This project was funded by the Natural Environment Council (NERC) with the grant reference - NE/M005119/1 - which was led by Professor Keith Haines (University of Reading).

The Atlantic Meridional Overturning Circulation (AMOC for short) transports warm and saline Atlantic surface waters northward to high polar latitudes where it is transformed into colder (and also fresher) water, which sinks and returns southward underneath the warm waters. The Gulf stream is the best known surface current that contributes to carrying the warm surface waters to the north. The strength of this exchange flow has been monitored now for 10 years at Latitude 26N (Florida-Africa). Models suggest that low frequency changes in this AMOC flow may (i) be predictable some years ahead, and (ii) lead to changes in North Atlantic weather and climate some years later due to changing the warm water distribution at the ocean surface. To realise the potential of the AMOC monitoring measurements at 26N these observations needed to be successfully assimilated into the ocean and climate prediction models being currently used at the Met Office and elsewhere. This project developed novel methods to introduce the AMOC observational data into the latest ocean and climate model environments, in combination with other complementary datasets that were available, e.g. from ocean profiling floats (ARGO), and from satellites measuring sea surface temperatures and sea level. The experiments were carried out in close collaboration with scientists from the Met Office and the European Centre for Medium Range Weather Forecasts (ECMWF).

The main objectives of the project were:
- To develop novel techniques for assimilating AMOC monitoring data compatible with sequential data assimilation schemes, in particular NEMOVAR, as used in operational centres.
- In particular to develop the methods of lagged covariances to develop remote AMOC driving fields and the use of coupled ocean wind covariances for assimilation in the subtropical Atlantic.
- To develop methods to identify the strongest RAPID signals for assimilation, using depth structure information, temporal correlations and other filters, in order to develop the appropriate covariance information for assimilation.
- To demonstrate the application of new assimilation methods in ideal model experiments where the true AMOC signal is known.
- To produce some estimates of uncertainties by applying ensemble methods to the reanalyses, varying forcings and observations within errors.
- To develop new reanalysis datasets over the RAPID period incorporating RAPID data along with other datasets in a combined product.
- To provide suitable output from these reanalyses for products to be useful as initial conditions for predictability experiments and for experiments on the budgets and transports of biogeochemical tracers in the N Atlantic, topics covered by other proposals in this call.