Applying a retrospective optimal interpolation to WRF

This study presents the results and characteristics of retrospective optimal interpolation with the SPEEDY AGCM (Simplified Parameterizations, primitivE-Equation Dynamics, Molteni 2003). Retrospective optimal interpolation which is one of retrospective data assimilation method iteratively assimilates observations at a post analysis time into a prior analysis at the analysis time. We can implement this method without using an adjoint model. In this study, retrospective optimal interpolation is applied to the SPEEDY AGCM which shows good computational efficiency while maintaining characteristics similar to the state-of-the-art AGCMs with complex physics. We use the SPEEDY AGCM for applying retrospective optimal interpolation as Kalnay et al. (2007) and Ham et al. (2009) used this model for the EnKF experiments. Through the perfect model experiments, we investigate the characteristics of retrospective optimal interpolation, and compare retrospective optimal interpolation with other data assimilation method.