3D-mask for the ETR Initialization in the NCEP Global Ensemble Forecast System

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 February 2014
Hall C3 (The Georgia World Congress Center )
Juhui Ma, Nanjing University of Information Science & Technology, Nanjing, Jiangsu, China; and Y. Zhu

A new three-dimensional regional rescaling (3DR) based on short statistics of the total energy norm from the ensemble Kalman filter (EnKF) analysis error is imposed into the ensemble transform with rescaling (ETR) method (ET_3DR). The new method has been tested and compared with the NCEP operational ETR initialization method (the ensemble transform with two-dimensional rescaling or ET_2DR). The experiments are performed from 11 September to 30 November 2012 using the NCEP operational Global Ensemble Forecast System (GEFS) version 9.0 (T254L42 for 0-196 hours, and T190L42 for 196-384 hours).

Based on the preliminary study, a rescaled distribution of initial perturbations is very important for the performance of ensemble forecast. Therefore, there is significant difference for the ET_2DR and ET_3DR initial perturbations. Relative to the ET_2DR, the horizontal distribution of the ET_3DR initial perturbations at 500hPa is more correlated to the baroclinic instability over the extra-tropic of the Northern Hemisphere and connected with the deep convection over the tropics. Furthermore, the maximum value of vertical distribution for the ET_3DR initial perturbations corresponds to the subtropical jet region and tropical easterlies jet region. Naturally, due to the better representation of horizontal and vertical structures of initial perturbations for the ET_3DR, the ensemble spread (or uncertainty) grows much faster than the ET_2DR, especially at the lower levels and over the tropics.