Our current knowledge concerning cloudiness in the Arctic, as has already noted been by Raatz (1981), Barry et al. (1987) and Serreze and Rehder (1990), is still remarkably poor. Of the existing 15 distinct global cloud climatologies reviewed by Hughes (1984), only two (Scherr et al. 1968 and Berlyand and Strokina 1980) provide information about both poles while a further four have information for only one or other of the poles. In the second half of the 1980s and at the beginning of the 1990s four new cloud climatologies became available; one based on surface observations (Warren et al. 1986, 1988) and three based on satellite radiation measurements (METEOR - Matveev and Titov 1985; Aristova and Gruza 1987; Mokhov and Schlesinger 1993, 1994; N1MBUS-7 - Stowe et al. 1988, 1989; ISCCP - Rossow and Schiffer 1991; Rossow and Garder 1992). Rossow (1992) comments that these climatologies are in excellent agreement on the geographic and seasonal cloud amount variations, and even on total cloud amounts (except for one), everywhere except in the Polar regions. Here large differences occur both in the average geographical distribution of cloud amounts and in their annual march. Moreover, McGuffie et al. (1988) report that "none of the existing global cloud climatologies provides comprehensive information for the Polar regions". Moreover, usually the scale and projection of maps do not allow results to be presented in sufficient detail.

In the 1980s and 1990s quite a large number of papers relating to cloudiness in the Arctic were published. The majority of them use satellite-derived radiation measurements to estimate cloudiness. Two groups of works can be distinguished: 1) those presenting possible analysis methods (manual methods and those using computcr-based automatic algorithms) of cloud analyses (Key and Barry 1989; Dutton et al. 1991; Carsey 1992; Curry and Ebert 1992; Key and Haefliger 1992; Robinson et al. 1992; Serreze et al. 1992; Francis 1994; Hahn et al. 1995), and 2) those presenting satellite-based cloud climatologies and their comparisons with surface climatologies (e.g. Barry et al. 1987; Kukla and Robinson 1988; McGuffie et al 1988; Serreze and Rehder 1990; Rossow 1992, 1995; Schweiger and Key 1992). One should add here that the satellite-derived cloud climatologies which arc presently available have many weaknesses: they are incomplete, particularly in seasonal coverage, they are based on short time periods, and their reliability is often questionable as a result of cloud-detection problems.

Intercomparative studies of satellite-derived and surface-based cloud climatologies have shown significant differences in the results obtained (e.g.,

McGuffie et al. 1988; Rossow 1992; Schweiger and Key 1992). Schwciger and Key (1992) found that satellite-derived estimates of cloud amounts are generally 5-35% lower than had been indicated by surface observations over the entire Arctic, and regional differences may reach up to 45%. In the annual march these differences are 2-3 times greater from May to October than they are in winter. These authors concluded that at present it is not possible to determine which cloud climatology is 'correct'. Certainly, such a situation is very unfavourable and requires extensive further investigation in the near future, because:

— Polar regions are considered to be of great importance for the global climate (see e.g., The Polar Group 1980 or Arctic Climate System Study 1994),

— cloud cover is a major component of the Arctic climate system through its influence on both energy and moisture exchange between the elements of the system, i.e. atmosphere, ocean, cryosphere, biosphere, and litosphere.

All hope lies in improvements to the analysis methods of satellite-derived radiation measurements, which in many cases reduce the quality of current analyses of available data sets (Rossow 1995),

From the above review, it seems that the old climatologies, mainly based on surface observations, are still the best source of information (Vowinckel 1962; Huschke 1969; Vowinckel and Orvig 1970; Gorshkov 1980). They show a broad agreement over much of the Arctic in regard to the seasonal cycle of total cloud amount. However, there arc some disagreements in the geographical distribution of cloud cover, particularly in the case of low cloud in winter (McGuffie et al. 1988). Crane and Ban^ (1984) also found significant differences in the mean values of cloudiness presented in the above climatologies. These differences are probably connected with the different data sets used (a denser or sparser network of stations, longer or shorter periods of observations). For example, a short period of observations can significantly obscure results, especially those of winter cloud climatologies because of difficulties observing clouds with little sky illumination (sec e.g., Schneider et al. 1989; Hahn etal. 1995).

0 0

Post a comment