4-DIMENSIONAL VARIATIONAL DATA ASSIMILATION EXPERIMENTS WITH A MULTILEVEL SEMI-LAGRANGIAN SEMIIMPLICIT GENERAL-CIRCULATION MODEL

Four-dimensional variational data assimilation (VDA) experiments have been carried out using the adiabatic version of the NASA/Goddard Labor atory for Atmospheres semi-Lagrangian semi-implicit (SLSI) multilevel general circulation model. The limited-memory quasi-Newton minimizatio n technique was used to find the minimum of the cost function. With mo del-generated observations, different first-guess initial conditions w ere used to carry out the experiments. The experiments included random ly perturbed initial conditions, as well as different weight matrices in the cost function. The results show that 4D VDA works well with var ious initial conditions as control variables. Scaling the gradient of the cost function proves to be an effective method of improving the co nvergence rate of the VDA minimization process. The impacts of the len gth of the assimilation interval and the time density of the observati ons on the convergence rate of the minimization have also been investi gated. An improved assimilation was obtained when observations were av ailable in selected segments of the assimilation window. Moreover, our 4D VDA experiments with the SLSI model confirm the results obtained b y Navon et al. and Li et al. concerning the impact of the length of th e assimilation window. The choice of an adequate time distribution of observations along with an appropriate length of assimilation interval is an important issue that will be further investigated.