Spatial Objective Analysis of Oceanographic Variables. An Application to the Ebro Delta Shelf/Slope Domain (North- Western Mediterranean)

Abstract

In this research we tested two data analyses methodologies, namely the empirical orthogonal functions (EOFs) to estimate hydrographic CTD data profiles, and a successive corrections algorithm (SC) to project the results onto two dimensional grids, whose superposition leads to a quasi-three dimensional array, on the Ebro Delta shelf/slope domain. The EOFs as an extrapolation procedure with shallow casts was also tested with thickness profiles, for an adequate set of extrapolated profiles might render reliable geostrophic currents on the shelf. The reference level of no motion for the relative geostrophic currents was 500 m. The resulting currents were contrasted with the ones obtained with a classical methodology for shelf domains proposed by Csanady (1979). Another important goal was to test the potential use of linearly independent basic profiles vector modes derived from historic data, to reproduce the data from the oceanographic campaigns. <br/>We used CTD data from four oceanographic campaigns on the Ebro Delta region under different conditions: FANS I - autumn, FANS II - winter, FANS III - summer and MEGO 94, a previous winter campaign, showed atypical winter conditions, with open ocean dynamics that differed significantly from the shelf ones. <br/>We find that the EOFs applied to hydrographic variables with the self campaign eigenvectors renders very good results (expressed as variance explained percentage), when a non-standardized analysis is used (in which the variability is assumed to be unevenly distributed in the water column, as occurs in the ocean). Good results are also obtained when the variability is assumed to be evenly distributed (standardized analysis). When the historic eigenvectors are used, together with the campaign data, a larger number of modes needs to be included. The results are very good for the summer-FANS III and winter-FANS II campaigns. On the other hand, some significantly large errors are obtained for EOF-standardized density with the autumn-FANS I and the winter-MEGO 94 campaings.<br/>The projection of the EOFs results onto a 2.5 x 2.5 km 2D grid were carried out through a SC algorithm. Some unreliable distributions are obtained with the standardized density analysis, due to the differences of the open ocean and shelf distributions, which in turn lead to a very wide range of mode amplitudes. This lead us to test the interpolation of the estimated profiles instead of the amplitude values, improving the results in a significant manner. <br/>The geostrophic circulation obtained with Csanady's method renders spurious along-transect values. While this method has been successfully used on other shallow regions, the abrupt shelf/slope topography of the Ebro Delta, together with the integration path, render it unreliable. The EOFs estimates with the self-campaign eigenvectors tend to approach well the dynamic height distributions of outer slope and open ocean, particularly the non-standardized analysis. The estimates on the shelf circulation seem to be reliable in two campaigns only. Since no current ADCP data is available in any of the campaigns, a quantitative evaluation is not possible. The use of historic eigenvectors does not contribute to a better estimate of the shelf circulation.