Previous: Introduction Next: Discussion and Possible Mechanisms Up: Ext. Abst.

Observations and Model Comparisons

Observational evidence for a quasi-decadal oscillation (QDO) of the lower stratosphere and upper troposphere that is approx- imately in phase with the 11-year solar cycle has previously been reported using both long-term meteorological data [4,5] and column ozone data [6,7,8]. At high latitudes in winter, it has been found necessary to separate the data according to the phase of the equatorial quasi-biennial wind oscillation (QBO) in order to demonstrate a correlation between lower strato- spheric meteorological parameters and solar UV proxies such as the 10.7 cm radio flux (e.g., ref. [4]). At low latitudes (! 50 ffi), no such division of the data according to QBO phase is necessary to investigate a possible solar response although the QBO is still a major source of interannual variability.

Evidence for an apparent , 11-year variation of upper stratospheric ozone, temperature, and zonal wind has previ- ously been reported based on satellite data sets extending over a minimum of one solar cycle [9,10,11,12]. Based mainly on Nimbus 7 SBUV ozone profile data, it was found that the observed ozone change from solar minimum to maximum in the 40-50 km altitude range was significantly larger (by about a factor of 2) than the predictions of two-dimensional strato- spheric models. In contrast, the observed change in the mid- dle stratosphere (30-40 km) was significantly less than model predictions. These results have been generally supported by later analyses of SBUV-SBUV/2 data as well as Stratospheric Aerosol and Gas Experiment (SAGE) I/II data extending over 18 years (see Figure 3.5 in SPARC Report No. 1, Assessment of Trends in the Vertical Distribution of Ozone, May, 1998).

Figure 1a shows the annual mean variation of ozone mix- ing ratio (in per cent) from solar minimum to maximum de- rived from a combination of Nimbus 7 SBUV and NOAA 11 SBUV/2 data covering a 16-year period (1979-1994). A mul- tiple regression statistical model including linear trend, QBO, and volcanic aerosol terms in addition to the solar term was applied to the data following the method of McCormack and Hood [13], The heavy dark line separates regions where so- lar regression coefficients are statistically significant at the 2oe level from regions where no significant solar variation could be measured. The largest variation of more than 3% is present in the upper stratosphere at altitudes greater than 45 km while the solar variation in the tropical middle stratosphere is not statistically significant. Figure 1b shows the ozone variation from solar minimum to maximum predicted by a representa- tive general circulation model with parameterized chemistry [3] (D. Shindell, priv. comm., 2000). The model ozone varia- tion reaches a maximum of , 2.5% in the middle stratosphere (30-40 km altitude).

Although early estimates of the solar cycle variation of temperature in the middle and upper stratosphere relied on NCEP Climate Prediction Center (CPC) data derived from measurements by the Stratospheric Sounding Units (SSU) on the NOAA operational satellites, these estimates are subject to errors introduced by satellite changes. As discussed in the 1998 WMO Assessment report [ref. 14, p. 5.30], adjustments to the SSU time series using overlap periods allow an improved combined data set to be constructed covering the 1979-1995 period (17 years).

Figure 2a shows the mean stratospheric temperature vari- ation from solar minimum to maximum as derived from the overlap adjusted SSU data by the SPARC Stratospheric Tem- perature Trends Assessment project. At the lower boundary, estimates are based on measurements with Channel 4 of the Microwave Sounding Units (MSU) on the TIROS-N NOAA operational satellites [15]. The heavy dark lines separate re- gions where regression coefficients were positive at the 95% confidence level from regions where the coefficients were not significantly different from zero. A maximum temperature variation of , 0.8 K is present near 40 km altitude in the tropics. The amplitude decreases with decreasing altitude but begins to increase again in the tropical lowermost stratosphere. Figure 2b shows annual mean temperature changes from so- lar minimum to maximum estimated with the GISS GCM (D. Shindell, priv. comm.). Maximum temperature changes oc- cur in the upper stratosphere and decrease continuously with decreasing altitude. No increase in amplitude occurs in the lowermost stratosphere.

In order to assess further the possible existence of a sig- nificant solar cycle variation in lower stratospheric temper- The Solar Component of Stratospheric Variability: Hood and Soukharev ature, we consider further the MSU Channel 4 temperature record noted above. Channel 4 of the MSU instrument yields brightness temperature measurements at a series of frequen- cies near the 60-GHz oxygen absorption band. These represent weighted mean atmospheric temperatures in a layer with peak power at 75 hPa and half-power at 40 and 120 hPa [15]. This pressure range is mainly in the lowermost stratosphere but in- cludes significant contributions from the upper troposphere in the tropics. Standard errors of measurement for 5-day means at individual grid points at 2.5 ffi horizontal resolution are less than 0.25 ffi for most of the globe and less than 0.15ffi for the tropics. These microwave soundings are not affected by the presence of aerosols. Correlations between MSU4 and radiosonde profiles for monthly anomalies are usually above 0.94 and approach unity in some cases. Monthly signal-to-noise ratios are usually over 500 at individual grid points.

The top panel of Figure 3 shows a simple average of the MSU4 data for the tropics (20 ffi S to 20ffi N) but with the equatorial data removed to minimize interannual variability associated with the QBO. In addition to positive temperature anomalies resulting from the El Chichon (C) and Pinatubo (P) volcanic aerosol injection events, a long-term, approximately decadal variation is also present. As indicated in the lower panel of Figure 3, this decadal variation is roughly in phase with the Mg II index, a satellite-based proxy for solar UV variations. (The Mg II data are based on Nimbus 7 SBUV measurements prior to January, 1994 and on UARS SUSIM measurements afterwards.) Specifically, ignoring the periods following major volcanic eruptions, the mean temperature de- viation during solar minima is 0.5-0.8 K lower than during solar maxima. (Application of a multiple regression statisti- cal model yields an estimated solar cycle variation of 0.70 \Sigma 0.18 K.) During the past several years as the solar UV flux has increased toward the next solar maximum, the mean tropical temperature has also increased significantly by more than 0.5 K. An increase of similar magnitude occurred between 1986 and the 1990 solar maximum. These temperature increases can not be attributed easily to volcanic influences and are most probably solar in origin.

In order to investigate whether a significant variation of ozone from solar minimum to maximum is present in the lowermost stratosphere, we consider here a column ozone record derived from a combination of Total Ozone Mapping Spectrometer (TOMS) data and measurements by the Solar Backscattered Ultraviolet (SBUV) instruments on the Nimbus 7, NOAA 9, NOAA 11, Meteor 3, and Earth Probe satellites [16]. A merged record covering the period from January 1979 through December 1998 (R. Stolarski, private communication, 1999) was specifically studied. The second panel of Figure 3 shows a tropical average of the merged record. The intercali- bration of the record is best during the Nimbus 7 period prior to January 1994. In addition to a long-term trend and possible ozone decreases associated with the El Chichon and Pinatubo eruptions, a decadal variation is also evident that is approx- imately in phase with the solar Mg II record. In particular, an increase in tropical mean ozone occurs between solar mini- mum in 1986 and solar maximum in 1990 that can not easily be attributed to volcanic or anthropogenic influences. Although no evidence for a solar variation is present in the data after 1994, the shortness of the record (ending in 1998) and remain- ing questions about instrument intercalibration errors suggest that these data should not be considered as definitive.