Adaptation Actions for a Changing Arctic: Perspectives from the Barents Area

71

Chapter 4 · Physical and socio-economic environment

observed in the lower atmosphere during summer and early autumn.Variability in cloud cover has been linked to sea-ice variability near the ice margin (Schweiger et al., 2008) and retreating sea ice may be associated with a response governed by several factors, with a decrease in low-level cloud amount and an increase in mid-level clouds (Sato et al., 2012). Stramler et al. (2011) found conditions such as overcast or cloudless skies had a strong effect on the Arctic atmosphere and on phenomena such as storms at the surface. Extensive cloud cover prevents surface heat loss and so reinforces warming for sea ice and snow, in contrast clear skies allow heat to be lost back to space, thus cooling the surface. These effects are particularly strong in the Arctic, because cold air holds less water vapor and thus absorbs less heat energy. Although frequent transitions between clear and cloudy states in winter have been associated with pronounced swings between warm and cold sea-ice temperatures, there seems to be no clear upward or downward trend in the past observations (Stramler et al., 2011). When seasonal cloudy skies begin to dominate in spring, temperatures begin to rise and the sea ice is brought closer to its melting temperature, even before the heating effect of increased sunlight takes effect. Hence, springtime clouds may bring forward the date for melt onset, and thus more frequent Arctic clouds under a warmer climate could accelerate sea-ice decline. Dobler et al. (2016) analyzed cloud characteristics in a set of projections and found a decrease in winter mean cloud cover over the Barents Sea and an increase over the nearby land area, dominated by changes in low-level clouds. They observed an increase in the convective cloud fraction over the Barents Sea as well as increased convective and total precipitation. Although these model results were derived using a single RCM, Koenigk et al. (2015) found a similar decline in cloudiness with a different RCM. There are natural hazards for which there is little published scientific literature, such as thick fog.There are few systematic observations of fog in the Arctic and it is difficult to model. It is mostly mentioned in connection with field trips and wildlife observation. Arctic fog appears to be more frequent in summer, and according to a Norwegian report (Loeng, 2008), is often caused by cold air extending over open sea during high pressure systems. The frequency of fog events is expected to fall in the central Barents Sea during mid- summer because the ice edge has receded. However, more fog is expected in other seasons when more sea ice is present with more polynyas (areas of persistent open water that are usually ice-covered). That GCMs are poor at simulating seasonal variation in Arctic cloudiness (Inoue et al., 2006) is a concern and may result in inaccurate predictions for sea-ice onset and duration (Stramler et al., 2011). Sea ice and cloudiness are mutually interdependent through a feedback, which makes it difficult to determine cause and response. Nevertheless, such shortcomings may help explain the spread in model projections and their underestimate of sea-ice decline in recent decades. Furthermore, even if GCMs can reproduce average conditions in the Arctic, it may be for the wrong reason if they do not capture sea-ice onset and duration.

The combination of relatively high sea surface temperature and low air temperatures drives strong vertical convection in both the lower atmosphere and upper ocean. Vertical heat flux reaches an annual average of ~80 W/m 2 , and in winter may exceed 200 W/m 2 over the open water in sea areas representative of conditions in the Barents area (Smedsrud et al., 2013). Frequent winter storms regularly increase the heat flux to 500 W/m 2 or more (Ivanov et al., 2003), which is higher than observed at tropical latitudes. Clouds require cloud condensation nuclei (aerosols) for droplets to form, and data from Arctic field studies suggest aerosol concentrations may be higher over the open sea and the ice-edge than over sea ice-covered regions (Leck and Svensson, 2015). Leck et al. (2002) identified two local aerosol sources: bubble bursting and oxidation products of dimethyl sulfide. As the Arctic climate warms, intensive convection follows the advance of open water, initiating regional and larger hemispheric impacts through teleconnections (Inoue et al., 2012; Smedsrud et al., 2013; Mori et al., 2014; Sato et al., 2014). There are some indications of more frequent convective cloud types over the past three decades (Esau and Chernokulsky, 2015), but strong interannual variability in winter cloudiness makes it difficult to identify trends (Stramler et al., 2011). Clouds are one of the most uncertain aspects of climate models (Boucher et al., 2013), and so projections are not yet possible. Esau and Chernokulsky (2015) analyzed cloud observations recorded at stations since 1880 and found a steady increase in the frequency of convective cloud types over the past three decades. Local vertical convection with latent heat release in convective clouds is critical to the observed increase in moisture content in the mid-troposphere (2–6 km) and precipitation/snowfall in the surrounding regions (Bulygina et al., 2011; Boisvert et al., 2013; Bintanja and Selten, 2014).These processes cannot be linearly extrapolated following the prescribed scenario of Arctic warming and sea-ice retreat. Esau and Chernokulsky (2015) argued that convection over the Barents Sea develops spatially in the form of convective fields, based on the analysis of Bruemmer and Pohlman (2000), and thus is controlled by the frequency of cold air outbreak events and the size of the open water area. As both factors are constrained, it is reasonable to suggest that the observed intensification of extreme winds and related dangerous events will peak and then decline through the 21st century. Such a reduction is seen in the regional climate projections according to the analysis of polar lows by Zahn and von Storch (2010). It should be noted that the convective fields, cold air outbreaks, and polar lows indicate a shift in extreme weather phenomena to the eastern Barents Sea where observations show they have previously been rare or absent. The question about cloud cover changes is not clear. On the one hand, Screen and Simmonds (2010) argued that past Arctic warming is mainly due to the decline in sea-ice extent with cloud cover playing a lesser role. However, changes in humidity may also have had some effect, and the decline in sea-ice extent is linked to the increase in humidity through the increase in open water area, and thus increased evaporation. On the other hand, higher water vapor concentrations may have enhanced the warming