Allowing fluxes through it, the model has short open boundaries, which are shifted by 5–20 km outside the narrowest
parts of the straits. The model equations were solved numerically using the finite difference method with an integration time step of 30 seconds. Because of the chosen time step and horizontal grid resolution, the numerical diffusion generated by the numerical scheme was relatively low. The 2D model performance had earlier been compared with a Helmholtztype model (Otsmann et al. 2001) Crizotinib and flow measurements in the straits from 1993 to 1995 (Kullas et al. 2000). Hindcast simulations for 1999 and 2005 proved the model’s success in simulating sea level (Suursaar et al., 2002 and Suursaar et al., 2006). Outside the straits, in situ flow measurements for model comparison were not available until this study. Although some gridded geostrophic wind or re-analysis data are in principle available, including the latest ERA-40 refinement for the Baltic Sea area known as BaltAn65 + (Luhamaa et al. 2011), such data cannot be used in this study for meteorological forcing. Covering 1965–2005 with a 6 h time step, the BaltAn65 + does not match our measurements from 2011. We used hourly wind and sea level time series measured by the Estonian Meteorological and Hydrological Institute (EMHI). Obtained from the Ristna
tide gauge, which is located just outside the Soela Strait (Figure 1), the hourly sea level data were applied in identical fashion at the four cuts of the open boundaries. Oxaprozin The wind stress was calculated see more from the wind data measured at the Kihnu meteorological station, located at the southern
tip of Kihnu Island. Although the Virtsu station is somewhat closer to both measuring sites, it lies in a far more sheltered position, and unlike Kihnu, it does not adequately represent marine winds (Keevallik et al. 2007). A spatially homogeneous wind was applied at the grid-points. The one-hour sustained wind speed data had a 1 h time step. Although the Kihnu station has been operational since 1931, EMHI digitized wind data have been available only since 1966. The completeness of the data set is very good. The time interval of the older data (until 2003) is 3 hours, but for hydrodynamic modelling they were subsequently interpolated into an applicable format. The value step was 1 m s− 1 until September 2003, and 0.1 m s− 1 thereafter. Wind directions were given in the 16-rhumb system in 1966–1976 (converted into degrees in the EMHI database), the resolution was 10° until 2003, and the currently used equipment providing a 1° resolution output. Thus, with regard to the potential homogeneity issue, three sub-sets can be distinguished over the study period. Wind speed was measured with a wind vane of Wild’s design during 1966–1976, a recording anemorhumbometer during 1976–2003, and the MILOS-520 automatic weather station after September 2003.