The global-mean tropospheric temperature is monitored by two independent observing systems: radiosondes and polar-orbiting satellites. The global radiosonde record began in 1958 and has the advantage of a relatively long time series of directly measured temperatures. The satellite record began in 1979 and has the advantages of 1) sampling the atmosphere with one type of instrument at any given time; and 2) global sampling, with most of the earth sampled twice-daily from each of two instruments flying concurrently on different satellites (Spencer et al. 1990). The radiosonde observations for 1995 (Fig. 8a) indicate a mean tropospheric temperature anomaly of 0.25°C, computed with respect to the 1961_90 base period (red bars). This value is the highest annual, global mean anomaly observed since the eruption of Mt. Pinatubo in 1991, but is below the extremely warm 1987_1991 period. The 1995 radiosonde-estimated anomaly is also larger than any value recorded prior to 1979, the beginning of the satellite record.
The satellite-based temperature anomalies (Fig. 8b), obtained from measurements taken by the Microwave Sounding Unit (MSU) channel 2R flown aboard the polar-orbiting satellites (Spencer et al. 1990) and calculated with respect to the 1982_91 period, show an annual global-mean anomaly of 0.03°C for 1995. This value is well below the magnitude of the positive anomalies experienced during the 1987 - 1991 period.
The radiosonde time series (red bars in Fig. 8a) appears to show substantially larger departures in tropospheric mean temperature than the MSU time series for the past two decades (Fig. 8b). However, these differences are largely attributed to the different base periods from which the anomalies are calculated. When the radiosonde and MSU anomalies are calculated from the same base period (1982_91), both time series (blue bars in Figs. 8a, b) become remarkably similar. These independent estimates lend credibility to both measurement techniques used in quantifying global-mean tropospheric temperature anomalies. Note also that there are large-amplitude swings in global-mean temperatures occurring over periods of decades. Neither time series is sufficiently long to accurately assess this variability.
The positive global tropospheric temperature anomaly for 1995 could be interpreted as a recovery from the Mt. Pinatubo-related tropospheric cooling, which dominated these time series for the previous three years. It also suggests that the developing cold episode in the equatorial Pacific was yet to be felt in the global, annual mean temperature. However, the influence of this cold episode is evident in the pattern of below-normal mean annual temperatures throughout the eastern equatorial Pacific (Fig. 9).
Elsewhere, the spatial pattern of annual mean tropospheric temperature anomalies shows a predominance of positive anomalies in the middle and high latitudes of the Northern Hemisphere. These conditions are in strong contrast to the mean negative anomalies observed in the Southern Hemisphere poleward of 30°S at most longitudes.
December 1995 featured a dramatic drop to below-normal global mean tropospheric temperatures (0.23°C). This drop resulted from a pronounced shift to below-normal temperatures in the Northern Hemisphere extratropics(Fig. 10), and from a continued decrease in temperatures throughout the global tropics,presumably in association with a strengthening of cold episode conditions in the tropical Pacific.
Global lower-stratospheric temperatures are estimated from radiosonde observations (Fig. 11), satellite observations (Fig. 12), and analyses obtained from a global data assimilation system that utilize both sources of data (Fig. 13 ) (Kalnay et al. 1996). During 1995, all three analysis techniques showed a continuation of well below-normal temperatures in the lower stratosphere. The radiosonde estimate shows 1995 to be the coldest year in the 37-year record and the tenth out of the past eleven years with negative temperature anomalies (only 1992 had a positive anomaly). One difference in the relative behavior of the three time series is that the satellite derived anomalies show 1995 to be slightly warmer than 1994, whereas both the radiosonde-derived anomalies and the reanalyzed data obtained from the NCAR/NCEP Reanalysis Project (Fig. 13 ) show 1995 to be colder than 1994.
Both the satellite and Reanalysis time series also show pronounced warm periods during 1982/83 and 1991/92. These periods are also relatively warm in the radiosonde data (Fig. 11 ) and immediately follow the major volcanic eruptions of El Chichon (April 1983) and Mt. Pinatubo (June 1991), respectively. These eruptions contributed to markedly increased aerosol concentrations, resulting in elevated temperatures in the lower stratosphere. This is in contrast to the cooler-than-normal lower-stratospheric temperatures that would have been expected during the 1982/83 and 1991/92 ENSO episodes.
In the years following these eruptions stratospheric temperatures decreased to lower values than were observed prior to the volcanic eruptions (Figs. 11, 12 , and 13). One possible explanation for this decrease is that the increased aerosol concentrations in the stratosphere contribute to a decrease in ozone, and subsequently to a cooling in that region. Ozone depletion would be expected to continue after a volcanic eruption until aerosol concentrations return to normal. Thereafter, a gradual increase in ozone and temperature would be expected.
