Severe magnetic storm effects in the ionosphere over Istanbul : a case study

The present study concentrates on the effects on the ionosphere of an individual severe magnetic Storm of the Sudden Commencement (SSC) type, with SSC taking place in the daytime hours. The storm started on 29 October 1968 and went on to 2 November 1968 with geomagnetic 3-hourly magnetic activity index reaching values of Kp ≥ 7. Interplanetary magnetic field polarities included A (field polarity Away from the solar wind) positive and T (field polarity Towards the solar wind) negative polarities. In these conditions, the local response of true height of Fregion (hF) ionization fails nonlinearly from fixed plasma densities. The interplanetary Bz, the magnetic field Hcomponent and Kp were examined for the 5 days following the sudden commencement. Due to intensive geomagnetic SC effects, the lower and upper limits of the F-regions were unbalanced because of gravity relaxation and solar wind pressure effects, until normal levels were restored. In the F-layer 70% deviations of critical frequencies ( f0 F) from median values, as well as hF level variations reaching hundreds of kilometers, were observed and were responsible for the destruction of communication channels. With a plasmapause location value L = 1.6 and with Kp ≥ 7 the protonosphere reservoir should take 1.1 days for its replenishment; one day was in fact insufficient for a full plasma recovery.


Introduction
A Geomagnetic Storm (GS) is the result of strong enhancement of Ring Currents (RC) in which the Earth's Magnetic Field (EMF) is usually depressed below its normal quiet day value.An Ionospheric Storm (IS) is the ionospheric effect of a GS that increases or decreases the electron density (N e ) and the Total Electron Content (N TEC ) of the F-region; please refer to Förster and Jakowski (2000) and references therein, for a review on these phenomena.There are two kinds of ISs known as negative ISs and Mailing address: Associate Prof. Dr. Harutyun Agopyan, Istanbul University, Vocational School of High Technology, 34320, Avcilar, Istanbul, Turkey.Former address: I.C.T.P., 34100 Trieste, Italy.e-mail: harutyun@istanbul.edu.trKey words geomagnetic activity − ionospheric positive storm − neutral gas concentration excess − radio reception − ring current positive ISs, depending on the Local Time (LT) of onset of GS, latitude and season.The response of critical frequency of the F-region ( f 0 F 2 ) depends markedly on season and solar and geomagnetic activity, and also on geomagnetic latitudes (Matsushita, 1959).Most of the winter storms show enhancements, whereas summer storms and equinox storms show depressions following short-lived enhancements.The Interplanetary Magnetic Field (IMF) during spring has an A (field polarity Away from the solar wind), anti-sunward polarity, inducing + B z values, and T (field polarity Toward the solar wind) during autumn it has a sunward polarity, inducing -B z values.Thus, inverse B z signs are most common for October storms.Solar flare-induced storms are weak but last a few days longer, because of the time taken for the stream to pass over the Earth.Several years ago, Dungey (1961) and Akasofu and Chapman (1963) claimed that the energy transfer mechanism from the solar wind to the magnetosphere is a magnetic reconnection between the southward IMF -B z and the Earth's dipole EMF.The non-linearity of the magnetosphere and ionosphere both depend on the RC energy injection rate (Akasofu, 1981).Mendillo and Klobuchar (1975) presented TEC storm behavior after the important discovery of a trough producing a negative phase that begins progressively later at sites with lower latitudes.Prölss (1987) investigated the temperature response of the upper thermosphere, and gave quantitative results on composition changes.He also published a thermospheric model but suggested that further studies are required for negative ISs at middle latitudes.Models differ significantly near 200 km where the nitrogen (N 2 ) density is important and oxygen (O) is depleted in the summer.Prediction of atomic O density in new solar cycles will require new satellite missions and revival satellite drag analyses, or improved methods of monitoring from the ground (Hedin, 1988).Rees et al. (1988) found that, at fixed pressure levels, the increase in mean molecular weight spread only a short distance from the auroral oval can be taken as evidence that negative ISs at lower latitudes cannot be due to changes in the neutral gas composition.The N 2 /O density ratio is important together with that of other minor constituents assuming that the pressure scale height is approximately constant above 280 km.Hewish and Duffet-Smith (1987) suggested a means of giving warning of impending high speed solar wind streams at 1AU.This would allow predictions of time intervals where there will be higher probabilities of MS occurrence (Tsurtani et al., 1988).
The present study aims to contribute to satellite communications and as well as to ionospheric model studies for especially high geomagnetic activity conditions.

Observation
Observations of the hourly behaviour of the critical frequency of the F 2 layer of the ionosphere have been analyzed for their percentage deviations from their monthly median values The above measure of ∆f 0 F 2 , peak height values of h p F 2 ,minimum height values of h min F and virtual height values of h′F for Istanbul (geographic coordinates 41°N; 29°E; geomagnetic coordinates 39°N;108°E;Mc Ilwain (1961) Earth's magnetic shell parameter L ≅ 1.6) have been analyzed to reveal the behavior of the ionosphere during strong GS disturbances.Observations were made during the high solar activity period from 29 October 1968 to 2 November 1968.During this interval, D(γ) ≥ 120 and quite high magnetic activity indices of K p ≥ 7 were observed extensively (Uyar, 1964;Bulat and Agopyan, 1980;Agopyan, 1986Agopyan, , 1988Agopyan, and 1996)).
Figure 1a shows the percentage deviation of f 0 F 2 (MHz) during the severe Magnetic Storm (MS) period of 5 days.It is seen that the f 0 F 2 values always show clear increases from the monthly medians of about 50% to 70% 8 or 10 h after the beginning of a SSC.However, the wireless Short Wave (SW) prediction 10 MHz for f 0 F 2 goes to 17 MHz during a severe GS.So it is advisable to tune to the higher frequencies.In other words, it would be better to lower the wavelength for SW radio reception (i.e.instead of noisy reception on the 31 m band, it would be better to tune to the 19 m band).In fig.1b, hourly values of h p F 2 and h min F are plotted.This shows the ionospheric plasma situation more clearly during a strong GS series (29/30/31 October -1/2 November 1968).During the first day of the storm, the height changes are over 250 km.It can be seen that no changes in height occur during the sunlit hours but they are significant during night hours.Nocturnal h p F 2 and h min F crossing on November 1 electric current increases and cross generations of EMF present ionospheric G condition existence respectively (Agopyan, 1988).
In fig.2, maximum electron density (N m F 2 ⋅ 10 5 e cm -3 ), frequency (f 0 F 2 ⋅ 1 MHz), monthly median (N MED F 2 ⋅ 10 5 e cm -3 ) and K p ranges are plotted against time for 28 October 1968 from 00 h 00 LT to 24 h 00 LT.The following days containing missing values have been omitted.The electron-density versus time curves are not suitable for studying the electron concentration changes at various heights because of strong day/ night differences during the storm.Therefore fig. 3 shows isoionic contour plots of electron densities derived from true height profiles.Here fixed plasma density variations show a decreasing amplitude of plasma wavelike behaviour as isoionic counter plots for the storm that started on 29 October, 1968 and continued for 5 days.
The increase in height for various electron density values during the night and their nonlinear behaviour are not only the result of Coronal Mass Ejections (CME) of the Sun but also partly due to the effects of tides, wind pressures and gravity relaxation occurred progressively.From the fig.3, most of the height changes of density values in between 1.5-6.5 ⋅ 10 5 e/cm 3 occur near 20 h 00 LT.The variation is from 500 km to nearly 700 km at the topside window of the ionosphere (Agopyan, 1988).Here, the electron concentration of O + ions is too low for a reliable determination of the height.The most convenient time of day seems to be in the evening at 17 h 00 LT, when height variations of no more than 50 km are seen (the height remains between 250 and 300 km).
The data have been subjected to spectral analysis.In fig.4, the power spectra including Hanning windowing (Claerbout, 1985 andPress et al., 1988) show the effects of planetary waves having about 12 hourly period.Lunar and solar hourly periods with periods around 5 h, 18 h, and 64 h.Furthermore, the power spectral density including Hanning windowing shows that the similar results are present during the storm period as is seen in fig. 5.Besides solar tide that can affect the acceleration of gravity over 280 km where pressure scale height becomes important on N 2 /O density ratio equilibrium not only plasma waves, minor constituents but also lunar and tidal local effects should be taken into consideration for further investigations.
In fig.6, K p , IMF (Bz) and the H-component of the magnetic field are plotted for the 5 days of the SC storms (Agopyan, 1986) observed worldwide from 29 October 1968 to 2 November 1968, including A, Antisunward positive and T, Toward the Sun negative polarities.If the IMF sector boundary crossings are effective with their A or T polarities, thus the variation of f 0 F 2 during T→A and A→T transitions could be explained by the solar wind speed (fig. 2 and fig.9; Bremer, 1988).There is a positive correlation between amplitudes and phases of sunspot cycle, K-index and EMF, but, particularly in equinoctial months, the first harmonic of the H-component varies irregularly with sunspot number.The seasonal variation of amplitudes and solar cycle variation of amplitude harmonics (smoothed) have been investigated for disturbed days together with the yearly variation and sunspot number and K-index, and also for the H-component (Is ¸ıkara, 1971).The K-index since 1952 from the Kandilli   observatory shows a statistically significant relation with the amplitudes of the solar and lunar variations of the EMF.All components of the EMF and K-index show an increase in amplitude generally with increasing sunspot number as N m F 2 , f 0 F 2 and subpeak electron content (Agopyan and Bulat, 1990).

Discussion of results and conclusions
The results presented in the graphs may be summarized as follows: a) The F-layer starts to rise up in a rather normal way near sunset.Coincident with this rising of the layer where there are sudden increases in ionization, which reaches a value about 60% to 70% larger than its normal level, for the three SSCs.
c) The negative phase for the third storm is very marked and coincides with a possible expansion of the F-layer.
Possible dominant mechanisms for the increase in ionization concentration about 8-10 h after the SSC could be: i) intensification of the neutral wind circulation equatorward bringing up ionization to the regions of lower recombination processes; ii) ionospheric feeding by plasma, flowing from the plasmasphere towards the equator due to its compression.This compression could be due to the E × B ions drifting from polar (higher) to equatorial (lower) L-shells causing further motion along magnetic flux tubes.The saturation of the flux tubes is reached after 8-10 h.
The reason for the rapid decrease following the peak of concentration could be due to: i) a sudden change in the structure of the electric field that causes downward drift to the F-layer as a whole to regions of higher recombination processes; ii) a sudden change in direction of the neutral winds.
The source of heating of the thermosphere at lower latitudes is more difficult to understand.It appears that the F-layers oscillate afterwards, as illustrated by its wavelike structure.The return to higher values of F-layer heights could be due to the following: i) the sudden vanishing of the special structure of the electric field; ii) the returning of neutral wind to normal conditions.
Nevertheless, the positive phase cannot be due to the increase of ionizing radiation from the Sun because in that case we would observe a positive peak rather closer to the SC of the storm.From the aeronomic point of view, in general, Istanbul (L = 1.6) experiences equatorial storm characteristics because of its location in the same longitudinal network as Athens (L = 1.4),Rome (L = 1.6) and Cape Kennedy (L = 1.8) and it also shows phase similarities, parallel morphologic features, and characteristic Mendillo Peak.For the geomagnetic latitude of Istanbul, with a plasmapause location value L = 1.6 for K p ≥ 7, the saturation of the protonospheric reservoir would be expected to take t s = 0.17 (L) 4 ≅ 1.1 days (Kamide and Richmond, 1986).Thus one day is insufficient for its replenishment.
The case is not similar for stations having similar coordinates but it is for equivalent L values.If the F 2 layer was completely solarcontrolled, the LT contour maps of f 0 F 2 would follow lines of geographic latitudes (Fox and McNamara, 1988).It should be kept in mind that when the Sun is fast, with a lead-time of about a quarter of an hour (i.e. the equation of time is positive), there is an extreme energy injection into the ionosphere by way of the magnetospheric plasma and vice versa for the high solar geomagnetic activity conditions.During the times with − B z solar wind, the magnetosphere energy transfer function derived by Perreault and Akasofu (Akasofu, 1981) is dominant near 180°d ue to the term sin 4 ν/2.Here ν is the polar angle of the IMF vector in the y-z plane of the solarmagnetosphere coordinate system.It is known that ε = vB 2 I 0 sin 4 ν/2 where v is the velocity of the solar wind, B is the magnitude of the IMF and I 0 = 7 Earth radii.The variation of the ε function gives evidence that the energy transfer from the solar wind into the magnetosphere is higher during IMF T pro sectors than during A anti sectors.After a CME only the variation of the solar wind speed determines the mean energy input, but the increase during the anti → pro sector transition is steeper than the decrease during the transition from pro → anti sector.Plasma-sheet distance depends on cross-tail potential drop, Φ T , or cross-polar cap potential, Φ cp , and Φ T in turn depends on ε via the relation with the correlation coefficient ∝ 0.92 where the power (ε) of solar windmagnetosphere dynamo changes as a function of time (Reiff et al., 1977).The sources of temperature increases and their physical explanations give the most reliable information.But magnetic activity variations still remain a critical area needing improvement for all models (Hedin, 1988).

Fig. 1 a
Fig. 1 a,b.a) Percentage critical frequency deviations from monthly medians [∆f 0 F 2 (MHz)] versus consecutive local time (LT = UT + 2) hours with storm sudden commencement (SSC) occurrences for 29 October to 2 November 1968; b) minimum to maximum height variation of the ionospheric F-layer versus local time (LT) hours.

Fig. 3 .
Fig. 3. Isoionic contours derived from N(h) analysis of ionograms for between 29 October to 2 November 1968 show true height variations h (km) of whole storm series throughout the 5 K p ≥ 7 days showing disturbed wavelike behaviour of plasma densities.

Fig. 4 .
Fig. 4. Power spectra of the electron density value of the ionospheric storm including Hanning windowing.

Fig. 5 .
Fig. 5. Power spectral density of the electron density value of the ionospheric storm including Hanning windowing.

Fig. 6 .
Fig. 6.Comparison of K p , IMF (B z ) and H-component magnetic field variations for the 5 days of the SC storms observed on 29, 30, 31 October -1, 2 November 1968, including A (field polarity Away from the solar wind) positive and T (field polarity Toward the solar wind) negative polarities.