The IPCC Fourth Assessment Report (AR4) undertook an evaluation of the evidence for and impacts of anthropogenic change worldwide where they concluded that human-induced climate change was occurring (IPCC, 2007). As part of the IPCC process, the results from several Ocean-Atmosphere Global Circulation Models were presented. The performance of 20 models for different Arctic regions, including the Barents Sea, was evaluated by Overland and Wang (2007).
Their assessment was based upon each model’s ability to simulate observed seasonal changes in ice concentrations for the period 1979-1999. For the Barents Sea, a limit of within 30% was used to determine acceptable models and those exceeding 30% were considered unacceptable. The reasoning was that the models should be able to hindcast the present day conditions if they are to do a good job on future projections. Most of the models produced too much ice in the Barents, as only 7 models met the acceptable criteria.
By 2050 using A1B scenario, 5 of these 7 models indicate a 40% or more loss of sea ice in the Barents Sea. The annual mean temperature in the Barents Sea at the end of the 21st century under the A1B scenarios based on the ensemble mean from all of the IPCC GMCs shows maximum temperature increase of around 7°C, but this is felt to be too high due to an overestimate of the albedo feedback caused by the removal of the present-day simulations’ excessive sea-ice cover (IPCC, 2007).
The Bergen Climate Model (BCM) was not one of the 7 models that performed well in the IPCC evaluation but an earlier version of that model had produced more realistic ice coverage. Using the earlier version of the BCM, Furevik et al. (2002) developed future climate scenarios for the Barents Sea. By 2080, they suggested surface ocean temperatures will warm 1° to 2°C (Figure 4.6.1), winter sea ice will almost disappear, Atlantic waters will spread farther eastward and northward, and the surface mixed-layer depth will increase due to stronger winds. Climate scenarios obtained from the regional climate model REMO of the Max Planck Institute for Meteorology in Hamburg, Germany and forced by a global climate model driven by a B2 scenario suggested a 25% increase in freshwater runoff to the Barents Sea and the snow season was projected to be 30-50 days shorter, with the peak spring discharge occurring about 2-3 weeks earlier than in the present day but remaining dominated by snowmelt (Dankers and Middelkoop, 2008). In spite of this, model studies are predicting an increase in salinity due to higher salinities in the Atlantic Water inflows, generated by higher evaporation in the tropics (Betke et al., 2006). Modelling studies by Ellingsen et al. (2008) suggested that higher temperatures in this inflow resulted in the fraction of water in the Barents Sea with temperatures >1°C increased by 25% between 1995 and 2059 (the same magnitude as the present seasonal change) but with high interannual and multi-decadal variability. They also noted that sea-ice coverage will decrease with the largest decline during the summer resulting in virtually ice free conditions by 2059. Huse and Ellingsen (2008) examined changes in the position of the Polar Front that separates the cold Arctic and warm Atlantic waters. The frontal position was projected not to change much in the western Barents, where it is tied to topographic features, but in the eastern Barents the front will move farther north and east (Figure 4.6.2).
Recently, Paul Budgell (IMR, personal communication) used the GISS Ocean-Atmosphere Model to downscale to a regional model of the Barents Sea based on ROMS (Regional Ocean Modeling System). The GISS OAM was chosen based on its selection by Overland and Wang (2007). Temperature results from ROMS for 1986-2000 (present) to 2051-2065 (future) for 0-50 m, 50-100 m, and >100 m to the bottom suggested increases in the sea temperatures throughout the Barents Sea were typically 1°C in both winter and summer. The largest increase occurred during summer with temperature increases of 2 to <4°C over the upper 50 m in the eastern (>30°E) and northern (>78°N) regions of the Barents Sea. In the 50-100 m layer, temperatures increased by the same amount but only in the eastern Barents Sea while in this same area in the layer from 100 m to the bottom, temperatures increased from 1° to <3°C. In winter, temperatures rose by 2 to 3°C but were restricted to the eastern region in the upper 50 m layer and about 1°C less in the layers below 50 m. While future projections of the summer distribution of ice indicated almost no ice left in the Barents Sea, there was still ice left in winter, including most of the northern region as well as a narrow band of ice immediately to the west of Novaya Zemlya. Comparison of ice concentrations with present conditions showed a decrease into the future with the largest changes in the eastern area of the Barents Sea and somewhat lower ice concentrations also in the north.
It must be cautioned that the atmospheric and ocean climate scenarios remain highly uncertain. Better regional models of the Barents through improved downscaling from the GCMs are required. There is a need to undertake the downscaling using several GCMs and then take an ensemble mean. This should provide a better estimate and indicate the uncertainty in the projections. Also, there is a need to couple the atmosphere and ocean for the regional models, which even in the recent modelling by Budgell has not been attempted. In a coupled model the changes in the ocean feedback to the atmosphere and influence it. In an uncoupled model, there is no feedback.