JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109, D05302, doi:10.1029/2003JD004243, 2004 Negligible long-term temperature trend in the upper atmosphere at 23°S Barclay R. Clemesha, Dale M. Simonich, and Paulo P. Batista Instituto Nacional de Pesquisas Espaciais, Sa˜o Jose´ dos Campos, Brazil Tomas Vondrak and John M. C. Plane School of Environmental Sciences, University of East Anglia, Norwich, UK Received 13 October 2003; revised 5 January 2004; accepted 13 January 2004; published 2 March 2004. [1] Measurements of the vertical distribution of atmospheric sodium, made at Sa˜o Jose´ dos Campos (23°S, 46°W) from 1972 to 2001 show a negligible net long-term trend in the layer over this time period. In view of the fact that cooling of the upper atmosphere, as indicated by some experimental studies and predicted by models, would be expected to influence the vertical distribution of sodium, this result can be used to estimate the limits of such cooling. In order to investigate the sensitivity of the sodium layer to changes in the atmospheric temperature profile, and thus estimate an upper limit to the changes it should be possible to detect, we have used a comprehensive model for the mesosphere and lower thermosphere (65–110 km) that includes all species and reactions thought to be relevant to the sodium chemistry. On the basis of this study we find that the sodium layer observations are consistent with model calculations for the global effects of increased CO2 but are not consistent with many of the experimental studies, which show cooling trends much larger than expected from theoretical studies. INDEX TERMS: 0317 Atmospheric Composition and Structure: Chemical kinetic and photochemical properties; 0350 Atmospheric Composition and Structure: Pressure, density, and temperature; 0340 Atmospheric Composition and Structure: Middle atmosphere—composition and chemistry; 1610 Global Change: Atmosphere (0315, 0325); KEYWORDS: global trend, atmospheric sodium layer Citation: Clemesha, B. R., D. M. Simonich, P. P. Batista, T. Vondrak, and J. M. C. Plane (2004), Negligible long-term temperature trend in the upper atmosphere at 23°S, J. Geophys. Res., 109, D05302, doi:10.1029/2003JD004243. 1. Introduction effects of the increase in CO2 known to have occurred over [2] Modeling studies suggest that the observed long-term the past 3–4 decades. For a change in CO2 concentration increase in the concentration of greenhouse gases in the from 331 to 360 ppm, they found temperature changes atmosphere should lead to atmospheric cooling at heights going from 0 at 20 km to about À10 K at 180 km. The trend above the tropopause. A number of workers have attempted did not increase monotonically with height, and was close to to model such an effect. These studies have generally aimed zero at 120 km. The trend modeled by Akmaev and at the so-called ‘‘doubled CO2 scenario.’’ One of the earliest Fomichev [2000] provides a convenient basis for compar- of such studies was that of Roble and Dickinson [1989]. For ison with measured trends. Portmann et al. [1995] studied a doubling in the concentration of atmospheric CO2, Roble the effects of including dynamical feedbacks in the Garcia and Dickinson [1989] found a temperature decrease of 15 K and Solomon [1985] two-dimensional model. They found at 50 km and more than 50 K at 400 km. that such feedbacks have a major effect, especially at the [3] In recent years a number of workers have developed summer pole, where the doubled-CO2 scenario produced more detailed atmospheric models allowing predictions to net heating of about 14 K without feedbacks, but cooling of be made for the effect of increased greenhouse gas concen- around 6 K with. Portmann et al. obtained typical trends, trations. Brasseur et al. [1990] presented a chemical- with dynamical feedbacks included, of 0 K at 20 km to dynamical-radiative model that includes Rossby wave À25 K at 110 km. absorption and gravity wave breaking. For the doubled [4] Experimental studies aimed at detecting long-term CO2 scenario they found maximum cooling at around trends in the MLT region have been made using various 50 km, with temperature changes between À16 K at the techniques. Rocket measurements [see, e.g., Keckhut et al., winter pole to À8 K at the equator. This model gave a small 1999] suggest trends of as much as À5 K/decade at 70 km. increase in temperature below 20 km, and zero effect at Rayleigh lidar [see, e.g., Keckhut et al., 2001] indicate a few 70 km. Akmaev and Fomichev [2000] have modeled the degrees per decade at heights between 50 and 70 km. Satellite measurements by Aikin et al. [1991] suggest a Copyright 2004 by the American Geophysical Union. cooling of 3.5 K per decade. These measurements were 0148-0227/04/2003JD004243 from the SSU 47X channel, corresponding to about 0.4 mB, D05302 1of8 D05302 CLEMESHA ET AL.: NEGLIGIBLE LONG-TERM TEMPERATURE TREND D05302 bution of sodium is also strongly influenced by atmospheric waves, including internal gravity waves, tides and planetary waves, but these effects can be expected to cancel out when averages are taken over long time periods. Long-term changes in wave amplitudes would, however, affect the eddy transport and consequently the sodium profile. Since the sodium profile depends on the vertical distribution of atmospheric constituents, both major and minor, it is to be expected that any changes in these distributions resulting from long-term global change might result in corresponding changes in sodium. Since we have sodium data taken over the past 30 years, we can search for such an effect. [7] Lidar measurements of the vertical distribution of atmospheric sodium have been made at Sa˜o Jose´ dos Campos since May 1972. Although numerous changes have been made to the equipment used for these measurements Figure 1. Annual variation of the Na layer centroid height. over the years, the determination of the relative vertical distribution of sodium is a straightforward measurement that does not depend on any difficult calibration process, and is or 55 km. OH rotational temperature measurements, which thus independent of equipment changes [Clemesha et al., refer to about 87 km, indicate trends that vary from À10 K/ 1992]. Determination of the absolute abundance of sodium decade [Golitsyn et al., 1996] to zero [Bittner et al., 2002]. is a much more difficult process, and we do not believe that Measurements of the reflection height of VLF radio signals our calibration of the absolute sensitivity of the lidar is have been interpreted by Taubenheim et al. [1997] as reliable enough for us to detect small long-term changes. indicating a À6 K/decade trend in the mesosphere. These For this reason, we restrict this study to the normalized measurements refer to the ‘‘column mean temperature vertical distribution of sodium. between the stratopause and 81.8 km.’’ [8] In two earlier papers we showed long-term trends in [5] These various experimental studies suffer from a the centroid height of the sodium layer from 1972 to 1986 variety of deficiencies. For example, many are based on [Clemesha et al., 1992] and 1972 to 1994 [Clemesha et al., comparatively short data series, less than 10 years in the 1997], respectively. In the first paper we found a linear trend À1 cases of some of the satellite and Rayleigh lidar measure- of À49 ± 12 m yr , and in the second this reduced to À37 ± À1 ments. The rocket measurements involve changes in instru- 9myr . In both analyses the trend is statistically signif- mentation that are difficult to allow for accurately, the OH icant, but it should be noted that extending the period rotational temperature measurements are based on small studied from 15 to 23 years resulted in a considerable changes in the relative strength of the components of the decrease in the trend obtained. rotational spectrum, and are thus sensitive to small changes [9] The sodium layer shows significant diurnal and sea- in instrumentation. Nevertheless, a majority of the measure- sonal variations, and these effects must be taken into ments do appear to indicate a negative trend in the temper- account when searching for any long-term trend. The ature of the MLT region. In view of the implications of this average diurnal variation shows a peak-to-peak oscillation fact, it is clearly important to investigate any long-term data of about 2 km in the height of the centroid of the layer series that have bearing on the subject. [Clemesha et al., 1982]. This variation can be allowed for either by appropriately displacing individual height profiles before including them in the long-term average or by 2. Atmospheric Sodium Profile restricting the analysis to a relatively short local time [6] The sodium layer has its origin in the ablation of interval, over which the layer is known not to show a strong meteors in the upper atmosphere, liberating free sodium variation. In our 1992 paper we tried both these methods atoms which subsequently participate in a complex chem- and found them to give almost the same final trend for the istry, both neutral and ion. The net result of this chemistry, centroid height. In subsequent work, including this paper, together with the ablation profile and eddy diffusion trans- we have simply restricted our analysis to the time interval port, is to produce a layer peaking at about 90 km, with 1900–2200 local time. See our earlier papers for details of densities falling to zero at about 105 km on the topside and this aspect of the analysis. 80 km on the bottom. The exact shape of this distribution [10] In the two earlier papers cited above we did not take depends on the vertical distribution of other atmospheric into account the seasonal variation in the height of the constituents, both major and minor.