Strongly Coupled Ocean-Atmosphere Data Assimilation with the Local Ensemble Transform Kalman Filter

Abstract

Current state-of-the-art coupled data assimilation systems handle the ocean and atmosphere separately when generating an analysis, even though ocean atmosphere models are subsequently run as a coupled system for forecasting. Previous research using simple 1-dimensional coupled models has shown that strongly coupled data assimilation (SCDA), whereby a coupled system is treated as a single entity when creating the analysis, reduces errors for both domains when using an ensemble Kalman filter. A prototype method for SCDA is developed with the local ensemble transform Kalman filter (LETKF). This system is able to use the cross-domain background error covariance from the coupled model ensemble to enable assimilation of atmospheric observations directly into the ocean. This system is tested first with the intermediate complexity SPEEDYNEMO model in an observing system simulation experiment (OSSE), and then with real observations and an operational coupled model, the Climate Forecasting System v2 (CFSv2). Finally, the development of a major upgrade to ocean data assimilation used at NCEP (the Hybrid-GODAS) is presented, and shown how this new system could help present a path forward to operational strongly coupled DA.