3C.7
A statistical model for forecasting tropical cyclogenesis over the Atlantic Basin
Christopher C. Hennon, UCAR Visiting Scientist Program, NOAA/NWS/NCEP/TPC/NHC, Miami, FL
A statistical model to forecast tropical cyclogenesis (TCG) over the Atlantic Basin up to 48 hours in advance is described. The model was trained on four seasons (1998-2001) of tropical cloud cluster activity during the Atlantic hurricane season. Eight large-scale predictors of TCG were derived from the NCEP-NCAR reanalysis and applied to each case.
Although the probability of detection was low (30-50%), false alarm rates were only 2-3%. Heidke skill scores ranged from 0.4 - 0.5 for each forecast period. Composite and case study results suggest three major areas for improvement: 1) The identification of more developing cases, 2) The addition of a climatological predictor, and 3) A better representation of the moisture conditions within the cloud cluster and its environment. Each will be discussed.
Session 3C, Tropical Cyclogenesis III
Monday, 3 May 2004, 1:30 PM-3:15 PM, Napoleon II Room
Previous paper