Data assimilation in ROAM-Ocean – performance assessment
Dr David Griffin1, Dr Emlyn Jones1, Dr Paul Sandery1
1Csiro, Hobart, Australia
ROAM (Relocatable Ocean Atmosphere Model) is a GUI-driven system for rapidly configuring a regional environmental modelling system. Here, we focus on ROAM-Ocean, which is the system for rapidly (e.g. minutes) setting up and running a regional ocean model to generate a short-term forecast (or hindcast), using the Bureau’s OceanMAPS global model for initial and boundary conditions, BoM ACCESS surface fluxes and pressure, and TPXO tidal constituents. A typical ROAM-Ocean grid will have ~2km horizontal resolution and span a 2-week period. Until recently, ROAM was purely a nested model, adding tides, inverse-barometer response, hourly resolution of internal waves and mixed-layer physics etc, and finer-resolution topographic influences to the parent global model. The inclusion of data assimilation into the system has been an important upgrade, soon to be operationalised. This talk will outline the data assimilation strategy and demonstrate the impact this has had on model performance, using a number of data sets that were not assimilated (various IMOS observations, and drifters) as well as data (satellites and Argo) that was.
Biography:
David Griffin has been at CSIRO as a research scientist (oceanographer) since 1994. Nowadays, he splits his time between Bluelink, IMOS and miscellaneous mysteries (or, more dryly: applications of models and observations).
