3. Interpretation of the Observational Data
[9] In this section we consider the problem on the correspondence of
the observational results obtained to the current concept of
formation of the ionosphere response to a magnetic storm.
According to this concept, approximately 10-20 min after the
onset of the storm, there begins an expansion and shift to lower
latitudes of the auroral oval, these events being accompanied by a
rapid heating of the high-latitude ionosphere proportionally to the
increase in the
AE index
[Emery et al., 1999].
As a result of this
pulse heating, strongly stretched along longitude large-scale
gravity waves with a period of 1-3 hours are generated in the
atmosphere. Propagating in the thermosphere from high to lower
latitudes, they change the relative composition of the thermosphere
and parameters of the neutral wind. According to measurements,
the meridional velocity of the wind can reach values
U 
500-800 m s-1 in
the periods of magnetic storms, whereas in quiet
conditions
U 100-200 m s-1
[Buonsanto et al., 1999a;
Emery et al., 1999;
Hagan, 1988;
Prolss and Ocko, 2000].
The
intensification of the equatorward wind leads to a lifting of the
F2 layer and to increase in the charged particle concentration within
its maximum. This increase is usually called "positive phase" of a
magnetic storm
[Rishbeth, 1998].
Besides the dynamical impact on
the
F2 layer, there is formed in the meridional plane a disturbed
("storm") circulation cell capable to change considerably the
midlatitude thermosphere composition. It happens, first, as a result
of the direct transport of the disturbed composition by the wind out
of the auroral oval zone, and, second, because of intensification of
vertical motions leading to a decrease (upward flows) or to
increase (downward flows) of the atomic oxygen concentration in
the lower thermosphere
[Buonsanto, 1999;
Burns et al., 1991;
Rishbeth, 1998].
Within the height interval 200-300 km where the
maximum of the
F2 region is formed, the electron concentration is
proportional to the ration of contents of O and neutral molecules,
i.e.,
NmF2 
R = [ O]/[ N2].
Usually, during the main phase of the
storm, the circulation leads to a decrease of
R. This leads to a
decrease in
NmF2 called the negative phase of an ionospheric storm
[Buonsanto, 1999].
Studying the ionospheric reaction to a
magnetic storm both at middle and high latitudes, one should
consider, beside thermospheric disturbances also the effects related
to the storm-time variations of magnetospheric sources such as the
convection electric field and fluxes of precipitating energetic
electrons
[Rodger et al., 1992].
It should be noted that the principal
difficulty in studying of disturbed ionosphere behavior is related to
the limitations in information on spatial-time variations of the
thermospheric and magnetospheric parameters during particular
magnetic storms. That is why in this paper two sets of model
simulations were performed for interpretation of ionospheric
observations in the periods of the considered geomagnetic
disturbances. In the first set some statistically mean (empirical)
models of magnetospheric sources and thermospheric parameters
were used. In the second set, a correction of these empirical
models was performed in order to obtain the best description of the
experimental data.
[10] The theoretical model of ionosphere-plasmasphere interactions
(developed in the Institute of Solar-Terrestrial Physics)
[Tashchilin and Romanova, 2002]
was used. This model is based on numerical
solution of the system of nonstationary equations of the balance of
particles and thermal plasma energy within closed geomagnetic
field tubes, their bases being located at a height of 100 km. It
assumes that the plasma consists of atomic O+, H+, N+, and He+ and
molecular N2+, O2+, and NO+ ions.
[11] Concentrations of all ions, except N2+, were calculated taking into
account the processes of photoionization, recombination, transport
along geomagnetic field lines under the action of the ambipolar
diffusion, and drawing of ions by the horizontal neutral wind. The
reference spectrum of the EUV radiation from
Richards et al. [1994]
was used for calculations of photoionization of
thermospheric constituents and energetic spectra of the primary
photoelectrons.
[12] Electron and ion temperatures were determined taking into account
the heat conductivity processes along geomagnetic field lines and
exchange of thermal energy between electrons, ions, and neutral
species due to elastic and inelastic collisions. The rate of the
thermal electron heating was calculated self-consistently by
solution of the kinetic equation of photoelectron transport in the
conjugated ionospheres taking into account the energy loss while
passing through the plasmasphere. The global empirical
thermospheric mode MSIS 1986 was used to describe spatial-time
variations of the temperature and concentration of the neutral
constituents O, O2, N2, H, and N. The velocities of the horizontal
thermospheric wind for the high-latitude stations Yakutsk and
Norilsk and equatorial station Khainan were determined from the
HWM 90 model. For the midlatitude stations Irkutsk and
Manzhouli, the velocities were calculated by the approximate
method of
Kohl and King [1967].
[13] The values of the integral flux and mean energy of the
precipitating electrons needed to calculate the auroral ionization
rates were taken from the global model of electron precipitation by
Hardy et al. [1987].
The electric field of magnetospheric
convection was determined according to the empirical model of the
potential distribution
[Sojka et al., 1986] and the
Richmond et al. [1980]
model for the high-latitude and equatorial models,
respectively.
[14] The reaction of the ionosphere to the considered magnetic storms
was reproduced calculating the variations of the plasma parameters
within the entire magnetic field tube thee basis of which in the
North Hemisphere was located in the points with the geographical
coordinates of ionospheric stations shown in Table 1. The general
algorithm of model equation solution consisted of three stages. At
the first stage, the trajectories of the drift were calculated by
integration of the equation of plasma tube motion back in time
from the given moment of UT to some initial moment UT0.
Variations of the electric field in time were taken into account via
real variations of the hourly values of geomagnetic activity indices
( Kp and
Ap ) and parameters of the interplanetary magnetic field
( Bz and
By ). The second stage included calculation of initial
distributions of the plasma concentrations and temperatures (at the
UT
0 moment) along the field line. At the third stage, the equations
of the ionospheric model were integrated in the right direction in
time (from UT0 to UT) along the calculated drift trajectory. The
variations in parameters of precipitations, neutral atmosphere, and
thermospheric wind were also taken into account using the real
variations of the hourly values of geomagnetic activity indices.
These calculations were performed for the three above indicated
ionospheric storms. In this paper the initial conditions were
determined for the moments: 0000 UT on 20 January 2004, 0000
UT on 16 June 2003, and 0000 UT on 12 October 2003. The initial
profiles were calculated in the same way by integrating the model
equations at the 120-hour time interval beginning from UT0. Such
choice of the calculation interval provides reaching of the degree
of filling in the plasmosphere corresponding to quiet magnetic
conditions at middle and low latitudes
[Krinberg and Tashchilin, 1984].
[15] Tashchilin et al. [2002a]
performed a preliminary study of the
reaction of the midlatitude ionosphere to an intense geomagnetic
storm on the basis of the comparison of the Irkutsk Incoherent
Scatter (IS) Radar measurements and the numerical simulation
results. It was found that the negative phase of an ionospheric
storm (as has been noted earlier
( Danilov and Belik [1991],
Rishbeth [1998],
and others)) is formed mainly because of
variations in the thermosphere composition. Besides this, the
preliminary theoretical analysis made it possible to perform a
correction of the thermospheric parameters calculated according to
the MSIS 86 model to the conditions of the storms in question.
Since the changes in the composition (the
[O]/[N2] ratio)
obtained
from the MSIS 86 model were not able to reproduce the observed
behavior of the ionosphere during magnetic storm, these changes
were multiplied by a factors of 0.6 and 1.6 for summer and fall,
and winter, conditions respectively. These correcting factors were
used in the presented below model calculations of the ionosphere
reaction to the chosen geomagnetic storms.
|
|
Figure 4
|
|
|
Figure 5
|
|
|
Figure 6
|
[16] The results of the simulation of the ionospheric behavior during the
storms are presented in Figures 4,
5, and
6. Solid curve shows the
calculated values of the logarithm of the electron concentration in
the
F2 -layer maximum
( lg NmF2) for two high-latitude, two
midlatitude and one equatorial stations, circles correspond to the
measured values, and dashed curves denote the quiet level
calculated over several quiet days. The model reproduce well the
variations of the critical frequency during the storms: the values of
|dfoF2| calculated by the model vary from 4% to 50%. The
peculiarities of the modeling results for particular storms are as
follows.
[17] 1. For summer conditions a good agreement between the measured
and modeled values of
NmF2 in the daytime both at high and
middle latitudes is obtained. The main differences are found in the
evening and nighttime periods of local time at high
and low latitudes.
[18] 2. For winter conditions the calculated values of
NmF2 also
satisfactorily correspond to the daytime measurements. The
simulation results at subauroral station Yakutsk agree with
observations better then at other stations both in the daytime and at
night. The strongest differences are found in the evening and
nighttime LT hours at Irkutsk midlatitude station.
[19] 3. For equinox conditions we succeeded in correcting the
calculated and measured variations in the daytime at auroral station
Norilsk. In the evening hours when a break in the diurnal behavior
was observed, and at night when the ionization at these latitudes is
determined by the fluxes of precipitating particles, the model
values
NmF2 do not correspond to the observed values. In Yakutsk
a satisfactory coincidence between the simulation results and
observations is obtained for the entire period except the evening
hours on 14 October during the break of the diurnal behavior at the
first negative peak in the
Dst index. At the midlatitude stations
Irkutsk and Manzhouli, the simulated and measured variations in
NmF2 agree well enough. The evening hours on 15 October in the
recovery phase, when very low values of
foF2 were observed,
present an exception.
[20] The results of model simulations were obtained because of the
correction of the MSIS 86 thermospheric model, the correction
being of a different sign for summer (fall) and winter conditions. In
the first case the value of
R was decreased by a factor of 2, and in
the second case it was increased by a factor of about 1.5. Such
variations in the neutral composition of the thermosphere are able
to lead us to the fact that the ionospheric storms are negative in
summer and positive in winter.
|
|
Figure 7
|
[21] The disagreement of the simulation results and measurements in
the evening and night hours at high-latitude stations shows that a
correction to the conditions of the considered magnetic storms is
needed not only of thermospheric parameters, but the empirical
models of magnetospheric sources as well. The effects of the
correction of the models of precipitation and convection electric
field were considered analyzing the situation of formation of the
main ionospheric trough (MIT) on 14 October 2003 according to
the data of Yakutsk and Norilsk stations. To do this, calculations of
two versions of global distributions of the electron concentration
for the moments 0800, 1000, and 1200 UT, the results being
presented in Figure 7. Version I corresponds to the calculations
without corrections of empirical models of magnetospheric
sources, whereas version II show the results of the following
corrections for disturbed conditions: the zone of the auroral
precipitations and magnetospheric convection was widened
equatorward by 5o, the electric potential was increased by 30% of
the value obtained from the empirical model. The correctness of
these changes was discussed by
Fuller-Rowell et al. [1994].
Figures 7a and 7e show the values of
lg Ne at a height of 300 km in
the coordinates: geomagnetic colatitude-MLT for versions I and II
of the calculations, respectively. Figures 7b and 7d show isolines
of the electric potential of convection (taking into account the
corotation) and the energy flux of the precipitating electrons also
for versions I and II. The positions of Norilsk and Yakutsk stations
are shown by circles. One can see in Figures 7a and 7e that the
main ionospheric trough (MIT) in version II of calculations is
better pronounced: the trough depth is a factor of 2.8 and 3.8 for
0800 and 1200 UT, respectively, the length in MLT is from 16 to 4
hours, whereas in version I the depth of MIT is a factor of 1.8-3.1
and the length in MLT is 18-5 hours. Stations Yakutsk and Norilsk
are located at the edge of the trough at 1000 UT and 1200 UT (see
|
|
Figure 8
|
Figure 7e) for version II of the calculations. Figure 8 shows the
diurnal behavior of the logarithm of the electron concentration in
the
F2 -layer maximum
( lg NmF2) calculated using versions I and II
(curves 1 and 2, respectively). The circles show the measured
values of
foF2. One can see that according to the diurnal behavior
of
foF2 at Norilsk station, the equatorial wall of the trough was
observed at 0800 UT, whereas according to the simulation results
it appeared at
1100 UT and
1200 UT for versions II and I,
respectively.
Therefore the correction performed brought the
calculation results closer to the observational data, but appeared to
be insufficient to provide their complete agreement.
The calculations for Yakutsk subauroral station showed the absence of
MIT in version I and its presence in version II. However, the
calculated time of the MIT equatorward wall appearance poorly
agree with the observational data. Thus at high latitudes where
magnetospheric convection and precipitation of energetic electrons
play an important role, the use of empirical models for calculation
of the auroral ionization and electromagnetic drift velocity does
not make it possible to simulate the real structure of MIT. A simple
correction of empirical models of magnetospheric sources provide
no significant improvement, this fact indicating to a need of
development of methods of adaptation of empirical models to real
geomagnetic disturbances.
[22] The strongest differences between the calculations and
measurements are seen for the winter storm at middle latitudes
(Irkutsk) in the evening and nighttime. One can see from Figure 5
that for the winter storm at midlatitude station Irkutsk in the
evening and nighttime, the calculated values of the electron
concentration exceed considerably the measured values of. This
can happen because of two causes. It is known that the ionization
level in the vicinity of the
F2 -layer maximum after sunset is
controlled, first, by the plasma input from the outer ionosphere
and, second, by the change in the
F2 -layer height caused by the
meridional component of the thermospheric wind. In order to
estimate the input of each of these factors into the deviation of
calculations from the observations, we first performed calculations
without the wind. As a result, the agreement of calculations and
observations was considerably improved. This fact shows that the
wind velocities are overestimated both in the calculations and in
the empirical model HWM 90. Then we considered an assumption
that the disagreement is caused by the input of ions from the
plasmasphere, the input value depending on plasmasphere filling
in. The calculation results presented in Figure 5 correspond to the
situation when the plasmasphere over Irkutsk is almost full and is
able to support high values of
Ne after sunset during the storm,
whereas in real conditions the ionosphere is not completely full. In
order to check these assumptions, two versions of calculations
were performed for Irkutsk station without any correction of the thermosphere (Figure 9).
|
|
Figure 9
|
Version III corresponds to the conditions
of filled plasmasphere (curve 1). Version IV corresponds to the
conditions of unfilled plasmasphere when the model equations
were integrated at the time interval of 24 hours (curve 2). In the
case of filled plasmasphere, in the evening and nighttime the ion
fluxes from the plasmasphere to ionosphere (from
-0.7 
107 cm-2 s-1 to
- 1.6 107 cm-2 s-1 )
are observed even during the storm
(Figure 9 (bottom), curve 1). At the same time in the case of
unfilled plasmasphere, the fluxes are much weaker
(from
-0.2 107 cm-2 s-1
to
- 0.5 107 cm-2 s-1,
Figure 9, curve 2). During the
storm on 22 January, the flux is directed from the ionosphere into
plasmasphere
(0.2 107 cm-2 s-1 )
what is close to the real
conditions. This fact explains the better coincidence of the
calculated values of the electron concentration with observations in
the evening and night hours obtained for version IV at the recovery
phase of the storm (from 22 to 26 January 2004). On quiet day 21
January the best agreement between the calculated and observed
values of
Ne is obtained for version III. This is due to the fact that
in the period 18-21 January 2004 (4 days) corresponds to quiet
geomagnetic conditions (when the total value of the
Kp index did
not exceed 22) and the plasmasphere was filled, because the
characteristic time of filling in of a plasma tube with
L 
2 is three
days (for Irkutsk
L = 1.74 )
[Krinberg and Tashchilin, 1984].
Thus
one can conclude that modeling the response of the midlatitude
ionosphere to geomagnetic storms, one has to take into account the
time variations of the degree of filling in of the plasmasphere, that
is, to consider the ionosphere and plasmasphere as a united system.
Powered by TeXWeb (Win32, v.2.0).