A03-K111 New developments in ocean modeling: data-driven nesting models linking open ocean and coastal area
A03-K111
Principal Investigator
| AOKI Kunihiro | Meteorological Research Institute |
As an alternative to numerical general circulation models, data-driven models based on deep neural network with high-speed computation are rapidly being developed as a pillar of next-generation forecasting research, but there is still no data-driven model capable of representing the coastal ocean environment with a sufficient resolution. In this study, a data-driven ocean forecasting model that connects the open ocean and the coastal ocean, consistent with the coastal ocean dynamics, will be developed, using a training data set covering the open ocean to the coastal ocean, which is comprised of long-term reanalysis data with a realistic field around Japan and a super-fine resolution regional circulation model output based on the reanalysis data. Based on this, our research aims
- To understand the characteristics of the model predictions, and to approach the issue of the predictability of coastal ocean phenomena.
- To establish the methods for a probabilistic forecasting and for identifying precursor of coastal phenomena.
- To discover the control mechanisms in the persistence and state transition processes of coastal ocean extremes.
