This project has ended | -
The project has resulted in a prototype coupled atmosphere-ocean assimilation system capable of climate quality reanalysis as well as for initialising medium-range coupled forecasts.
A comparative study has been carried out over short time periods to assess the differences between the analyses produced by CERA on the one hand and by an uncoupled system based on the uncoupled operational ECMWF scheme on the other. Comparing the behaviour of the two systems serves to highlight some potential benefits of the CERA system: producing a consistent coupled ocean–atmosphere analysis; making better use of near-surface measurements; and reducing initialisation shocks.
The development of the CERA system initially funded by ESA will continue with the EU-funded ERA-CLIM2 project that aims to produce a 20th century coupled reanalysis.
Publications:
- Laloyaux,P., M. Balmaseda, D. Dee, K. Mogensen, and P. Janssen, A coupled data assimilation system for climate reanalysis. Quarterly Journal of Royal Meteorological Society, In press.
- , D. P. , , and , Origin and impact of initialization shocks in coupled atmosphere-ocean forecasts Monthly Weather Review, In press.
Coupled ECMWF ReAnalysis (CERA) is funded by ESA and led by the University of Reading
The aim of this project is to build a coupled data assimilation system that ingests simultaneously ocean and atmospheric observations in the ECMWF coupled model. This development will be a major advance, with wide ranging applications for climate reanalysis and weather forecasting.
The CERA system is a coupled assimilation system which uses the ECMWF coupled model to compute the misfits with the observations in the outer loop, but runs separate inner loops for the ocean and the atmosphere. The atmospheric increments are computed through a 4D-Var technique comparable to the one used to produce the ERA-Interim uncoupled atmospheric reanalysis. The computation of the ocean increments relies on the NEMOVAR system based on the 3D- FGAT technique used for the production of the ORAS4 uncoupled ocean reanalysis.
There are two fundamental differences in the design of the CERA system compared to the uncoupled operational ECMWF data assimilation system. The first difference is the use of a coupled model in the variational method and in the forecast that carries the analysis forward in time. The second difference is the treatment of the atmospheric boundary condition, which evolves dynamically in the coupled model by receiving the SST fields computed in the NEMO model through the coupled framework.