* 1951-1980

The variability coefficient of annual P in the Arctic, which was calculated using data from the period 1951-1990, was highest (over 30%) in the east of SIBR, in PACR, in a small part of CANSRn and BAFR, and on the northeast coast of Greenland. Also the eastern part of ATLSRn, the northern part of ATLSRe, and the far western part of SIBR are characterised by a substantial variability (ca. 30%). These are the areas with the lowest P in the Arctic (cf. Figures 6.5 and 6.1). The most stable P (v < 20%) is observed in the southern parts of ATLSRs, which are characterised by the highest P in the Arctic and are situated on the most frequent route of cyclones moving along the Iceland-Kara trough. According to Kozuchowski (1985), P in Poland is also more stable in the areas where moving cyclones are more frequent.

Figure 6.5. The spatial distribution of the coefficient of variability in annual P over the period 1951-1990 in the Arctic.

Figure 6.5. The spatial distribution of the coefficient of variability in annual P over the period 1951-1990 in the Arctic.

The dispersion of seasonal P totals is much greater than that of annual totals. The highest v values are observed in the seasons which are characterised by the lowest P, namely in winter and spring. In most of the Arctic, v exceeds 50% (Figure 6.6), and the area is bigger in winter. In spring it is much more diversified, from ca. 100% in Alaska to ca. 30%-35% in the south of ATLSRs. In winter the highest v does not exceed 70% and it occurs dis-continuously in small areas (Figure 6.6); the lowest v is slightly higher than the spring v, and can be observed in the areas characterised by the greatest cyclone frequency, which are the south of ATLSRs and the southeastern part of BAFR. In summer and autumn, which are the periods characterised by the highest P in the Arctic, its variability is much lower than in the aforementioned seasons. A variability coefficient higher than 50% can be observed only in small areas of the Arctic. The autumn P is more stable, and its v is less than 30% in a considerable area of the southern parts of ATLR and SIBR. In summer such an insignificant P variability can be observed in a smaller area, mainly in the southeastern part of ATLSRw and in the south of BAFR. Similar to the case with annual totals, the seasonal P dispersion is greater in the areas where P is the lowest. It follows that a low P is usually more 'sensitive' to the fluctuations of factors conducive to precipitation, a point which was also confirmed by Kozuchowski (1985) when analysing v for precipitation in Poland.

Figure 6.6. The spatial distribution of the coefficient of variability in winter, spring, summer, and autumn P over the period 1951-1990 in the Arctic.

Aside from the diversification of P variability in the Arctic (Figures 6.5 and 6.6), it is important to establish whether the differences of v are statistically significant. To this end, standard errors of the differences of extreme v values were calculated for annual P, and they are as follows: v/ = 15.1% (Bjornoya), and v2 = 40.8% (Danmarkshavn). The formula presented by Gregory (1976) was used:


bs - standard error of the difference of variability coefficients, np n2 - sample sizes.

The error for the stations that were analysed is 4.9%. As it possesses properties of a normal distribution (Gregory 1976), for the 0.05 level the significant differences are those which exceed its double value, namely 9.8%. The difference between v in Danmarkshavn and Bjornoya is 25.7%, so it is significant. Calculations showed that variability of annual P in Danmarkshavn is statistically significantly different from its variability in different stations when v in those stations is lower than 30%. These areas include ATLR (without ATLSRw), the western part of SIBR, most of the Canadian Arctic, and the Atlantic part of IARCR (Figure 6.5). For maximum v, which ranges around 30% and 20%, significant differences are those which exceed 8% and 5% respectively.

It follows that there is a substantial spatial diversification in v of annual P in the Arctic. This is also true for seasonal P, whose variability diversification is much greater.

Aside from the changes in P totals, climatologists are also increasingly interested in so-called 'time-dependent changes of variability'. The increase in the variability of some climatic elements that was observed in the 1980s was believed to have been caused by the growing greenhouse effect. We have stated so far that the behaviour of P in the Arctic (in terms of its mean values) is inconsistent with the changes to which this element should be subject in the circumstances of global warming. Let us check then whether there is any consistency as far as P variability is concerned. In order to do this we calculated v of winter, summer, and annual P in moving decades for 9 stations which represent particular climatic regions and sub-regions of the Arctic (Figure 6.7). The courses of P variability presented in this way allow us to calculate the periods with their high and low values, to establish when their tendencies change, etc. Figure 6.7 shows that v changed to a considerable extent in all the stations. Usually the variability of seasonal P is greater than that of annual P. Among the analysed stations, the greatest range of changes of v of annual P was observed in Clyde A in 1951-1990. In the decade 1951-1960 v was 57.6%, and from 1977-1986 it was only 17.8%. Its range was slightly smaller in Barrow, from 49.9% (1961-1970) to 14.6% (1977-1986).

Figure 6.7. Variability of coefficients of winter, summer, and annual totals of P in running decades in selected Arctic stations.

The course of curves which present fluctuations of v in selected stations is quite diverse. If we analyse the behaviour of v in the period from 1960 to 1990 we will notice that a clear increase in the variability of annual and seasonal P occurred only in ATLSRn (Ostrov Vize). In most of the stations a downward trend in v was observed. The trend is particularly visible in the area from Alaska to Greenland, with the exception of the northern parts of the Canadian Arctic, where the decrease of v is constant but inconsiderable (Fig ure 6.7). A type of v change similar to that in CANSRn (Resolute A) occurs also in ATLSRs (Jan Mayen) and, to the largest extent, in SIBR (Ostrov Kotelny).

Summer and winter P are characterised by very irregular and sudden changes in v, which quite often disrupt certain tendencies. In most of the stations their course is roughly similar to the courses of annual P (Figure 6.7). The fluctuations of v of summer and winter P are approximately analogous in some ofthe stations (e.g. in Jan Mayen, Ostrov Vize, Ostrov Kotelny, or Coral Harbour A). In the remaining stations the courses are less similar, and often the tendency is even reverse (e.g. in Danmarkshavn, Barrow, or Mys Kamenny). It is also worth noticing that in the areas characterised by an advectional origin of P (Jan Mayen, Clyde A) fluctuations of v of annual and winter P are very similar, whereas in the areas dominated by anticyclones there is more similarity between the changes in variability of annual and summer P (e.g. Danmarkshavn, Barrow, Ostrov Kotelny).

The data from Barrow and Jan Mayen stations (Figure 6.7) lead us to assume that P variability in the Arctic was highest in the period from 1921 to 1950. In Barrow it was particularly high between 1921 and 1945, whereas in Jan Mayen the period of high v lasted until the early 1950s, decreasing slightly at the turn of the 1920s and 1930s.

We proved above the existence of considerable changes of v over time. It should be checked, however, whether they are statistically significant. In order to do this we use the aforementioned formula for a standard error of v difference, using extreme values of v in calculations. It turns out that at the level 0.05, significant differences occurred in all the stations.

It should be concluded that a significant decrease in the variability of annual P occurred throughout most of the area of the Arctic in the last few decades of the period of observations. Thus, this characteristic of P is not consistent with the expected changes which should occur along with the global warming determined by a growing concentration of C02 and other trace gases. For Poland, Kozuchowski (1985) observed a significant increase in the variability of annual P in the period from 1881 to 1980, which persisted in the last few decades of his research period. The results prove the existence of different tendencies of P changes and P variability changes in the Arctic and in Poland, and perhaps also in the whole of Europe.

Was this article helpful?

0 0

Post a comment