Using an adjoint model to improve an optimum interpolation-based data-assimilation system
Källén, E. R.
- Publisher: Co-Action Publishing
(issn: 1600-0870, eissn: 0280-6495)
A highly simplified 4-dimensional variational data assimilation (4DVAR) system is formulated aiming at an early operational implementation on current supercomputers. The system is a hybrid system based on both an intermittent data assimilation, including an Optimal Interpolation (OI) analysis scheme, and a 4DVAR setup. The idea is to use the adjoint model to produce an improved first-guess for the OI scheme. A 5-day data assimilation is chosen to demonstrate the applicability of our method. It is shown that the analysis increments are reduced and the forecast of a rapid cyclone development is improved.DOI: 10.1034/j.1600-0870.1997.t01-1-00001.x