Observations and data assimilation
Observations
BRAN and OceanMAPS depend on the provision of high-quality ocean observations. Many sources of data have been used to underpin ocean reanalyses and ocean forecasts. These include:
- Satellite altimetry: Bluelink systems have always assimilated along-track satellite altimetry. We source data from RADS.
- Argo: Argo data are the primary source of in situ ocean observations, particularly from about 2000 onwards, that help constrain our reanalyses and forecasts. Argo data are accessed in NRT from the GTS and in delayed-mode from the two Argo GDACS.
- In Situ profiles: Before the “Argo period”, in situ data used for ocean reanalyses includes CTD measurements and XBT profiles. These have been obtained from a few sources, including the archive that was prepared for CARS, ENACT/ENSEMBLES (ENS2), and most recently, CORA.
- Satellite SST: Bluelink systems assimilate both AVHRR data and microwave SST. We’ve obtained SST data from Pathfinder, NAVO SST, and Remote Sensing Systems.
Data Assimilation
The primary data assimilation method used under Bluelink is called Ensemble Optimal Interpolation (EnOI). EnOI was first developed by Oke et al. (2002) for a coastal application, and was later described by Evensen (2003), who coined the name. EnOI is a relatively inexpensive ensemble-based data assimilation method that is feasible for large, global applications. The first EnOI-based system developed under Bluelink was called the Bluelink Ocean Data Assimilation System (BODAS; Oke et al. 2008). BODAS was used for many BRAN experiments, and underpinned OceanMAPS from 2007 (when it first became operational) to 2019. In 2019, a new code – called EnKF-C – was adopted. EnKF-C applied the same algorithms and techniques as BODAS, but was more flexible and more elegant. EnKF-C also includes a variety of algorithms, including the Deterministic Ensemble Kalman Filter (Sakov & Oke 2008) – a new flavour of ensemble Kalman filter that was developed under Bluelink, and is now quite widely used.