Variational Data Assimilation in the Tropics Using Precipitation Data. Part II: 3D Model

Abstract

A global primitive equation model is used to examine the performance of four-dimensional variational data assimilation (4D-VAR) with moist processes and to assess the impact of assimilating precipitation data in the Tropics. Several types of discontinuity in the parameterization schemes of moist processes are removed. In the assimilation experiments, truth and observations are provided by the full-physics model, while the assimilation model and the corresponding adjoint model include moist processes, horizontal diffusion, and simplified surface friction only. An idealized observation network that is sparse in the Tropics and the Southern Hemisphere is used. It is demonstrated that the addition of a penalty term for suppressing gravity wave noise increases the efficiency of 4D-VAR with moist processes by avoiding locally large gradients in the cost function during the minimization process. It is found that 4D-VAR with moist processes included yields a much better analysis in the Tropics despite a slower convergence rate than 4D-VAR without the moist processes. 4D-VAR assimilates the simulated precipitation data quite well. Inclusion of the moist processes and assimilation of precipitation data improve the analyses of divergence, moisture, and lower-tropospheric vorticity. In particular, the wind field in the tropical planetary boundary layer is better analyzed, and the structure of a tropical cyclone is well retrieved. A 72-h forecast experiment shows that assimilation of precipitation data improves the precipitation forecast.