Analysis of hydrodynamic model outputs characterising the SE Tasmanian coastal region using a self-organizing feature map

The Coastal Environmental Modelling Team at CSIRO Marine and Atmospheric Research, in Hobart, Tasmania, has been modelling hydrodynamic conditions (including water temperature, salinity and ocean current components) within the coastal environment of south-east Tasmania for several years. Output from this modelling process, taken over one year from August 2009 to August 2010, has been analysed in an effort to identify prototype hydrodynamic states (i.e., frequently encountered typical hydrodynamic situations) exhibited over that period. A competitive-learning neural network, the Self- Organizing Feature Map (SOM), was used in an effort to identify these prototype states. Once such a network has been trained, each node in its output layer represents a particular pattern in the input data and nodes representing similar patterns are located close to each other on a two-dimensional output grid, while those representing dissimilar patterns are located further apart. Estimated daily average surface values for salinity, water temperature and ocean current components were derived from output produced by the hydrodynamic model and these were analysed using the MATLAB SOM Toolbox to create a number of SOM grids, of various sizes, and depicting different hydrodynamic variables. Subsequent inspection of the SOM grids enabled a number of prototypical hydrodynamic states to be identified from the model outputs. These prototype states were then interpreted, using prior knowledge of the dominant dynamics of the region. The dominant prototype states highlight the strong influence and interaction between the Zeehan Current and the East Australia Current and the inshore region of freshwater influence. The results obtained with the SOM were consistent with those expected by domain specialists.