P-10

Applications of Ensemble Singular Vectors for the LETKF system combined with a global NWP model

S. Shin (KIAPS), J.-S. Kang (KIAPS), and S.-C. Yang (National Central University)

We have recently implemented the KIAPS-LETKF system to the global numerical weather prediction (NWP) model newly developed at the KIAPS. This model has cubed-sphere geometry and is based on the spectral element method for horizontal spatial discretization. The coupled system has been verified by examining analysis increments in Observing System Simulation Experiments (OSSEs). Besides further tests using real observations, applications of Ensemble Singular Vectors (ESV) for the LETKF system have been investigated to improve the skill in representing background error covariance. We compute ESVs following Enomoto et al. (2006) and examine if leading ESV can capture directions of fast growing forecast errors of the global NWP model. It is shown that leading ESV can represent uncertainty associated with fast growing modes in most locations of large ensemble variance. We also consider using the ESV for the LETKF data assimilation system as an additive inflation to better estimate background error covariance.