Sources of uncertainty of Baltic Sea future projections

In an ensemble of regional scenarios for the Baltic Sea we analyzed the sources of uncertainty in climate indices and environmental quality indicators. The ensemble is based on 32 regionalized scenarios where four different external drivers have been varied. Climate is represented by four different Earth System Models (ESMs). Uncertain future greenhouse gas emissions are represented by two different Representative Concentration Pathways (RCPs). Two nutrient load scenarios, broadly equivalent to two Shared Socio-economic Pathways (SSPs), describe two distinct evolutions of the regional population development, agricultural practices and food demand and two scenarios for global mean sea level rise (GMSL) measure the impact of the water level on the biogeochemical cycle in the Baltic Sea. The volume averaged temperature increase at the end of the century relative to the reference period 1976-2005 is 1.3 to 2.2 K (RCP 4.5) and 2.9 to 4.2 K (RCP 8.5). Averaged salinity changes by -2.1 and +0.2 g/kg (RCP 4.5) and -3.2 and -0.2 g/kg (RCP 8.5). For temperature, uncertainties before 2080 are dominated by natural variability and ESM biases. After 2080 the largest source of uncertainty is related to the unknown greenhouse gas concentrations. As expected, uncertainties related to either SLR or nutrient loads are negligible. For salinity, the dominating source of uncertainty during the entire 21st century is explained by the biases of the ESMs. However, natural variability and, in particular by the end of the century, uncertainties due to unknown greenhouse gas concentrations and sea level rises are important as well. For hypoxic area, uncertainties before 2040 are dominated by ESM biases. After 2040 the largest source of uncertainty is related to the unknown nutrient loads (SSPs). However, ESM biases, natural variability, unknown greenhouse gas concentrations and unknown sea level rises play an important role as well. Hence, the predictability of hypoxic area on long time scales requires accurate knowledge of various drivers and accurate quality of ESMs.