engleski

The Adriatic sea tidal open boundary conditions via data assimilation : exploring the altimetry contribution

Satellite altimetry aided prediction of ocean tides has clearly demonstrated its usefulness on the global scale. In coastal seas sampling constraints (e.g. separation of satellite ground tracks) and small scale tidal features (e.g. large harmonic amplitude gradients) make altimetry application more challenging. In the present work we explore the impact of assimilating local tide gauge and altimetric data on the quality of predicted tides. In particular, we address the question of properly forcing the co-oscillating, Adriatic tides, an elongated land-locked Mediterranean sub-basin with a fairly complex tidal structure, although not of exceptional hannonic amplitudes. In order to examine the Adriatic response to major tidal harmonics (M2, S2, KI, 01) we rely on a series of numerical models and the use of appropriately processed Topex/Poseldon data. We start with a non-linear dynamical model forced at the Straits of Otranto by the results of a Mediterranean tidal modeling study (Tsimplis et al., 1995). A 2D linear barotropic model is then used to assimilate the Adriatic tide gauge data and obtain optimal boundary conditions. Typically, such a model can fit the sea level data rather well, but does not provide information about the 3D current structure. Hence, we then use a 3D finite element model (Quoddy - Lynch et al., 1996) forced at the Otranto by a literature-derived set of empirical boundary conditions (Polli, 1960). Although in line with known Adriatic tidal behavior these results warrant further considerations of the boundary conditions. Consequently, in the final step we use a combination of a forward 3D model (Quoddy) and 2D linear assimilative model, following the so called incremental approach (Thompson et al., 1998). The goal has been to obtain the optimal boundary conditions for the 3D model, by assimilating the tide gauge data via a simpler 2D inverse. First, the observations are assimilated into the 2D model, which then provides the open boundary conditions to force the Quoddy. The residuals from the 3D model are then assimilated back into the 2D model to provide a new set of the open boundary conditions for Quoddy. The procedure is iterated until the convergence criterion is met. The whole procedure is then repeated with harmonic constants derived from altimetry instead of gauge data. The constants were derived in a previous work (Kuzmić and Wagner, 1997) for each of the 84 points along 4 Adriatic UP segments, by constrained least squares fit to altimetry-derived time series. In the concluding step, the results obtained from each assimilation (altimeter, tide gauge data) are validated against current and sea level data that were not assimilated.

Adriatic; tides; data assimilation; altimetry

nije evidentirano