Impact of satellite observations and forecast model improvements on tropical cyclone track forecasts

The tropical cyclone (TC) track forecasts of the Navy Operational Global Atmospheric Prediction System (NOGAPS) were evaluated for a number of data assimilation experiments conducted using observations from August 14-September 30, 2004. This was a particularly active period with 12 hurricanes (including Charley, Frances, Ivan, and Jeanne), 5 typhoons, and 7 tropical storms. The current operational configuration of NOGAPS consisting of the NRL Atmospheric Variational Data Assimilation System (NAVDAS) with the assimilation of all available conventional and satellite observations and a T239L30 global spectral model with Emanuel cumulus parameterization was used as the control run. Two sets of data assimilation experiments were conducted. The first set was designed to illustrate the impact on the NOGAPS TC track forecasts of the assimilation of different types of satellite observations while the second set was designed to illustrate the impact of improvements to the NOGAPS global spectral model. From the first set of experiments, it was found that the satellite observations with the most impact upon TC track forecasts were feature-track winds from geostationary and polar-orbiting satellites, AMSU-A radiances, and SSM/I precipitable water. The assimilation of the feature-track winds resulted in the largest improvement in the 24-h to 72-h forecasts while the assimilation of the AMSU-A radiances and SSM/I precipitable water resulted in the largest improvements in the 96-h and 120-h forecasts, respectively. The overall TC track forecast improvement due to the assimilation of satellite observations was 18% at 24 h, 27% at 48 h, 23% at 72 h, 19% at 96 h, and 19% at 120 h. From the second set of experiments, it was found that increasing the resolution of the NOGAPS global spectral forecast model from T79L18 to T159L24 resulted in the largest improvement in the 72-h to 120-h forecasts while replacing the relaxed Arakawa-Schubert cumulus parameterization with the Emanuel cumulus parameterization resulted in the largest improvement in the 24-h and 48-h forecasts. The overall TC track forecast improvement due to forecast model improvements was 16% at 24 h, 23% at 48 h, 26% at 72 h, 38% at 96 h, and 46% at 120 h.