Ocena dokładności stosowanych metod obliczania średnich i ekstremalnych dobowych wartości temperatury powietrza w Arktyce Kanadyjskiej w XIX wieku

W artykule omówiono różne metody obliczania średnich dobowych temperatury powietrza w Arktyce Amerykańskiej w XIX wieku. Oceniono dokładność stosowania tych metod biorąc pod uwagę jako średnią wzorcową tzw. średnią dobową rzeczywistą temperaturę powietrza obliczaną z 24 danych cogodzinnych. Drugim problemem badawczych, który podjęto w artykule, jest oszacowanie wielkości błędów jakie się popełnia wybierając z różnych zbiorów danych godzinowych (co 1-, 2-, 3-godziny itd.) najwyższe i najniższe dobowe temperatury powietrza. Jako wzorzec w tym przypadku wykorzystano wartości temperatur maksymalnych i minimalnych powietrza odczytane z termometrów ekstremalnych. Podobną analizę przeprowadzono także dla amplitudy dobowej temperatury powietrza. Dla wszystkich analizowanych parametrów termicznych i dla wszystkich metod obliczania/wyznaczania średnich dobowych temperatury powietrza i temperatur ekstremalnych obliczono m. in. przeciętne błędy estymacji ich średnich miesięcznych wartości.

Knowledge about the history of climate in the Arctic is more and more important and necessary, especially at present when we are approaching the Fourth International Polar Year 2007-2008. Generally speaking, the history of the climate in this area during the 20th century is quite well known. On the other hand, little is known about the climate in the 19th century. Moreover, while we have extensive meteorological data for this period, in particular for the American Arctic, these data have many errors and biases. One of the most important biases is connected with the way in which daily mean air temperature has been calcu-lated. In the American Arctic during the 19th century nine different methods (m1-m9) were used. For the analysis we also added two presently used methods (m10-m11). The main aim of this paper is to estimate the magnitudes of errors which are connected with the use of the above methods of calculating daily means. As a real daily mean, the mean calculated using hourly data (m1) was used. Because in the American Arctic the mean daily air temperature is still calculated using formulae m11, we also calculated differences relative to this mean. Another issue which we undertake in the present paper concerns the estimation of errors which are the result of the method which was adopted to determine extreme temperatures (Tmax and Tmin) and the diurnal temperature range (DTR). We checked this for ten different methods (nine used in the 19th century) which used hourly, 2-, 3-, and 4-hourly etc. readings of air temperature for the purposes of calculation (see formulas m1-m10 for more details). As a base, real data, temperature readings from the extreme thermometers were used. For the analysis, hourly temperature data as well as daily Tmax and Tmin for the period 1979-1983 were used for the four meteorological stations (Eureka, Resolute, Coral Harbour and Iqaluit) located in the American Arctic. The results of our investigations are presented in Tables 1-5. The main conclusions can be summarized as follows: 1. Mean monthly temperatures obtained using methods m2-m5 and m9 of daily mean temperature calculation do not need to be corrected. The greatest errors (overestimation by 0.5 to 1.5°C) were found for the methods m6-m8 (owing to a lack of measurements during the night hours). The method m11 also produces significant errors. Generally, using this method, the mean monthly temperatures are most often lower (by 0.2 to 0.7°C) in relation to all methods analysed in the present paper (see Tables 1 and 2). 2. In accordance with expectations, mean monthly Tmax and Tmin determined using different methods are lower/higher than the respective monthly means calculated based on the readings from the maximum and minimum thermometers. When we determine Tmax and Tmin using hourly, 2-, and 3-hourly data, their monthly means are lower/higher, though generally by no more than 1.0°C. Greater errors are more clearly seen in the cold half-year than in the warm half-year (see Tables 3 and 4), 3. Mean monthly DTR calculated using hourly, 2-, and 3-hourly temperature data are lower than real values by about 0.5 to 2.0°C. For other methods of DTR calculations their errors are significantly greater - lower by about 3.0 to 4.0°C (see Table 5).