Key as in Table 5.13
The analysis presented is very general, so in order to make it more precise, an examination was carried out on the frequency of occurrence (in the periods analysed) of the situations in which a greater upward trend (or weaker downward trend) of T . was observed than that of T . The calcula' mm max tions showed that in the period 1951-1990 the situation described above occurred in 76% of cases of mean winter, spring, and summer values, in 64% of cases of autumn values, and 72% of annual values. The trends of extreme temperatures calculated for the period 1961-1990 confirm even more strongly a greater warming of Tmin. Mean winter, autumn, and annual Tm.n values were characterised by a greater upward trend or a smaller downward trend than were T values in as many as 88% of stations, whereas for spring and summer the figures were 81% and 73% of stations respectively. Only in the eastern part of PACR, in the west of CANSRn and in the east of CANSRs was this not the case for annual values. In the remaining area of the Canadian
Arctic the differences in the magnitudes of trends were also significantly smaller than in other regions of the Arctic. This marked dominance of an increase of Tm.n became significantly weaker in the last 20 years of the period of observations. For annual and autumn means, a greater increase of Tmin than that of T was observed only in 58% of the stations. For winter, spring, and summer means, Tmin manifested a greater warming or a weaker cooling than did T in 54%, 50% and 46% of the stations respectively. These changes between 1971 and 1990 occurred mostly in the Canadian Arctic, on the western coast of Greenland, in the central part of ATLSRs, and in the western and central parts of ATLSRn. The above results and those concerning T. show that in the period 1971-1990, the factors shaping the climate in the region of the Arctic and, in all likelihood in the whole Northern Hemisphere, must have changed significantly. One such factor may be atmospheric circulation, which - as follows from Chapter 4 - has been undergoing a major reorganisation since the mid-1970s.
Figure 5.26. The spatial distribution of the mean seasonal trends of Tmai (left panels) and Tm.n (right panels) (in °C/10 years) in the Arctic over the period 1951-1990. Negative trends are hatched.
Similar to T., seasonal and annual trends of extreme temperatures are statistically insignificant in almost all Arctic stations (Tables 5.13 and 5.14). Their share in the general variability of the relevant thermal parameters does not, as a rule, exceed 10%. The calculations of upper limits of confidence intervals of coefficients a of regression equations for extreme temperatures revealed the existence of positive values in all of the 9 stations analysed (the stations are the same as those listed in Table 5.12). The lower limits, on the other hand, were always negative, except for the summer and annual means of Tmin in Danmarkshavn. Out of these stations and the seasons analysed, the only significant trend turned out to be the trend of mean summer in Danmarkshavn. The remaining coefficients of regression equations are so small that there is no basis for considering them different from zero. The magnitude of a 40-year increase in summer Tmtn in Danmarkshavn falls with a 95% probability within the range 0.56-1.92°C. The linear trend of summer Tmin in this station accounts for ca. 30% of its general variance.
On the basis of the analysis presented it is possible to conclude that in the Arctic, similar to most areas of the Northern Hemisphere, there is a perceptible tendency towards a smaller decrease or a greater increase in Tmin than Tmax since the 1950s (Tables 5.13 and 5.14, Figure 5.27).
A question arises as to the reason for this phenomenon. One of the causes may be the increase in cloud cover, as suggested by Frich (1992) and Karl et al. (1993a). The review of literature concerning long-term changes in cloud cover in the Arctic revealed the scarcity of publications of this kind. Two works deserve to be mentioned: Raatz (1981) and Mokhov (1991). The former, using data from only five Arctic stations from 1921-1978, did not determine any trends in cloud cover. The latter, using satellite data from 1971-1985, focuses mostly on the analysis of the relationships between the cloud cover in the Arctic and the temperature of the Northern Hemisphere. Unfortunately, the author does not specify the tendency in cloud cover in the period examined. The above publications are of little use for the problem in hand as they pursue different research aims. Consequently, it was necessary to carry out special research on the variability of cloud cover in the Arctic and its relationship to T.In order to do this, mean seasonal and annual values of cloud cover obtained in 19 Arctic stations from various periods were used (Figure 5.28). Using these data, calculations were made of the magnitudes of linear trends from the period 1961-1990 (Table 5.15, Figure 5.28) and curvilinear trends (5-year moving averages, Figure 5.28). Positive trends of cloud cover were observed in most of the area of ATLR, SIBR, and BAFR, whereas in PACR and CANR the trends were negative. Significant trends of mean winter and annual cloud cover occurred only in the area of ATLSRn, in the northern part of ATLSRe and in the western part of SIBR.
In order to establish the relations between cloud cover and extreme T, the conformity between their trends was examined. It was determined that they are most compatible in the winter (ca. 70-80%) and spring (ca. 60%) and least compatible in the autumn (37-47%). The comparison of the trends of mean annual values of extreme T and cloud cover yielded 58% of compatible trends for and 53% for These results do not allow an unequivo-
max mm 1
cal evaluation of the relationship obtaining between the elements examined. The blurring ofthe picture may result from the fact that most trends analysed, both those of extreme Tand those of cloud cover, are not statistically significant. Taking this into consideration, the only trends examined for compatibility were the ones from the stations which were characterised by statistically significant cloud cover trends between 1961 and 1990. In this case, it turned out that the increase in cloud cover is almost always accompanied by an increase in extreme T (for annual means, this was true in 92% of cases and for seasonal means in 82% of cases). As stated above, the decrease in the Diurnal Temperature Range (DTR) between 1961 and 1990 was most manifest in the area of the Arctic with the exclusion of Alaska and the Canadian Arctic, i.e., in the area characterised predominantly by the positive trends of cloud cover. In Alaska and the Canadian Arctic on the other hand, areas characterised by a decrease in cloud cover in the 30-year period examined, the DTR did not change considerably. In order to document this important relationship, the mean difference was calculated between the trends of and in particu-
mm max r lar stations located in these two Arctic regions. This difference amounted to 0.16°C/10 years for the area of the Arctic characterised predominantly by increasing cloud cover and -0.03°C/10 years for the area characterised by decreasing cloud cover. This appears to be sufficient evidence to confirm the significant role of cloud cover in the process of diminishing the DTR in the Arctic. It should be added, however, that there are areas of the Arctic (especially the southernmost fragments of the continental Russian Arctic) where this relationship does not manifest itself. The decrease in cloud cover is accompanied here by a decrease in the DTR. These areas are the most economically exploited parts of the Arctic and they are located in relatively close proximity to the industrial areas of Europe and Asia. Thus, the decrease in the DTR may be caused by the effect of urbanisation and the increase in sulphate aerosol and greenhouse gases.
Was this article helpful?