Investigating interdecadal salinity changes in the Baltic Sea in a 1850–2008 hindcast simulation

Abstract

Listen

Translate article

Abstract. Interdecadal variability in the salinity of the Baltic Sea is dominated by a 30-year cycle with a peak-to-peak amplitude of around 0.4 g kg−1 at the surface. Such changes may have substantial consequences for the ecosystem, since species are adapted to a suitable salinity range and may experience habitat shifts. It is therefore important to understand the drivers of such changes. We use both analysis of empirical data and a numerical model reconstruction for the period of 1850–2008 to explain these interdecadal changes. The model explains 93 % and 52 % of the variance in the observed interdecadal salinity changes at the surface and the bottom, respectively, at an oceanographic station at Gotland Deep. It is known that the 30-year periodicity coincides with a variability in river runoff. Periods of enhanced runoff are followed by lower salinities. We demonstrate, however, that the drop in mean salinity cannot be understood as a simple dilution of the Baltic Sea water by freshwater. Rather, the 30-year periodicity in river runoff occurs synchronously with a substantial variation in salt water import across Darss Sill. Fewer strong inflow events occur in periods of enhanced river runoff. This reduction in the import of high-salinity water is the main reason for the freshening of the water below the permanent halocline. In the bottom waters, the variation in salinity is larger than at the surface. As a consequence, the surface layer salinity variation is caused by a combination of both effects: a direct dilution by river water and a reduced upward diffusion of salt as a consequence of reduced inflow activity. Our findings suggest that the direct dilution effect is responsible for 27 % of the salinity variations only. It remains unclear whether the covariation in river runoff and inflow activity are only a coincidental correlation during the historical period or whether a mechanistic link exists between the two quantities, e.g. whether both are caused by the same atmospheric patterns.