Sensitivity of the Baltic Sea level prediction to spatial model resolution

- High resolution model was used to calculate Baltic Sea levels in 2010-2015.
- Sea levels were calculated using a spatial resolution varying from 3 NM to 0.5 NM.
- Resolution increase resulted in a more realistic reflection of sea level variability.
- 0.5 NM model resolution is recommended for modelling storm surges in the Baltic Sea.
- Parallel calculations in the PM3D model reduced the computation time of simulations.

The three-dimensional hydrodynamic model of the Baltic Sea (M3D) and its new parallel version (PM3D), developed at the Institute of Oceanography, University of Gdańsk in Poland, was tested to establish a grid resolution adequate for the Baltic Sea level prediction. Four outputs of the M3D/PM3D, calculated with spatial resolution varying from 3 NM to 0.5 NM, were validated by comparing the results with hourly sea level readings collected at 9 Baltic gauges in 2010-2015. The spatial resolution of 1 NM applied to the Baltic Sea resulted in a distinct improvement of agreement between the calculated and observed distributions of data. An increase in the resolution to 0.5 NM in the southern Baltic Sea improved the model quality further, as indicated by the lowest variability, the highest correlation and the highest percentage of water level simulations within the range of ± 0.15 m difference relative to readings. The increase in horizontal resolution allowed to improve the fit between the observed water levels and those calculated by the PM3D in the cases of rapid sea level fluctuations, such as those registered in January 2012. The model performed slightly worse for stations with larger ranges of water level oscillations. As parallel calculations were used in the PM3D, the time necessary for computing the simulations was significantly reduced, which allowed to apply the high-resolution grid also to the operational version of the model.

447