Data Assimilation Planning and Testing for Version 16 of the NCEP Global Forecast System

Following the successful implementation of version 15 of the NCEP Global Forecast System (GFS) and transition to the Finite Volume on the Cubed Sphere (FV3) dynamic core, work is underway toward the next suite of data assimilation upgrades to be included as part of version 16. Version 16 will continue to utilize a hybrid 4DEnVar solver based on the Gridpoint Statistical Interpolation (GSI), doubling the number of vertical layers in the model and raising the model top to approximately 80km. Additionally, many science changes are under consideration for the data assimilation component of the package. Some of the proposed changes include replacing the Ensemble Serial Filter with the Local Ensemble Transform Kalman Filter (LETKF) for the ensemble perturbation update, 4D incremental analysis update (IAU), improvements to the inline near sea surface temperature analysis, and semi-coupled land analysis as forced by observed precipitation. Additional changes will be incorporated to account for the new model layer specification, including modified normal mode initialization, changes to stratospheric humidity increments, and satellite channel selection and bias correction.
This presentation will provide details regarding many of the components under consideration and testing for inclusion as part of GFSv16. Results from low resolution prototypes as well as full-resolution pre-implementation testing will be discussed.