Modeling how eutrophication in northern Baltic coastal zone is driven by new nutrient inputs, internal loading, and 3D hydrodynamics

The nutrient load and the resulting eutrophication responses in the Archipelago Sea, a northern Baltic Sea basin, were assessed by developing the FICOS model system (Finnish Coastal Nutrient Load Model). FICOS includes a simple but sufficiently mechanistic biogeochemical (BGC) model with two functional phytoplankton groups (strict autotrophs and diazotrophic N 2 fixing cyanobacteria) accounting for key processes in the Baltic nitrogen‑phosphorus cycle. The BGC model is linked to a high-resolution 3D hydrodynamic coastal model and includes all important nutrient sources. We applied Bayesian inference on time series of dissolved inorganic phosphorus (DIP), dissolved inorganic nitrogen (DIN), chlorophyll a (Chla), and algal biomass observations to estimate five unknown phytoplankton growth parameters of the BGC model, and the uncertainties associated with them, as well as to calibrate the model predictions. In general, the calibrated BGC model simulated well the time courses of the observations at three intensive monitoring stations in the Archipelago Sea. Main sources of new nutrients to the Archipelago Sea include catchment area, atmosphere, point load (e.g. fish farming), N 2 fixation by cyanobacteria and background transport from open sea. An important and characteristic additional nutrient source in the shallow area is recycling from bottom sediments. During the season of phytoplankton growth, this internal load was the most important DIP source for the productive surface layer. In the sheltered basin coupled with the main river Aurajoki (area 1221 km 2 ), about half of the DIP released from sediments was transported to the surface. For DIN in the Aurajoki basin, atmospheric deposition and catchment load were the most important sources for phytoplankton, both accounting for about 30 % of total surface inputs. In the whole Archipelago Sea, annual inputs of DIP and total P were dominated by internal load while atmospheric load was the most important DIN source. Import from the northern Baltic proper was the other main source of both forms of N and P. During the growth season in the Aurajoki basin, high internal DIP load led to low surface DIN:DIP input ratio, indicating primary N-limitation of phytoplankton. The internal DIP load and catchment area DIN load were the local inputs with the widest impact areas in model simulations, suggesting that local inorganic nutrient loads were exhausted within the coastal zone during the growth season. Furthermore, Bayesian uncertainty analysis of nutrient load scenario predictions suggested that good environmental status (below 2.5–3.0 μg Chla L −1 ) is achievable in the Aurajoki basin only with drastic reductions in practically all anthropogenic loads. • Nutrient load and eutrophication responses were assessed in the Archipelago Sea. • Atmospheric and internal loads dominated inorganic N and P inputs, respectively. • Simple but sufficiently mechanistic biogeochemical model was developed. • Novel Bayesian inference algorithm was applied to deal with model uncertainties. • Good environmental status is achievable only after drastic load reductions.