Mauricio Magalhães Mata

Statistical description of the East Australian Current low-frequency variability from the WOCE PCM3 array

The in situ dataset used in the current study consists of the Pacific Current Meter 3 (PCM3) array, which was a significant part of the Australian contribution to the World Ocean Circulation Experiment to study the variability of the East Australian Current (EAC), and was operational between September 1991 and March 1994. Area-preserving spectral analysis has been used to investigate the typical time scales observed by the current meters. As a general rule, the spectra from the top layers of the shallow (1, 2 and 3) and the deep (4, 5 and 6) moorings have a distinct peak in the temporal mesoscale band (periods between 70 and 170 days), with a general redistribution of energy towards the higher-frequencies near the ocean floor. This peak has been linked with eddy variability of the EAC system, which influences the fluctuations of the current main jet. The vertical modes of the velocity profile show that the strong surface-intensified baroclinic signal of the EAC dominated the variability at mooring 4 location. Further offshore the predominant configuration resembles more closely the barotropic mode. Ultimately, spatial empirical orthogonal functions (EOF) analysis point out the impact of the presence/absence of the EAC jet in the array.

Dimensional reduction of large image datasets using non-linear principal components

In this paper we apply a Neural Network (NN) to reduce image dataset, distilling the massive datasets down to a new space of smaller dimension. Due to the possibility of these data have nonlinearities, traditional multivariate analysis, like the Principal Component Analysis (PCA), may not represent reality. Alternatively, Nonlinear Principal Component Analysis (NLPCA) can be performed by a NN model to fulfill that deficiency. However, when the dimension of the image increases, NN may easily saturate. This work presents an original methodology associated with the use of a set of cascaded multi-layer NN with a bottleneck structure to extract nonlinear information of the large set of image data. We illustrate its good performance with a set of tests against comparisons using this methodology and PCA in the treatment of oceanographic data associated with mesoscale variability of an oceanic boundary current.