4. Results of Observations and Modeling: Data Analysis

[10]  Traditionally, considerable deviation ( ge 20% ) of the F2 -layer critical frequencies from the median values over a long period of time (from half a day to 2-3 days) usually accompanying a magnetic storm is called an ionospheric storm. It is known [see, e.g., Buonsanto, 1999; Prölss, 1995] that the ionospheric and thermospheric effects of a storm are closely coupled. In this section we present a brief description of ionospheric and thermospheric disturbances over Kharkov accompanying three magnetic storms with a different character.

2005GI000125-fig01
Figure 1

4.1. The magnetic storm on 25 September 1998

[11]  The magnetic storm on 25 September 1998 was associated with the M6/3B solar flare that occurred during 0644-1009 UT on 23 September 1998. The parameters of heliogeophysical situation are shown in Figure 1 ( For this and other figures the dates are shown at the horizontal axis, fluxes of protons are taken with energies greater than 10, 50, and 100 MeV and of electrons with energies greater than 2 MeV.) (http://sec.noaa.gov, http://swdcwww.kugi.kyoto-u.ac.jp/index.html). The first three observation days on 21-23 September 1998 that preceded the storm interval were weakly disturbed days (  Ap=14, 10, 14). A storm began on 24 September (maximum index Kp approx 5, Ap=28 ) (see Figure 1). The disturbance rose to severe magnetic storm on 25 September. The storm was initiated by the arrival of interplanetary shock associated with the M6/3B flare and registered with the ACE satellite about 2300 UT on 24 September. The shock was followed by the southward turning of the IMF Bz component with a maximum deflection of -18 nT at 2333 UT. The solar wind parameters changed (see Figure 1), the temperature T increased up to about 7 times105 K and the speed from 440 to 880 km s-1. The dynamic pressure of solar wind reached 10 nPa value and the energy e, transferred to the magnetosphere comprised 75-100 GJ s -1. The solar flare was accompanied by the ejections of energy particles. A greater than 10 MeV energy proton flux reached its maximal level after the midnight on 25 September. The maximum precipitations of electrons were observed during the main storm phase. The variations in the Hp component of the geomagnetic field were significant on 25 September. Index Dst during the main phase of the storm rapidly decreased to -202 nT value at 0800 UT and stayed at the level about -200 nT until 1100 UT. The Kp index reached a maximum value of 8+ during 0600-0900 UT. The recovery phase began after 1100 UT and continued at least to the end of observation. Against a background of the storm the sequence of the intense substorms with index AE=1200-2000 nT was registered at night 24-25 and on 25 September. During the days under consideration, the solar activity was moderate with F10.7=139 on 25 September, and 81-day average F10.7a =130.

[12]  The IS radar measurements were conducted from 1300 UT on 21 September to 1500 UT on 25 September. The results of the analysis of the ionosphere processes over Kharkov that accompanied the geomagnetic storm on 25 September 1998 were described in detail in some publications, for example, by Chernogor et al. [2002a, 2002b], Grigorenko et al. [2003a, 2003b], Mishin et al. [2001, 2002], Taran [2001], and Taran et al. [1999]. Here we consider briefly the main results and their interpretation.

2005GI000125-fig02
Figure 2
[13]  The very strong negative ionosphere storm commenced soon after the local midnight on 25 September and persisted at least until the end of the measurements (here with Kp ge7 ). For the Kharkov IS radar ( l=36.3 o E) local Daylight Saving time corresponds to LT approx ( UT + 0325). The storm was accompanied by a decrease in the peak of the electron density NmF2 during the main magnetic storm phase approximately by a factor up to 3-3.5 as compared to the reference day, for which the data averaged over the previous weakly disturbed days on 21-23 September were selected (Figure 2a). The maximum decrease of NmF2 was observed in the morning time (approximately during 0130-0800 UT). Then NmF2 gradually increased, reached and even exceeded (during short time) the reference values NmF2 near noon, whereupon again began to decrease up to a factor of 1.4 about 1500 UT. The height of the electron density peak hmF2 increased at about 100 km at night and at 50 km near the noon (Figure 2b). In the morning during the main storm phase, the main electron density Ne peak descended to the F1 region (below 200 km) where the molecular ions dominate.

2005GI000125-fig03
Figure 3
[14]  To study peculiarities of the ionospheric storm, we consider the vertical profiles of Ne at successive moments of time (every 15 min) during the main magnetic storm phase (Figure 3). Within the time interval of 0430-0830 UT a significant deformation of the Ne(h) profiles was observed: the electron density in the F2-layer maximum decreased, the height and the thickness of the layer increased, and the shape of the profiles changed. These effects could be caused by large-scale disturbance in the neutral composition with a depletion of the ratio p= N( O)/(N( N2)+N( O2)) and rebuilding in the global thermospheric circulation with an increase of equatorward neutral wind velocity [see, e.g., Brunelly and Namgaladze, 1988; Buonsanto, 1999]. It is known that such events are associated with the high-latitude heating of the thermosphere due to the enhancement of the auroral currents and energetic particle precipitation during the magnetic disturbances [e.g., Danilov and Morozova, 1985; Serebryakov, 1982] that were observed during this storm (see Figure 1). Changes in the neutral composition and thermospheric wind could be also transferred from high to middle latitudes by the traveling atmospheric disturbances (TADs) generated by enhancements of the auroral electrojets during magnetospheric substorms [see, e.g., Buonsanto, 1999; Prölss, 1993a, 1993b]. Such substorms with the increase of index AE up to about 2000 nT were registered during the main phase of the storm in consideration (see Figure 1, where hourly AE index values are taken). For example, at night on 24-25 September in the absence of ionization production sudden rises in the height of the F2 layer were observed at about 2230, 0100, and 0200 UT (see Figure 2b). One can suppose that they are associated with pulse-like increases of the AE index that reached the values of 1400, 2000, and 1600 nT at about 2145, 2345, and 0030 UT, respectively (see (http://swdcwww.kugi.kyoto-u.ac.jp/index.html), and AE index values with 1-min time resolution). A similar case was considered, for example, by Prölss [1993a]. From the observed time delay in the disturbances of hmF2 height (of about 45, 75, and 90 min in these cases) and under assumption that the maximum of the source of high-latitude heating occurs near 70o, i.e., at a distance of approximately 2000 km from the Kharkov radar, one can obtain the velocities of TAD propagating toward the equator of about 740, 440 and 370 m s-1, respectively. Such values are typical for the horizontal phase velocity component of the internal gravity waves (IGWs) related to the large-scale TADs.

[15]  In the course of the F2 -layer deformation, the density Ne in the F1 layer varied slightly. As a result, the NmF2/NmF1 ratio became less than 1 over 0630-0730 UT. The so-called G condition happened, when in ionograms the F2 layer was shielded by the F1 layer ( foF2 le foF1 ). Similar effects were described, for example, by Buonsanto [1995a] and Mikhailov and Foster [1997].

2005GI000125-fig04
Figure 4
[16]  The simulation using the NRLMSISE-00 empirical model of the atmosphere [Picone et al., 2002] showed that the development of a deep Ne depression in the ionospheric F2 layer and the F2-layer decay (the G  condition), only partly could be explained by the changes in the neutral composition. For example, at a height of 300 km in the daytime (around 0730 UT) the parameter p from the NRLMSISE-00 model decreased by a factor of 1.1 as compared, for example, to the quiet day on 22 September (Figure 4), whereas the depletion of Ne was by a factor of 3.8 (see Figure 2a). Thus it is required to attract some additional factors to explain the Ne depression. Among these factors, there can be: a correction of the neutral composition taken from the NRLMSISE-00 model for the conditions of geomagnetic disturbances and taking into account possible contribution to the increase of the O+ ions loss rate of such factors as the atmosphere heating, enhancement of the electric fields, and the excitation of vibrational levels of N2 and O2 molecules [Buonsanto, 1995a; Mikhailov and Foster, 1997; Mikhailov and Förster, 1999; Pavlov, 1998; Pavlov and Buonsanto, 1996; Pavlov et al., 1999; Richards et al., 1994; Schlesier and Buonsanto, 1999]. It is known that the character of a magnetic storm is determined by the complicated interaction of a complex of processes in the near-Earth environment. Therefore one should expect that the observed features in the behavior of the disturbed ionosphere are the result of superposition of the effects caused by different disturbance sources, and their contribution changes during the storm. One of such sources (together with the mentioned above) could be the equatorward shift of the main ionospheric trough caused by the enhancement of the electric field of the magnetospheric convection during the main phase of the magnetic storm. The trough shift is confirmed by the analysis of the maps of global distribution of the total electron content (TEC) obtained from the GPS navigation system data [Afraimovich et al., 2002]. It follows from this analysis that on 25 September during the main phase of the storm, the low-latitude wall of the trough in the European region reached a geographic latitude j=50-40o and the Kharkov radar could enter into the trough (in the night and dawn sectors).

2005GI000125-fig05
Figure 5
[17]  The time variations of electron Te and ion Ti temperatures at heights of 250-500 km are shown in the Figure 5. Under quiet conditions the electron temperature at midlatitudes is determined by a balance between heating by photoelectrons, thermal conduction along the magnetic field lines, and cooling due to collisions with ions and neutrals. After the storm commencement the increase in Te was noticed. The enhancement in Te increased with altitude and at sim 0200 UT and 500 km altitude it reached sim 700 K as compared to a quiet day on September 23 (see Figure 5). Increase in Te could be due to Joule heating associated with penetration of magnetospheric electric fields to midlatitudes and energetic particle precipitations, and also decreased cooling due to very low electron density in the morning time [see, e.g., Buonsanto, 1995a]. The shift toward equator of the precipitation zone could be indirectly confirmed by the maximum values of the POES auroral activity level equal to 10, which were registered on board the NOAA POES 12, 14 and 15 satellites on 24-25 September during 2138 UT-1255 UT interval [http://www.sec.noaa.gov/ftpdir/lists/hpi/power_1998.txt]. The statistical pattern describing the auroral oval is appropriate to the Auroral Activity Level determined from the particle power flux observed during the most recent polar satellite pass. The value 10 of this parameter confirms that the equatorward boundary of the auroral oval could shift to geomagnetic latitudes fapprox 51-45o [http://www.sec.noaa.gov/Aurora/index.html]. Thus the Kharkov radar ( f = 45.7o ) that was close to the midnight sector during the storm main phase could be situated within the trough (see above) and not far from the equatorward boundary of the auroral oval.

[18]  The peculiarity of the electron temperature behavior was a decrease in Te after the sunrise (for the Kharkov radar the sunrise was nearly 0325 UT on the Earth's surface) that was by about of 500 K at a height of 250 km around 0600 UT and decreased with the altitude growth. This decrease could be caused by several factors, including the intensification of the cooling of the electron gas due to the started morning increase in Ne (see Figure 2a), etc.

[19]  Ion temperature under quiet conditions depends on a balance between heating due to collisions with electrons and cooling via collisions with the neutrals. At night, when the heating of electrons is reduced, electron, ion and neutral temperatures tend toward the common value. During this storm the increase in the ion temperature Ti (approximately by 300 K about 0600 UT at 300 km as compared to a quiet day on 23 September) till the end of the measurements (see Figure 5) could be the result of the Joule and the frictional heating associated with the intensification of the ionospheric currents (see below) [Buonsanto, 1995a; Richards et al., 1994].

2005GI000125-fig06
Figure 6
[20]  The temperature Tn of the neutral gas was derived from the IS data by solving the heat balance equation [Salah and Evans, 1973; Salah et al., 1976] and using the NRLMSISE-00 model of the atmosphere. The calculations showed that during magnetic storm the increase in Tn (Figure 6) was on the average nearly 200 K at a height of 300 km. The heating of neutrals could be related to both, nonlocal source of the heat transported from the region of the high-latitude heating of the thermosphere [Danilov and Morozova, 1985; Mikhailov and Foster, 1997; Serebryakov, 1982] and local Joule heating due to the penetration of the magnetospheric electric fields to midlatitude (see below) [Mikhailov and Foster, 1997]. Probably both phenomena could occur in the ionosphere over Kharkov during the main phase of the storm. As the obtained results showed (see Figure 6 and Grigorenko et al. [2003b]), the nonlocal heating apparently dominated, because the Tn disturbances were propagating from the above with a velocity of about 50 m s -1 (the time delay in the 400-250 km altitude range was about 50 min) and this value could correspond to the vertical component of the IGWs velocity related to TADs. The friction heating due to the ion drift with respect to the neutral gas could provide also some contribution to the increase of Tn during the electric field enhancement over radar [Buonsanto, 1995a]. It should be also noted that the NRLMSISE-00 model (see Figure 6) and MSIS 86 model [see Grigorenko et al., 2003b] give underestimated values of Tn (at a height of 300 km by 450 and 350 K, respectively) in disturbed conditions and require correction. This disagreement (with MSIS 86 model) has been also noted by other authors [see, e.g., Buonsanto, 1995a; Mikhailov and Förster, 1999; Richards et al., 1994].

[21]  It should be noted that during quiet days Tn at the heights of 250 and 300 km, apparently, reached the exospheric temperature value and changed slightly, whereas above 300 km the calculated Tn value decreased with the height growth (see Figure 6). This could manifest that the method applied for calculation of Tn at large heights is incorrect and requires taking into account thermal conductivities of ion and neutral gases. At the same time on the disturbed day 25 September thermal conductivity effects can be neglected at least up to the height of 450 km (see Figure 6), that probably can be explained by the disturbance in the neutral composition (increase in N2 and O2 concentrations) [Prölss, 1993a, 1993b] (see Figure 4) and in collision frequencies of charged and neutral gas species.

[22]  As was mentioned above, the technique we applied to calculate Ti and Te insufficiently reflects changes in ion composition. Mikhailov and Schlegel [1997] and Mikhailov and Foster [1997] indicated that incorrect consideration of ion composition could result, e.g., in underestimation of Ti and Te up to 50% during geomagnetic disturbances. Therefore we should anticipate that the calculated values of Tn (as well as other derived parameters, e.g., energy input rate to the electron gas Q/Ne, the heat flux density PT transferred by electrons from the plasmasphere, etc.) only qualitatively describe the behavior of the disturbed atmosphere.

2005GI000125-fig07
Figure 7
[23]  The mechanisms of the ionospheric disturbance considered above can be attracted for explanation of the observed reversal in the vertical plasma drift velocity and plasma flux during the main phase of the magnetic storm (Figure 7). In the morning hours (near 0400 UT) of the disturbed day, for example, at altitudes 250-350 km, the vertical velocity and flux density of plasma were Vz approx 50-35 m s-1 and F p approx(4-3)times 1012 m-2 s-1, respectively, whereas on the quiet day (on 23 September) these values were Vz approx -(25-20) m s -1 and F p approx-(8-4)times 1012 m-2 s-1, respectively.

[24]  It is known that at midlatitudes the ion drift velocity is determined by the influence of three mechanisms: ambipolar diffusion along the geomagnetic field lines, neutral wind and Etimes B drift of ions. Near the peak of the F2 layer where the O+ ions dominate the vertical ion velocity may be written as [see, e.g., Brunelly and Namgaladze, 1988]

eq002.gif(2)

where (Vd|)z , (Vn|)z , (Vperp)z are the components of the vertical ion velocity due to ambipolar diffusion, neutral wind and electromagnetic drift, respectively (subscript parallels and perpendiculars relate to the ion velocities parallel and perpendicular to the geomagnetic field induction vector). Substituting the value of each term in the expression (2), we obtain the velocity Vz in the form

eq003.gif

eq004.gif

eq005.gif(3)

where

eq006.gif(4)

is the vertical velocity component due to ambipolar diffusion, Da= kTp/mi Snin is the ambipolar diffusion coefficient, Vnx, Vny, Vnz are the meridional, zonal and vertical components of the neutral wind velocity positive in the north hemisphere when they are directed toward the geographic south, the east and at zenith, respectively, nin is the ion-neutral collision frequency, Hp=kTp/mi g is the scale height of plasma, Tp=Te+Ti is the plasma temperature, I and D are the geomagnetic field inclination and declination (for Kharkov I = 66.4o, D = 6.7o ); Ex and Ey are the components of electric field intensity vector, directed toward the geographic south and the east in the north hemisphere, B is the absolute value of the geomagnetic field induction vector, mi is the ion mass (near the peak of the F2 layer, predominant ion is O+ ).

2005GI000125-fig08
Figure 8
[25]  From the radar measurements of Vz and calculation of the diffusion velocity Vdz, the velocity W that includes both electric field and neutral wind effects can be detected:

eq007.gif(5)

If the declination D effects in the expression of Vz (equation (3)) are neglected due to their smallness the velocities Vn|z and Vperp z may be written as

eq008.gif(6)

eq009.gif(7)

In the morning hours (near 0400 UT) on the disturbed day 25 September W =100 m s-1 at a height of 300 km, whereas on a quiet day Wapprox 0 (Figure 8). One of the reasons of this disturbance W and Vz could be equatorward surge in the neutral wind Vnx caused by the high-latitude heating and TAD [Buonsanto, 1995a; Buonsanto et al., 1999; Prölss, 1993a, 1993b; Richards et al., 1994]. If the electric fields are neglected, that is correct for the magnetic quiet conditions (as on 21-23 September when Ap= 14, 10, 14), the meridional (equatorward positive) component Vnx of the thermospheric wind velocity is presented in the form

eq010.gif(8)

This velocity should have a value of sim270 m s -1 (see Figure 8). The other reason could be a penetration into midlatitudes of the nonstationary magnetospheric electric field [Buonsanto et al., 1999; Foster and Rich, 1998; Foster et al., 1998a] with the zonal component Ey = 12-17 mV m-1 (determination of Ey from radar measurement of hmF2 see below) capable also to provide W approx Vperp z =100-130 m s-1. This case is the limiting one that neglects the neutral wind effects and gives the upper estimations of Ey and Vperp z. In expression (7), B approx 5 times 10-5 T was taken for Kharkov. Probably, both factors contribute to the increase in W and Vz. The high substorm activity (in the auroral region the AE index reached 800-1200 nT at 0300-0340 UT interval) and also the upward propagation of the disturbance in Vz with a velocity of about 100 m s-1 (the delay of the disturbance in Vz at 250-500 km altitude range was about 40 min; see Grigorenko et al. [2003b] and Figure 7) could testify to the predominance of the electric field pulse effects over Kharkov. The disturbance in Vz could be caused by the local Joule heating of the atmosphere at the dynamo-region heights (100-110 km), related to the disturbance in the electric field over Kharkov, and upward motion of the gas (similarly to the high-latitude heating source [Mikhailov and Förster, 1999]). The decrease of Vz with the height growth is likely associated with the dissipation of kinetic energy of gas due to viscosity and thermal conductivity.

[26]  The effects of the electric field penetration (together with the enhancement of the equatorward meridional winds due to the intensification of the high-latitude thermosphere heating source and with the trough shift to midlatitudes) could be one of the reasons of the long-lasting increase in the height of the electron density peak hmF2 by about 100 km at night and 50 km around the noon as compared to the quiet day on 23 September (see Figure 2b). However, these effects are not related to the nonstationary magnetospheric electric fields as in the case considered above. These effects could be caused by long-lasting precipitation of energetic particles registered during the storm (see Figure 1). Particle precipitations lead to an increase in the conductivity of the underlying auroral ionosphere and short circuiting of the shielding polarization field, and promote a penetration of the magnetospheric electric fields to midlatitudes [see, e.g., Brunelly and Namgaladze, 1988; Gonzales et al., 1983]. Estimations of Ey were derived from hmF2 deviations during magnetic disturbances according to simplified empirical dependence given by Brunelly and Namgaladze [1988]. They showed that over Kharkov such fields should be eastward directed, should have the values of Ey approx 17 and 12 mV m -1 at night and in the daytime, respectively, and can contribute into the observed uplifting of the F2 layer.

2005GI000125-fig09
Figure 9

4.2. The magnetic storm on 29-30 May 2003

[27]  The magnetic storm on 29-30 May 2003 was caused by the arrival of two interplanetary shocks from the X1.3 and X3.6 flares on 27-28 May. The main parameters of heliogeophysical situation are presented in Figure 9. The first shock passed the NASA/ACE spacecraft on 29 May at 1150 UT. The second and stronger shock passed ACE at 1830 UT with 125 km s-1 increase in the solar wind speed up to over 800 km s-1 and Bz deflections that ranged between -20 nT and +25 nT. The solar wind temperature increased, the dynamic pressure exceeded 15 nPa, e energy reached the value of 50 GJ s-1 at night on 29-30 May. Hp component of geomagnetic field varied in the 0-200 nT range. The greater than 10 MeV proton fluxes were observed during several days. The maximum precipitations of electrons began in the premidnight sector on 29 May and continued till local noon on 31 May. Magnetospheric substorms with index AE value greater than 2000 nT were registered during storm. The geomagnetic response to these events was severe storm with maximum indices Ap=89, Kp = 8+. The storm commenced suddenly on 29 May about 1225 UT. The main phase developed slowly. The Dst index rapidly decreased to -108 nT at 2300 UT on 29 May and remained at the level of -(116-131) nT until 0300 UT on 30 May, which was followed by the recovery phase related to northward turn of the IMF Bz component. For this storm solar activity was moderate with F10.7=138 and 117 on 29 and 30 May, respectively, and F10.7 a=124. The Kharkov IS radar was operated on 30-31 May according to the international program Low/High Latitude.

2005GI000125-fig10
Figure 10
2005GI000125-fig11
Figure 11
2005GI000125-fig12
Figure 12
[28]  The magnetic storm was accompanied by a strong ionospheric storm. Detailed description of the results of observation obtained by the Kharkov IS radar was published by Grigorenko et al. [2005a, 2005c]. Substantial effects of a negative ionospheric disturbance were revealed (Figure 10). Among them there were: a depletion of NmF2 by a factor up to 4 during the storm main phase (Figures 10 and 11a); unusual plasma heating at night on 29-30 May when the ion and electron temperatures at altitudes of about 300-800 km increased up to the daytime values of 1200-2400 K and 2000-3200 K, respectively, whereas during quiet conditions the values of these temperatures at night were about 800 K (Figures 10 and 12). One of the reasons of these disturbances could be the shift of the main ionospheric trough to midlatitudes and also the shift of the hot zone together with the plasmapause to lower L shells [Buonsanto, 1995a, 1995b; Richards et al., 1994]. Such phenomena in the ionosphere related to the inner plasmasphere (for Kharkov Lapprox 1.9 ) occur rarely. The trough equatorward shift was indirectly confirmed by the maximum values of the POES Auroral Activity Level equal to 10, which were registered on board the NOAA POES 14, 15, 16, and 17 satellites during the storm main phase (approximately from 2223 UT on 29 May till 0233 UT on 30 May) and could manifest the shift of the auroral oval equatorward boundary toward geomagnetic latitudes fapprox 51-45o (similar to magnetic storm on 25 September 1998). Thus the oval was able to approach the Kharkov radar latitude near local midnight.

[29]  It should be noted that we took as a reference data of the averaged values of foF2 on quiet days 19 and 20 May 2003 obtained by the ionosonde at San Vito (the geographic and geomagnetic coordinates are 40o N, 17o E and 39.7o, 96.4o, respectively), and also the results of ionosphere measurements on 26-27 May 1998 and 23-24 June 1998 (due to the absence of closer quiet periods). The latter periods were on the rising branch of the current solar cycle 23 but similar to the considered period with respect to the parameters of the heliogeophysical conditions (summer, moderate solar activity).

[30]  An increase in the hmF2 height by about 160 km during the storm main phase (at night) and by 70 km near noon as compared to reference day on 26-27 May 1998 was registered (Figures 10 and 11b). The lifting of the F2 layer is probably explained by the joint action of several factors. They include expansion of the thermosphere, the increase in the equatorward meridional velocity of the thermospheric wind, and the trough shift to middle latitudes. Along with the above reasons, the penetration of magnetospheric electric fields to midlatitudes, similar to the magnetic storm on 25 September 1998, could contribute to an increase in hmF2 due to long-lasting (for more than a day) precipitation of energetic protons and electrons. The latter fact could be manifested in the increase in the flux density of these particles registered on board the GOES 8 and GOES 12 satellites (see Figure 9). In the case when the effects of electric fields are predominant, the maximal values of the eastward zonal field component over Kharkov (determined from the upper Ey estimates obtained based on a change in hmF2 ) are approximately equal to 25 and 20 mV m -1 at night and in daytime, respectively.

2005GI000125-fig13
Figure 13
[31]  It should be referred to an unusual phenomenon that was observed near the sunrise on 30 May. It included a quasiperiodic disturbance in the velocity Vz for about 0200-0400 UT (Figure 13) against a background of the unusual morning decrease in NmF2 (see Figure 11a), a sharp decrease and the following increase in hmF2 by about 160 km (see Figure 11b, the time of these perturbations is shown by horizontal segments), and deformation of the Ne profile
2005GI000125-fig14
Figure 14
(Figure 14) (see Grigorenko et al. [2005a, 2005c] for details). One should note that such phenomena also could result from the superposition of the effects of different disturbance sources [see, e.g., Buonsanto, 1999]. One of these sources could be the penetration to midlatitudes of the nonstationary magnetospheric electric fields [Gonzales et al., 1983; Foster and Rich, 1998; Foster et al., 1998]. Thus an increase in hmF2 by about 90 km at 0330-0400 UT and later (ignoring hmF2 changes due to Ne profile stratification, see Figures 11b and 14) could be caused by a pulse of the electric field in the ionosphere over Kharkov with the eastward zonal component Ey cong 20 mV m-1. The sharp turn of the IMF Bz component from the south to the north and change in the dynamical pressure of the solar wind after midnight (see Figure 9) could be the sources of the electric field pulse in the magnetosphere.

2005GI000125-fig15
Figure 15
[32]  The decrease in the relative density of hydrogen ions N( H+)/ Ne at altitudes of 1000-1500 km more than by an order of magnitude during the storm main phase (at night on 29-30 May), as compared to a reference day on 26-27 May 1998, with its following increase in the daytime on 30 May during the recovery phase (Figure 15) points out the processes of emptying and further filling of the magnetic flux tube over Kharkov radar [Bailey et al., 1979; Brunelly and Namgaladze, 1988; Krinberg and Tashchilin, 1984; Naghmoosh and Murphy, 1983]. The tube is located in the inner plasmasphere and usually is slightly influenced by magnetic disturbances. These effects could be related to the equatorward shift of the main ionospheric trough and the light ion trough and, probably, are accompanied by a change of the processes of ionosphere-magnetosphere interaction. The development of the process of filling in of the magnetic flux tube is defined by a decrease of the height ht, where N( O+) = N( H+)
2005GI000125-fig16
Figure 16
(Figure 16). At night on 30 May, ht exceeded 1500 km and on 31 May ht decreased down to near 1000 km. During quiet conditions ht approx 700-850 km, that is, the ht height did not reach the level preceding the beginning of the storm. It should have been expected, because the process of filling in of magnetic flux tubes proceeds slowly with a time constant proportional to L4 [Brunelly and Namgaladze, 1988; Krinberg and Tashchilin, 1984; Saenko et al., 1982].

2005GI000125-fig17
Figure 17
[33]  The magnetic storm was also accompanied by thermospheric disturbances. The calculations using the NRLMSISE-00 model showed that on 30 May, for example, in the daytime about 0800 UT, the changes in the neutral composition led to the depletion of the p parameter by a factor of 1.4 as compared with quiet days on 19-20 May (Figure 17). However, it could not provide the observed for this time depletion of NmF2 by a factor of 2.5. In the same way as in the case of the 25 September 1998 storm, a correction of the model, or attraction of other factors considered above were required. The distinction of the neutral composition and parameter p during the reference days on 19 and 20 May and before the storm commencement on 29 May (at 0000-1200 UT) can be explained by the fact that before the reference day the conditions were quiet (for 16-18 May Ap= 10, 9, and 9, maximal indices Kpapprox 3 ), and for 29 May the conditions were disturbed (on 26-28 May Ap = 18, 26, and 36, maximal indices Kp approx 4, 5, and 6, respectively). It is known that in the model NRLMSISE-00 the values of the 3 hours Ap indices are taken with the "history" (within 72 hours).

2005GI000125-fig18
Figure 18
[34]  The neutral temperature Tn, as calculations showed (see Grigorenko et al. [2005a, 2005c] and Figure 18), during the storm main phase ( Kp approx 8 ), when an unusual plasma heating was observed against a background of deep Ne depression in the F region, was about 1000-1350 K at altitudes of 220-470 km, respectively. For comparison, we note that the Tn values decreased by approximately 200-350 K at the same altitudes on the next night during the recovery phase ( Kp approx 5 ). The heating of the atmosphere led to the increase of the thermopause height during the storm main phase at least up to 400 km, whereas on a quiet day its height was about 300 km.

2005GI000125-fig19
Figure 19
[35]  Considerable variations in the thermal regime of plasma accompanied this magnetic storm. The calculations showed that on the disturbed day about noon, the rate of the energy input to electron gas Q/Ne decreased as compared with a quiet day by a factor of 1.6 (Figure 19). This energy is determined from the thermal balance equation for electrons. Thermal electrons are heated in the process of thermalization of suprathermal electrons, and this process in the lower ionosphere ( h le 300-350 km) is local because the mean free paths of these electrons are small. The main mechanisms of electron gas cooling at these altitudes are Coulomb collisions of electrons with ions and excitation of the fine structure levels of oxygen atoms [Banks, 1966; Shunk and Nagy, 1978]. In such a case, the electron energy balance equation in the SI system can be written in the following form for stationary conditions [Banks, 1966; Dalgarno and Degges, 1968]:

eq011.gif(9)

eq012.gif(10)

eq013.gif(11)

where Q is the rate of the energy transfer to thermal electrons during Coulomb collisions with suprathermal electrons; Lei is the energy loss rate during electron-ion collisions; and Le is the rate of the energy loss by excitation of the fine structure of oxygen atoms (  Q, Lei, and Le are the corresponding energy values per unit time reduced to unit volume). Figure 19 presents results of calculating the Q/Ne energy transferred to electron per unit time, as well as the components of the electron gas energy loss during heat exchange with ions Lei/Ne and neutrals Le/Ne. Figure 19 indicates that under the quiet and disturbed conditions the contributions of Lei/Ne and Le/Ne components to the process of electron gas cooling were different.

2005GI000125-fig20
Figure 20
[36]  The decrease in Q/Ne on the disturbed day was accompanied by an increase in the heat flux density PT, transferred from the plasmasphere due to electron thermal conductivity by a factor of 1.2 (Figure 20). The vertical component of the heat flux density is

eq014.gif

where ke=2.082 k Ne Te/ m nei is the coefficient of the electron gas thermal conductivity, k is the Boltzmann constant, m is the electron mass, and nei is the electron-ion collision frequency. In the SI system [Ginzburg, 1967]:

eq015.gif

Substantial values of the energy input rate Q/Ne and the heat flux density PT during the storm main phase (at night on 29-30 May) became the specific feature of the thermal regime of the ionosphere rarely observed at midlatitudes. Under quiet conditions, these nighttime values were close to zero. These effects manifest the change in the processes controlling the heat balance in the ionosphere-plasmasphere system during a storm [Banks, 1966; Shunk and Nagy, 1978].

2005GI000125-fig21
Figure 21

4.3. The magnetic storm on 20-21 March 2003

[37]  The magnetic storm on 20-21 March 2003 proceeded against a background of high flare activity on the Sun. However, the geoefficiency of the flares was low, and the flares resulted in a minor magnetic storm on 20-21 March 2003 (maximal index Kp =5 ). Variations in heliogeophysical parameters are presented in Figure 21. After a sudden storm commencement at 0445 UT on 20 March with Dst increasing up to 15 nT at 0600 UT, Dst decreased to -57 nT at 2000 UT. The recovery phase began after 2100 UT and continued till the end of the observations. Solar wind parameters: the temperature and the speed changed weakly, the dynamic pressure did not exceed 4 nT value, the value of the Akasofu function e was less than 30 GJ s -1. The energetic proton fluxes and precipitations of electrons were not practically registered. The variations of the Hp component of geomagnetic field were insignificant. The high substorm activity with the values of index AE=1000-1500 nT was observed during the sunset period on 20 March. The storm occurred during moderate solar activity, with the values of index F10.7=97 and 91 on 20 and 21 March, respectively, and an 81-day average value of index F10.7 a =132. The measurements were conducted on 19-23 March according to the Storms/TIMED/LTCS program. The radar operated in two-pulse sounding mode in the 100-550 height range with the altitude resolution of about 10 km. The effects of this geomagnetic storm in the ionospheric F region and upper thermosphere over Kharkov were described in detail by Grigorenko et al. [2005b, 2005d]. Here we consider briefly the main results of the study.

2005GI000125-fig22
Figure 22
[38]  The magnetic storm was accompanied by a two-phase ionospheric storm (Figure 22a). A peculiarity of the latter was that its strong negative phase occurred against a background of a minor disturbance of the geomagnetic field ( Kp approx 5). An increase in NmF2 by about a factor of 1.5 during the positive phase of the storm and a decrease in NmF2 by a factor up to 5 during the negative phase (in the morning hours) as compared to the reference day were registered (Figure 22b). As reference data the foF2 values were taken during the quiet day on 19 May 2003. They were obtained by the ionospheric stations at Kharkov within 1230-2400 UT and at San Vito within 0000-0730 UT. The information from other
2005GI000125-fig23
Figure 23
(closer) stations was absent for this period. Figure 23 illustrates the behavior of the electron density Ne and other ionospheric parameters at altitudes of sim 100-550 km during the storm. A significant distinction of the daytime electron temperatures Te during the positive (1300 K at an altitude of 300 km) and negative (2400 K) storm phases from the value Te=1700 K on a quiet day was detected. The distinction in Te values during considered days is explained by the different cooling rates of the electron gas in the process of elastic heat exchange with ions, the cooling being proportional to Ne2. The ion temperature Ti increase in the daytime at altitudes of 250-300 km was about 50 K and 100-150 K on 20 and 21 March, respectively. Apparently, the Ti growth was related to a high-latitude source of atmospheric heating which was not considerable during the minor magnetic storm and also to the increased electron-ion heat exchange due to the substantial difference in the temperatures of electrons and ions on 21 March.

2005GI000125-fig24
Figure 24
[39]  The positive storm phase on 20 March had a character of a long-duration disturbance and lasted approximately for 6 hours. It could be caused by the enhanced equatorward meridional wind Vnx, related to the high-latitude thermosphere heating [Buonsanto, 1998; Danilov and Morozova, 1985; Danilov et al., 1985a]. A comparison with the quiet period (22 and 23 September 1998 which were chosen as reference days due to absent of any other, more suitable period) indicated that the downward velocity Vz at altitude of 300 km decreased on average by approximately 10 m s -1 near local noon on 20 March 2003 (Figure 24). For Kharkov LT approx (UT + 0225) in March. If we neglect the contribution of additional electric fields during minor storm and the change of diffusion velocity, we obtain the upper estimate of the DW approx 10 m s-1 that lead to the observed increase in hmF2 by about 20 km. In this case an additional velocity value DVnx should be about 25 m s-1.

[40]  Let us consider more in detail the processes accompanied the ionospheric storm phase reversal. It occurred in the sunset period and was accompanied by a decrease in hmF2 by 50 km during 1700-1800 UT with a subsequent ascent of the layer by almost 200 km from 1800 to 1900 UT (Figure 22c). On the reference days of 22-23 September 1998 [see Grigorenko et al., 2003b] and 19 March 2003 (see Figure 22c) a change in the hmF2 at the sunset was much smaller (about 50 km). At the same time on 20 March, hmF2 rapidly (for not more than 30 min) decreased by 40 km. Fifteen minutes later, hmF2 increased by about 90 km during one record of measurements, i.e., within 15 min (the data processing was performed at a 15-min signal integration).

[41]  It is known [see, e.g., Foster and Rich, 1998; Foster et al., 1998] that rapid changes in hmF2 (together with Vz, see Figure 24) can be caused by nonstationary disturbances of magnetospheric electric fields and by the penetration of these fields to midlatitudes. Another mechanism explaining such variations can be related to propagation of TADs generated by an enhancement of auroral electrojets during a substorm [see, e.g., Prölss, 1993a, 1993b]. Such a substorm with a maximal AE value of 1500 nT at auroral latitudes was registered about 1800 UT on the considered day (see Figure 21), i.e., during the ionospheric storm phase reversal. However, TAD effects cannot explain the initial decrease of hmF2 by 50 km since TADs are characterized by the transport of the equatorward meridional wind, as a result of which the F2 layer ascends and hmF2 increases both in daytime and at night. Moreover, TAD (which propagates from the auroral region with a velocity of 400-700 m s -1 ) appears at midlatitudes with a delay of about 1.5-1 hours. In our case this delay relative to the time of the substorm intensity maximum was almost absent.

[42]  Proceeding from the aforesaid, we can assume that an initial decrease in hmF2 could be caused by the penetration into the ionosphere over Kharkov of the magnetospheric electric field with a westward zonal component of Ey approx - 10 mV m -1. The estimation of Ey value was obtained from the hmF2 variations [Brunelly and Namgaladze, 1988]. Such cases of the hmF2 decrease are considered, e.g., by Reddy and Nishida [1992] and Prölss [1993b] and are explained by increased magnetospheric convection before a substorm, which generates the westward electric field.

[43]  The following increase of hmF2 could be caused by the regular equatorward turning of Vnx during the sunset, when DhmF2approx 50 km as on quiet days, and by the additional change in Vnx (DVnx approx 25 m s -1 ), which occurred in the preceding hours and was one of the reasons of the initial positive phase (see above). This additional change in Vnx could explain the F2 -layer lifting by about 20 km more.

[44]  Now we consider the mechanisms that can explain the remaining increase of hmF2 by about 130 km. One of these mechanisms could be the eastward electric field. If we assume that the registered uplifting of the F2 layer by 90 km during 15 min was caused by this electric field effects, we obtain Ey approx 15 mV m-1. The cases of such rapid eastward-westward switching of the electric field related to changes in the electrodynamic conditions during a magnetospheric substorm (gradients of conductivity, electric fields and currents at the edges of the westward auroral electrojet, nonstationary magnetospheric convection, etc.) were discussed, e.g., by Reddy and Nishida [1992]. One of the reasons of nonstationary magnetospheric convection could be the abrupt change of the solar wind parameters: the speed V sw, IMF  Bz component, the dynamic pressure p sw around 1800 UT (see Figure 21).

[45]  The second mechanism could be related to TAD, which results in an increase in the hmF2 after switching off of the previously acting westward electric field. TAD was able to provide the remaining 40-km lifting of the F2 layer. Such examples of a successive action of the electric field and TAD on the ionosphere were described, for example, by Reddy and Nishida [1992] and Prölss [1993b]. However, TAD effects should have been related to the earlier substorms (see Figure 21) with smaller intensity ( AE =600-900 nT). Nevertheless, the TAD effects were found out. They were observed as a delay of the velocity Vz disturbance that propagated from top to bottom (see Figure 24b). The delay was 80 min in altitude range of 400-200 km. This corresponded to a velocity of about 40 m s -1, which is typical of the vertical component of IGW velocity. Besides, the Vz disturbance amplitude increased up to an altitude of 350 km and then began to decrease. Such features of the wave energy dissipation with an increase of height are also typical of IGWs. From this we can conclude that an increase of hmF2 by about 130 km could result from the superposition of the effects of two sources: the eastward electric field, which is related to the intense substorm that occurred at 1800 UT ( AE=1500 nT), and TAD generated by the earlier substorm, e.g., after 1600 UT ( AE=600-900 nT).

[46]  Thus an unusual behavior of hmF2 during the sunset on 20 March can be explained in the following way. The 50-km decrease and the subsequent 90-km rapid increase in hmF2 (within the total approximately 200-km increase) were probably caused by the nonstationary electric field, with the zonal component changing its direction from westward to eastward ( Ey = -10 and +15 mV m -1 ), which penetrated to midlatitudes. It should be mentioned that the hmF2 variations correlated with the Vz changes from +10 to -35  m s -1 and, later on, to +20 m s-1 at an altitude of 300 km (see Figures 22c and 24b) in the time interval approximately 1700-1900 UT with a delay of about 20 min, which is typical for the effects caused by electric field disturbances in the ionosphere [Foster and Rich, 1998; Foster et al., 1998a]. These disturbances were probably related to the intense substorm that occurred at 1800 UT (see Figure 21) when at auroral latitudes the electric field strength could reach sim 70-100 m V m-1 at the values of index AE=1000-1500 nT [Krinberg and Tashchilin, 1984; Serebryakov, 1982]. Penetrating to midlatitudes, such a field could reach the calculated value |Ey| approx (10-15) mV m -1 and destabilize the behavior of the ionospheric F2 layer. Besides, TAD, which could be caused by a less intensive substorm occurred after 1600 UT, also could contribute to an additional increase in hmF2 by about 40 km.

2005GI000125-fig25
Figure 25
[47]  At the ionospheric storm phase reversal, the NmF2 decrease approximately by a factor of 2 (whereas on the quiet day such decrease is by about 20%) became a beginning of a deep negative disturbance. The estimates showed [Grigorenko et al., 2005b, 2005d] that the decrease in NmF2 could be caused by the increase in the downward plasma drift velocity Vz (see Figure 24b) and the change in the velocity W by -25 m s -1 near F2 -layer peak against a background, of b( O+) loss coefficient increase by a factor of almost 5 at a decrease in hmF2 from 280 to 230 km (see Figure 22c). The further development in the depression in Ne (by a factor of 4-5 in the morning on 21 March), could be related to the change of neutral composition and increase in the N( N2) and N( O2) concentrations that should be maximum in the dawn sector. However, the NRLMSISE-00 model data did not confirm this assumption: N( N2) and N( O2) concentrations about noon on 21 March, on the contrary, decreased by a factor of 1.7 and p parameter increased by a factor of nearly 1.3 as compared to a quiet day on 19 March (Figure 25). Possibly, model requires the correction of the values of N( N2) and N( O2) as it was shown, e.g., by Pavlov et al. [2004]. Besides, one more mechanism (vibrational excitation of N2 and O2 molecules [Buonsanto, 1999; Pavlov, 1998; Pavlov et al., 1999 Richards et al., 1994]) could begin to operate at the sunrise when plasma rapidly warmed up ( Te increased to 2000-3500 K) against a background of low Ne values. This mechanism becomes substantial at Te ge 2000 K [Banks, 1969; Brunelly and Namgaladze, 1988; Pavlov et al., 1999; Shunk and Nagy, 1978]. The contribution of excited N2 and O2 molecules also accelerates the loss of O+ ions.

2005GI000125-fig26
Figure 26
[48]  The considered minor magnetic storm did not cause considerable thermospheric disturbances. The daytime neutral temperature Tn as indicated calculations based on the radar data and NRLMSISE-00 model increased by approximately 50 and 100 K during the positive and negative ionospheric storm phases, respectively, as compared to quiet days on 19 and 23 March and the height of the thermopause where the atmosphere becomes isothermal was about 300 km (Figure 26). We established that the daytime Tn values obtained from NRLMSISE-00 model are lower than the calculated values. Under slightly disturbed conditions on 22 March, the differences were about 80 K at an altitude of 300 km, whereas during the storm, they were up to 130 K on 20 March, and about 180 K on 21 March.

2005GI000125-fig27
Figure 27
[49]  At the same time, the plasma thermal regime appeared to be sensitive to ionospheric disturbances. The calculations demonstrated that the rate of the energy input to the electron gas (per one electron) Q/Ne near noon at an altitude of 300 km during the storm negative phase (21 March) was factors of 2.5-4 and up to 2 higher than the Q/Ne value during the positive phase (20 March) and on the slightly disturbed day (22 March), respectively (Figure 27). Simultaneously the absolute value of the heat flux density PT transferred by electrons from the plasmasphere also reached the
2005GI000125-fig28
Figure 28
maximal value during the negative phase (Figure 28). This is explained by the fact that PT depends strongly on TePTsim Te5/2 dTe/dz ). The observed effects could manifest the changes in the ionosphere-plasmasphere thermal balance during the storm.


AGU

Powered by TeXWeb (Win32, v.2.0).