[Section 2.4] [Section 3] [Section 2] [Introduction] [Documents] [Data] [Start]

2.3 Trends in the Annual Mean


2.3 Trends in the Annual Mean

2.3.1 Temperature Trends at the North Pole since 1956

The time series of the 30-hPa annual mean temperatures (C) at the North Pole is shown in Fig.19. It is based on the data given in Table 4. If a linear trend is calculated for the whole series, the trend is -0.50 K/decade with a probability of 99%. But again, as discussed above, the interannual variability is large even in the annual mean and it is difficult to decide if this calculation is the correct answer to the question about temperature trend in the stratosphere. If one takes the first part of the time series, as indicated with the trend until 1979, there is no trend in the temperatures. If one starts with the data of 1979, the year when many satellite data became available and many trend analyses start, then a significant negative trend of -1.36 K/decade is obtained. As can be seen below this trend appears to be connected to changes in the winters/springs in the Arctic, see also discussions in Section 2.1. It is interesting to note that there are periods with a clear bi-annual signal and that the temperatures were relatively high in the beginning of the series, i.e. during the very strong solar maximum around 1958.

[30-hPa Annual Mean North Pole Temperatures]

Figure 19: Time series of the 30-hPa annual mean temperatures at the North Pole, 1956-2000. Linear trends have been calculated for different periods. Data: Free University Berlin

2.3.2 Hemispheric Maps of Annual Trends at 30 hPa

Maps of the linear trend of the annual mean 30-hPa temperatures and heights are shown in Fig.20, for the two periods: 1965-2000 (top) and 1979-2000 (bottom). As pointed out above, many studies on trends start with the year 1979, because satellite data became more frequent then. Using the whole data set, 36 years, it is obvious that over most of the maps the temperature and height trends are negative and the statistical significance is mostly above the 95% confidence level, Fig.21. The structure of the trends is very regular, indicating an intensification of the zonal winds over middle and high latitudes (Fig.20, upper right map) and two regions of cooling: one in the arctic and a second, stronger one over the subtropics, around 30N (Fig.20, upper left map). This agrees with the discussion of Fig.22.

During the shorter period, 22 years, the trends are generally larger but less organized, especially the trends of the heights, and single disturbed winters probably mask the general trend. One should therefore consider the shorter trends with caution. With this in mind, it is of interest to note that the double structure of the cooling is much more pronounced in the shorter, more recent period and that there is even an indication of warming over Siberia. This hints to an intensification of wave number one and of the Aleutian high which in turn could have a negative feedback on the polar vortex (a typical example of the AO).

[30-hPa: Trend of Annual Mean Temperatures]

Figure 20: Left: The linear trend of the annual mean 30-hPa temperatures (K/decade): top for the period 1965-2000, bottom for the period 1979-2000; right: the same for the geopotential heights (geopot. decameters/decade); (outer latitude is 10N). Data: Free University Berlin

[30-hPa: Probability of Annual Mean Temperature Trends]

Figure 21: Probabilities for the maps in Fig.20

2.3.3 Height and Temperature Trends between 100 and 30 hPa, 1965 - 2000

Because the temperature analyses started only in summer of 1964 annual temperatures are available starting with 1965, exept for the North Pole. Therefore the following discussions concentrate on this period, also for the geopotential heights.

The negative temperature trend in the stratosphere peaked at 50 hPa at all latitudes, mainly in summer and early winter (cf. Section 2.2.2), and therefore in the annual mean, Fig.22, upper part, and the trend weakens upwards of 50 hPa. It is of interest to note that two regions with larger negative trends exist: the polar region and the subtropics, divided by a belt with weaker cooling. This hints to an interplay between radiative (in the polar region, likely connected to the ozone decrease in the arctic) and dynamical effects.

The trends in the geopotential heights (Fig.22, lower part) reflect the tropospheric warming in the tropics and subtropics with a positive trend in the heights from the equator to about 55N, up to 50 hPa. With such a tropospheric warming the troposphere expands and the level of the tropopause rises, which in turn leads to a cooling above the tropopause at about 50 hPa; this may explain the observed cooling in the subtropical lower stratosphere.

Polewards the trend of the heights is negative, in hydrostatic balance with the temperature changes. The structure of the height trends indicates an intensification of the polar night jet in the annual mean, which is largely connected with the winter and spring, (Section 2.2.2). This intensified jet could be connected with a weaker transport of heat and momentum into the polar vortex thus leading to a cooling in the arctic, similar to the cooling connected with the ozone decrease.

[Trends of the annual mean, zonally averaged]

Figure 22: Trends of the annual mean, zonally averaged temperatures (top) and geopotential heights (bottom) over the Northern Hemisphere between 100 and 30 hpa (16 to 24 km) for the period 1965-2000. (Labitzke and van Loon, 1994, updated). Data: Free University Berlin


[Section 2.4] [Section 3] [Section 2] [Introduction] [Documents] [Data] [Start]

Stratospheric Research Group
Last modified: Wed Sep 11 22:32:26 MST 2002