Report on the use of 3D or 4D-Var for the ocean component in the coupled data assimilation context

Arthur Vidard (1)

1 AIRSEA - Mathematics and computing applied to oceanic and atmospheric flows

This report describes the Inria contribution to WP2 regarding the use of 4D-Var in the ocean component of the reanalysis system. This task was twofold: 1. To study the relevance of the use of 4D-Var respect to 3D-Var. 2. To propose ways to reduce the computing cost, without altering the results. For the first part it turns out that 4D-Var is not really relevant for the ocean of the CERA reanalysis. Indeed at coarse resolution and with the use of observations of subsurface temperature and salinity only, the 3D-Var approximation is efficient enough. Getting a noticeable impact would require a data assimilation longer than 5 days, which is not feasible in coupled mode. For the CERA-SAT reanalysis however, mostly due to its use of SSH observations, 4D-Var does become a better option. The second part of the study shows that multi-incremental implementation of 4D-Var, even though challenging in the ocean, is a promising solution to significantly reduce the cost of 4D-Var, and could be used in future coupled reanalysis In addition to the scientific study of the above-mentioned points, it has required some significant technical developments. The first one was to include the ocean 4D-Var capabilities in the CERA system and was performed successfully in collaboration with P. Laloyaux at an early stage of the ERACLIM2 project. The second one was to develop the alternatives to full 4D-Var proposed in section 3. This was done in the framework of the NEMOVAR collaborative environment and is therefore available to the partners through, for instance, the ECMWF git repository more details about software transfer are given in appendix.

Domaines: Optimisation et contrôle [math.OC]; Océan, Atmosphère

Arthur Vidard. Report on the use of 3D or 4D-Var for the ocean component in the coupled data assimilation context. 2017. ⟨hal-01667507⟩