Indication of shrinking atmosphere above Tromsù (698N, 198E)

Transcription

1 Atmospheric Science Letters (2001) doi: /asle Indication of shrinking atmosphere above Tromsù (698N, 198E) C. M. Hall 1 and P. S. Cannon 2 1 Tromsù Geophysical Observatory, University of Tromsù, Norway 2 Defence Evaluation and Research Agency, Malvern, U.K. Abstract: At Tromsù (698N; 198E), ionospheric data from 1952 onwards indicate that the mean altitude of the F2 layer is dropping by 4 km per decade on average. This effect is thought to be attributable to middle atmosphere cooling due to increasing concentrations of anthropogenic carbon dioxide and methane. *c 2001 Royal Meteorological Society Keywords: Climate change, ionosphere. 1. INTRODUCTION 1.1 Background The so-called greenhouse gases, primarily carbon dioxide (CO 2 ) and methane (CH 4 ), with sources near the earth's surface emit ef ciently in the infrared. Thus, absorption of short-wave radiation by the atmosphere is partly dissipated as heat by infrared emission. In the troposphere, this infrared radiation is trapped, as in a greenhouse, and thus the equilibrium temperature increases with greenhouse gas concentration. In the middle atmosphere, these greenhouse gases are still present, and heat is similarly dissipated by this long-wave radiation. However, the middle atmosphere is largely outside the ``roof'' of the ``greenhouse'' such that the re-emitted radiation is free to escape into space. In this way, CO 2 and CH 4 effectively remove heat from the middle atmosphereðin fact from any part of the atmosphere above the cloud layersðand cool it. Roble and Dickinson (1989) describe the process in detail. To summarize one thinks of ``greenhouse warming'' in the troposphere, but of ``greenhouse cooling'' in the middle and upper atmosphere. As the atmosphere cools it occupies less space; it shrinks. As a consequence, lower air densities will occur at lower altitudes, and it may be deduced that ionization similarly drops to a lower altitude. The effect on the ionospheric layers is summarized by Rishbeth (1990) and an excellent review containing many references may be found in Rishbeth and Clilverd (1999). This cooling effect, and its consequences for the heights of the ionospheric layers, have been modelled by Rishbeth and Roble (1992). Experimental con rmation of this cooling possibly comes from the increasing frequency of noctilucent clouds (NLC), This effect could be due to increasing water vapour concentration, due to oxidisation of methane, or lowering of the mesopause temperature (Thomas et al., 1989; Thomas, 1996). In such studies it is easy to rule out possible effects due to changes in solar ux, since sunspot number and 10 cm ux (F 10.7 ) measurements exist as long time-series. We are thus faced with the conclusion that the cause is most probably anthropogenic X *c 2001 Royal Meteorological Society

2 There is also observational evidence of this cooling from ionospheric soundings, as proposed by Bremer (1992). At high latitude (67.48N; 26.78E), Ulich and Turunen (1997) have identi ed a systematic decrease in the height of the ionospheric F2 layer using a sequence of observations from At mid-latitude (54.68N; 13.48E), similarly unambiguous results have been reported by Bremer (1998) and, in the southern hemisphere, extensive work has been carried out by Jarvis et al. (1998). Data from a variety of stations including low-latitude ones are given by Upadhyay and Mahajan (1998). The above references are not intended to be exhaustive; nevertheless, many references to the phenomena described here can be found therein. While ionosondes measure only the virtual heights of the atmospheric layers (due to group delay of the radio wave in the underlying ionosphere), the studies mentioned here employ a (frequency-related) parameter combined with models to estimate the true F2 layer height. 1.2 The Tromsù dataset Ionospheric soundings at the Auroral Observatory in Tromsù were rst performed during the Second Polar Year, 1932±1933, by a British expedition under the direction of Sir Edward Appleton, and simultaneously by a German expedition from Berlin. Soundings of the ionosphere on a regular basis in Tromsù (69.78N; 18.98E) commenced in April 1935 as described by Harang (1937) using a copy of the ionosonde described by Appleton et al. (1937). During the rst years, only the critical frequencies were noted and not the echo height, and observations were only made at 09, 11 and 13 UT. Observations were somewhat erratic during the war years 1939±1945, but became regular from 1951 following the acquisition of a NPL (National Physical Laboratory, Slough, U.K.) Mk II ionosonde. Hourly measurements were made by virtue of the more automatic instrument and heights were also recorded. Importantly, for this study, the M(3000)F2 parameter (de ned later) was also determined, from February These results were assembled in regular reports until In 1968 the NPL Mk II instrument was replaced by a faster ionosonde manufactured by Magnetic AB of Sweden. The period from 1980 until 1992, while yielding much data of good quality, was devoid of scaling; nevertheless, the photographic data remain, and the digitally recorded data is also being preserved. To scale the 4000 or so ionograms for inclusion in this study was beyond the current resources, however. The present instrument, yielding suitable data since July 1993, is located at Ramfjordmoen (69.68N; 19.28E), a few kilometres to the east of the site of the original 1935 installation, and is a University of Massachusetts Lowell DPS-4 Digisonde, owned and operated jointly by the Defence Evaluation and Research Agency of the U.K. and the University of Tromsù. Details of this instrument may be found at ulcar.uml.edu/. For the purposes of this study we have selected monthly means of the noon soundings. The reasons for this are twofold. First, it has been necessary to reduce the data quantity for purely practical purposes, and noon soundings are more likely to yield results during winter night at high latitude. Second, in the early years soundings were only performed (manually) up to three times per day, at 10, 12 and 14 LT, and we will attempt to extend the dataset used here to include these early data in a later study. 1.3 Analysis of the Tromsù time-series In order to compare with results of other workers, we have used the analysis paradigm described by Ulich and Turunen (1997). It is based on a time series of hmf2, the semi-empirical estimate of the true height of the F2 maximum, this being determined from M(3000)F2. The latter is the so-called maximum-usable frequency (MUF) factor (as distinct from the MUF itself) pertaining to oblique transmission over a

3 3000 km path with a single re ection from the F2 layer. It is a parameter which has been regularly derived from ionograms since the early years and, therefore, forms a useful time series. hmf2 can be determined using (Bilitza et al., 1979): hmf2 ˆ M 3000 F2 DM 1 where DM is a function of E- and F-region critical frequencies, sunspot number, R, and geomagnetic latitude, y. When DM is zero, eqn (1) is identical to that given by Shimazaki (1955) where the F region height is referred to as h p F2. The sunspot number R may be found and incorporated into hmf2, but it adds a random nature without apparently changing our nal result, we do not show this modi cation of the time series here. Furthermore, over the period in question, ve decades, y is also a variable. If, at a later date, we are able to extend the time series, either by inclusion of new data, or as a result of re-analysis of ionograms prior to 1952, or both, the variation of y might have to be included (we keep y xed here to facilitate comparison with other workers' results). The Shimazaki method tends to overestimate the height of the F-region compared to other M(3000)F2 methods. Despite the lack of sophistication, we felt it simpler at this point to merely use the Shimazaki (1955) expression and avoid over complication. The result is shown in Figure 1. Here the dominant variation is that of the solar cycle and any trend is hard to see. Since the time series contains a non-integral number of solar cycles, and to rule out external effects, solar activity must be removed (Ulich and Turunen, 1997). A linear regression was performed: h p F2 0 ˆ a F 107 b 2 shown by the line in Figure 2. The corresponding effect on H p F2 is shown in Figure 3, and this was then subtracted from the original time series. The result still contains seasonal variations, but, since many occur within the total period of observations, these will not affect the analysis. The result is shown in Figure 4, which includes the least-squares linear t to the revised time-series. The trend was determined to be 4.16 km decade 1 with a standard deviation of 0.11 km decade 1. This trend, visible to the eye from the gure, is undeniable and our result agrees very favourably with that of Ulich and Turunen (1997) for SodankyaÈ, Finland (678N). It is important to note that for both instruments we have used a common technique to convert from M(3000) to hmf2, however, the derivation of M(3000) from the ionograms differs slightly between the two instruments. This may account for the small apparent offset between the two data sets. The automatic scaling, via ARTIST, ( used in the present ionosonde is not thought to introduce any signi cant errors since all of the ionograms have been checked manually. Restricting ourselves to the manually scaled subset of the data yielded a gradient of 1.12 km year with a 2% uncertainty. Although a scaling-method bias can exist, the trend is still irrefutable. We also considered the possibility of differentiating between seasons as Ulich and Turunen (1997) have done. However, examination of the data shows how winter ionograms have often yielded few or no parameters, especially in the early years with a consequential danger of bias. Indeed Ulich and Turunen (1997 report a disagreement with the mid-latitude results of Bremer (1992) (558N) with respect to differing seasonal variations. This discrepancy may be due to data quality since SodankylaÈ is often intersected by the auroral oval (like Tromsù) such that measurements of the F2 layer are sometimes affected by particle precipitation. Also it should be noted that Bremer (1992) averages the entire 24-hour period, in contrast to Ulich and Turunen (1997) (5 hours spanning midday) and this study (midday

4 Figure 1. Estimated true height, h p F2, derived from the monthly average of the local noon M(3000)F2 observations at Tromsù, using the semi-empirical model of Shimazaki (1955). Figure 2. Monthly average local noon h p F2 against monthly mean f10 7 ux ( ) and the linear regression (solid line).

5 Figure 3. The solar ux controlled component of the h p F2 variation corresponding to Figure 1. Figure 4. The residual: the result of subtracting Figure 3 from Figure 1. The variations that remain describe the seasonal variations and the overall trend. Although the seasonal variation is signi cant, the trend is clearly visible: the F2 falls by 4 km per decade, with a 2.5% uncertainty (1-sigma).

6 only). Jarvis et al. (1998) have shown that changes in the thermospheric winds may induce LT-dependent variations in long-term F-region peak altitude trends. A general problem with ionospheric trend studies is that there is no standardised way of how the F2-layer peak height (hmf2) should be obtained from standard ionogram parameters. Furthermore, there is no standard recipe for extracting the long-term trend from the hmf2 time series. Therefore, while general comparison with similar works is interesting, the results of the present paper can be compared directly only with the results of Ulich and Turunen (1997), who employ the very same analysis method for their station, which is only a few 100 km away from the Tromsù site. Indeed, particularly at mid-latitudes the trend is reported as being reversed for certain stations (Bremer, 1998; Upadhyay and Mahajan, 1998). Differential vertical transport of gases between low and high latitude may lead to a latitudinal gradient in the ``fall'' of the F2 layer. Thus, it will be important for future studies to standardise the analysis strategy for each station and attempt to map the phenomenon. 2. CONCLUSIONS At the time of writing, it remains an unresolved problem as to whether effects of the enhanced greenhouse gas concentrations can be observed in F2 layer height (hmf2) time series. This question is re ected in the fact that many of the observed negative hmf2 trends (lowering) are much stronger than predicted by, e.g. Rishbeth (1990). Anthropogenic changes of chemical composition of the atmosphere presumably do affect the ionosphere and they are in principle visible in ionosonde data. However there are other long-term changes affecting hmf2, which at some locations dominate over the ``greenhouse cooling''. The question we have to ask today is what are the possible causes of long-term trends and what are their relative effect, i.e. which one is stronger, which one weaker, and do these relationships depend on time and/or location? An interesting paper published by Mikhailov and Marin (2000) indicates that, e.g. changes related to the geomagnetic eld might be responsible for the long-term trends. We have demonstrated that the height of the ionospheric F2 layer maximum is occurring at progressively lower altitudes: approximately 4 km per decade. While considering the caveats above, it remains likely that this phenomenon is attributed to the climatic cooling of the middle atmosphere, which, in turn is a consequence of increases in methane and carbon dioxide with anthropogenic origins. Ionospheric sounders (ionosondes) may be usefully employed to attempt to monitor such climatic changes in addition to giving information on radio transmission conditions. it is, therefore, of paramount importance that long-term time series of ionospheric parameters are maintained, and from a variety of geographic locations. Acknowledgements The authors wish to thank the respective staffs of the Defence Evaluation and Research Agency and the Tromsù Geophysical Observatory for continued maintenance of the Tromsù ionosonde. The dedicated work of the personnel of the Auroral Observatory in Tromsù dating from 1932 should not be forgotten. Thanks also go to the referees of this paper.

7 REFERENCES Appleton, E. V., Naismith, R. and Ingram, L. J., British radio observations during the international polar year 1932±33. Phil. Trans. A., 236, 191±259. Bilitza, D., Sheikh, N. M. and Eyrig, R., A global model for the height of the f2-peak using M3000 values from the CCIR numerical map. Telecom. J., 46, 549±553. Bremer, J., Ionospheric trends in mid-latitudes as a possible indicator of the atmospheric greenhouse effect. J. Atmos. Terr. Phys., 54, 1505±1511. Bremer, J., Trends in the ionospheric E and F regions over Europe. Ann. Geophys., 16, 986±996. Harang, L., The Auroral Observatory at TromsùÐResults of radio echo observations for the years 1935 to In: Publikasjoner fra det norske Institutt for kosmisk fysikk. Bergen, Norway: A.S. John Greig. 11, 24 pp (English). Jarvis, M. J., Jenkins, B. and Rodgers, G. A., Southern hemisphere observations of a long-term decrease in F-region altitude and thermospheric wind providing possible evidence for global thermospheric cooling. J. Geophys. Res., 103, 20774± Mikhailov, A. V. and Darin, D., Geomagnetic control of the fof2 long-term trends. Ann. Geophys., 18, 653±665. Rishbeth, H., A greenhouse effect in the ionosphere?. Planet. Space Sci., 38, 945±948. Rishbeth, H. and Clilverd, M., Long-term change in the upper atmosphere. Astronomy and Geophysics, 40, 3.26±3.28. Rishbeth, H. and Roble, R. G., Cooling of the upper atmosphere by enhanced greenhouse gasesðmodelling of thermospheric and ionospheric effects. Planet. Space Sci., 40, 1011±1026. Roble, R. G. and Dickinson, R. E., How will changes in carbon-dioxide and methane modify the mean structure of the mesosphere and thermosphere?. Geophys. Res. Lett., 16, 1441±1444. Shimazaki, T., World-wide variations in the height of the maximum electron density of the ionospheric F2 layer. Radio Res. Labs. Japan, 2, 85±97. Thomas, G. E., Global change in the mesosphere-lower thermosphere region: has it already arrived?. J. Atmos. Terr. Phys., 58, 1629±1656. Thomas, G. E., Olivero, J. J., Jensen, E. J., Schroeder, W. and Toon, O. B., Relation between increasing methane and the presence of ice clouds at the mesopause. Nature, 338, 490±492. Ulich, Th. and Turunen, E., Evidence for long-term cooling of the upper atmosphere. Geophys. Res. Lett., 24, 1103±1106. Upadhyay, H. O. and Majahan, K. K., Atmospheric greenhouse effect and ionospheric trends. Geophy. Res. Lett., 25, 3375±3378.