Data assimilation in a high-resolution ocean model off the Norwegian coast
The Norwegian Meteorological Institute (MET) does operational ocean forecasting for the Arctic Ocean, Nordic Seas and Norwegian coastal waters. The predictions from these forecast models are important for all kinds of off-shore operations and also for emergency response (oil spills, search and rescue). Just like weather forecast models the ocean prediction systems rely on data assimilation (‘DA; the tuning of model predictions against available observations) to produce good forecasts. Today the large-scale models (of e.g. the entire Nordic Seas) use very advanced DA methods (‘4D-var’) that require huge computational resources. These DA methods are too heavy for use in the high-resolution coastal ocean models—the models that are perhaps most important for Norwegian industry and society.
In this project we will develop, implement and test an assimilation technique called ‘spectral nudging’. With this technique ocean fields (temperature, salinity, velocity) from a larger-scale, lower-resolution, model will be imposed (‘nudged’) into a high-resolution coastal model. But we will nudge only the spatial scales resolved by the lower-resolution model. The fine-scale details resolved by the high-resolution model (but not by the lower-resolution model) will be left alone. In other words, only part of the wavenumber spectrum (low wavenumbers) will be transferred to the fine-scale coastal model.
MET uses the ROMS ocean model (www.myroms.org), and the student will need to learn how ocean models work in general and how ROMS works in particular. Spectral nudging will either be done ‘on-line’ (be built into the model itself—in which case the programming will need to be done in Fortran) or ‘off-line’ (outside of the model—using Python).
Getting spectral nudging implemented in MET’s coastal ocean models is a high priority for the institute. This student will be involved in all stages of the development, implementation and testing and will work closely with researchers from MET.