tropospheric temperature in the extratropical Southern Hemisphere also fails to reproduce the pronounced warming that is evident in the surface air temperature record, as shown in Figure 11.
Figure 12 shows a comparison of the same surface air temperature and NMC upper-air temperature time series for the extratropical Northern Hemisphere, based on a longer period of record. Other discrepancies are evident around 1955 and 1963, when the mean 1,000 to 500 hPa temperature in the NMC analyses dropped substantially, while the surface temperatures cooled by a lesser amount. Lambert (1990) has suggested that both these features are reflections of changes in analysis procedures, while Shabbar et al.
(1990) have linked the second to a change in climate regimes.
There are many reasons to be suspicious of the sharp downward temperature trend during the late 1950s and early 1960s indicated by the NMC analyses. Changes in radiosonde sensors, reporting times, and analysis procedures that occurred during this period could have introduced spurious trends or discontinuities in the record. Yet in view of the marked differences between temperature trends in the MSU-2 and DOE data sets during the 1980s, it would be imprudent to dismiss the possibility that the cooling aloft might have been somewhat more pronounced than that at the earth's surface. A tendency toward stronger cooling aloft than at the surface would be indicative of a long-term decrease in the stratification of the lower troposphere.
A contributing factor may be the shift in the predominant polarity of the wintertime PNA pattern between the early 1950s and the 1980s. This change was marked by cooling of several degrees aloft over the North Pacific and warming by a comparable amount at the earth's surface over western Canada and much of Siberia (Gutzler et al., 1988; Wallace et al., 1993). Since this effect was observed only during the cold season over limited regions of the hemisphere, it seems unlikely that it could be the main reason for the difference between the time series of surface air temperature and temperatures aloft over the Northern Hemisphere. Of course, it also would not account for the difference between the recent temperature trends in the MSU-2 and DOE time series in the Southern Hemisphere, which is in the same sense.
A downward trend in static stability is also suggested by the marked difference between the long-term trends in daily maximum and daily minimum temperatures since the 1940s (Plantico et al., 1990; Karl et al., 1995). Minimum temperature exhibits a pronounced upward trend relative to maximum temperature, resulting in a decrease of several percent in the mean diurnal temperature range during this period of record. The decrease is observed during all seasons and over most geographical regions. Higher daily minimum surface air temperatures could be an indication of weaker nighttime inversions, which would contribute to a decrease in the mean static stability of the lower troposphere.
Figure 13 shows the time series of lower stratospheric temperatures based on Channel 4 of the MSU, whose level of unit optical depth is near 70 hPa. The global curve is the same as the one published by Spencer and Christy (1993). Consistent with earlier studies based on radiosonde data (Newell, 1970; Angell and Korshover, 1978), it shows distinct signatures of the two major volcanic eruptions during this period (El Chichon in April 1982 and Pinatubo in June 1991) whose plumes reached stratospheric levels. The