A Three-Dimension Variational Data Assimilation Method for A Nonhydorstatic Storm-scale Model with equation constraints

In this paper, a 3DVAR data assimilation scheme for a Nondydrostatic storm-scale model, ARPS, has been developed in which a cost function is defined as a sum of background field constraint, observation constraint and some penalty terms. The background field is provided by a previous ARPS model forecast, or a relatively larger-scale model, such as RUC operational numerical model from NCEP. Observations include: single-level surface data (such as Mesonet), multiple-level or upper-air observations (such as rawinsondes and wind profilers) and raw Doppler radar observation. The background error covariance matrix based on recursive filter is used as a preconditioning. In the further developed system, a dynamic constraint based on ARPS equations and the anelastic mass continuity equation serve as two weak constraints. We consider this an important feature for storm-scale data assimilation since by doing so, the retrieval process for wind field and thermodynamic field may also be included in the variational data assimilation system. Some numerical experiments will be performed based on this new scheme and will be reported on the conference.