

After the 2005 season, the National Hurricane Center (NHC) discontinued the issuance of Tropical Cyclone (TC) strike probability forecasts. Since then, no official probabilistic information regarding the position of forecast TCs has been made available. Instead, the NHC provides an error cone associated with its single value track prediction product. In this presentation we discuss the concept of Storm Position Probability Density Function (SPPDF). SPPDF describes the probability density of a storm's occurrence at any point in 2-dimensional geographical space and can be considered as a generalization of strike probability.

A method is proposed to create SPPDF based on an ensemble of forecasts. Following Roulston and Smith (2003), error statistics associated with the ensemble member with the smallest overall track error (i.e., a posteriori “best member” selected separately for each case in a sample) are used to specify kernels placed on each member of the real time ensemble.

Error statistics are assessed for Atlantic TCs from 2009 for both the 20-member NCEP GEFS global ensemble and the 6-member ensemble of unperturbed high-resolution forecasts from various centers that contribute to the consensus TCVN guidance. SPPDF forecasts are then prepared and evaluated for 2010 TCs. An advantage of SPPDF is that from it the probability of a storm striking any arbitrary, user defined region can be easily determined. Error cone information reflecting case dependent uncertainty indicated by the ensemble is also derived from the SPPDF and is compared with the NHC's official and static error cone forecasts.

Supplementary URL: