Abstract
In the Baltic modelling research, assimilation techniques were developed with advance. They were concerned to model assimilated basic parameters and observed them directly. In present paper, the most important was the assimilation of surface information and its projection deep into temperature and salinity fields. In oceanic investigations altimetry viewed from satellite was the sea level changes projected far inside and predetermined surface-to-subsurface correlations. To obtain improved modelled hydrophysical fields, sea level variations measured at coastal gauges and efficient data assimilation were taken into account. A data assimilation algorithm has been developed and used in conjunction with a three-dimensional baroclinic model of the Baltic Sea. It was based on a time and space weighted nudging technique. The sea level data were inserted continuously by updating the model solution every time step. Several sensitivity experiments with different values of time and spatial weighting scales were performed. In first series of experiments, only sea level data (SL) were assimilated. In the next simulations, seawater temperature (SWT) and seawater salinity (SWS) related directly to SL were assimilated. To evaluate the effectiveness of the assimilation scheme, modelled sea level series and vertical profiles of seawater temperature and salinity in selected coastal gauges in the Gdansk Basin were examined. Evidently low but statistically essential correlation coefficients indicated nonlinear character of vertical mixing and transfer processes. Decreasing errors obtained while comparing the model results to a control case without assimilation confirmed a real transfer of surface information deep and usefulness of such approach in modelling.