SD o

Figure 4.3.

Change in air temperature in the Antarctic from data on the isotopic composition of ice cores at Vostok station. The dashed line indicates measured air temperature.

-452

i—|—I—i—i—i—i—|—i—i—i—i—|—i—r—i—i—|—i—i—i—i--

1800 1850 1900 1950 2000

Years

cycles of interannual fluctuations within ±2°C. These fluctuations occurred against the background of an even longer change, a positive linear trend showing a gradual increase of 0.8-0.9° in Arctic air temperature during the century. This trend may fit into one of the multi-century climate fluctuations that have been observed in Earth's history (Monin and Sonechkin, 2005). Calculations of the wavelet-spectrum of a 400-year series of reconstructed anomalies of mean annual air temperature in the region from 17.5°N to 87.5°N for the period 1579-1983 (data taken from Bashkirtsev and Mashnich, 2004) show significant peaks at the frequencies corresponding to 200 and 100 years (Figure 4.4, see color section).

A stable cycle with an average duration of about 210 years is also found in data on beryllium-10 isotope concentration (responding to air temperature changes) contained in dendrologic evidence from the northern Eurasian forestry boundary (Raspopov et al., 2004). It is possible that part of this cycle contributes to the linear trend in twentieth century air temperature.

An analysis of consistency among the main components of SAT changes in the Arctic and in the hemisphere in general is of great interest. The correlation coefficients characterizing this consistency are quite large: 0.59 (1900-2003) and 0.70 (1971-2003). Table 4.1 compares the characteristics of the linear trends and the "50-60-year" fluctuations in three geographical areas of the Northern Hemisphere in the twentieth century. The assessment methodology is similar to that used in Sections 2.2 and 2.3. The table shows much greater variability in air temperature and its two main climatic components in the Arctic than in the temperate latitudes and over much of the Northern Hemisphere. The contribution of the linear trend of mean annual temperature, averaged over the corresponding area, increases with decreasing latitude, while the "50-60-year" cycle mean annual temperature decreases. The possible causes of these changes will be considered in Section 5.4.

The fact that weather and climate variability increases with latitude is known as "polar amplification." A model proposed by Alekseev and Svyashchennikov (1991)

Table 4.1. Characteristics of mean annual twentieth century air temperature variation in three zones.

Region

rms

Trend coefficient (deg/year)

Average cycle amplitude

Trend contribution to dispersion

"50-60-year"

cycle contribution to variance

70-85oN

0.78°

0.0097

0.65°

13%

39%

40-65oN

0.34°

0.0048

0.25°

17%

27%

17.5-87.5°N

0.26°

0.0069

0.17°

59%

23%

explains this phenomenon by taking into account heat advection in the atmosphere that results in air mixing between adjoining latitudinal zones. Zakharov (1996) examined the relationship between the maximum air temperature variability zone and the location of the frontal area between the Arctic and the marine polar air masses. This allowed him to conclude that polar amplification is a simple result of the mobility of the polar front and its fluctuations in time, and furthermore, he was also able to explain localization of the most significant climatic changes in the sub-Atlantic Arctic. For most of the year, the mobile ice edge is located in this region, where the Arctic front passes between 70° and 80°N, and the horizontal temperature gradients are most pronounced near it. However, recognizing the important role of the North Atlantic in generating low-frequency climate fluctuations, Polyakov et al. (2002) express doubts that the observed air temperature trends in the Arctic confirm the hypotheses of polar forcing of global warming.

Alekseyev et al. (2004) show convincingly that seasonal twentieth century climate changes occurred extremely irregularly over the Earth's surface. The spatial non-uniformity of air temperature changes was connected with both geographical latitude and the longitude of the region. It is important to note that a negative correlation was revealed in this study between the mean zonal air temperatures at high and middle latitudes in some seasons.

Increased air temperature variability in temperate latitudes of the Eurasian and North American continents is typical of the winter months. The signs of temperature anomalies over the oceans and the continents are usually opposing. Such temperature field structures were called COWL (cold ocean warm land) by Wallace et al. (1995), who accurately related these phenomena to increased west-to-east transfers in the atmosphere during warming periods and their attenuation during cooling periods.

Klimenko (2007) presents similar findings. He charts the differences between mean annual and seasonal air temperatures in the Northern Hemisphere during the warmest 20-year period (1986-2005) and the coldest 20-year period (19111930) and shows the maximum warming to cover the temperate latitudes of Eurasia and North America. The charted differences in mean annual temperatures at the warming epicenters exceed 1.5°C (5°C in the winter season); that is, a tenfold increase in the average global signal. This is much greater than warming in the Arctic region adjoining the North Atlantic; however, a stricter approach to estimating the values under consideration requires a preliminary exclusion of cyclic fluctuations.

Klimenko (2007) reports that no warming was observed during the same period in the northern parts of the Atlantic and Pacific Oceans. Hassol (2004) shows most of the North Atlantic in the zone of decreased (by 1°C) mean annual and winter air temperature during 1954-2003. There is further evidence in 3,000 years of data on surface temperatures in the Sargasso Sea based on the oxygen ratio in the remains of plankton organisms buried in bottom sediments (Sorokhtin, 2001, Keigwin, 1996).

A direct cause of the patterns noted is, undoubtedly, intensification of zonal (west-to-east) transfers in the atmosphere of temperate latitudes during periods of climate warming, as discussed below in Sections 4.2 and 4.7. A corresponding increase in heat advection from the oceans to the continents plays an important role, as does moisture advection, which is accompanied by increased cloudiness, resulting in the increase of both long-wave counter-radiation in the atmosphere and temperatures in the lower atmosphere. As expected, loss of heat from upper ocean layers, especially in winter, is confirmed by the data presented in the aforementioned studies.

In addition to a positive trend and a "50-year" SAT cycle, the Arctic SAT exhibits significant interannual variations. Figure 4.5 shows that when the linear trend and the "60-year" cycle are excluded from the mean annual temperature anomalies considered above, the largest contribution to these variations comes from relatively high-frequency, variable-amplitude, cyclic fluctuations lasting 2-3 years; these are overlain by longer cyclic fluctuations, which are approximately represented by 5-year running averages. A spectral analysis of this curve reveals the dominance of a "20-year" cycle (Figure 4.6, see color section).

A periodogram analysis of the data smoothed by 5-year periods allowed us to estimate the amplitude of the "20-year" cycle and its contribution of about 5% to the variance in Arctic air temperature. The data summarized in Table 4.1 indicate that

Figure 4.5.

Annual air temperature anomalies in the 70-85°N zone. Linear trend and 50-60 year fluctuations are excluded. The heavy line represents a 5-year running-mean time series.

Figure 4.5.

Annual air temperature anomalies in the 70-85°N zone. Linear trend and 50-60 year fluctuations are excluded. The heavy line represents a 5-year running-mean time series.

the mean annual Arctic SAT changes appear to be mainly due to fluctuations lasting more than a century (a linear trend) as well as 50-60 year and 20-year cycles. Their total contribution to the variance comprises 57%, and hence the contribution of relatively high-frequency fluctuations is 43%.

Was this article helpful?

0 0

Post a comment