Modeling the impact of reduced sea ice cover in future climate on the Baltic Sea biogeochemistry


Corresponding author: K. Eilola, Swedish Meteorological and Hydrological Institute, Sven Källfelts gata 15, SE-426 71 V Frölunda, Sweden. (


[1] In a warming future climate, the sea ice cover is expected to decrease, with very likely large consequences for the marine ecosystem. We investigated the impact of future sea ice retreat on the Baltic Sea biogeochemistry at the end of the century, using an ensemble of regionalized global climate simulations. We found that the spring bloom will start by up to one month earlier and winds and wave-induced resuspension will increase, causing an increased transport of nutrients from the productive coastal zone into the deeper areas. The internal nutrient fluxes do not necessarily increase because they also depend on oxygen and temperature conditions of the bottom water. Winter mixing increases in areas having reduced ice cover and in areas having reduced stratification due to increased freshwater supply. The reduced sea ice cover therefore partly counteracts eutrophication because increased vertical mixing improves oxygen conditions in lower layers.

1 Introduction

[2] Satellite observations indicate that primary production in the Arctic Ocean has increased since 2003 [Arrigo et al., 2008]. This increase is attributable both to decreased summer sea ice extent and to a longer phytoplankton growing season due to warmer water temperature. Scenario simulations of future climate match these findings and suggest that in the future, climate species composition may change considerably [e.g., Slagstad et al., 2011]. Hence, we may expect large changes for the Arctic ecosystem in the near future if the unprecedented shrinking of sea ice continues at today's observed rate [Doney et al., 2012].

[3] Although biogeochemical cycles in the Baltic Sea differ considerable from those in the Arctic Ocean, we expect also in the Baltic significant changes of primary and secondary production as a consequence of warming climate and shrinking ice cover [cf. Granskog et al., 2006]. It is likely that the Baltic Sea ecosystems may enter alternative states if tipping points are reached [Duarte et al., 2012]. For instance, algae and bacteria living within the sea ice, as well as seals that breed on ice, may completely lose their habitats. In the present climate, the Baltic Sea ice season lasts normally 5–7 months, from November to May. During a mild winter, ice occurs only in the Bothnian Bay, but during a cold winter the whole Baltic Sea is ice covered [e.g., Vihma and Haapala, 2009]. Because the interannual variability of the ice cover in the Baltic Sea is large [e.g., Vihma and Haapala, 2009], trends in biogeochemistry caused by shrinking ice cover are more difficult to detect than in the Arctic Ocean. In addition, the biogeochemical response to changes in forcing is slow. Finally, due to the complexity of coupled physical-biogeochemical processes in the water and sediment, it is difficult to separate and quantify specific effects of a sea ice retreat directly from observations [e.g., Cooper et al., 2012]. Hence, model studies are needed to investigate possible impacts of the changing climate in the Baltic Sea [e.g. Meier et al., 2012].

[4] Scenario simulations suggest that the annual maximum sea ice extent will decrease between 50 and 80% at the end of the century [Meier et al., 2004]. The projected reduction of future sea ice cover depends mainly on air temperature changes during winter. Both dynamical and statistical modeling suggest that the relationship between annual maximum ice extent and winter mean air temperature changes are nonlinear [Meier et al., 2004; Jylhä et al., 2008]. Hence, in the northern Baltic, sea ice can even be expected in the future climate (Figure 1).

Figure 1.

(a) Mean seasonal cycle of sea ice extent (103 km2) during 1969–1998 and mean sea ice concentration in March during (b) 1969–1998 and (c) 2070–2099. In Figure 1a, model results and observations are depicted by the black and red lines, respectively. The grey shaded area indicates the ±1 standard deviation of observations data. Locations of the major sub-basins of the Baltic Sea are indicated in Figure 1b.

[5] The effects of increased water temperature and decreased salinity in the future climate on the biogeochemistry have been investigated in numerous modeling studies mainly with focus on today's eutrophied Baltic proper [e.g., Meier et al., 2012]. However, the impact of the shrinking ice cover has not been investigated yet.

[6] In this study, we investigate the potential impact of reduced sea ice cover on the biogeochemistry in the northern parts of the Baltic Sea, i.e., Bothnian Bay, Bothnian Sea and Gulf of Finland, using an ensemble of four different regionalized global climate simulations calculated with a coupled of three-dimensional high-resolution physical-biogeochemical model. Because all scenario simulations differ only quantitatively, we focus on one experiment to illustrate and discuss the basic processes.

2 Methods

2.1 Coupled Ice-Ocean Model

[7] We used the three-dimensional circulation model RCO, the Rossby Centre Ocean model [Meier et al., 2003]. Subgrid-scale vertical mixing is parameterized using a turbulence closure scheme of the k-ε type with flux boundary conditions to include the effect of a turbulence enhanced layer due to breaking surface gravity waves and a parameterization for breaking internal waves. The ocean model is coupled to a Hibler-type sea ice model with elastic-viscous-plastic rheology, with explicitly resolved ice thickness distribution, i.e., ice concentrations of variable thickness categories, mechanical redistribution of the ice due to deformations and ice strength, and with a more detailed description of the ice strength. RCO is used with a horizontal resolution of 3.7 km (2 nautical miles) and with 83 vertical levels with layer thicknesses of 3 m. For further details of the RCO model and the multicategory sea ice model, the reader is referred to Meier et al. [2003] and Mårtensson et al. [2012], respectively.

2.2 Biogeochemical Model

[8] The Swedish Coastal and Ocean Biogeochemical model [Eilola et al., 2009] describes the dynamics of nitrate, ammonium, phosphate, phytoplankton, zooplankton, detritus, and oxygen. Phytoplankton consists of three algal groups representing diatoms, flagellates and others, and cyanobacteria (corresponding to large, small, and nitrogen fixing cells). The growth rates depend on nutrient concentrations, irradiance, and water temperature. The modeled cyanobacteria also have the ability to fix molecular nitrogen. Organic matter sinks and enters the sediment containing benthic nitrogen and phosphorus. The sediment processes include oxygen-dependent nutrient regeneration and denitrification as well as permanent burial of nutrients. With the help of a simplified wave model, the combined effect of waves and current induced shear stress is considered to calculate resuspension of organic matter [Almroth-Rosell et al., 2011]. Organic matter that has not previously been in contact with the sediment is here called “not resuspended matter”.

2.3 Regional Climate Data Sets

[9] The forcing of the ocean model is calculated from the atmospheric surface fields from four climate change scenario simulations. These data were computed applying a dynamical downscaling approach using the regional climate model RCAO (Rossby Centre Atmosphere Ocean model, see Döscher et al. [2002]) with lateral boundary data from two global general circulation models, HadCM3 and ECHAM5, forced with two greenhouse gas emission scenarios [Meier et al., 2012, and references therein]. In this study, we focus on results driven by HadCM3 and the A1B greenhouse gas emission scenario. For further details of the downscaling method, results from our mini-ensemble, and the quality of the atmospheric forcing fields the reader is referred to Meier et al. [2011, 2012]. An evaluation of the RCO-Swedish Coastal and Ocean Biogeochemical model results in the control period is presented by Meier et al. [2012].

2.4 Experimental Strategy

[10] Mean changes were calculated from the differences between a future (2070–2099) and a control (1969–1998) period. Thus, the present investigation is based on averaged results relevant to time scales of changing climate (~30 years), whereas systematic changes in the large variations between individual years were not analyzed further. Spring is defined as the period from March to May. A two-sample T-test was used on the means from the 30 individual years in each period to investigate the significance of the mean changes shown in Figures 2 and 3. The null hypothesis is rejected at 1% significance level (p < 0.01). The well-mixed surface layer (WML) depth, defined by a density change of 0.125 from the surface [Levitus, 1982], is used to visualize the effect of changing stratification. The day of the spring bloom initiation in each year is here defined as the first day when the 10 day running mean of the vertically integrated primary production (PP, mg C m–2 day–1) shows an increasing trend in spring and exceeds 50% of the annual mean PP in the actual year. Hence, by this definition we focus on changes in the productivity in spring, which is the driving mechanism behind drastic increases in phytoplankton concentrations in the water. To further elucidate the direct role of reduced ice cover on the biogeochemistry, we performed a sensitivity experiment where during the control period the impact of sea ice on biogeochemical variables was completely removed. By this we modified the air-sea exchange of oxygen, the illumination, and the effect of wind-wave driven bottom stresses on sediment resuspension. Processes of the physical model were not changed.

Figure 2.

Mean spring changes of (a) mean significant wave height (cm), (b) mean underwater (surface) solar radiation (Wm–2), (c) mean WML depth (m), (d) average sediment concentration of organic matter (mmol N m–2), and (e) fraction of resuspended matter in the sediments (%). Only statistically significant changes (p < 0.01) are depicted. (f) The daily rate of primary production relative to the annual mean of the daily values during 1969–1998 (black dots) at the station F9 (64°42.5′N, 22°4′E) in the Bothnian Bay. A value of 100% corresponds to the annual mean value. The black and red solid lines show the 10 day running mean values of the control and future climates, respectively.

3 Results

3.1 Sea Ice Cover

[11] Although atmospheric forcing fields regionalized from general circulation model simulations have considerable biases compared to observed climate [Meier et al., 2011], the sea ice cover during the control period is relatively well simulated (Figure 1a). However, in particular during spring, simulated sea ice cover is overestimated. In the present climate, the Bothnian Sea, Bothnian Bay, and Gulf of Finland are in the mean ice covered every winter (Figure 1b). In the future climate, approximately all areas, except the Bothnian Bay, are projected to be ice free during early spring (Figure 1c), in agreement with earlier studies [Meier et al., 2004].

3.2 Waves, Light, and Stratification

[12] The change from ice covered to open water conditions has significant impact on both wave climate (Figure 2a) and underwater light conditions (Figure 2b). In spring, the mean significant wave height increases by 30 to more than 50 cm in large parts of the Gulf of Finland, Bothnian Sea, and the Bothnian Bay. The mean irradiance in spring increases by about 40–50 Wm–2 in the previously ice-covered coastal areas in the Gulf of Finland and the Bothnian Sea, while the increase is even larger in the Bothnian Bay. Significantly increased WML depths are found in the main parts of the Bothnian Bay and the Gulf of Finland (Figure 2c). Shallower WML depths are found in the Baltic proper, and in a few shallow areas in the north and in the Gulf of Riga.

3.3 Sediment Resuspension

[13] Increased wave-induced bottom stress causes increased resuspension of organic matter from the sediments [cf. Almroth-Rosell et al., 2011] in areas that previously were sheltered by the ice cover during most of the winter. For example, the area where resuspension occurs on average more than 1 day per month in spring increases from 13% to 24% of the total bottom area in the Gulf of Finland, and from 2% to 44% in the Bothnian Bay (not shown). In addition, the average frequency of resuspension events increases, e.g., in shallow areas (<6 m) from less than 2 (0.5) days per month to more than 4 (3) days per month in the Gulf of Finland (Bothnian Bay).

3.4 Sediment Concentrations

[14] With the exception of some deeper locations in the previously ice-covered gulfs, the amount of organic matter in the sediment generally decreases (Figure 2d). The increased sediment concentrations in the deeper Bothnian Sea and Bothnian Bay are caused by an increased deposition of both resuspended sediments and not resuspended matter (not shown). In the Gulf of Finland, however, only the amount of resuspended matter increases while the concentrations of not resuspended matter generally decrease (not shown). The effect of enhanced resuspension is reflected in the relative fraction of resuspended sediments in the deeper areas that increases especially in the Gulf of Finland (Figure 2e).

3.5 Start of the Spring Bloom

[15] In the Bothnian Bay the spring bloom starts up to 20–30 days earlier, with a more intense peak of the production (Figure 2f), in areas that were previously ice covered (Figure 3a). A similar change of the start of the spring bloom is also found in the northern parts of the Gulf of Finland while the changes are small in areas having less severe ice conditions under the control climate. Also, the Kattegat and the Belt Sea show significant changes of the start of the spring bloom but these changes are not connected to the changing ice conditions, as the results of the sensitivity experiment show (Figure 3b).

Figure 3.

Mean changes of the day of the spring bloom initiation (number of days) in the (a) scenario and (b) sensitivity simulations. Only statistically significant changes (p < 0.01) are depicted.

3.6 Sensitivity Experiment

[16] The phytoplankton bloom development usually starts when surface layer stratification increases and light conditions improve. Because in the sensitivity experiment mixing and stratification are unchanged when the ice is removed, the earlier start of the spring bloom is in this case only caused by the increased availability of light (Figure 3b). The changes in winter nutrient concentrations, sediment resuspension, and phytoplankton production are quite small and statistically not significant (not shown). The increase in significant wave height is much smaller than in the future projection (Figure 2a).

3.7 Ensemble Simulations

[17] Qualitatively, similar changes as described above are obtained in all four projections from Meier et al. [2012], but the impact from sea ice retreat is smaller in ECHAM5-driven simulations compared to HadCM3 driven simulations due to a warm bias in the north causing too small sea ice cover in the control period as compared to observations [Meier et al., 2011]. Consequently, the largest changes in wind and significant wave height are found in the HadCM3-driven scenario simulations.

4 Discussion

[18] The amount of organic matter in the sediment decreases generally in the future Baltic Sea due to an intensified nutrient mineralization caused by increased water temperatures [Meier et al., 2012]. However, we found a significant increase especially in the deeper bottoms of the Bothnian Sea and Bothnian Bay due to increased transports of resuspended matter and increased productivity; the latter forced by improved light conditions and increased nutrient loads from land and from the Baltic proper [Meier et al., 2012]. Although significant wave heights increase in the sensitivity experiment, these changes are small compared to the future projection. In warmer climate the wind over ice-free areas is significantly increased due to a decreased stability in the planetary boundary layer over sea [Meier et al., 2012]. Organic matter may therefore be deposited in larger areas and deeper lying sediments due to more frequent wave-induced resuspension. In particular, in the Gulf of Finland the fraction of resuspended organic matter increases significantly when the ice cover retreats. Thus, one might expect a decrease of the bottom oxygen concentrations here because of an increased oxygen demand for the oxidation of organic matter. This effect is, however, overshadowed by the impact of a weaker vertical stratification [Meier et al., 2012]. Increased wind-induced mixing and river runoff in warmer climate cause a weaker vertical stratification and, consequently, an enlarged ventilation of the bottom water with oxygen-rich surface water. Actually, in the deep Bothnian Sea the bottom water oxygen concentrations are projected to decrease [Meier et al., 2012], but quantification of the impact from increased resuspension is not straightforward because other effects, like changing stratification and a larger productivity, are also important. The decreased fraction of resuspended sediments observed on the eastern side of the Baltic proper, Bothnian Sea, Bothnian Bay, and the northern parts of the Gulf of Riga, are mainly caused by an increased sedimentation of not resuspended matter (not shown). The enhanced productivity and sedimentation in these coastal areas is likely supported by the intensified exchange of nutrients between shallow and deeper waters in warmer climate, as suggested by Eilola et al. [2012]. Other studies suggest that the increased availability of light in previously light limited regions increases the phytoplankton production, e.g., by a deepening of the euphotic layer [Yun et al., 2012]. Indeed, in our sensitivity experiment the productivity increased during the early spring period when previously ice-covered areas became ice free, but changes of the average spring bloom or annual production are found insignificant. The integrated production is instead limited by the availability of nutrients, which is not significantly affected in our sensitivity experiment. However, according to the results of our scenario simulations, the availability of nutrients in the future climate will change mainly because in shallow seas other drivers, like changing hydrographic conditions and external nutrient supply, are more important. However, in other parts of the World Ocean that are today permanently ice covered, like the Arctic Ocean, the direct impact of reduced ice cover on biogeochemical cycles might be larger.

[19] The start of the spring bloom in the future scenario occurs later than in the sensitivity experiment due to shading by ice in some areas and a deeper WML in spring caused by the weaker stratification. In these areas with a deeper WML larger illumination is therefore needed before a positive net phytoplankton community growth may happen. The earlier onset of the spring bloom in the Kattegat and the Belt Sea, however, is not connected to the changing ice conditions. Instead, the changes might possibly be explained by a combination of a shallower WML and strengthened stratification, changing light conditions, nutrient limitation, and water temperature, but further analysis of this issue are beyond the scope of the present study. The weaker stratification in ice-free areas in spring likely benefit blooms of diatoms, while increased temperatures and strengthened thermal stratification in summer [Hordoir and Meier, 2012] may benefit smaller phytoplankton and potentially increase the occurrence of cyanobacteria blooms in the north. The sea ice retreat prolongs the “after spring bloom” period with low nutrient concentrations in the surface layers, when primary production is based mainly on regenerated nutrients. In combination with increased illumination and nutrient concentrations, the longer growing season may affect the phytoplankton communities and the autotrophic vs. heterotrophic food web structure in the northern Baltic Sea [Jaanus et al., 2011; Dahlgren et al., 2010]. An increased supply of dissolved organic carbon (reduced water transparency) and increased freshwater runoff (deeper halocline) will also have an impact on the food web structure [Berglund et al., 2007]. The increased illumination benefits the phytoplankton production and the transfer of energy to higher trophic levels while reduced water transparency and an increased bacteria-based food web may reduce pelagic productivity at higher trophic levels. Detailed studies of these complex interactions in the northern Baltic require models that include, for instance, the dynamics of dissolved organic nutrients, the microbial food web, and sea ice ecosystems [Klais et al., 2011; Tedesco et al., 2010; Sibert et al., 2010], which are still not available [Eilola et al., 2011].

5 Conclusions

[20] We conclude that cause-and-effect studies with coupled physical-biogeochemical models are useful tools to disentangle and understand the chain of processes involved in the response to climate change. As a consequence of the shrinking ice cover in warmer climate, we found that the spring bloom in the Baltic Sea will start and end considerably earlier. Other changes in biogeochemical cycling cannot unambiguously be attributed to the sea ice retreat because the impacts of changing water temperature, salinity, horizontal nutrient transport, and external nutrient supply might be important as well. We suggest that the results of our process study might also be applicable for other ice-covered marginal seas.


[21] This work was part of the project ECOSUPPORT (Advanced modeling tool for scenarios of the Baltic Sea ECOsystem to SUPPORT decision making) and has received funding from the European Community's Seventh Framework Program (FTP/2007-2013) under grant agreement 217246 made with BONUS, the joint Baltic Sea research and development program, and from the Swedish Environmental Protection Agency (08/381). Additional support came from the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (FORMAS) within the project “Impact of accelerated future global mean sea level rise on the phosphorus cycle in the Baltic Sea” (grant 214-2009-577) and from the Swedish Research Council (Vetenskapsrådet) within the project “Modeling climate variability of the Arctic Ocean in past and future climates with special focus on changing sea ice” (grant 621-2006-5030). Numerical simulations were partly performed on the climate computing resources ‘Vagn’ and ‘Ekman’ that are operated by the National Supercomputer Centre (NSC) at Linköping University and the Centre for High Performance Computing (PDC) at the Royal Institute of Technology in Stockholm, respectively. These computing resources are funded by a grant from the Knut and Alice Wallenberg foundation.