A three-dimensional variational data assimilation system for a climate model - Basic scheme and tests

In this study, a scheme of three-dimensional variational data assimilation (3D-Var) is introduced for a grid-point Atmospheric General Circulation Model, developed by the Institute of Atmospheric Physics, Chinese Academy of Sciences (IAP/CAS-AGCM). The scheme adopts stream function, unbalanced potential velocity, unbalanced geopotential height and relative humidity as analysis variables which are independent of each other. In order to avoid the complexity of background error covariance (B) and reduce the tremendous cost of minimization, variable transform and Conjugate-Gradient Method are used here. Meanwhile, a relationship among multivariable is considered. To test the validity of system and dynamical constraints between mass field and wind field, many experiments including single observation and continuous assimilation added in model are performed, The results show that the relationship among multivariable is reasonable and improvement because of the assimilation is obvious.