Figure 5.23. Lines of regression and curves of confidence at a : rof RV over the period 1951-1990.
Taking this into consideration, it should be emphasised that the greatest changes of T., relative to its variability observed in a given season, occurred in the summer. These results do not support Chapman and Walsh's conclusion (1993) that the averaged trend in the Arctic in the summer between 1961 and 1990 approximated zero. The latest climatic models, which take into account not only but also anthropogenic aerosol created as a result of the emission of sulphur compounds into the atmosphere, greatly improved the correspondence of their results with the observational data (Kiehl & Briegleb 1993; Charlson & Wigley 1994; Taylor & Penner 1994; Hegerl et al. 1996; Santer et al. 1995). The last of these works demonstrated that in the past 50 years the trends of T caused by the influence of and sulphate aerosol are significant in the summer and autumn. It follows that they can be discernible in the series of summer T in the Arctic. The same obtains for the analysis of the trends between 1971 and 1990 (Table 5.11). Moreover, it was in the summer that a positive trend in T. occurred for the first time in BAFR. Comparing the magnitudes of the trends from the periods 1971-1990 and 1961-1990, it is apparent that in the past 20 years warming decreased in those regions of the Arctic which in the 30-year period were characterised by the greatest trends (i.e. in ATLR and SIBR), whereas a marked warming occurred in PACR and CANR. The rate of cooling in BAFR decreased from -0.5°C/10 years to -0.34°C/10 years. The averaged T. from the zone between 65°N and 85°N behave differently in certain seasons. This applies mostly to the summer, for which no trend was determined in the 20-year period examined, and to the winter, for which the trend more than doubled (up to 1.18°C/10 years) and became statistically significant at the level of 0.05. It should be observed that the mean trend of annual T. in the Arctic is the same for the periods 1961-1990 and 1971-1990 whereas the trends of mean calculated for the past 20 years are many times greater and statistically significant in all seasons (Table 5.11). For instance, the averaged trend of for the data collected only on land almost doubled, and for the data including also SST the trend increased three-fold. Thus the greatest disparity between the T of the Northern Hemisphere and that of the Arctic occurred between 1971 and 1990 (Figure 5.12).
Establishing the reason for this disparity is an important research problem. Is it caused by the different reaction-time of the climate in the two areas to the abrupt increase, from ca. 1960, in the concentration of greenhouse gases (mostly C02) and sulphate aerosols? Or is this behaviour determined by natural factors? In the case of the first reason, what needs to be explained is the mechanism responsible for the delay in the reaction-time of the Arctic climate relative to the global warming observed. The lack of any warming (or presence of only a very slight warming) observed in the past 20-30 years in the Arctic is inconsistent with the results provided by the models of general atmospheric circulation. It follows, as Kahl et al. conclude (1993a, b), that these models inadequately describe the physical processes taking place in the polar regions. The results of the research published recently by Santer et al. (1995) shed new light on the above interpretation. The authors demonstrate that the cooling connected with the increase in the concentration of anthropogenic sulphate aerosol is very high, although lower than the warming determined by the increase in the concentration of C02. Using the general circulation model (NCAR CCM1) coupled with the chemical model (describing the quantitative changes of sulphates in the troposphere), the authors calculated the difference of temperatures corresponding to the change in the concentration of and the amount of anthropogenic sulphate aerosol since pre-industrial times up to the present moment. The results show that if only these factors were to have influenced the changes of T, then T (in annual means) should have increased most in the area of the Antarctic Peninsula (> 3°C) and around the Antarctic (1.5-2.5°C). The warming in the remaining area of the Southern Hemisphere should be within the range 0.5-1.0°C. The Northern Hemisphere is characterised by a significantly lower warming. In its southern part, the warming does not exceed 0.5°C, and a large part of moderate and polar regions (ATLR, part of PACR, the Central Arctic, the northern Atlantic, and almost all of Europe except for the Iberian Peninsula and the area around the Baltic Sea) should even cool down. The remaining continental areas (including most of Greenland) should get warmer. It is worth adding that, according to this research, polar regions get cooled mostly through the action of sulphate aerosols and at the same time they get warmed mostly as a result of the increase in the concentration of This is particularly evident in the cold half-year, whereas in the summer (but only in the Northern Hemisphere) this zone of activity shifts towards the area of moderate latitudes, which can probably be connected with the southward movement of the Arctic front.
Thus, it is probable that the significant climatic effect of sulphate aerosols in the Arctic completely (or at least to a high degree) neutralises the greenhouse effect connected with C02. As a result, T in this area does not manifest major changes. However, it is not certain that this is so as the model still has many deficiencies. Firstly, as the authors of the publication point out, the assessment of the climatic effect of sulphate aerosols may be burdened with serious errors. Secondly, the model employed does not take into account, among other things, the dynamics of the ocean, the indirect influence of sulphate aerosols on climate (which serve as condensation nuclei, thus increasing the optical thickness of clouds, their albedo and life-time - processes which, according to Karl etal. (1995), lead to cooling), and greenhouse effects caused by other trace gases (apart from and aerosols created in the process of burning biomass, fossil fuels, and in industrial processes.
Another possibility is that the Arctic climatic system, characterised by considerable inertia due to large sea-water masses (the Arctic Ocean) as well as sea and land ice, has not yet reacted perceptibly to the warming occurring in the lower latitudes. Research results obtained by Aleksandrov and Lubarski (1988) may provide some support for this hypothesis. They show that in the phase of global warming, the increase of T in the Arctic occurs with a delay relative to lower latitudes whereas in the cooling phase of the globe, the decrease of T occurs first in the Arctic and only later in the lower latitudes. These results, on the other hand, are contradicted by the fact that the climatic warming between 1930 and 1940 occurred initially - and most apparently - in the Arctic (Figure 5.12).
As for the significant trends of summer T. described in the present work, they point to the fact that at least part of this warming may be caused by the increase in the concentration of trace gases, even more so because it is in this season that the sensitivity of the thermal regime to the changes of concentration of these gases is the greatest and weather noise is the smallest (Alekseev et al. 1991).
On the other hand, there is considerable evidence to support the conclusion that the slight changes of T observed in the past decades in the Arctic are, to a large degree, the result of natural factors responsible for climatic changes, mostly the fluctuations of the advection of air masses. This thesis is supported by, among others, Barnett (1986), Alekseev et al. (1991), Alekseev and Svyashchennikov (1991), and Przybylak and Usowicz (1994). Alekseev et al. (1991) maintain that the trends of winter T in the Arctic are the most apparent because the fluctuations of the advection of air masses are the greatest at this time. They add that in the summer, when (as mentioned above) the sensitivity of the thermal regime to the changes in the concentration of trace gases is the greatest, the trends are slight. The above statement is plausible if we take into consideration the magnitudes of trends from particular stations and climatic regions. On the other hand, it is still true that, as emphasised above, the statistical assessment of the magnitude of summer trends is superior. More statistically significant trends are observed in this season than in the winter (Table 5.11). It is worth adding that the spatial distribution of the trends of summer is, out of all seasons, the least changeable and the most consistent (i.e. warming can be discerned in the largest area of the Arctic). As a result, the calculated mean trend of summer in the Arctic amounted to 0.16°C/10 years (1971-1990) and was lower than only the mean spring trend. As follows from Chapter 4 and the relevant literature (Atlas Arktiki 1985; Serreze & Barry 1988; Walsh & Chapman 1990; Przybylak 1992a; Serreze et al. 1993; and others), the atmospheric circulation is the most intensive in ATLR and BAFR. It is there, then, that its influence should be the most evident if it indeed had some influence on the changes of T in the Arctic. Figure 4.4a suggests that around the mid-1970s a significant increase in the frequency of occurrence of the W circulation macrotype began and this process has con tinued up to the 1990s. The analyses of the changes in the zonal index carried out by Kozuchowski (1993) and Jonson and Barring (1994), yielded similar results. According to Dmitriev (1994), the E circulation epoch ended in 1992, when the W epoch started. As follows from Tables 5.20a-d and Figure 5.34a-e, this circulation macrotype corresponds to negative anomalies of T in ATLR and BAFR. Thus, if circulation indeed has an influence on T in these areas, a decrease in the magnitude of trends should be observed there between the periods 1961-1990 and 1971-1990. This phenomenon is indeed readily observable in the area where cyclones pass most frequently, i.e., in the area along the Iceland-Kara Sea trough (Table 5.11). This applies first and foremost to summer trends because decreases of winter and spring T. were already noted in the whole area of ATLR except for the south-eastern part of ATLSRe. The mean winter trend changed for the periods discussed from 0.38°C/10 years to -0.27°C/10 years whereas the mean spring trend changed from 0.53°C/10 years to 0.16°C/10 years (Table 5.11). On the basis of the examination of atmospheric pressure and temperature changes between the periods 1967-1976 and 1977-1985, Walsh and Chapman (1990) conclude that at least some changes of Tin high latitudes were caused by changes in atmospheric circulation and the corresponding advections of air masses. Alekseev et al. (1991) present still further evidence supporting the thesis ofthe advective nature of the increase of T in the Arctic in the cold half-year. They observed that positive anomalies of T in the Arctic correspond to negative anomalies in lower latitudes, from the equator to 40-50°N.
Summing up the above discussion, it should be stated that there is no conclusive evidence to support the thesis that the slight warming observed in the Arctic during the last years of the observation period is a consequence of the increase in the concentration of trace gases. Alekseev et al. (1991), Kahl et al. (1993a, b), and Przybylak and Usowicz (1993, 1994) have expressed a similar opinion. The warming may be reduced, as demonstrated by Santer et al. (1995), by the cooling connected with the increase of the amount of sulphate aerosol in the atmosphere. Further detailed research is needed in order to establish the validity of this hypothesis. Thus, there is some evidence in favour of the thesis that the joint influence of anthropogenic climatic factors is minimal in the Arctic. Therefore, in all likelihood the discrepancy in the behaviour of T in the Northern Hemisphere and T of the Arctic observed in the past 20 years is caused by the change in the atmospheric circulation which occurred in the mid-1970s.
Sidorienkov and Svirienko (1983) state that in 1972 there ended a period in which the rotation movement of the Earth slowed down - a process which had begun in 1935. From 1973 to ca. 2000-2010, the increase of the velocity of this movement should continue. What is more, these researchers established that in the periods when the velocity of the rotation of the Earth increases, the frequency of occurrence of the C circulation macrotype drops below the norm, whereas the frequency of the combined macrotypes W + E increases above the norm. On the basis of the above, they formulated the forecast that such a change in atmospheric circulation should also occur between 1973 and 2005 ± 5 years. However, measurements of the velocity of earth's rotation showed that the increase in the velocity continued only up to 1986 (International Earth Rotation Service 1994). The mean duration of a day in that year was longer by 1.23 milliseconds as compared to the so-called 'standard day' and in 1972 and 1993 it was longer by 3.13 and 2.37 milliseconds respectively. Thus, if the relationship established by Sidorienkov and Svirienko (1983) between the changes in the velocity of the Earth's rotation and the frequency of the occurrence of certain circulation macrotypes actually obtains, then the trend of circulation changes that started in the mid-1970s may reverse ca. 10 years earlier than anticipated. It is known, however, that it did not change up to the beginning of the 1990s (cf. Dmitriev 1994 and Figure 4.4a-c). This change in the circulation causes a gradual decrease in the transportation of warmth into the Arctic from lower latitudes, and, consequently, the cooling of the climate in this region.
Research conducted by Stanhill (1995) proved that between 1950 and 1994 there was a substantial decrease in the inflow of solar irradiance of the order of 0.36 W/m2/year. It was most frequent in the spring and in the western part of the Arctic, where the air is the most polluted and where the haze termed 'Arctic Haze' occurs frequently. Stanhill, having no data concerning the variability of cloud cover, considered the increase of the inflow of pollution over the Arctic as the reason for the decrease in radiation in the area. It is worth observing, however, that there has been a significant increase of cloud cover over most of the Arctic (Table 5.15, Figure 5.28) which, although to a certain extent is probably connected with the increase in air pollution, may nevertheless be the effect of natural fluctuations. Thus, it appears that, aside from the increase in the air pollution, the increase in cloud cover was also conducive to the negative trend in the solar irradiance in the Arctic in the 40 years of the observation period. Stanhill (1995) also observed the existence of a permanent, statistically insignificant, negative trend in the radiation balance in the Arctic between 1962 and 1981. These changes in the amount of radiation inflowing into the Arctic led to its cooling. Another natural factor which amplifies the aforementioned climatic effects caused by changes in atmospheric circulation and solar irradiation is solar activity. Voskresensky et al. (1991) established that periods of low solar activity correspond to a decrease in T in the Arctic. Since 1957, when the secular maximum of solar activity occurred, its downward trend has been observed. On this basis it can be said that this factor leads to climatic cooling. According to Charvatova and Strestik (1993), the movement of the sun along a chaotic orbit around the barycentre of the solar system, which started in 1990 and will last till 2040, will cause a lowering of solar activity and a lowering of T over the entire globe.
It follows from the above analysis that all the natural factors mentioned lead to the cooling of the Arctic climate. Their effect is amplified by the increase in the amount of anthropogenic sulphate aerosol in the atmosphere (Santer et al. 1995). If their joint influence is stronger than the greenhouse effect, then the slight warming of the Arctic noted at the end of the period of observations should give way to cooling.
The analysis presented above shows that the magnitude of linear trends and their sign (±) depend on the period chosen for analysis. Of particular importance is establishing a starting point for calculations. Definite conclusions should not be based on the calculations of trends from a single period, as was done, for example, by Chapman and Walsh (1993). Using the data from the period 1961-1990, they asserted that the Arctic had undergone warming, and that seasonal and spatial distributions of T fields roughly correspond to the picture obtained from climatic models. However, if a longer period is examined (e.g. 1951-1990) or a shorter one (e.g. 1971-1990) the picture changes significantly. A considerable discrepancy is revealed between models and empirical observations. Extreme caution is advised when drawing con-elusions based on linear trends. It should be borne in mind that linear trends, as Kozuchowski affirms (1985), are only a most general characteristic of the variability of a given climatic element in the period under examination, a characteristic emerging from the maximal "smoothing" of the course of the element over long-term periods.
A more accurate picture of the variability of T in the Arctic can be obtained using curvilinear trends, such as moving averages. With moving averages, only those oscillations whose periods are shorter than the assumed period of averaging get smoothed. In the present work, 10-year moving averages have been used for regional mean T. and mean T. of the Arctic (Figures 5.14-5.19). Curves representing moving averages have a far smoother course than "raw" values. However, a significant diversity in fluctuations of both in particular climatic regions and over the whole Arctic is still visible. Figure 5.14 shows that the warm period of 1930s and 1940s ended ca. 1962 (in a similar vein, see Dmitriev 1994). The period characterised by negative anomalies of T. then began and continued practically till 1990, which is particularly evident in the data for autumn and winter (Figure 5.14). The greatest negative deviations occurred in 1966 and since then an increase in T. has been visible. This increase was most evident up to the mid-1970s and then decreased significantly, probably due to the change in atmospheric circulation mentioned above. The series of T represented in the figures analysed are too short for the long-term cycles of their changes to be determined. This is possible only for summer (Figure 5.14), which has two maximums and two minimums separated by ca. 16 years. For the remaining seasons, those cycles are probably far longer. This problem will be analysed in detail in the next sub-chapter. The general nature of the fluctuations of in particular seasons is similar. Their most divergent course, relative to the averaged one, occurs in the summer. In the seasons analysed there are significant differences in the magnitudes of the amplitudes of 10-year anomalies of
The fluctuations of anomalies of in particular climatic regions of the Arctic differ from one another (Figures 5.15-5.19). As follows from the above figures, the warming of 1930s and 1940s was strongest in ATLR, and weakest in PACR. Also fluctuations ofT. are the greatest in ATLR because this is the area which is most affected by atmospheric circulation. In contrast, the fluctuations of in CANR are the weakest, especially in the winter when high synoptic pressure centres develop intensively in this area (Serreze et al. 1993). Deviations from the long-term mean fall within the range of-3°C and 3°C. Slightly greater fluctuations of anomalies ofT. occur in SIBR. The courses of anomalies in ATLR and SIBR (Figures 5.15 and 5.16) are the most congruent with the courses of mean Arctic anomalies (Figure 5.14). This is caused by the fact that the share of variability of T in these two regions in the general variability of the T field in the Arctic reaches 70% in the warm half-year and even 80% in the cold half-year (Aleksandrov et al. 1986). Fluctuations of anomalies in the remaining regions are different and the greatest differences, relative to their averaged course, occur in PACR and BAFR. The former is characterised by the most rhythmical course of T. anomalies which oscillating around 0°C. This is why the trends plotted for this region are the smallest. It is also worth noticing that there was climatic warming in this region in 1960s, whereas this period saw the greatest cooling in the past 70-80 years in ATLR and SIBR. A characteristic feature of BAFR is the clear occurrence of a warm period in 1950s and 1960s. High values of mean annual T. were predominantly the result of a significant warming of the winter (Figure 5.19). Since that warm period a stable clear downward trend of T. has been observed (except for the second half of 1970s). At the same time the year-to-year variability of T. has increased markedly.
In all regions of the Arctic the greatest anomalies of occur in the winter. Mean 10-year anomalies in ATLR in this season oscillated from ca. 6°C to -1°C (Figure 5.15). They are lower in transient seasons but are still quite high (from ca. 3°C to -1°C). The lowest variability of anomalies can be clearly observed in the summer, when their range of changes in ATLR oscil lated from ca. 1°C to -0.5°C. For mean T. of the Arctic, it was ascertained that the course of their anomalies in all seasons is generally similar to that of annual means. We can now check if this regularity is true of particular climatic regions of the Arctic. An analysis of Figures 5.15-5.19 will show that in most cases it is not so. It is only in ATLR that the course of 10-years anomalies of T. is consistent. Similar to the whole Arctic, the course of sumi '
mer anomalies is the most divergent.
In SIBR, significantly similar are the courses of winter and autumn anomalies of on the one hand, and of summer and spring, on the other (Figure 5.16). Analysing the fluctuations of winter and summer anomalies, one notices that their courses are different. There is even a tendency towards resolving their anomalies (e.g. between 1936 and 1950, and between 1971 and 1990).
In PACR, the courses of winter anomalies of T. are similar to those of ' i the year as a whole. In the remaining three seasons, the fluctuations of anomalies are irregular and no similarity can be discerned between them. It is worth observing, however, that the changes of 10-year anomalies in all seasons, as well as for the annual mean, fall within the range ca. -1°C to 1°C, though, of course, the greatest deviations of from the norm still occur in the winter and the smallest, in the summer (Figure 5.17).
A certain similarity in the courses of 10-year anomalies of T. of the spring, summer, and the year as a whole can be noticed in CANR (Figure 5.18). This means that in this area of the Arctic the temperatures of the warm half-year are of greater importance for the year-to-year changes of annual means of T.. An analysis of the fluctuations of T. of the summer and winter brings a certain tendency towards resolving their anomalies.
Anomalies of T. in BAFR fluctuate differently in each season. There is some similarity between the courses of anomalies of the summer and spring. The course of their annual anomalies, on the other hand, is undoubtedly highly dependent on the behaviour of winter T. (Figure 5.19). This may clearly be seen in the case of 10-year anomalies and annual anomalies.
Fluctuations are often characterised through calculating cumulative deviations from the norm (Figure 5.24). Their ordinates in the present work were calculated using the following formula:
mean annual air temperature in a given year mean long-term air temperature.
Figure 5.24. Cumulated deviations of the mean annual T of five climatic regions (RI-RV), the Arctic (ARCTIC 1, ARCTIC 2, and ARCTIC 3), the Northern Hemisphere (TNH-1 and TNH-2), as well as the Barents Sea water temperature and the index of zonal circulation (Zi) over the period 19511990.
ARCTIC 1 - means have been calculated from 27 stations; ARCTIC 2 - after Alekseev and Svyashchennikov (1991); ARCTIC 3 - after Jones (1994); TNH-1 - for land only; TNH-2 - for land and sea.
Was this article helpful?