Sensitivity of Hudson Bay Sea ice and ocean climate to atmospheric temperature forcing
A regional sea-ice–ocean model was used to investigate the response of sea ice and oceanic heat storage in the Hudson Bay system to a climate-warming scenario. Projections of air temperature (for the years 2041–2070; effective CO2 concentration of 707–950 ppmv) obtained from the Canadian Regional Climate Model (CRCM 4.2.3), driven by the third-generation coupled global climate model (CGCM 3) for lateral atmospheric and land and ocean surface boundaries, were used to drive a single sensitivity experiment with the delta-change approach. The projected change in air temperature varies from 0.8°C (summer) to 10°C (winter), with a mean warming of 3.9°C. The hydrologic forcing in the warmer climate scenario was identical to the one used for the present climate simulation. Under this warmer climate scenario, the sea-ice season is reduced by 7–9 weeks. The highest change in summer sea-surface temperature, up to 5°C, is found in southeastern Hudson Bay, along the Nunavik coast and in James Bay. In central Hudson Bay, sea-surface temperature increases by over 3°C. Analysis of the heat content stored in the water column revealed an accumulation of additional heat, exceeding 3 MJ m−3, trapped along the eastern shore of James and Hudson bays during winter. Despite the stratification due to meltwater and river runoff during summer, the shallow coastal regions demonstrate a higher capacity of heat storage. The maximum volume of dense water produced at the end of winter was halved under the climate-warming perturbation. The maximum volume of sea ice is reduced by 31% (592 km³) while the difference in the maximum cover is only 2.6% (32,350 km2). Overall, the depletion of sea-ice thickness in Hudson Bay follows a southeast–northwest gradient. Sea-ice thickness in Hudson Strait and Ungava Bay is 50% thinner than in present climate conditions during wintertime. The model indicates that the greatest changes in both sea-ice climate and heat content would occur in southeastern Hudson Bay, James Bay, and Hudson Strait.
Numerical models provide a useful tool to simulate the response of climate to changing atmospheric forcing. The freshwater budgets and sea-ice variability of Hudson Bay and James Bay were first studied with the aid of numerical models during the 1980s to quantify the oceanic response to runoff modifications associated with the development of hydroelectric facilities in northern Quebec and Manitoba (Prinsenberg 1980, 1991). Later, the impact of global warming on the Hudson Bay marine system was assessed with a first generation Global Climate Model (GCM) (Ingram et al. 1996), followed by several climate change impact studies using projections from different generations of GCMs and different warming scenarios (Gough 1998; Gough and Allakhverdova 1999; Gough and Wolfe 2001; Gagnon and Gough 2005). Future projections derived from the first generation of coupled atmosphere–ocean GCMs include a yearly increase in air temperature over Hudson Bay ranging from 2.5 to 4.5°C in an increased CO2 scenario (from year 2040 to 2069). The largest rise occurs in winter (Gagnon and Gough 2005). Changes in sea-ice cover were highly variable depending on the GCM used. However, these projections were made using a relatively coarse resolution of the oceanic domain, represented as an enclosed sea of from 5 to 31 ocean grid points depending on the GCM used. The coarse resolution of Hudson Bay in GCMs justifies the use of a regional model with higher resolution to resolve regional scales of interest.
High-resolution sea-ice (Wang et al. 1994a, b) and ocean circulation models of Hudson Bay (Saucier and Dionne 1998; Saucier et al. 2004) have been developed. In the first version of the sea-ice–ocean model, simple sensitivity experiments investigated the sea-ice–ocean response to modified atmospheric and hydrologic forcing (Saucier and Dionne 1998). A uniform warming of 2°C produced a 20% reduction in winter sea-ice volume, an increase of 4°C in summer sea-surface temperature, and an increase of 30 days in the ice-free period (Saucier and Dionne 1998). However, the specific bathymetry of Hudson Bay was not considered, and tidal mixing and mesoscale dynamics were not resolved in the model by Saucier and Dionne (1998). Moreover, the prescribed warming did not include spatial and seasonal variability. The latest version of this three-dimensional regional sea-ice–ocean model can produce multi-year simulations with good agreement to observations (Saucier et al. 2004). The mesoscale dynamics observed in Hudson Bay are reproduced with the 10 km horizontal resolution of the model. The objective of the present study is to investigate the impact of a warmer climate scenario on the Hudson Bay marine system using the sea-ice–ocean model presented in Saucier et al. (2004).
2.1 Numerical model
The numerical model used in this study is a high-resolution regional oceanic model developed by Saucier et al. (2004). The model uses the hydrostatic shallow-water incompressible formulation from Backhaus (1983, 1985), Stronach et al. (1993), and Saucier et al. (2003). The ocean model is coupled with a dynamic (Hunke and Dukowicz 1997) and thermodynamic (Semtner 1976) two-layer sea-ice model with snow cover as an independent layer. Turbulence is incorporated as a level 2.5 turbulent kinetic energy equation. Heat, salt, and momentum fluxes at the ocean–ice–atmosphere interfaces are represented with the bulk aerodynamic formula of Parkinson and Washington (1979). The conservation equations of momentum and mass are discretized on an Arakawa C grid. The model utilizes a 10 km horizontal resolution projected onto a polar stereographic grid and has a 10 m vertical resolution for up to 36 depth intervals. Surface and bottom layer thickness are a function of local water level and bathymetry that are based on realistic bathymetry data (Jackobsson et al. 1996; Sandwell et al. 2000). The model domain covers Hudson Bay, James Bay, Foxe Basin, and Hudson Strait, with open boundaries at the Hudson Strait–Labrador Sea connection to the east and at Fury and Hecla Strait in the northwest (Fig. 1). Atmospheric forcing includes 2 m air and dew point deficit, cloud cover fraction, precipitation rate, and 10 m winds. Atmospheric, hydrological, and tidal forcing as well as the initial and boundary conditions for temperature and salinity were those used in Saucier et al. (2004). The time step of the ocean model is 300 s, while a 600 s step is used for the sea-ice model.
2.2 Numerical experiments
Most of the results presented in this study and their interpretations focus on the anomalies between the future scenario and the present simulations for oceanographic conditions in salt and temperature. All the time series data illustrated herein are full domain-averaged, which includes Foxe Basin, James and Hudson bays, and Hudson Strait.
3.1 Temperature and salinity variability
3.2 Sea-ice sensitivity and spatio-temporal variability
Median dates of freeze-up and break-up for Foxe Basin, Hudson Bay, and James Bay for the present climate simulation and the warmer climate scenario
New sea-ice season length
~6 ½ months
~5 ½ months
We performed a coupled sea-ice–ocean simulation of the Hudson Bay marine system for a current and a warmer climate following a RCM-derived climate scenario, as used in Music and Caya (2007). Our objective was to quantify the response of Hudson Bay sea-ice cover and its variability to a warmer atmospheric forcing keeping all other forcings such as wind or river run-off unchanged. The use of a regional sea-ice–ocean model was necessary to integrate the specific estuarine circulation and the local thermohaline processes of a freezing estuary into the global response of the Hudson Bay system.
Regional climate change scenarios of subarctic environments such as the Baltic Sea have been widely studied (Meier 2002a, b, 2006; Meier and Döscher 2002; Döscher and Meier 2004; Meier et al. 2004). These earlier works provide a reliable methodology to develop a robust climate change scenario. In Meier (2002a), the introduction of a spin-up strategy to initialize future oceanic conditions is discussed. The spin-up time scale is closely related to the residence time of the modelled system. In addition, the spin-up experiment produces new conditions of stratification that integrate future climate variability. In Meier (2006), the simulation period (96 years) was long enough to allow almost 60 years of integration to spin-up the stratification in their scenario simulations. However, their surface atmospheric forcing for the 1902–1998 reference period has been reconstructed using a statistical model that produces monthly mean atmospheric fields except for the high frequency input of sea-level pressure (Kauker and Meier 2003). This approach to reconstruct a 96-year time period has not yet been performed for Hudson Bay. Our coupled sea-ice–ocean model uses high spatial and temporal resolution, atmospheric forcing, and river runoff data for the 2001–2005 period. The delta-change approach has been successfully used to transfer the signal of climate change either to the ocean system (Meier 2006) or to the drainage basin (Graham 2004) of the Baltic Sea. Meier (2006) calculated monthly mean changes of atmospheric forcing from present and future 30-year time slice experiments. The same approach was successfully applied in our study.
The simulation of a warmer climate over the Hudson Bay marine system shows a lengthening of the ice-free season by more than 2 months. A simple analysis of the internal mass and energy structure of seawater revealed that these temporal changes are associated with a modification in the seasonal variability of water stratification. These simultaneous changes revealed the strong coupling between sea ice and the surface layer stratification (Manak and Mysak 1989; Backhaus and Kämpf 1999; Steele and Boyd 1999). It is worth noting that a perturbation of the atmospheric forcing results in a significant change of the freshwater balance in terms of both timing and intensity. In an estuarine environment, such as the Hudson Bay system, where density stratification is primarily controlled by salinity, stabilization of the mixed layer through a large seasonal input of freshwater is a dominant mechanism for controlling atmospheric heat loss and sea-ice freezing (Rudels et al. 1999). In Hudson Bay, a lower amount of spring meltwater weakens the halocline, thus enhancing the downward mixing of heat during summer and fall. Haline convection at the beginning of winter takes place within a deeper mixed layer, which delays surface freezing. Moreover, the heat content analysis shows a warmer water column year-round. The model shows that heat storage at depth occurs in stable waters along the eastern shores of James and Hudson bays. Thermodynamic growth of sea ice, which lowers surface buoyancy fluxes through brine rejection, is also delayed and reduced in the warmer climate scenario. The simulated atmospheric perturbation sustains a positive feedback that decreases sea-ice production. Results from the present study suggest that the preconditioning of the mixed layer is a dominant factor influencing the thermodynamic growth of sea ice in Hudson Bay. This evidence supports previous findings on the positive correlation between river runoff and sea-ice extent (Manak and Mysak 1989; Wang et al.1994b) or thickness (Saucier and Dionne 1998) inside the Hudson Bay system.
We found that southern and coastal regions as well as Hudson Strait show greater sensitivity to a warmer climate in terms of sea ice and water column heat content. The storage of additional heat along the coast during fall is explained by winds. The pattern of the heat content anomaly in fall perfectly matches the mean sea-surface slope generated by wind stress in Hudson and James bays (Wang et al. 1994c). Warm surface waters are advected toward the coast, removing additional heat from central regions. Our results are different from those of a similar regional climate study conducted over the Baltic Sea (Meier 2002a, b, 2006)—a European Inland Sea located at a similar latitude. The highest changes in summer SST for the Baltic Sea were projected in the northernmost basins (see Fig. 9 in Meier 2002a) and mostly in the central Bothnian Sea compared to shallow coastal areas (see Fig. 13 in Meier 2006). Coastal areas in the Baltic Sea were found to be less sensitive to the warming signal than were central parts of the basins (Meier 2006). This author also concluded that the sensitivity of sea ice to salinity was relatively small in the Baltic Sea (Meier 2002b). In our study, the greatest oceanic warming signal observed is correlated to the wind-driven and buoyancy-driven circulation along the coast of Nunavik. Sea-ice redistribution by the winds in Hudson Bay (the polynya to the northwest and accumulation of sea ice to the southeast (Saucier et al. 2004)) is a reason why the Hudson Bay is different from the Baltic Sea.
Southern regions of Hudson Bay undergo the largest changes in the simulations. The strong decline in sea-ice thickness in James Bay and southeastern Hudson Bay can be explained by a reduction of both thermodynamic and dynamic growth of sea ice. The southward drift of sea ice by winds (Maxwell 1986) leads to the accumulation of pack ice in southern Hudson Bay along the Ontario coast (Wu et al. 2006) and particularly in James Bay, where ridging was found to be important during April (Saucier et al. 2004). Pressure ridges act as stores for ice mass and constitute additional sources of freshwater at the onset of melting. The confined geography of James Bay facilitates the formation of pressure ridges. The warmer climate lowers the ridge production rate in terms of dynamic height increase. However, the relative decrease in ridging is not as large as the relative thinning of the mean sea-ice cover. Because thin ice is preferentially used before thicker ice as a primer in ridging schemes (Thorndike et al. 1975; Amundrud et al. 2004; Lipscomb et al. 2007), one could hypothesize that ridging events in our simulation are more frequent with thinner pack ice although the net volume production is lower. This hypothesis is difficult to demonstrate with a field survey since the observed thinning of the Arctic sea-ice cover is controlled by large-scale atmospheric and oceanic processes that act simultaneously (Rothrock et al. 1999). Ridging has more often been studied as a causal mechanism to explain changes in Arctic sea-ice thickness (Shoutilin et al. 2005) rather than the reverse. In modelling results of the Arctic basin, Makshtas et al. (2003) found that changes in atmospheric circulation decreased the concentration of ridges, which in turn was responsible for the observed thinning of Arctic sea ice (Rothrock et al. 1999).
The production of cold, brine-enriched waters occurs inside wind-driven coastal polynyas mainly located in western Foxe Basin (Defossez et al. 2008) and northwestern Hudson Bay (Markham 1986). These areas exhibit high rates of sea-ice formation and brine rejection that produce convecting cold and salty water masses, as demonstrated by observations and model simulations (Saucier et al. 2004; Defossez et al. 2008). Our results suggest that a warmer atmosphere slows the ventilation of bottom waters in Hudson Bay via the slowdown of the polynya activity. While the overall 3.9°C warming simulated in our study reduces the maximum volume of dense water produced by 50%, a sensitivity study conducted on a coastal polynya in Antarctic showed that dense water export was reduced by 40% for a surface temperature increase of 2°C (Marsland et al. 2007). These similar results from a shallow Canadian Inland Sea and the east Antarctic continental shelf illustrate the strong connection between atmospheric surface processes and the buoyancy-driven ventilation of bottom waters. Although the ventilation of bottom waters inside a shallow basin such as the Hudson Bay marine system differ significantly from deep ocean ventilation, the seasonal variability of vertical salt fluxes is greatly affected, which in turn altered the baroclinic circulation at all depths.
The decrease in maximum sea-ice volume by 31% contrasted greatly with the 2.6% change in maximum sea-ice extent. The greater impact on sea-ice thickness rather than extent reveals the strong polar influence that prevents a large reduction in the ice sheet during winter. It also reflects the sea-ice thickness sensitivity from water column preconditioning. These results also demonstrate the relative protection of the Hudson Bay marine system from large lateral oceanic heat input compared to the Arctic Ocean, which underwent an accelerated decline of its ice cover faster than was anticipated by numerical models (Stroeve et al. 2007).
This numerical experiment has shown that changes in sea-surface temperature could exceed the change from air temperature forcing. The input of thermal energy from shortwave radiation is greatly increased through the lengthening of the ice-free season. In the present climate simulations, the persistence of the ice cover around summer solstice is sufficient to reflect a large proportion of incident solar radiation. Solar heat input is therefore inhibited through a decoupling between the peak of solar radiative forcing and the highest proportion of low albedo surface. Sea ice plays a significant role in the absorption and partitioning of solar radiative energy in the atmosphere, sea ice, and ocean (Jin et al. 1994). The simulated warmer climate produces thinner sea ice, but this thinning trend allows higher input of solar energy (Jin et al. 1994; Perovich 2005). The increasing absorption of solar energy through an extended ice-free period and the thinning of sea ice increase the positive ice–albedo feedback. Maykut and McPhee (1995) demonstrated that shortwave radiation, absorbed below the bottom of the pack ice, is the main energy source in the oceanic heat flux between water and ice. Therefore, the increase of solar heat input into the ocean unbalances the fragile thermodynamic equilibrium and accelerates warming at high latitudes. We have provided a general pattern of changes in the sea-ice climate associated with the variability in temperature and salinity profiles for the Hudson Bay marine system. However, we note in closing that the impact of global warming at a regional scale requires further investigation, specifically regarding the freshwater export toward the Labrador Sea in the perspective of changing river discharge for the Hudson Bay drainage system (Déry et al. 2005).
The authors wish to thank James Caveen and François Roy for technical support and Dr. Christopher-John Mundy for helpful advice and encouragement. Hydrological data were provided by HydroQuébec, Water Survey of Canada (Environment Canada), and the Ministère du Développement Durable, de l’Environnement et des Parcs (Government of Québec). The authors are members of the Canadian ArcticNet Program. This work is a contribution to the ArcticNet Program, funded in part by the Network Centres of Excellence (NCE) Canada.
- Backhaus JO (1983) A semi-implicit scheme for the shallow water equations for application to shelf sea modeling. Cont Shelf Res 2:234–254Google Scholar
- de Ramón Elía, Caya Daniel, Côté Hélène, Frigon Anne, Biner Sébastien, Giguère Michel, Paquin Dominique, Harvey Richard, Plummer David (2007) Uncertainty study of an ensemble of CRCM regional climate simulations over North America. Clim Dyn 30:113–132Google Scholar
- Döscher R, Meier HEM (2004) Simulated sea surface temperature and heat fluxes in different climates of the Baltic Sea. Ambio 33:242–248Google Scholar
- Gough WA, Wolfe E (2001) Climate change scenarios for Hudson Bay, Canada, from general circulation models. Arctic 54:142–148Google Scholar
- Graham LP (2004) Climate change effects on river flow to the Baltic Sea. Ambio 33:235–241Google Scholar
- Ingram RG, Wang J, Lin C, Legendre L, Fortier L (1996) Impact of freshwater on a subarctic coastal ecosystem under seasonal sea-ice cover (Southeastern Hudson Bay, Canada), I, Interannual variability and predicted global warming influence on river plume dynamics and sea ice. J Mar Syst 7:221–231CrossRefGoogle Scholar
- Manak DK, Mysak LA (1989) On the relationship between Arctic sea-ice anomalies and fluctuations in northern Canadian air temperature and river discharge. Atmos Ocean 27:682–691Google Scholar
- Meier HEM, Döscher R (2002) Simulated water and heat cycles of the Baltic Sea using a 3D coupled atmosphere-ice-ocean model. Boreal Environ Res 7:327–334Google Scholar
- Meier HEM, Döscher R, Halkka A (2004) Simulated distributions of Baltic Sea-ice in warming climate and consequences for the winter habitat of the Baltic ringed seal. Ambio 33:249–256Google Scholar
- Mysak LA, Ingram RG, Wang J, Van der Baaren A (1996) The anomalous sea-ice extent in Hudson Bay, Baffin Bay and the Labrador Sea during three simultaneous NAO and ENSO episodes. Atmos Ocean 34:313–343Google Scholar
- Prinsenberg SJ (1991) Effects of hydro-electric projects on Hudson Bay’s marine and ice environments. James Bay Publication, James Bay, Series No. 2, pp 1–8Google Scholar
- Sandwell DT, Walter H, Smith F (2000) Bathymetric estimation. In: Fu LL, Cazenave A (eds) Satellite altimetry and earth sciences: a handbook of techniques and applications. International Geophysics Series, vol 69. Academic Press, San Diego, USAGoogle Scholar
- Steele M, Boyd T (1999) Retreat of the cold halocline layer in the Arctic Ocean. J Geophys Res 103(C05):10419–10435Google Scholar
- Stronach JA, Backhaus JO, Murty TS (1993) An update on the numerical simulation of oceanographic processes in the waters between Vancouver Island and the mainland: the GF8 model. Oceanogr Mar Biol Annu Rev 31:1–86Google Scholar
- Tivy A, Alt B, Howell S, Wilson K, Yackel J (2006) On the relationship between ENSO, the NAO, the PNA and anomalous spring ice cover in Hudson Bay. Proceedings of the ArcticNet Annual Conference, Banff, 13–16 December 2005Google Scholar
- Wang J, Mysak LA, Ingram RG (1994b) Interannual variability of sea-ice cover in Hudson Bay, Baffin Bay and the Labrador Sea. Atmos Ocean 32:421–447Google Scholar
- Wu W, Barber D, Iacozza J, Mosscrop D (2006) Trends of spatial and temporal patterns of sea ice concentrations in Hudson Bay region over the period 1971 to 2004. Proceedings of the ArcticNet Annual Conference, Banff, 13–16 December 2005Google Scholar
- Zhang X, Vincent LA, Hogg WD, Niitsoo A (2000) Temperature and precipitation trends in Canada during the 20th century. Atmos Ocean 38:395–429Google Scholar