Sensitivity Study of Tropical Cyclone Initial Perturbations for GEFS

Friday, 22 April 2016: 1:15 PM
Ponce de Leon B (The Condado Hilton Plaza)
Xiaqiong Zhou, IMSG at NOAA/NWS/NCEP/EMC, College Park, MD; and B. Fu and Y. Zhu

In the recent GEFS implementation, ETR is replaced with EnKF to generate initial ensemble for the medium-range ensemble forecast. ENKF 6hr forecast ensemble instead of the analysis are used since only the EnKF from the previous cycle is available at the time when GEFS starts in the NCEP operational environment. The environment and tropical cyclone (TC) components in GEFS are perturbed separately. To get TC perturbations, EnKF forecast is separated into TC and environment components. The TC perturbations are calculated after the TC component in each ensemble member and the control is relocated to the observed location. An artificial scaling factor is used to control the amplitude of TC perturbations. The TC scheme reduces the TC track spread as well as the mean track forecast errors.

Sensitivity experiments were performed to investigate the effect of the amplitude of TC perturbations on TC track forecasting to find an optimal solution. It is found that the TC track forecast errors are sensitive to the amplitude of initial TC perturbations. The spread of TC tracks is generally under-dispersive but larger amplitude of initial TC perturbations leads to slightly larger spread as well as larger ensemble mean track forecast errors.

- Indicates paper has been withdrawn from meeting
- Submission entered in competition