Long-term trends of foE and geomagnetic activity variations

Abstract. A relationship between foE trends and geomagnetic activity long-term variations has been revealed for the first time. By analogy with earlier obtained results on the foF2 trends it is possible to speak about the geomagnetic control of the foE long-term trends as well. Periods of increasing geomagnetic activity correspond to negative foE trends, while these trends are positive for the decreasing phase of geomagnetic activity. This "natural" relationship breaks down around 1970 (on some stations later) when pronounced positive foE trends have appeared on most of the stations considered. The dependence of foE trends on geomagnetic activity can be related with nitric oxide variations at the E-layer heights. The positive foE trends that appeared after the "break down" effect may also be explained by the [NO] decrease which is not related to geomagnetic activity variations. But negative trends or irregular foE variations on some stations for the same time period require some different mechanism. Chemical pollution of the lower thermosphere due to the anthropogenic activity may be responsible for such abnormal foE behavior after the end of the 1960s. Key words. Ionosphere (ionosphere-atmosphere interactions; ionospheric disturbances)


Introduction
Ionospheric parameter long-term trends have been widely discussed during the last decade. This interest is due to possible anthropogenic impact on the Earth's atmosphere, and the ionospheric trends may serve as an indicator of such changes in the upper atmosphere. The interest was greatly stimulated by model calculations by Roble and Dickinson (1989), Rishbeth (1990) and Rishbeth and Roble (1992), who predicted changes in the neutral atmosphere and related ionospheric effects under the increase in the atmosphere greenhouse gas concentrations. Since then, researchers have been trying to reveal and confirm the predicted ionospheric effects related Correspondence to: A. V. Mikhailov (avm71@orc.ru) to the thermosphere cooling (Bremer, 1992(Bremer, , 1998(Bremer, , 2001Givishvili and Leshchenko, 1994;Ulich and Turunen, 1997;Jarvis et al., 1998;Upadhyay and Mahajan, 1998;Sharma et al., 1999). But the greenhouse hypothesis encounters serious problems at least with the F2-region parameter long-term trends Marin, 2000, 2001;Mikhailov et al., 2002;Mikhailov, 2002).
The main efforts have been directed towards the F2-region parameter long-term trends analysis, since these observations are the most abundant and consistent, while trends in the Eregion were considered only by some researchers (Givishvili and Leshchenko, 1993Bremer, 1998Bremer, , 2001Sharma et al., 1999). The result of these analyses is that positive foE trends are the most probable and are presumably due to a decrease in the neutral NO concentration at the E-region heights (Danilov and Smirnova, 1997;Bremer, 1998;Danilov, 2001). The "NO mechanism" is strongly supported by [NO + ]/[O + 2 ] rocket mass-spectrometer observations analyzed by Danilov (1997), Danilov and Smirnova (1997). Since the late 1950s, this ratio demonstrates a pronounced decrease which can be related only to a [NO] decrease at the E-region heights.
The nitric oxide concentration in the E-region is known to be strongly dependent on the geomagnetic activity level (e.g. Titheridge, 1997;Solomon and Barth, 1999;Ridley et al., 1999). Therefore, in the framework of the NO mechanism, in principle, one may expect foE trends of different signs for the periods of rising and falling geomagnetic activity, and this would be a good test for the proposed explanation. Besides this mechanism of natural origin, an anthropogenic impact on the neutral atmosphere cannot be excluded as well. Among such artificial factors are the earlier mentioned greenhouse effect and the atmosphere pollution due to the increasing rate of rocket and satellite launchings (Kozlov and Smirnova, 1999;Adushkin et al., 2000). Unfortunately, earlier proposed methods for the E-region trend analysis cannot be used for such investigations, and a new, more accurate approach should be developed to reveal and separate natural and anthropogenic (if it exists) parts in the foE long-term variations.
The aim of the paper is to develop a method which would be able to remove solar and geomagnetic activity effects from the observed foE long-term variations, to analyze the dependence of foE trends on the phase of geomagnetic activity and to check for any unnatural effects present in the foE trends revealed.

Method description
A general method for the F 2-layer trends analysis was described by Mikhailov et al. (2002), which, with some modifications, is used in the present study. As earlier we proceed from an assumption that natural foE long-term variations are due to solar and long-term geomagnetic activity variations which may be presented by R 12 and 11-year running mean A p indices. The method includes the following steps: 1. Observed mid-latitude monthly median foE values for 10:00, 11:00, 12:00, 13:00, 14:00 LT are reduced to a 12:00 LT moment, to give average noon foE values. The dependence foE ∝ (cos χ o ) p , where χ o -solar zenith angle and p = 0.6 (Muggletone, 1972), is used for this reduction. The use of an average over 5 values increases the reliability of the analyzed noon foE values.

A regression of this noon foE with R 12
foE reg = a 0 + a 1 R α

12
(1) is used to define monthly relative deviations We, as initially it was proposed by Mikhailov (1998, 1999), we analyze relative rather than absolute δfoE deviations normally considered in the ionospheric trend analyses by most of the authors. As far as we know, relative deviations were considered only by Deminov et al. (2000), to analyze foF2 trends. Relative deviations allows us to combine different months and obtain annual mean δfoE which are used in the analysis, with the final method being based on the 11-year running mean δfoE values. A simple arithmetic running mean smoothing with an 11-year gate is applied everywhere .
The optimal 12 different values of α (for each month of the year) are specified to provide the least standard deviation (SD) after a regression (see later) of 11year smoothed δfoE values with Ap 132 (11-year running mean A p indices). The 11-year δfoE smoothing requires all 12 values of α to be available simultaneously at each step of the SD minimization. This implies an application of special multi-regressional methods (Press et al., 1992) to solve the problem considered.
The expression (1) is of a general type and depending on α, it can describe both the linear and nonlinear relationship of foE with R 12 . The regression coefficients a i are specified by the least-squares method for each month and a given α value. It should be stressed that the expression (1) does not provide the best approximation of the observed foE versus R 12 dependence (other dependencies may give less sum of residuals), but it should be considered in terms of the following δfoE 132 regression with Ap 132 to find the minimal SD (see later). Therefore, the regression (1) is not a "model" in the usual sense of this word, as it is accepted in other approaches. This regression is used to remove the solar activity part from the observed foE variations, since a "pure" foE dependence on solar activity (presented by the R 12 index) a priori is not known for each month. The application of this general approach has shown that all 12 values of α turned out to be close to unity, and this strongly simplifies the calculation. It should be noted that monthly α vary in a wide range in the case of the foF2 trends analysis . The reason for α to be close to unity is considered in the Discussion.
3. One-hour gaps in foE within the 10:00-14:00 LT interval are filled by spline interpolation, but large gaps in observations are not filled. If the number of months with available foE values for a given year is less than 6, then the year is marked as "zero". During the 11-year δfoE smoothing, the arithmetic mean is calculated over the non-zero years only. Due to this smoothing, even oneyear gaps do not introduce any visible deviation in the foE 132 variation. But large gaps result in noticeable perturbations in the δfoE 132 , variations and an additional analysis is needed for each station to select the period for analysis. Therefore, only clear enough periods were left for further analysis on each station.
4. The geomagnetic activity effect is removed from the 11-year running mean δfoE variation using a regression with Ap 132 where δfoE 132 and Ap 132 are 11-year running mean values, β is a fitting parameter, n = 0÷−5 is a time shift in years of Ap 132 with respect to δfoE 132 variations, both parameters being specified to give the least SD for the residuals after Eq. (3). The regression coefficients b i are found by the least squares-method.
5. Our previous analysis  has shown that the best results (the least SD) can be obtained if an additional smoothing is applied to δfoE 132 and Ap 132 variations. Such smoothing is made by a 5-order polynomial approximation of these parameter variations.
6. The residual linear trend with the slope K r (in 10 −4 per year) may be estimated over the residuals after the regression (3). 7. The test of significance for the linear trend parameter K r (the slope) is made with Fisher's F criterion (Pollard, 1977) where r is the correlation coefficient and N is the number of pairs considered. Keeping in mind that we work with smoothed variations, we put the number of degrees of freedom (N − 2) = 4 (the 5th order polynomial is defined by 6 coefficients).  , Tomsk (1945Tomsk ( -1997, St. Petersburg (1946Petersburg ( -1999, and Juliusruh  were used in these calculations ( Fig. 1, middle panel). The 11-year smoothing applied to δfoE reduces the available periods by 10 years as shown in Fig. 1. Negative foE trends are seen to take place before 1953-1955 for the period of increasing ge-  -1966-1977 omagnetic activity (Fig. 1, top), while they are positive until 1965, in accordance with the decrease in geomagnetic activity. Close δfoE 132 variations are seen to take place for the stations until 1965, while this coherence breaks down after the end of the 1960s. A tendency to switch to a negative foE trend is clearly seen for Tomsk and Juliusruh after 1965, in accordance with a positive phase in the Ap 132 variation, but something overpowers this tendency, making the trends positive. The δfoE 132 variations are seen to be different at different stations after 1970. Therefore, the "natural" type of foE 132 dependence on geomagnetic activity seems to break down after the end of the 1960s. Figure 2 shows the foE 132 versus Ap 132 dependence in a explicit way for Tomsk, Rome, and Khabarovsk. The inverse relationship takes place between these parameters before 1970 (Rome-1972), but it switches abruptly to a direct one around 1970. The selected stations demonstrate the most pronounced cases of such a changeover in the type of dependence. The results of such analysis for 24 mid-latitude ionosonde stations are given in Table 1. Unfortunately, observations on many stations start only after 1964-65 (1969-70 after 11-year smoothing); therefore, it is only possible to check the sign of this dependence for the years available. For the long observing stations only the discussed period is considered in Table 1. Table 1 shows that most of the stations demonstrate a direct δfoE 132 relationship with Ap 132 after 1968-1972, but there are stations (the end of Table 1) which exhibit the inverse type of this dependence for the same period. Variations of δfoE 132 for these stations are given in Fig. 1 (bottom). Although foE observations are absent for earlier years on these stations, they clearly show the "natural" type of behavior, that is the inverse δfoE 132 versus Ap 132 dependence. Ashkhabad and Irkutsk present the best cases with positive foE trends before 1965 and negative ones afterwards, in accordance with the Ap 132 long-term variation (Fig. 1, top). At Ashkhabad this "natural" dependence breaks down after 1976. This analysis shows that not all stations were subjected to that breakdown of the "natural" foE behavior after the end of the 1960s. The reason(s) for this is not clear now, and further analysis is needed to explain this interesting result. An important conclusion of this analysis is that foE trends similar to foF2 trends are subjected to geomagnetic control (Mikhailov and Marin, 2000;Mikhailov, 2002), that is the sign of the foE trend depends on the phase of geomagnetic activity long-term variations (Fig. 1, top).

Complementary and residual foE trends
The observed δfoE 132 long-term variations show a clear dependence on geomagnetic activity at least for the period before 1970-1972, while some stations demonstrate this dependence for later years as well (Fig. 1). Therefore, we may try to remove this "natural" dependence on geomagnetic activity and analyze the residual foE trend. Figure 3 shows the δfoE 132 versus Ap 132 dependence for Slough and Ashkhabad, where the "natural" type of foE behavior takes place for different years. Two branches are seen in this dependence: on Slough -before and after 1955, on Ashkhabad -before 1965 and after 1967 (Fig. 3, left-hand panels). The inverse type of δfoE 132 versus Ap 132 dependence takes place for both branches, but the curves are shifted. It seems as if the "efficiency" of geomagnetic activity has been increasing with time as the same δfoE 132 values correspond to lower Ap 132 after the end of the 1950s on Slough and after 1967 on Ashkhabad. The same effect takes place at Tomsk (Fig. 2  for early ages), Moscow, and St. Petersburg, not shown in the plot. It should be stressed that Ashkhabad, which was not subjected to the 1970-1972 "break down" effect, demonstrates the same type of δfoF2 132 versus Ap 132 dependence and for later years as well. (Fig. 3, left-hand bottom panel).
The ambiguity in this dependence can be removed to a great extent by adding a complementary positive linear trend K c to the δfoE 132 variations. This approach was used earlier for the foF2 long-term trend analysis  where the same situation takes place. By a complete analogy with the foF2 trend results, the least SD (the best fitting) is obtained if the complementary trend is applied for the whole analyzed period, starting from the first year. The optimum complementary trends K c = +5.13 × 10 −4 per year for Slough and K c = +11.8 × 10 −4 per year for Ashkhabad, being added to δfoE 132 variations squeeze the loops in the δfoE 132 versus Ap 132 dependence practically to one curve (Fig. 3, right-hand panels), while the curves after 1967 for Slough and after 1975 for Ashkhabad exhibit quite different types of variation. Figure 4 shows observed (smoothed) and calculated δfoE 132 variations, as well as their difference for Slough and Ashkhadad. A good quality of model fitting practically results in zero residual trends on Slough (K r = −0.22× −4 per year for the period 1936-1967) and on Ashkhadad (K r = 0.04 × 10 −4 per year over [1962][1963][1964][1965][1966][1967][1968][1969][1970][1971][1972][1973][1974][1975], the trends being insignificant in both cases. This means that the "natural" δfoE 132 dependence on geomagnetic activity can be efficiently removed and there is no residual foE trend left. On the other hand, strong positive foE trends are seen for the two stations after the years mentioned.

Discussion
The proposed approach to the foE trend analysis has shown a close relationship between δfoE 132 and Ap 132 long-term variations. The "natural" type of this dependence is the inverse one, that is positive foE trends correspond to decreasing geomagnetic activity and vice verse. This is a new result which was not mentioned in earlier publications on the foE trends. However, it should be mentioned that Givishvili and Leshchenko (1996), using quite different methods were the first to reveal the foE trends of different signs before and after the end of the 1950s without any explanation of this effect. Due to this dependence on the phase of geomagnetic activity, one should be careful with the selection of the periods for trend analysis and not put together years belonging to different (rising/falling) periods of geomagnetic activity. Unfortunately, this is not taken into account in other publications devoted to the ionospheric trends and this (as one of the reasons) results in a chaos of various signs and magnitudes of the trends at various stations.
Our approach allows us to remove to a great extent solar and geomagnetic activity effects from foE long-term variations and to show that the residual foE trends are close to zero for the years before the "break down" effect occurs in the δfoE 132 versus Ap 132 dependence (see earlier). From a physical point of view, the obtained result is interesting, telling us that practically all observed foE long-term variations may be attributed to the variations in solar and geomagnetic activity -that is they are of a natural origin. An additional (presumably of a manmade origin) effect is also clearly seen in the δfoE 132 variations after the end of the 1960s on most of the stations and for later years on some other stations. The obtained results are mainly due to the efficiency of the method applied and some comments are required in this relation. The first one concerns the procedure of the solar activity effect removal. According to our general approach to the trend analysis , first, we used a general type of foE relationship with R 12 (see Eq. 1), to remove the solar activity part from foE long-term variations. But unlike our previous results on the foF2 trend analysis , where monthly α values varied in a wide range 1.8 < α < 3, in the case of foE, the α parameter turned out to be close to unity and this strongly simplified the calculations. This result may be explained as follows. As the mid-latitude daytime E-layer is produced via the ionization of neutral O 2 by two close EUV lines λ = 977Å (CIII) and λ = 1025.7Å (HLyβ), the classical Chapman theory (Chapman, 1931) may be applied in this case with a sufficient accuracy (Ivanov-Kholodny and Nusinov, 1979). The ionization rate in the E-layer maximum may be written as where I ∞ -the incident ionizing flux, H -scale height of neutral O 2 , χ o -solar zenith angle, σ i,a -ionization and ab-sorption cross sections. Critical frequency, foE is proportional to where α eff is the effective dissociative recombination coefficient for the molecular ions NO + and O + 2 . The intensity of solar EUV emission is proportional to F p 10.7 , where p = 1 ÷ 0.67 (Nusinov, 1984(Nusinov, , 1992Tobiska et al., 2000, and references therein). Our method gave a linear foE relationship with R 12 . This is also valid for F 10.7 , as annual mean F 10.7 and R 12 indices are known to be highly correlated (the correlation coefficient is 0.991 being significant at the 99% confidence level). This means that under the root sign we have to have an expression depending on solar activity as F 4 10.7 , or the H α eff product should be proportional to F −3 10.7 , at least. The only possibility of obtaining such a strong dependence on solar activity is to take into account a strong dependence of T e /T n on F 10.7 , revealed by rocket probe measurements in the E-region at the 110 km height (Duhau and Azpiazu, 1985;also Oyama et al., 2000). We used this idea earlier to explain seasonal variations in the E-region . The dependence T e /T n versus F 10.7 (Fig. 2 in Duhau and Azpiazu, 1985) is well approximated by cubic polynomial on F 10.7 . Keeping in mind Fig. 4. Polynomial approximated observed, calculated δfoE 132 and their difference, resulting in practically zero residual foE trends on Slough and Ashkhabad for the period before the "break down" effect occurs. Calculations were performed with the complementary trends given on the plots. that α eff is proportional to T −0.8 e , this practically gives the required dependence on F 10.7 under the root sign in (5), while T n is highly compensated in the H α eff product as H ∝ T n .
The other problem concerns the removal of geomagnetic activity effects from the foE long-term variations. This problem was discussed by Mikhailov et al. (2002) with respect to foF2 trends. Here we used the same approach with the only difference -the dependence on β in the regression (3) is of a general type as a priori we do not have any information on this dependence. Similar to F2-layer trends, a time shift n (in years) of Ap 132 with respect to δfoE 132 variations is required to obtain the least SD for the residuals after the regression (3). In our case the average time shift is 1 year. As in the F2-layer case, no physical explanation can be proposed now for this delayed thermosphere reaction to the geomagnetic activity variations, and a special consideration is required to understand this result. Such time delay implies the whole Earth's atmosphere to be involved with the processes provoked by geomagnetic activity. Changes in the global atmospheric circulation and/or in eddy diffusion and related variations in the thermospheric neutral composition and temperature is the most probable mechanism. On the other hand, one should keep in mind that Ap 132 maybe is not the most adequate proxy for solar wind influence on long-term trend studies. Unfortunately, A p , as well as R 12 , are the only indices which have been observed long enough to be used for the long-term trend analyses.
The revealed relationship between foE trends and geomagnetic activity may be explained by the variations of nitric oxide, NO, whose concentration in the E-region is strongly dependent on the geomagnetic activity level (e.g. Titheridge, 1997;Solomon and Barth, 1999;Ridley et al., 1999 After this, the standard scheme (e.g. Danilov, 1994) of chemical processes may be used. The total ion concentration (n e = n i ) in the E-region is presented by molecular ions O + 2 and NO + which disappear via the reactions of dissociative recombination NO + + e → N + O α = 4.5 × 10 −7 (300/T e ) 0.83 The key point of the "NO mechanism" is the difference in the reaction rate constants of these two reactions. Under increasing geomagnetic activity (before 1955 and after 1967, Fig. 1, top), resulting in the neutral [NO] increase, the share of NO + ions increases via the reaction (6). Since the recombination rate for NO + ions is higher (Eqs. 7 and 8), the electron concentration decreases and we have negative trends in foE (Fig. 1, middle and bottom panels). During the decreasing phase of geomagnetic activity (1955)(1956)(1957)(1958)(1959)(1960)(1961)(1962)(1963)(1964)(1965)(1966)(1967), the situation inverses and we have positive trends in foE (Fig. 1). This mechanism explains the "natural" relationship of foE trends with geomagnetic activity.
An inclusion of a complementary linear trend to our analysis restores the unambiguity in the δfoE 132 versus Ap 132 dependence (Fig. 3) and practically results in zero residual trends (Fig. 4) at least for the years when a "natural" relationship between δfoE 132 and Ap 132 is valid. By a complete analogy with the results of the foF2 trend analysis , the only plausible explanation (as it is seen from now) of the complementary trend is a compensation of a negative foE trend, which is due to a very long-term increase in geomagnetic activity during the 20th century. This increase is seen even for the analyzed period (Fig. 1, top) where a positive trend with K = 0.02 per year exists in the observed Ap 132 variation. Figure 1 (top) is only a fragment of the general picture showing the increase in geomagnetic activity in the course of the 20th century (e.g. Clilverd et al., 1998). This long-term effect in geomagnetic activity cannot be removed using conventional indices and smoother Ap 132 indices are required for its description. But this is a very delicate question which requires a special consideration and is not discussed here. The same NO mechanism can be used to explain the negative background foE trend related to this very long-term increase in geomagnetic activity.
But around 1970±2 (on some stations later), this "natural" inverse relationship of foE trends with geomagnetic activity has broken down. A well-pronounced positive foE trend not related to geomagnetic activity variations has appeared on most of the stations analyzed ( Fig. 2 and Table 1). In some cases this hardly can be considered as a trend -just a positive δfoE 132 upsurge lasting for some years followed by a drop in the δfoE 132 variation (Table 1, Fig. 1). But in many cases, this is a prolonged positive effect. The only plausible possibility is to relate this effect with the anthropogenic impact on the upper atmosphere. Among such mechanisms may be considered the increasing rate of rocket and satellite launchings which leads to the thermosphere chemical pollution (Kozlov and Smirnova, 1999;Adushkin et al., 2000) and the greenhouse effect mentioned in the Introduction.
Since the E-region formation mechanism is relatively simple, some explanations may be proposed for this positive effect in the δfoE 132 variations. A decrease in the molecular oxygen neutral scale height H will increase the production rate and foE, correspondingly (see Eqs. 4 and 5). The scale height H may be decreased by an intensified downward air motion at the E-region heights, as model calculations show (Mikhailov, 1983). Another possible channel of the foE increase is via a decrease in the effective dissociative recombination coefficient α eff (Eq. 5) due to a decrease in [NO] at the E-region heights. The decrease in α eff in this case is due to the [NO + ]/[O + 2 ] ratio decrease, as explained earlier. This mechanism of positive foE trends was considered by Danilov (2001), who has found a strong confirmation of it in both the [NO + ]/[O + 2 ] ratio trend at 120 km and in the positive trend of electron concentration in the D-region, where the ionization of NO plays the dominate role in the total ionization rate. Both results may be explained by the vertical transfer of NO from the E-to D-region due to intensified downward air motion or eddy diffusion. The latter was obtained by Kalgin (1998), analyzing rocket mass-spectrometer data on the [Ar]/[N 2 ] ratio at the E-region heights for the 1966-1991 period. But vertical air motion seems to be more preferable, since it decreases the neutral scale height H and transfers NO downward, thus, increasing the positive effect in foE.
Such explanations come from normal processes taking place in the lower thermosphere. But chemical pollution due to rocket launching may result in quite a different scheme of chemical processes, since the rocket fuel comprises exotic for the upper atmosphere components. Long-living "holes" in the electron concentration is a well-known effect accompanying heavy rocket launchings (Adushkin et al., 2000). Such chemical pollution looks as a very probable mechanism to explain the break down effect in the natural δfoE 132 versus Ap 132 dependence around 1970-1972. Figure 5 gives the number of rocket launchings for the 1957-1999 period. Data may be found in Aviation Week and Space Technology for 1957-1991, Rocket-Space Technique GONTI-1 for 1975-2000 (in Russian), (Adushkin et al., 2000). Maximum occurrence of rocket launchings is seen to take place in the second half of the 1960s. Keeping in mind that some time is necessary for the accumulation of the effect, we obtain the discussed "break down" time around 1970.
With regard to the greenhouse hypothesis widely used in attempts to explain the ionospheric trends, the following may be noted. There are serious problems with this hypothesis in the F2-region (Mikhailov and Marin, 2001;Mikhailov 2002), since it cannot be reconciled with the observed F2-layer parameter trends. Positive foE trends in the E-region observed for some periods seem to be in qualitative agreement with this hypothesis, but the observed trends already are much larger than predicted (Bremer, 2001), although we are still very far from the doubling of greenhouse gases in the atmosphere (Keeling et al., 1995;Houghton et al., 1996). On the other hand, positive foE trends, usually related with the worldwide greenhouse effect, in fact does not take place at all stations -the sign of trends may be different ( Fig. 1 and Table 1). One can say about the spotty global pattern with an unsystematic foE behavior at different stations after 1970 (cf. Slough and Juliusruh in Fig. 1, middle panel). It is only possible to conclude that since the beginning of the 1970s, there has appeared an additional factor in the lower thermosphere which has broken down the normal foE dependence on geomagnetic activity on a long-term time scale. Chemical pollution of the upper atmosphere due to the rocket launchings and perhaps the greenhouse effect look like the most probable reasons.

Conclusions
The main results of our analysis may be summarized as follows: 1. Using a newly proposed approach to the foE trend analysis, it was shown for the first time the relationship between foE trends and geomagnetic activity long-term variations. By a complete analogy with the earlier obtained results on the foF2 trends, the periods of increasing geomagnetic activity correspond to negative foE trends, while these trends are positive for the decreasing phase in geomagnetic activity. Therefore, it is possible to speak about the geomagnetic control of the foE long-term trends as well.
2. Similar to the foF2 trends, there exists a background negative foE trend which may be considered as a manifestation of a very long-term geomagnetic activity increase which took place during the 20th century (e.g. Clilverd et al., 1998). This effect is seen in the δfoE 132 versus Ap 132 dependence before the breaking down of "natural" dependence around 1970 (on some stations later). After removal of this background effect the residual foE trends are close to zero and insignificant. This means that observed "natural" foE long-term variations (trends) have a natural origin and may be attributed to solar and geomagnetic activity long-term variations. 4. Positive foE trends (or more or less prolonged positive upsurges) appearing after the "break down" effect around 1970 may be related with the [NO] decrease at the E-region heights due to the intensification of the downward air motion (Danilov, 2001). But negative trends or irregular foE variations also take place on some stations for the same time period, and this tells us about some other additional mechanism. Chemical pollution of the lower thermosphere due to the increasing rate of the rocket launchings and/or the greenhouse effect may be responsible for such abnormal foE behavior since the beginning of the 1970s.