Ice Thickness Variations

Regular measurements of Arctic ice thickness began approximately in the middle of the 1930s in the vicinity of a number of polar stations, some of which were closed in the 1990s. To investigate the changes in landfast ice thickness for this study, we chose 11 stations with approximately equal lengths of observational time series. Five of these stations were located in the Kara Sea (Beliy Island, Dikson Island, Uyedineniya Island, Cape Sterlegov, and Cape Cheluskin), and the others in the Laptev Sea (Tiksi Bay, Kotel'ny Island, Sannikov Island), East-Siberian Sea (Cape Shalaurov, Chetyrekhstolbovoy Island) and in the Chukchi Sea (Wrangel Island).

Figure 3.1 (see color section) plots temporal variability of maximum ice thickness for the period of ice growth and its spectral structure for 1936-2000 based on wavelet-analysis data. Figure 3.1a shows that the maximum thickness of landfast ice in the Kara Sea increased from 1936 to the late 1960s, and then decreased through the end of the twentieth century, but not to its 1936 value. These variations are approximately the same as ice extent variations in the western sector seas (Greenland, Barents, Kara) associated with the "60-year" cycle (Figure 3.1b) discussed in the previous section and are statistically significant in the total wavelet spectrum (Figure 3.1c). The linear trend coefficient for this time series is +0.140 cm/year with a 95% confidence interval of —0.064 ... +0.344 cm/year. This trend contribution to the interannual variations is only about 3%. The positive trend is noted at four of the five stations in this sea.

The linear trend coefficient for the landfast ice thickness series in the eastern region is very close to zero (Figure 3.1a) and is —0.003 cm/year with a 95% confidence interval of —0.124 ... +0.119 cm/year; its contribution to interannual variability is negligibly small (0.004%). The linear trend sign is positive at three of the seven eastern region stations and negative at the other four stations. The temporal changes in landfast ice thickness in this region are characterized by the influence of shorter

Table 3.1. Average thickness (cm) of landfast ice during different climate periods in the Arctic Seas, with the anomaly relative to the average value of the observation series in parentheses

Period in years

Kara Sea

Eastern Seas

1936-1957

165 (-8)

199 (0)

1958-1983

181 (+8)

200 (+1)

1984-2000

174 (+1)

197 (-2)

cycles (Figure 3.1b), which are statistically insignificant except for the ^7-year variability (Figure 3.1c).

Table 3.1 presents average values of landfast ice thickness of the "warm" and "cold" epochs for two regions under consideration. Whereas in the Kara Sea the "cold" epoch is distinguished by an insignificantly increased average ice thickness, there are practically no differences between the epochs in the eastern region.

The data presented in the table show that the variations in landfast ice thickness in the Arctic Seas were insignificant during much of the twentieth century. This is also indicated by the analysis of multiyear changes in landfast ice thickness in the Arctic Seas by Buzuyev and Dubovtsev (2002), which does not "confirm climate warming in the area of the Siberian shelf in the 1960s-1990s."

As noted above, the thickness of first-year landfast ice over much of the area of the Arctic Seas comprises 180-200 cm by the end of winter. Karelin (1951) and Nikolaeva and Shesterikov (1970) show that due to the influence of the heat flux from deep Atlantic water, the thickness of drifting ice of the same age in the deep-water part of the Arctic Basin is 15-20 cm less (given the same snow thickness and sums of negative degree-days). The ice growth rate decreases in response to variations in Atlantic water temperature and depth during periods of Arctic warming. Observations made during the drifting expedition of the icebreaker G. Sedov indicate that the first-year ice thickness at the end of winter in 1939 was 20% less than that observed during the Fram expedition of 1895 in approximately the same region (Buinitsky, 1951).

An analysis of sonar data collected by submarines in the Arctic Basin (Rothrock et al., 1999) showed that the ice thickness there decreased by 1.0-1.5m, i.e., approximately 40%, from the middle of the 1970s to the beginning of the 1990s. This thinning of the ice cover was attributed to the influence of anthropogenic greenhouse warming. However, analyses of the same data performed by a number of scientists (Shy and Walsh, 1996; McLaren et al., 1994), did not confirm such changes, while others (Wadhams, 1990, 1994) explain this phenomenon by ice ridging at the approaches to Greenland.

The partial concentration of old or multi-year ice (MY) from the ice charts may be used as another proxy for ice thickness data in the Arctic Basin. Though much less accurate than sonar or drilling information, it covers more area and more time intervals. Smolyanitsky (2003) analyzed gridded fields of multi-year partial concentration extracted from the AARI 10-day ice charts in the SIGRID format from the "Global Digital Sea Ice Data Bank" (GDSIDB). The World Meteorological Organization (WMO) Commission on Marine Meteorology (now the Joint WMO/ Intergovernmental Oceanographic Commission for Oceanography and Marine Meteorology, or JCOMM) established the GDSIDB of digital sea ice chart information from the operational ice forecasting centers of participating nations in November 1986. The nominal resolution of a SIGRID grid is 15 minutes latitude. The left column in Figure 3.2 (see color section) shows robust mean MY concentration values for August averaged for 1933-1992 and three sub-periods close to two decades in length: 1940-1959, 1960-1979, 1980-1992. (Air reconnaissance, which ended in 1992, was the prime source of information for AARI ice charts. No calculations were carried out prior to 1940 because of gaps in data taken before that year.) To facilitate interpretation of MY decadal variability, the right column in Figure 3.2 presents differences between three sub-periods and the whole period, or "climatology" from 1933 to 1992. During the first sub-period of the 1940s and 1950s, which at first corresponds to a warmer and then to a colder period in the Arctic, a mixed pattern of MY decrease and increase compared to the climatology is observed in the Eurasian Arctic: decrease in the Kara Sea and the western part of the East Siberian Sea, increase in the northern part of the Barents and Laptev Seas and the eastern parts of the East Siberian and Chukchi Seas. During the second sub-period of the 1960s and 1970s, i.e. during a pronounced colder period in the Arctic, a decrease in MY relative to climatology is observed in the Laptev and eastern part of the East Siberian and Chukchi Seas with an increase in MY relative to climatology in the Kara Sea and the western part of the East Siberian Sea, etc. During the third sub-period, i.e. during a warmer period of the Arctic temperature regime, an increase in MY is observed in the whole East Siberian Sea and the western part of the Chukchi Sea with a decrease in MY in the Barents, Kara and Laptev Seas. It is reasonable to conclude that since the same sign of MY variability is observed in most parts of specific seas, the MY edge also varies in the whole area of seas on decadal time scales and the thermal factor can not be the only cause of its variability.

Recent AARI studies (Gudkovich and Kovalev, 2002a,b) indicate that the sea ice thickness anomalies reported by Rothrock and Maykut (1999) and by Wadhams (1990) are caused by dynamic processes, rather than by thermodynamic processes inherent in global anthropogenic warming. The latter are connected with comparatively short-term changes in atmospheric circulation that control the processes of sea ice advection, ridging, and divergence.

Gudkovich and Klyachkin used a 2-dimensional polynomial to the power of 3 approximation of fields of long-term ice drift vectors, observed by manned stations and automated DARMS buoys for 1937-1975, to study changes in the Arctic ice thickness on annual scales. Their calculations show that the area of decreased end-of-summer ice extent formed in the seas east of Severnaya Zemlya with first- and second-year ice with thicknesses of 1.5-2.5 m, can be transferred as a result of 1-2-year drift to the near-pole area where multiyear ice with a thickness of 3-4 m is usually located.

As a result, a significant decrease in ice thickness is observed, which is then replaced again in 2-3 years by a restored multiyear ice cover typical of this region.

Gudkovich and Guzenko (2007) studied annual changes of the ice thickness in the Eurasian Arctic by tracking displacement of the zone of the former ice cover boundary in the Arctic Basin using ice drift vectors observed by IABP buoys. To that effect resulting vectors of buoy drift recorded for annual intervals were interpolated to positions of the ice cover boundary in order to track its further displacement on annual scale. Figure 3.3 (see color section) presents results of such calculations for a two-year interval from October 1995 to September 1997, which are based on the information from 7 buoys active during October 1995-September 1996 and 9 buoys active during October 1996-September 1997. In spite of significant ice drift anomalies for the period 1995-1997, the result of the calculations confirmed the conclusion above: the boundary of multiyear ice in 1997 was located much farther to the north and east of its mean multiyear location, which explains a decrease in Arctic drifting ice thickness detected in some years by sonar.

Such ice thickness variations depend not only on drift speed anomalies but also to a significant degree on initial ice cover distribution (location of the residual ice edge at the end of summer, boundaries of the ice massifs, etc.). In addition, the alternation of anticyclonic and cyclonic regimes discussed below (Section 4.2) is accompanied by recurring changes in the processes of sea ice convergence and divergence changes. The latter influences sea ice concentration and ridging and hence is responsible for sea ice thickness temporal and spatial variability (Losev et al., 2005; Porubayev, 2000; Makshtas, 2001). Climatic changes can only determine the probability of the formation of the corresponding conditions (initial ice distribution, anomalous character of the baric fields, etc.). It also appears (Makshtas et al., 2002) that the air temperature influences ice thickness but changes in heat fluxes from ocean affected by low-frequency variations in temperature and location of deep Atlantic water in the Arctic Basin play a specific role in ice thickness variations (see Section 4.6).

In order to clarify the problem of the real change in thickness of drifting ice in the Arctic Basin during arctic warming in 2000s, we analyzed the process of ice growth using observations from three Russian drifting stations: NP-32 (2003-2004), NP-33 (2004-2005), and NP-34 (2005-2006). The drift of all three stations was mainly within the near-pole area bounded by parallel 85°N. The data from these stations were made available courtesy of the heads of the stations, V. S. Koshelev, A. A. Visnevsky, and T. V. Petrovsky. It is interesting to compare these data with observations from the drift of the icebreaker G. Sedov (1937-1940) during the first twentieth-century Arctic warming. It is important to note that the G. Sedov drift in the winter of 1938-1939 was predominantly to the north of 85°N, and hence quite reasonable to compare with the three recent expeditions.

The rate of sea ice growth is known to depend on a number of factors (air temperature, ice thickness and snow cover, their thermal-physical characteristics, etc.). According to Buinitsky (1951), daily ice thickness growth per 1° of average air temperature is determined, at least for first-year ice, predominantly by average ice thickness. Figure 3.4 presents Buinitsky's plot of an empirical dependence using average daily ice growth values per 1 ° of mean air temperature. These values were

Figure 3.4. The thick black line shows daily ice thickness growth per 1° of average air temperature in G. Sedov observation data for 1937-1938, plotted by Buinitsky (1951). Gray filled circles denote data from drifting station NP-32 for 2003-2004.

also obtained from intervals of observations of young ice growth at the NP-32 station. These observations were unfortunately interrupted due to breakup of the station's ice floe at the beginning of January 2004.

The values shown confirm the general character of the dependence. However, they were 0.01cm/(°K • day) lower, on average, compared to the empirical curve, which corresponds to an ice growth deficit of 7 cm/month. This is explained to a great extent by increased average snow cover thickness (23 cm at the NP-32 station compared to 6 cm in the data of the expedition onboard the G. Sedov). Calculations of residual (first- and second-year) ice growth require accounting for snow cover depth as well as ice thickness. The heat conductivity coefficient of snow (0.3W/(m -°K) is known to be 7 times less than that of ice (2.2W/(m -°K) (Sea ice, 1997). That is why snow cover with a thickness of 0.2m (for example) has the same influence on the growth of 1m thick ice as 2.4 m thick ice. This is quite accurately accounted for in the following equation (Nikolayeva and Shesterikov, 1970; Frolov et al, 2005):

H =-7.0h + ^(7.0h + H0)2 + 0.00122(0 - Ts)r - Fw • t/(L • p), (3.1)

where H0 and H are the initial and final ice thickness, respectively; h is snow thickness; Ts is average snow surface temperature; 0 is the freezing temperature of water near the lower ice surface; and r is the time interval. Here, ice and snow thickness are expressed in meters and r in days. The third term is responsible for the influence of heat flux from water, where Fw is heat flux from water, and L and p are the heat of melting and the density of ice, respectively.

Table 3.2 presents the initial data and the results of ice thickness calculations for monthly time intervals using Equation 3.1. The calculations were based on data and

Table 3.2. Observation data on the ice thickness growth (m) during the icebreaker G. Sedov expedition and at the NP-32, NP-33, and NP-34 drifting stations, and the results of calculations using these data

Expedition

Year

Months

1 avg

H0

H

havg

SH*

Notes

G. Sedov

1937/38

XI-V

—20.9

0.33

1.95

0.06

+0.04

First-year

G. Sedov

1938/39

XI-V

—26.5

0.64

2.05

0.19

+0.06

First-year

G. Sedov

1938/39

XII-V

—27.3

0.99

2.02

0.23

—0.04

First-year

G. Sedov

1938/39

XII-V

—27.3

1.46

2.16

0.31

—0.01

Second-year

NP-32

2003/04

IX-XII

—19.7

0.00

1.05

0.23

+0.07

First-year

NP-32

2003/04

X-II

—29.8

1.55

1.92

0.33

—0.04

Second-year

NP-33

2004/05

XI-V

—25.0

1.97

2.42

0.52

—0.01

Second-year

NP-34

2005/06

X-IV

—22.3

1.00

1.92

0.31

+ 0.02

First-year

* SH is the average difference between the observed and the calculated ice thickness for the end of the calculation month.

* SH is the average difference between the observed and the calculated ice thickness for the end of the calculation month.

information from the G. Sedov expedition and the NP-32, NP-33, and NP-34 drifting stations. The value of 0 is assumed to be equal to —1.7°C (Anon. (F)). Instead of Ts values, air temperature Ta was used. The differences between them are noticeable at ice thicknesses up to 0.5 m with an insignificant snow cover.

As Table 3.2 shows, the relative error in ice thickness calculations at the end of winter is only 0.5-3%. If we assume that the average excess of the calculated ice thickness values over the measured values in winter of 2003-2004, when the NP-32 station was closer to Fram Strait, is determined by the influence of the heat flux from deep Atlantic water, then it is simple to estimate the value of this flux using the method described above. The calculations show that total heat flux from the ocean for the winter could not be more than 100,000 kJ/m2 (about 2.5 kcal/cm2). According to Panov and Shpaikher (1963), the maximum value of this flux near the continental slope of the Siberian Arctic Seas is 5-6 kcal/cm2, and from better documented estimates by Nikolayeva and Shesterikov (1970), 4.0 kcal/cm2 (up to 1.0-1.2 kcal/ cm2 in the deep-water part of the Arctic Basin).

In addition to heat from the ocean, an important factor slowing ice growth is the presence of melt-ponds. Their full-depth freezing delays the beginning of growth on the bottom surface of the ice cover, sometimes until the beginning to the middle of December. According to observations at the drifting stations, the area of the melt-ponds by the beginning of freeze-up comprised up to 40% of the surface of the ice which survived the summer melt. Therefore, and given the absence of significant error in calculated ice thickness, it can be considered that the influence on the ice growth of heat flux from the water during the epochs under consideration (1937-1939 and 2003-2006) was not significant. It is likely that in the continental slope zone to the north of the Eurasian Arctic Seas shelf, where the main flow of deep Atlantic water is located, the influence of this heat on ice growth is more obvious than in the near-pole area.

As the data in Table 3.2 show, average wintertime air temperatures during the first and second warming in the Arctic differ insignificantly, and there is no significant difference in ice thickness at the end of winter. Note again that according to Vize (1951) and Buinitsky (1951), winter air temperature during the drift of the icebreak-ing vessel G. Sedov was 6-8 degrees higher than at the time of the Fram drift. The sum of negative degree-days at the end of the 1930s, similar to the beginning of the twenty-first century, was 21% less than at the end of the nineteenth century. Ice growth in the near-pole area has decreased by exactly the same value (Karelin, 1951). So, the results presented above contradict estimates of catastrophic (almost twofold) ice thickness decrease in the Arctic Basin during the last decades of the twentieth century (Shimada et al. (2006), Anon. (A), Anon. (B) and Anon. (C)). It should be noted that in the Vize (1951) and Buinitsky (1951) data, the mean monthly air temperatures in springsummer of G. Sedov drift were 0.2-0.5°C lower than at the time of Fram drift, which indicates the secondary importance of ice melting compared to ice growth in ice thickness variations in the Arctic Basin.

Was this article helpful?

0 0

Post a comment