Per Pemberton

For two future scenarios on the expansion of offshore wind power in the Baltic Sea and the North Sea, SMHI has investigated how the hydrography, i.e. temperatures, salinities, currents and stratification, may be affected. Effects were induced by wind stress reductions on the sea surface and by the increased friction and turbulence in the water from wind turbine foundations.

The results show that an expansion of wind power in the Baltic Sea in general will cause a shallowed halocline, and increased deep water salinities and temperatures, due to decreasing winds behind the wind farms that lead to decreasing vertical mixing in the Baltic Sea. However, the magnitude of changes shows a strong sensitivity to assumptions about the wind stress reduction at the sea surface, and the size of wind power expansion.

The wind farm scenarios are prepared in collaboration with the Swedish Agency for Marine and Water Management (SwAM) and are based on marine plans from Sweden’s neighbouring countries as well as new proposals for suitable wind power areas that SwAM will present to the government in 2024. In one scenario, Scenario 1, it is assumed that there will be offshore wind in all proposed areas, while in the second scenario, Scenario 2, it is assumed that only 50% of areas will be developed. Both scenarios represent large offshore wind power developments that will probably not be realized in reality. The scenarios have been investigated by running an ocean model for the Baltic Sea and the North Sea with and without wind power for the period 1985 – 2016 to evaluate how different the sea would have looked if the wind power had been built in 1985 according to the scenarios.

There is still lack of knowledge about how wind farms affect the wind at the sea surface, so this work is based on studies of existing wind farms in the North Sea, where studies show a reduction of the wind by around 8% and an area that extends about 30 km behind the wind farm under stable atmospheric conditions. When the atmosphere is unstable, which it often is in winter, the reduction is less. In order to get an estimate of the largest and smallest possible impact of wind power on the sea, we have therefore, for both scenarios, assumed that the reduction of wind only exists in summer and no reduction during winter (minimum possible impact), or that the reduction exists all year round (upper limit of impact).

The magnitude of expected changes is very dependent on the assumptions on the wind wakes, and the response is much smaller for the minimum possible impact than for the upper limit impact. The real response for these scenarios probably lays somewhere in between these estimates.

For the scenario with less wind farms in Swedish waters (Scenario 2), the influences on salinity, temperature, and halocline are reduced relative to Scenario 1 in a manner that may be expected from the difference in total wind farm areas in the Baltic Sea in the two scenarios.

The model results also show that the wind power foundations (modelled as bottom mounted) cause a salinity decrease in the Baltic Sea deep water, probably due to increased friction and mixing in the entrance region to the Baltic Sea. This effect is much smaller than the wind wake effect when it is active during the whole year.

The Baltic Sea surface salinity, surface temperature, and currents show much smaller and less robust changes than the salinity and temperature changes in the deepwater.

SMHI has analysed how sea ice conditions in the Bothnian Bay, Bothnian Sea, Åland Sea and northern Baltic Proper may change in a 20 and 50 year perspective relative to 2020. The study is focused on seven indicators describing different aspects of sea ice change. The indicators were identified jointly with the Swedish Maritime Administration (SMA), and chosen based on available data and relevance to ice breaking.The study is based on historical observations from SMHI, the Finnish Meteorological Institute (FMI) and SMA, and climate scenario data from previous projects.Climate scenarios representing two different representative concentration pathways (RCP4.5 and RCP8.5) have been analysed based on a total of ten different climate model simulations. Scenarios based on the lower representative concentration pathway (RCP2.6) are absent because existing datasets for this pathway do not have sufficient quality for sea ice parameters. The time frame for this assignment did not allow for new climate scenario simulations to be produced.The results show that future winters will gradually, on average, have a smaller maximum ice extent compared to the control period (1975-2004). Ice seasons will also get shorter, with the largest differences in the southern areas. None of the scenarios yield ice free winters, and at least Bothnian Bay is expected to become fully ice covered on average, also during future winters. However, in the RCP8.5 scenario, ice with an average thickness of 10 cm or more disappears from the southern Bothnian Bay.In a 20-year perspective, changes in maximum ice extent are less distinct due to large inter-annual variations. In a 50-year perspective the change becomes more distinct and shows decreasing ice extents and smaller inter-annual variations.Level ice is expected to get thinner on average in all analysed areas, and the presence of heavily deformed ice is expected to decrease. However, models lack the ability to simulate brash ice barriers, which are formed when thin ice is pressed against a thicker ice edge or land by wind and waves. These types of barriers can be problematic for ships even in mild winters, and are expected to occur also in the future. Thinner and less dense ice fields also lead to increased ice drift in the Bothnian Bay and Bothnian Sea.The number of days with ice class based traffic restrictions for Swedish harbours are expected to decrease as sea ice thickness become thinner and ice seasons become shorter. The distribution of restrictions will also change, mainly in the Bothnian Bay where days with heavier ice classes (1A/B) decrease and days with lighter ice classes (1C/II) increase.Changes in maximum ice extent, length of ice season and average level ice thickness are judged to have a low uncertainty as the results are supported by both historical observations, and by the fact that model simulations are relatively close to the observations during the historical period. Changes in ice deformation, ice thickness distribution, and ice drift are judged to have a higher degree of uncertainty as there are no or very few observations to support model results.The study is partly limited by the lack of data for the lower RCP2.6 and by lacking analyses of possible changes in meteorological conditions. Another limiting factor is the relatively low number of regional climate model simulations with reliable ice parameters used in the study.