N. P. Sergeenko and G. V. Givishvili
Institute of Terrestrial Magnetism, Ionosphere, and Radio Wave Propagation, Troitsk, Moscow Region, Russia
Publications that indicate a multiyear decrease of the temperature and total concentration in the upper and middle atmosphere have recently appeared (see, for example, Semenov ). Such atmospheric evolution could have led to climatic ionospheric changes not related to solar and geomagnetic activity changes. Analysis of long series of measurements carried out at some ionospheric observatories of the former USSR [Givishvili and Leshchenko, 1993] and abroad [Bremer, 1992] shows that the F2 layer critical frequencies decreased during about 40 years of observations. Studies of long-term changes in ionospheric disturbance parameters also discovered climatic-scale trends not related to secular or cyclical variations of solar and geomagnetic activity [Sergeenko and Kuleshova, 1994, 1995]. It is supposed that long-term changes in atmospheric parameters (which may be due to both natural and anthropogenic causes) are the reason for the above mentioned electron concentration variations. The aim of this paper is to estimate background changes of the quiet thermosphere which lead to observed changes of ionospheric disturbance properties.
The vertical ionospheric sounding data at the IZMIRAN observatory (Moscow Region) from 1950 to 1990 were used in this work. The disturbance degree of the F2 ionospheric layer is usually described by the relative deviation of the current critical frequency foF2d from the running median value foF2q to get rid of effects related to solar activity variations:
It is believed that at middle latitudes, the F2 layer is disturbed if | dfoF2 | 15%. A catalog of ionospheric storms was composed for the entire observational period, the storms being classified according to the following signs: (1) the intensity and type of the magnetic storm commencement (sudden or gradual), if ionospheric and magnetic disturbances occur simultaneously; (2) the disturbance sign, positive or negative; (3) season; and (4) the type and intensity of the ionospheric storm. Two disturbance classes are the most numerous within this classification:
1. The first is negative single-phase ionospheric disturbances, which occur simultaneously with a magnetic storm with gradual commencement. Since the upper boundary of small and moderate disturbances is limited by the values of | dfoF2 | < 25% and |dfoF2 | < 35%, respectively, a study of the amplitude trend of ionospheric storms should be carried out for high-intensity disturbances. The number of these disturbances during the observational period was ~150. Analysis of long-term changes in characteristics of disturbances of this class has confirmed the presence of multiyear trends in their amplitude and duration. In summer and during equinoxes their amplitude increased by ~10% and the mean duration increased by 1.5 times: from ~24 hours to 36 hours, the standard deviation being 8 hours in both cases [Sergeenko and Kuleshova, 1994].
2. The second is ionospheric disturbances, which are not related to such geoeffective sources on the Sun as flares, coronal holes, and fine filaments. As a rule, these disturbances occur against a quiet geomagnetic background. The number of these disturbances varies from 50 to 100 events per year. Sergeenko and Kuleshova  showed that there is a meaningful trend of this class of ionospheric disturbance during the past 15-20 years, the trend being directed to an increase for negative disturbances and a decrease for positive disturbances.
Figure 1 shows examples (solid curves) of two intense summer storms of class 1, which illustrate the above mentioned multiyear increase of the disturbance amplitude in 1981 as compared with 1957 and the increase of the disturbance duration under identical solar and geophysical conditions (maximum Kp indices are 8.3 and 8.0, respectively). Here the storm duration DT is determined at a level of dfoF2 -15%, as is marked in Figure 1. The maximum disturbance intensity during the main storm phase is characterized by the dfoF2max parameter. According to Figure 1, DT during the indicated 24-year period of observations increased by ~12 hours. Herein dfoF2max increases from -51% to -60%. Note that these two storms demonstrate a mean picture of the long-term changes in ionospheric disturbance properties.
The effects observed are most probably caused by a decrease of the temperature and partial concentrations of neutral gases at altitudes of the F2 layer. To confirm this assumption, let us use analytical relations between the F2 layer critical frequency and the thermospheric parameters at corresponding altitudes. Shubin and Annakuliev  showed that the electron concentration variations in the F2 layer depend on neutral atmosphere composition and temperature in all seasons both in the daytime and at night via the parameter
where b = K1[ N2] + K2[ O2] ; K1 = 8 10-14(T/300)2 ; K2 =10-9T-0.7 ; a = m( O)(X + 1)/[m( N2)X + m( O2)] ; X =K1[ N2]/K2[ O2] ; [O], [N 2 ], and [O 2 ] are the concentrations of atomic oxygen, molecular nitrogen, and molecular oxygen, respectively; m(j) is the mass of the j th component of the gas; and T is the temperature (it is assumed that Ti = Te = Tn in spite of the fact that in the daytime always Te > Tn, Ti).
Shubin and Annakuliev  noted that during disturbances
where Rd and Rq are determined by (1) for corresponding quiet and disturbed conditions. To estimate temporal changes of disturbance parameters, the following empirical relations between the ionospheric storm intensity and the maximum parameters of magnetospheric disturbances ( AE, Dst, and Kp indices) may be used with allowance for the fact that dfoF2 depends on magnetic activity during the previous 3-6 hours:
where t is the current time of the ionospheric storm; DT is its duration; tm is the time of the maximum storm activity counted from its beginning; t = exp(-3/tm) is the characteristic time of the ionospheric reaction to magnetic activity changes; n = 0, 1, and 2 characterizes the dfoF2 delay relative to the Kp index; and A, B, C, m, a, b, and d are the empirical coefficients presented by Zevakina et al.  for various latitudes. It follows from (3) and (4) and from the condition for dfoF2 maximum that the temporal characteristics of ionospheric storms are related to each other:
Equation (2) provides the possibility of obtaining changes in the F2 layer critical frequency during a storm as a function of variations of the exospheric temperature and atmospheric gas composition. Formulae (3) and (4) show the dependence of the tm and DT temporal characteristics of ionospheric disturbances on thermospheric conditions.
To estimate the influence of long-term trends in the quiet thermosphere on the occurrence character and parameters of F2 layer disturbances, here we calculated by using (2) the values of dfoF2 for intense negative storms in different years and seasons using the mass spectrometer/incoherent scatter (MSIS) model [Hedin, 1987]. Figure 1 shows the results of these calculations for the two above mentioned storms (dashed lines). One can see that for the 1957 storm, the calculated results agree well with the experimental behavior of dfoF2. However, for the 1981 storm the calculated curve differs significantly from the empirical one. To correct the calculations, experimental data of Semenov , which show a temperature decrease at F2 layer altitudes with a rate of -3 K per year (this rate for the period considered corresponds to about -50 K), were used. According to the Roble and Dickinson  model calculations, this temperature decrease led to a decrease of the atomic oxygen concentration by ~40% and a decrease of the molecular nitrogen and oxygen concentrations by ~60%. For such atmospheric modification the dfoF2 behavior calculated using (2) and (3) for the 1981 disturbance becomes adequate for the observed picture (dotted curve in Figure 1).
It is known that the occurrence character and intensity of ionospheric storms depend to a high degree on neutral composition changes during ionospheric disturbances and especially on the [O]/[N 2 ] ratio. Therefore (paradoxically) the mentioned effects of the multiyear changes in ionospheric disturbance properties are evidently due to a single cause: to a decrease of atomic oxygen content at F2 layer altitudes in normal conditions. Actually, since the storms taken as an example occurred under similar (solar and magnetospheric) conditions, the higher intensity of the 1981 storm should have been due to the lower [O]/[N 2 ] ratio as compared with the 1957 storm.
At first sight, this contradicts the data of Semenov  and Roble and Dickinson , according to which the background (undisturbed) values of [N 2 ] decreased during recent decades more significantly than the [O] values. However, during such intense storms, such as those discussed in this paper, the neutral gas temperature increases by ~100-400 K as a result of atmospheric heating. According to numerous experimental data, this leads to an increase in [N 2 ] at middle latitudes by 5-10 times under almost unchanged concentration of O [Shchepkin and Klimov, 1980]. In other words, the multiyear decrease of the background content of N 2 is compensated by its many-times increase during a storm, whereas there is no such compensating mechanism for O. For this very reason, the [O] background concentration decrease plays a much more important role in multiyear trends of [O]/[N 2 ] under disturbed conditions than the corresponding decrease of [N 2 ].
Changes in the annual number of ionospheric disturbances, which occur against a quiet geomagnetic background, confirm this conclusion. Actually, one can easily see from (1) and (2) that the observed increase of negative disturbances and simultaneous decrease of positive disturbances are explained by the same decrease of the background concentration of O.
Finally, one more important circumstance should be noted. The Roble and Dickinson  model calculations were performed under the assumption of doubling of the CO 2 and CH 4 greenhouse gas concentration near the Earth's surface. However, according to World Meteorological Organization  data, the average increase of these gas concentrations is ~12% for carbon dioxide from 1958 to 1990 and ~11% for methane from 1978 to 1989. Therefore if the temperature decrease and the proportional changes in O, O 2, and N 2 content were only due to the increase of CO 2 and CH 4 concentrations, the corresponding ionospheric response should have been approximately an order of magnitude weaker. Thus there exist either some additional (not related to CO 2 and CH 4 ) source of disturbances of the thermal regime and composition of the mesosphere and thermosphere gas or mechanisms (still unaccounted for) of the influence of CO 2 and CH 4 content on the middle and upper atmosphere of the Earth.
An analysis of climatic trends in the atmosphere has shown that these trends are an order of magnitude higher in the upper atmosphere than in the lower atmosphere. Taking such changes into account is important not only ecologically aspect but also for applied problems related to satellite lifetimes and various radiophysical system operations. The effects discovered raise a question on prediction of the evolution of these changes in upcoming years and decades. This problem is closely related to the modeling of processes in the near-Earth space environment and requires corrections of existing atmospheric and ionospheric models. Corrections with allowance for the experimental data of recent years should be introduced in empirical models, and changes in lower boundary conditions should be planned in theoretical models.
Bremer, J., Ionospheric trends in mid-latitudes as a possible indicator of the atmospheric greenhouse effect, J. Atmos. Terr. Phys., 54 (11/12), 1505, 1992.
Givishvili, G. V., and L. N. Leshchenko, Long-term trends of the midlatitude ionospheric and thermospheric properties, Dokl. Ros. Akad. Nauk, 333 (1), 86, 1993.
Hedin, A. E., MSIS-86 thermospheric model, J. Geophys. Res., 92 (A5), 4649, 1987.
Roble, R. G., and R. E. Dickinson, How will changes in carbon dioxide and methane modify the mean structure of the mesosphere and thermosphere?, Geophys. Res. Lett., 16 (12), 1441, 1989.
Semenov, A. I., The lower thermosphere temperature regime from emission measurements during the recent decades, Geomagn. Aeron., 36 (5), 90, 1996.
Sergeenko, N. P., and V. P. Kuleshova, Climatic changes in disturbance properties in the ionosphere and upper atmosphere, Dokl. Ros. Akad. Nauk, 338 (4), 534, 1994.
Sergeenko, N. P., and V. P. Kuleshova, Multiyear trends of ionospheric disturbances in the F2 layer, Geomagn. Aeron., 35 (5), 128, 1995.
Shchepkin, L. A., and N. N. Klimov, The Earth's Thermosphere, 220 pp., Nauka, Moscow, 1980.
Shubin, V. N., and S. K. Annakuliev, Model of the ionospheric storm negative phase at middle latitudes, Geomagn. Aeron., 35 (3), 79, 1995.
World Meteorological Organization, The Global Climatic System: Climate System Monitoring, December 1988-May 1991, p. 71, World Clim. Data and Monit. Programme, United Nations Environ. Program, Centrooffset, Siena, Italy, 1992.
Zevakina, R. A., E. M. Zhulina, G. N. Nosova, and N. P. Sergeenko, Guide-Book on Short-Term Ionospheric Predictions, 71 pp., MGK Akad. Nauk SSSR, Moscow, 1990.