Mean Monthly Seasonal and Annual Air Temperature

Air temperature is the most important, and therefore also most often studied, climatological clement. This is as true for the Arctic as it is for everywhere else. For this reason, our knowledge about this element, in comparison with others, is the best, but is still not sufficient for some parts of the Arctic (e.g. the central Arctic and the inner part of Greenland).

The instrumental records of Arctic temperature arc brief and geographically sparse. Only five records (Upernavik: commenced 1874; Jakobshavn: 1874; Godth&b: 1876; Ivigtut: 1880 and Angmagssalik: 1895) extend back to the second half of the 19lh century. As can be seen, all climatic stations operating during the nineteenth century were located in Greenland. Outside of Greenland, the first station was established in Spitsbergen in 1911 (Green Harbour). In the 1920s, the next seven stations came into operation, mainly in the Atlantic region of the Arctic. Following the Second International Polar Year (1932/1933) most of the Russian stations were established, while most of the Canadian stations were founded after World War II. For this reason, both spatial distribution and reliable estimates of air temperature characteristics in the Arctic are only possible for the last 40-50 years.

Besides the stations, extensive meteorological data have also been gathered during the well-known Fram (1893-1896) and Maud (1918-1925) expeditions. Later on, since 1937 Soviet drifting stations have supplied a large stream of different kinds of meteorological data for the central part of the Arctic Ocean. Unfortunately, this long-term project ended in 1991. Luckily, however, the Polar Science Center at the University of Washington ran a new research project "The Arctic Ocean Buoy Program" in 1979. Early that year an array of automatic data buoys was established in the Arctic Ocean. The main objectives of this programme are to provide measurement of surface atmospheric pressure over the Arctic Ocean and to define the large-scale field of motion of the sea ice. The temperature sensors are installed inside the buoys and therefore give only rough information about this element. Since 1992, however, they have been mounted outside, so the quality of data is significantly greater. Satellites also constitute new and extremely powerful sources of information about temperature in the Arctic.

Although, as has been seen, the first reliable climatological estimation of the spatial air temperature distribution in the Arctic was practically possible only in the second half of 1950s, there were also some earlier attempts to do this. Mohn (1905) was probably first to publish maps showing the spatial distribution of mean air temperatures for all months and for the year. The next proposition was given by Brown (1927), though only for January and July. The maps are, of course, in both cases very schematic. For example, Brown (1927) drew isotherms every 30°F. In the central Arctic, temperatures in January and July were estimated to be about -30°F (-34.4°C) and 30°F (-1.1°C), respectively. More accurate maps (with isotherms every 5°C) were published by Mecking (1928). In his maps the lowest temperatures in the Arctic occur near the North Pole and on the Greenland Ice Sheet, where they reach about -40°C (January) and 0°C (July). Sverdrup (1935) presented a significantly more detailed spatial distribution of air temperature (isotherms every 2°C) for almost the whole Arctic and for every second month (starting from January). The coldest month was January with the mean temperature below 36°C, including the area spreading from the North Pole to the northern parts of Greenland and the Canadian Arctic Archipelago. In comparison with Mecking's map, Sverdrup assumed that the mean temperature in July (near 0°C) is not only present around the North Pole, but over the whole Arctic Ocean, including large parts of the Arctic shelf seas.

Reviewing the climatological literature after World War II, we find only a few more propositions (besides those presented above) showing the spatial distribution of the temperature in the Arctic (Pctterssen et al. 1956; Prik 1959; Baird 1964; Central Intelligence Agency 1978; Herman 1986; Parkinson et al. 1987 (based on data from Crutcherand Meserve 1970) and Przybylak 1996a, b, 2000a). Most other sources (e.g. Prik 1960; Sater 1969; Vowinckel and Orvig 1970; Donina 1971; Sater et al. 1971; Barry and Hare 1974; Sugden 1982; Martyn 1985) mainly reprint some of Prik's (1959) maps. In turn, maps published in well-known Russian atlases (Gorshkov 1980 and Atlas Arktiki 1985) are also updated versions of maps which had been authored and edited by Prik (1959). The majority of the sources cited, as Przybylak (1996a) has noted, present only the spatial distribution of temperature for January and July. Only Przybylak (1996a) has published maps for the four meteorological seasons popularly used in climatology (Dcc.-Feb., March-May etc.) and for the year as a whole. About 40 homogeneous continuous series of temperature from the period 1951-1990 have been used to draw these maps. Out of the regional research, one should mention such works as those by Rae (1951), Donina (1971), Maxwell (1980). Barry and Kiladis (1982), Ohmura (1987), Calanca et al. (2000).

In this monograph, the maps published by Przybylak (1996a) are presented because: 1) they show actual mean seasonal and annual temperature conditions in the Arctic, 2) Prik's maps are easily available in the existing literature, and 3) Ohmura (1987) showed that the distribution of temperature over Greenland in Prik's maps contains "serious climatological errors". I must add that these errors are eliminated in the maps presented in Atlas Arktiki (1985), which is, however, a less available source. Przybylak (1996a) does not give the temperature for Greenland. Therefore, the temperature maps for Greenland after Ohmura (1987) and Calanca (personal communication) are also presented.

4.1.1 Annual and Daily Cycles of Temperature

The annual cycle of air temperature is a result of

1. changes in the amount of energy received from the sun, which depends on the geographical latitude and season of the year,

2. changes in atmospheric circulation,

3. changes in the physical properties of the underlying surface.

Petterssen et al. (1956) have distinguished three well-defined types in the annual cycle of temperatures in the Arctic: 1) maritime, 2) coastal, and 3) continental. These types can be seen in Figure 4.1, Jan Mayen has a maritime type, Malye Karmakuly and Egedesminde have a coastal type, and the rest of the stations have a continental type. From the map presenting the thermic continentality of the Arctic climate (Figure 4.2), we can see that Jan Mayen has a continentality below 20%, Malye Karmakuly and Egedesminde about 40-50%, and the rest of the stations above 60%. Thus, we can assume that regions of the Arctic with a degree of continentality lower than 30% probably have a maritime type, areas with a continentality ranging from 30% to about 55% have a coastal type, and regions with a continentality > 55% have a continental type. From Figure 4.2 it is evident that the continental type is the most common type, occurring in almost 80% of the Arctic, excluding mainly the Atlantic region, the southern part of Baffin Bay, and probably also the Pacific region. The second most common variety is the maritime type. What are the main features of these types?

Maritime type (Jan Mayen). Very small annua! range of temperature (i.e. difference between the highest and the lowest mean monthly temperature) slightly exceeding 10°C. A curve presenting the mean temperatures of the summer months (June-August) shows a small variation only between 4-5°C. There is a similar situation in the winter months (from December to March) when the temperature oscillates between -5°C and -6°C. The maximum temperature is shifted to August, and the minimum temperature to March.

— a —b —c —d
-e---f -g
JFMAMJJASOND

Figure 4.1. Annual course of air temperature (according to monthly means) in the selected Arctic stations, 1961-1990 (after Przybylak 1996b). a - Danmarkshavn, b .ian Mayen, c - Malye Karmakuly, d - Polar GMO E.T. Krenkelya, e Ostrov Dikson, f - Ostrov Kotelny, g - Mys Shinidta, h Resolute A, i - Cora! Harbour A, j - Clyde A, k - Egedesminde, t - average temperature for the 65-85"N zone for the period 1961 1986 (after Alckseev and Svyashchennikov 1991).

Figure 4.2. Thermic continentality of the climate in the Arctic (after Ewerl 1997).

Coastal type (Malye Karmakulv. EgedesmindeV This type, in principle, can be called "the transitional type" between the first (maritime) and the third (continental) types. The annual range of temperature (20°C) here is about twice that of the first type and half that of the continental type. Air temperature in winter is markedly lower than in the maritime type and also significantly greater than in the continental type. The minimum temperature is often shifted to February and sometimes, like in the case of Egedesminde, even to March. In summer, air temperatures are similar or higher than in the maritime type. The maximum can sometimes be delayed as Petterssen et al. (1956) note, but in our case such a situation does not occur.

Continental type (rest of stations). This type is characterised by the highest annual range of temperature (about 40°C), the lowest winter temperature (monthly means oscillate between -30°C and -35°C) occurring mostly in Janu ary and rarely in February. Summer temperatures are relatively very high, especially in southern parts of the Arctic and can reach values near I0°C. In the high Arctic, however, they are reduced to only l-3°C (Figures 4.1a, d, g, and h). At the North Pole, the mean temperature of the warmest month (July) is only equal to -0.5°C (see Table 4.1). In the central part of the Greenland Ice Sheet (Eismitte) the mean monthly temperatures oscillate from about -42°C (February) to about -12°C (July) (Donina 1971).

Table 4.1. Monthly and annua! mean air temperature (°C) (after Radionov el al. 1997)

Region

Monlh

Jan.

Feb.

March

April

May

June

July

Aug.

Sept.

Ocl.

Nov.

Dec.

Year

North Pole

-32.3

-35.4

-33.8

-25.8

-!2.1

-2.4

-0.5

-2.2

-9.5

-19

-28.1

-31.5

-19.4

Siberian

-31.5

-31.8

-31.4

-24.9

-10.8

-1.8

-0.1

-1.3

-7.2

-17

-25.3

-30.9

-17.8

Pacific

-30.8

-31.2

-29.7

-22.6

-10.5

-2.2

-0.1

-1

-i.5

-18.3

-25.7

-29

-17.3

Ocean Central

-32.4

-34.4

-32.8

-25.9

-12.1

-2.3

-0.3

-1.7

-8.9

-19.4

-28.3

-31.2

-19.1

According to the presented temperature data from stations, the wannest month is July. Out of 11 stations representing different climatic regions of the Arctic, this was the case in as many as 9 stations. Only in Jan Mayen and Ostrov Dikson was the warmest monthly temperature in August. The coldest month is most often February (6 stations), January (3 stations), or March (2 stations). These results are in line both with recently reported areally-averaged monthly temperatures for the Siberian, Pacific, and Central Ocean regions and with results from the North Pole (see Table 4.1), where clearly the highest temperatures occur in July. The lowest temperatures are noted in all regions in February, but in the Siberian and Pacific regions the temperature differences between Febniary and January arc very small (< 0.5°C). The annual cycle of mean temperature from the latitude band 65-85°N shows similar results (Figure 4.11)

The mean monthly daily courses of air temperature have a clear asymmetric course throughout the year, except during the polar night. The second half of the day is usually wanner. On average, the highest temperature occurs between 13°° and 15°° (Figure 4.3). In the polar night, Przybylak (1992a) distinguished five basic types of daily courses of temperature in Hornsund (Spitsbergen): 1) a "normal" pattern with a maximum temperature in "daytime" hours and minimum values at "night"; 2) a "reverse" pattern with a maximum temperature at "night" and minimum in "daytime" hours; 3) with a tendency towards increasing temperatures throughout the entire 24 hours; 4) with a tendency towards decreasing temperatures throughout the entire 24 hours and 5) with a nearly constant temperature throughout the entire 24 hours. During the four winter months (Nov. - Feb.) of the period 1978-1983, types 4 (25.9%) and 2 & 3 (23.3% each) were most frequent. The "reverse" daily course of temperature occurred mainly in December (25.3%) and January (17.4%). This was most apparent on fine days (Przybytak 1992a).

XI

Figure 4.3, Mean monthly daily courses of air temperature in Hornsund (Spitsbergen), 19791983 (after Przybylak 1992a). 1 - January, II - February, III - March etc.

Figure 4.3, Mean monthly daily courses of air temperature in Hornsund (Spitsbergen), 19791983 (after Przybylak 1992a). 1 - January, II - February, III - March etc.

Przybylak (1992a) also found that the occurrence of the "reverse" daily course of temperature during the polar night in Hornsund is usually accidental and mainly connected with nonperiodic changes of temperature resulting from intensive cyclonic activity. Even if we assume that there are other factors favouring the occurrence of the "reverse" daily course of temperature (e.g. daily periodicity of radiation balance and the influence of ozone on it, or daily changes of geomagnetic activity) they may manifest themselves only in some synoptic situations (usually anticyclonic, non-advectional) which, however, occur very rarely during this period.

The clearest daily courses occur on days with little cloudiness. This can be seen in all seasons, except during the polar night, when the differentiation role of the cloudiness is negligible (see Figure 15 in Przybylak 1992a).

4.1.2 Spatial Temperature Patterns

Air temperature in the Arctic shows a great spatial variability in al! seasons, but particularly in the cold half-year (see Figures 4.4-4.6). The well-developed atmospheric circulation (and cyclonic activity in particular) during this period is the main factor responsible for this situation. The coldest placc in the Arctic in all seasons is the northern part of the Greenland Ice Sheet. The thermal regime of this part of Greenland is shaped by high elevation above sea level, the character of the underlying surface (snow and ice), and the occurrence of quasi-stationary anticyclone circulation.

Figure 4.4. Spalial distribution of mean seasonal (DJF, MAM etc.) air temperature (in °C) in the Arctic, 1951-1990 (after Przybylak 1996a, modified for the central Arctic».

In winter the mean temperature in Greenland drops below 40°C (Figure 4.6a), reaching almost -45°C. The secondary minimum temperature is shifted from the North Pole towards the Canadian Arctic Archipelago and Greenland, where the mean winter temperature is around -36°C in the vicin-

ity of the Eureka station (Figure 4.4). A belt of low temperature (< -30°C) spreads from this area through the North Pole towards the central part of the Siberian region. The balance between loss of heat from the snow and ice surface by radiation and the gain in energy conducted by the surface from the water under the ice and transported from the lower latitudes by the atmospheric circulation can explain the existence of this large homogeneously cold area. The highest temperatures in the Arctic occur in the southernmost parts of the Atlantic and Baffin Bay regions, where they oscillate between -2°C and -6°C. These high temperatures are the result of the transport of warm air masses within the very intense cyclonic activity connected with the Icelandic tow (see Figure 2.3a). Cyclones which enter the Pacific region are weaker than Atlantic cyclones, but their warming effect is also evident (see the shape of isotherms). Generally, as rightly noted by Radionov et al. (1997), the patterns of the isotherms and of the isobars are in good agreement with one another (compare Figure 2.2a with Figure 4.4 or see Atlas Arktiki 1985).

Figure 4.5. Spatial distributions of mean annual maximum (Tmt), minimum (Tmia) and average daily (TmeJ air temperature (in °C) in the Arctic, 195l-199o"'(after Przybylak " 1996a).

In spring and autumn, the general patterns of temperature distribution in the Arctic are very similar to those in winter (Figure 4.4). The main difference lies in the magnitude of the temperatures, which, of coursc, are significantly higher in transitional seasons by about I0-15°C. Comparing, however, temperatures in spring and autumn, one can see that spring temperatures are markedly lower by about 6--8°C. Such a situation is not only observable in the central part of the plateau of the Greenland Ice Sheet, where temperatures are similar or even colder in autumn (compare Figures 4.6b and 4.6d). At the Eismitte station the mean temperatures in spring and autumn, calculated according to data published by Ohmura (1987), are equal to -30.9°C and -31.0°C, respectively.

In summer, the distribution of temperature is clearly more dependent upon the insolation (polar day) than upon the atmospheric circulation. As a result, the courses of the isotherms are more latitudinal. In comparison with other seasons, the smallest spatial air temperahire variation is also noted (the smallest horizontal temperature gradients). As mentioned earlier, the lowest mean summer temperature occurs in the northern part of the Greenland Ice Sheet (about -15°C), but, in comparison with other seasons, this temperature may be seen to shift significantly to the South, to the region of the highest elevation in Greenland (Figure 4.6c). In July the temperature drops slightly here below -10°C, and in June and August below -15°C. The secondary temperature minimum occurs in the central Arctic, where the prevailing melting of snow and ice holds the surface temperature slightly below freezing point (Figure 4.4). The highest summer temperature (> 8°C) is observable in the central continental parts of the Canadian and Russian Arctic. Areas where strong cyclonic activity prevails (the Atlantic and Baffin Bay regions) tend to be relatively cold (4 8°C).

Annual mean temperatures depend mainly on temperatures occurring in the cold half-year. Therefore, the patterns of annual, winter, autumn, and spring temperature in the Arctic arc very similar (compare Figure 4.5 with Figure 4.4). The lowest temperatures occur in the Greenland Ice Sheet (> 2000 m a.s.l.), where the mean annual ones above 70°N are below -20°C. In the central part, this minimum drops even below -30°C. Outside Greenland, the lowest temperatures (< -18°C) occur in the same region as in winter, spring, and autumn, i.e., in the north-eastern part of the Canadian Arctic Archipelago. The Eureka station has noted the lowest mean annual temperature (-I9.7°C). Only a slightly warmer temperature occurs around the North Pole. The mean annual temperature at the North Pole for the period 1954-1991 was equal to -19.4°C (Table 4.1). Slightly wanner conditions occur in the central part of the Arctic (above 85°N from the Atlantic side and above 80°N from the Pacific side), where mean annual temperatures oscillate between -16°C and -18°C. The warmest parts of the Arctic are those where cyclonic activity is greatest, i.e., firstly the regions spreading from Iceland to the Kara Sea (Atlantic region) and then the Baffin Bay and Pacific regions. In all these areas the isotherms are bent to the north.

Figure 4.6. Spatial distribution of mean seasonal (a - DJF, b - MAM, c - JJA, d - SON, after Calanca, personal communication) and annual (e, after Ohmura 1987) air temperature (in °C) in Greenland.

I
Figure 4.7. Spatial distribution of the standard deviations (in °C) of winter (DJF), summer (JJA), and annual (Year) air temperature in the Arctic, 1951 1990 (after Przybylak 1996a).

The variability of the annual mean temperature (Figure 4.7) is the greatest (a > 1.5°C) in the area between Spitsbergen, Zemlya Frantsa Josifa, and Novaya Zemlya, and the smallest (cf < 1,0°C) in the greater part of the Siberian region, in the north of ihe Canadian Arctic Archipelago, and probably also in the central Arctic Ocean, particularly from the Pacific side. Significantly, the mean winter temperature has the greatest variability. Przybylak (1996a) has distinguished three regions of the highest variability (o > 2.5°C): 1) the central and eastern part of the Atlantic region, 2) the belt encompassing the southern part of the Baffin Bay region and the south-eastern part of the Canadian Arctic, 3) the eastern part of the Pacific region (mainly Alaska). The reason for (his high variability here is, without doubt, the vigorous cyclonic activity. The cyclones bring into the Arctic warm air masses from the lower latitudes. The greatest variability, however, does not occur in these areas where the cyclone activity is most common (see Figure 2.3a). This has been noted in the areas where changes of different kinds of air masses most often occur, e.g. air masses of maritime origin (warm) transported by cyclones and of Arctic or polar-continental origin (cold) flowing in from the northern sector. Other Arctic areas which are most often occupied throughout the year by either cold (central Arctic) or warm (seas in contact with the mid-latitudes) air masses, have the lowest variability. Mean summer temperatures have the lowest variability (a rarely exceeding 1.5°C). This occurs only in some areas of the Russian Arctic (Figure 4.7). The greatest stability of summer temperature may be seen in the region from Spitsbergen to Severnaya Zemlya (C < 0.5°C). The smallest summer temperature variability can be explained by (Przybylak 1996a): 1) the lowest thermal differentiation of inflowing air masses (Przybylak 1992a), 2) small daily differences in the altitude of the sun (polar day), and 3) the stabilising influence of the melting of snow and sea ice, which is especially strong in the Arctic Ocean.

Figure 4.8. Isocorrclates of mean annua! air temperature in the Arctic in relation to Svalbard Lufthavn, Ostrov Kotelny, and Resolute A stations, 1951 1990 (after Przybylak 1997b). I - positive correlations statistically significant at the level 0.05, 2 - negative correlations statistically significant at the level 0.05, 3 - meteorological stations. 4 - station in relation to which the correlation coefficients of temperature were computed.

Alckseev and Svyashchennikov (1991) and Przybylak (1997b) found that air temperature in the Arctic is most spatially correlated in winter and spring, and least in summer. Mean annual temperatures reveal a slightly stronger correlation than winter temperatures (see Przybylak 1997b). The radius of the extent of statistically significant coefficients of correlation of temperature changes around the stations Svalbard Lufthavn (Atlantic region), Ostrov Kotelny (Siberian region), and Resolute A (Canadian region) is equal to 2500-3000 km for annual values, 2000-2500 km for winter, and 1500-2000 km for summer (Figure 4.8). From Figure 4.8 it can be seen that of the three analysed climatic regions, the highest spatial correlation of temperature occurred in the Canadian region, probably due to the highest stability of atmospheric circulation, especially in the wmter and spring (Serreze etal. 1993).

The strong correlation of the winter temperature in the Atlantic and Baffin Bay regions seems to be due to very common vigorous cyclonic activity (Baranowski 1977b; Niedzwiedz 1987, 1993; Przybylak 1992a; Serreze et al. 1993). This circulation which carries warm and humid air masses from the lower latitudes, diminishes local and even regional features of climatic variations. Cyclones move most often along the Iceland-Kara Sea trough. As a result, the isocorrelates in the eastern Atlantic region have a north-eastern bend. This bend is also present in isocorrelates of the annual temperature (see Figure 4.8). The correlation of winter temperature changes in these regions is also undoubtedly caused by a lack of solar radiation over most of the area. In the other Arctic regions, the strong correlation of temperature change is probably also determined by the predominance of anticyclonic circulation as well as by a high uniformity of the ground. In spring - almost over the whole Arctic - high coefficients of correlation of temperature change are most probably connected with the highest simultaneous homogeneity of the ground (the largest part of (he Arctic is covered by snow and sea ice) which, moreover, favours the development and upholding of anticyclones. The low correlation of the summer temperature change is probably caused by: (i) the greatest differentiation of the ground during this season, (ii) weak and evenly distributed cyclonic and anticyclonic circulation (see Figure 2.3), (iii) the influence of local conditions, which are remarkable during this season (it is at this time that the highest values of incoming solar radiation can be noted during a polar day).

Working from the least geometrical distances and using the dendrite method, Przybylak (1997b) delimited 9 groups of stations with the most similar (coherent) mean annual temperature in the Arctic. The schematic distribution of these regions is presented in Figure 7 (Przybylak 1997b). It can be seen that most regions consist of two isolated areas. For example, the southeastern part of the Canadian Arctic has a similar annual temperature to the area of the south-western Kara Sea and its surroundings in the Atlantic re gion, and the Pacific region has the same annual temperature as the northeastern part of Greenland.

Przybylak (1997b) found that the coefficients of correlation between seasonal and annual areally averaged Arctic (and also Arctic climatic regions) and Northern Hemisphere (two series: land only and land+sea, after Jones 1994) temperatures are not strong. The average Arctic temperature computed from 27 stations is statistically insignificantly positively correlated with both series of Northern Hemisphere temperature (for annual values r = 0.18), Out of the five series of the mean annual regional temperatures, the highest correspondence with the previously mentioned hemispheric series may be noted with the temperature of the Canadian region. The temperature of the Pacific region also has a high correlation, but only for the Northern Hemisphere temperature computed from the land stations (/• = 0.40). Such relations are typical for almost all seasons. The examined series of temperatures are most strongly correlated in spring and (especially) summer. For the latter season, the statistically significant correlations were computed between the Northern Hemisphere temperature and temperatures of the Atlantic and Canadian regions (see Table 4 in Przybylak 1997b).

0 0

Post a comment