Description: Highlights • A new description of sediment phosphorus dynamics was implemented in a 3D-model. • Oxygen consumption affects oxygen penetration in coastal sediments. • Low oxygen concentrations determine the oxygen penetration in deeper water sediments. • More than 80% of the phosphorus loads (1980–2008) are retained in the Baltic Sea. • Phosphorus is released from anoxic sediments and retained in oxic sediments. Abstract The new approach to model the oxygen dependent phosphate release by implementing formulations of the oxygen penetration depths (OPD) and mineral bound inorganic phosphorus pools to the Swedish Coastal and Ocean Biogeochemical model (SCOBI) is described. The phosphorus dynamics and the oxygen concentrations in the Baltic proper sediment are studied during the period 1980–2008 using SCOBI coupled to the 3D-Rossby Centre Ocean model. Model data are compared to observations from monitoring stations and experiments. The impact from oxygen consumption on the determination of the OPD is found to be largest in the coastal zones where also the largest OPD are found. In the deep water the low oxygen concentrations mainly determine the OPD. Highest modelled release rate of phosphate from the sediment is about 59 × 103 t P year− 1 and is found on anoxic sediment at depths between 60–150 m, corresponding to 17% of the Baltic proper total area. The deposition of organic and inorganic phosphorus on sediments with oxic bottom water is larger than the release of phosphorus, about 43 × 103 t P year− 1. For anoxic bottoms the release of total phosphorus during the investigated period is larger than the deposition, about 19 × 103 t P year− 1. In total the net Baltic proper sediment sink is about 23.7 × 103 t P year− 1. The estimated phosphorus sink efficiency of the entire Baltic Sea is on average about 83% during the period.

Description: Climate model results for the Baltic Sea region from an ensemble of eight simulations using the Rossby Centre Atmosphere model version 3 (RCA3) driven with lateral boundary data from global climate models (GCMs) are compared with results from a downscaled ERA40 simulation and gridded observations from 1980-2006. The results showed that data from RCA3 scenario simulations should not be used as forcing for Baltic Sea models in climate change impact studies because biases of the control climate significantly affect the simulated changes of future projections. For instance, biases of the sea ice cover in RCA3 in the present climate affect the sensitivity of the model's response to changing climate due to the ice-albedo feedback. From the large ensemble of available RCA3 scenario simulations two GCMs with good performance in downscaling experiments during the control period 1980-2006 were selected. In this study, only the quality of atmospheric surface fields over the Baltic Sea was chosen as a selection criterion. For the greenhouse gas emission scenario A1B two transient simulations for 1961-2100 driven by these two GCMs were performed using the regional, fully coupled atmosphere-ice-ocean model RCAO. It was shown that RCAO has the potential to improve the results in downscaling experiments driven by GCMs considerably, because sea surface temperatures and sea ice concentrations are calculated more realistically with RCAO than when RCA3 has been forced with surface boundary data from GCMs. For instance, the seasonal 2 m air temperature cycle is closer to observations in RCAO than in RCA3 downscaling simulations. However, the parameterizations of air-sea fluxes in RCAO need to be improved.