CC BY
57
High-latitude version of the global numerical model of the Earth's upper atmosphere
A.A. Namgaladze12, O.V. Martynenko12, M.A. Volkov1'2, A.N. Namgaladze2, R.Yu. Yurik1,2
1Physics Chair of the Electromechanics Department of the MSTU 2
Polar Geophysical Institute of the Kola Science Centre of the RAS
Abstract. The global numerical model describing the thermosphere, ionosphere and protonosphere of the Earth as a single system has been modified for the polar upper atmosphere studies. The spatial and time resolution of the model has been significantly enhanced by the use of the variable latitudinal steps of numerical integration. The model is being developed to encompass modelling of the inner part of the magnetosphere confined by the closed geomagnetic field lines and the mesosphere. The results of the model calculations for the quiet magnetic conditions have been compared with the data of the empirical ionospheric and thermospheric models as well as with the EISCAT data and, in general, reasonable agreement between theoretical and empirical data has been found. The new high-latitude version of the model has been applied as well to the investigations of the disturbed behaviour of the Earth's upper atmosphere during geomagnetic substorms and storms and during disturbances in the cusp region. The physical mechanisms of the upper atmosphere responses to the solar wind and magnetospheric forcings have been understood by the use of the model in several case studies.
1. Introduction
The global numerical model of the Earth's upper atmosphere has been constructed at the Kaliningrad Observatory of IZMIRAN (Namgaladze et al, 1988, 1990, 1991, 1994) on the basis of the previous numerical models of the mid-latitude ionosphere (Namgaladze et al., 1977), equatorial ionosphere (Surotkin et al, 1979), protonosphere (Klimenko and Namgaladze, 1980) and thermosphere (Karpov et al., 1985). The model describes the thermosphere, ionosphere and protonosphere of the Earth as a single system by means of numerical integration of the corresponding time-dependent three-dimensional continuity, momentum and heat balance equations for neutral, ion and electron gases as well as the equation for the electric field potential. It is the main difference of this global model from many others (e.g., Fuller-Rowell and Rees, 1980, 1983; Dickinson et al., 1981, 1984; Fuller-Rowell et al, 1984, 1987, 1988; Roble et al., 1988b; Schunk, 1988; Sojka, 1989; Sojka and Schunk, 1988, 1989; Richmond et al., 1992; Roble and Ridley, 1994) that it calculates not only winds, gas densities and temperatures of the thermosphere and ionosphere, but electric fields both of thermospheric dynamo and magnetospheric origin and protonospheric parameters as well.
In last years this model was modified for the studies of the high-latitude phenomena at the Polar Geophysical Institute and Murmansk State Technical University (Namgaladze et al., 1995b, 1996b; Volkov et al, 1996a; Hall et al, 1997b). The spatial and time resolution of the model has been significantly enhanced by the use of the variable latitudinal steps of numerical integration (Namgaladze et al., 1995b, 1996b). The latitudinal steps can be taken as small as 1 degree or even less at high latitudes instead of 5-10 degrees used in previous versions of the model, and the longitudinal steps can be decreased from 15 to 4 degrees. A new MHD magnetospheric block has been incorporated in the model to calculate the zone 2 field-aligned currents instead of using them as input of the model (Volkov et al., 1996a). Some investigations of the high-latitude ionospheric and thermospheric variations at heights of the E and F regions of the ionosphere were performed using the high latitude version of the global model of the Earth's upper atmosphere both for quiet and disturbed conditions (Namgaladze et al., 1995a,b; 1996a,b,c; Volkov and Namgaladze, 1996a,b; Volkov et al, 1996a,b). As a next step of development of the model, the mesosphere has been included in consideration to use the model for interpretation of the mesosphere and lower thermosphere observation data (Hall et al., 1997b).
Very often, when studying the mesosphere and lower thermosphere variations, it is not necessary to have detail information concerning the upper ionosphere, protonosphere and magnetosphere behaviour. So, to economize the calculation time, the physical parameters of these height regions in the new version of the model can be "frozen" by keeping them as calculated globally, for example, for one day or one UT moment. It is important to have such a possibility because of large time scales of the processes in the mesosphere and lower thermosphere requiring the long time computer runs. Further, previously we used the empirical model of the
thermosphere MSIS-86 (Hedin, 1987) as initial and lower boundary conditions to calculate neutral gas temperature and density, but the lower limit of this model is 85 km and to extent the model to 80 km an extrapolation was used that was a rather rough approximation. Now, the MSISE-90 (Hedin, 1991) extending MSIS-86 to lower atmosphere is incorporated in the model to use it, first, as initial and lower boundary conditions (now lower boundary can be taken at any height between 60 and 80 km) and, second, for global calculations of the neutral gas temperature, density and composition at any height in parallels with (or instead of) self-consistent theoretical calculations of these parameters to compare the results obtained by the use of theoretical and empirical models of the neutral atmosphere. The possibility of using the empirical model atmosphere of Lubken and von Zahn (1990) is also envisaged.
Most of the equations solved in the high-latitude version of the global model has the same form as in previous versions of the model (excepting the magnetospheric block). As for the neutral atmosphere, the important difference is that now the barometric law for the molecular nitrogen density is used to calculate it above the turbopause level instead of calculating it as the difference between the total mass density and mass density of molecular and atomic oxygen because of great errors which arise when this difference is small. The turbopause level is determined by the height profile of the eddy diffusion coefficient, K, so we must also incorporate an estimate of this important parameter into the model. To do this, we employ the maximum value, Km, and its height, zm, as functions of season and latitude which we obtain from Danilov and Kalgin (1996), these being the newest available at the time of writing.
The aim of this paper is to give a review of the modern state of the high-latitude version of the global numerical model of the Earth's upper atmosphere including the description of the modelling equation system, coordinates and numerical grids used, initial and boundary conditions, inputs and outputs, as well as to demonstrate the capabilities of the model to reproduce numerically various situations in the Earth's upper atmosphere and to explain physically its behaviour in quiet and disturbed conditions. The great amounts of the results of the model calculations will be presented and discussed on the base of the published and unpublished investigations.
2. The structure of the model
Fig.1. Inputs of the model, main computation blocks and their outputs.
The main peculiarities of the new high-latitude version of the model are the following. The model considers the thermosphere, ionosphere and protonosphere of the Earth as a single system. It covers the height range from 60 km up to 15 Earth radii of geocentric distance and takes into account the offset between the geomagnetic and geographic axes of the Earth and consists of four main blocks (see Fig.1):
1) neutral atmosphere and lower ionosphere block which calculates neutral atmosphere temperature, mass density, neutral gas composition and winds as well as ion and electron temperature, molecular ion density and velocity at heights 60 to 520 km;
2) ionospheric F2 region and protonospheric block which calculates atomic ion O+ and H+ densities, velocities and temperatures as well as electron temperature at heights from 175 km to 15,RB of geocentric distance;
3) electric field block which calculates the electric field potential both of magnetospheric and thermospheric (dynamo) origin assuming that geomagnetic field lines are equipotential at heights above 175 km and
4) magnetospheric block which calculates magnetospheric plasma-sheet ion density, velocity, pressure and field-aligned currents at the same heights as in the second block.
In these blocks the corresponding hydrodynamical continuity, momentum and heat balance equations for the neutral, electron and ion gases as well as the equation for the electric field potential are all solved numerically by the use of the finite difference methods, using different coordinate systems and different spatial grids of numerical integration. The height steps of numerical integration are variable. They vary from 1-3 km at heights below 100 km to 30 km and more at heights above 400 km. Spherical geomagnetic coordinate system is used in the neutral atmosphere and lower ionosphere block and geomagnetic dipole coordinate system is used in other blocks. The exchange of information between the blocks is carried out at every time step of the numerical integration of the modelling equations.
2.1. Neutral atmosphere and lower ionosphere block
2.1.1. Neutral atmosphere
In the neutral atmosphere section of the block, the neutral gas temperature Tn , mass density p, thermospheric wind velocity vector V and number densities nn of the main neutral gas components N2 , O2 and O are calculated for the height range from 60 to 520 km using the spherical geomagnetic coordinate system. We can perform our calculations either by solving the full system of hydrodynamical equations for the neutral gas or by using the empirical thermospheric models such as MSISE-90 (Hedin, 1991) to calculate the temperature and number densities of the main neutral gas components. The three-dimensional thermospheric circulation is calculated from the solution of the momentum and continuity equations in all cases.
The following system of the continuity, momentum and energy balance equations for the neutral gases
is solved in the fully self-consistent variant:
dnn/dt + V[ nn(V+VJn) ] = Qn - Ln, (1)
p [ dV / dt + (V,V) V + 2flxV ]hor = - (VpW -1 vm nn (V-V, \or + r, (V2V )hor, (2)
i n
pg = - dp /dr , (3)
dp/dt + V(pV ) = 0 , (4)
p = I nn Mn , (5)
n
p = Z nnkT, (6)
n
pcv [ dT/dt + (V,V)T ] + p VV = V(2nVT ) + PnQUV + PnQJ + PnQC - PnL . (7)
In these equations nn is the concentration of the n-th neutral component; V is the neutral wind velocity vector; Vdn is the diffusion velocity vector which has only a vertical component equal to sum of molecular and eddy diffusion velocities; Qn, Ln are the production and loss rates of the n-th neutral component taking into account dissociation of O2 and reactions of recombination for O and O2 ; the index "hor" stands for horizontal vector components; p , p are the mean mass density and pressure of the neutral gas; Q is the Earth's angular velocity vector; pni , vni are the reduced mass and frequency of collision between the neutral and ion components
of the atmosphere; Vt is the ion velocity vector; ij is the coefficient of viscosity; g is the sum of gravity and centrifugal accelerations; r is a geocentric distance; mn is the mass of the n-th neutral component; k is the Boltzmann's constant; T is the temperature of the neutral gas; cv is the specific heat at constant volume; X„ is the thermal conductivity coefficient of the neutral gas; P„quv , P„qj , P„qc are the rates of heating of the neutral gas by UV and EUV solar radiation, Joule heating and heating by precipitating energetic particles; P„l is the rate of heat loss of the neutral gas due to radiation. The detailed expressions for all coefficients and terms of the equations (1)-(7) and their form in a spherical geomagnetic coordinate system can be found in the paper of Namgaladze et al. (1988) and in the book of Brunelli and Namgaladze (1988).
We use the equations (1) to compute the O and O2 concentrations, the total mass density p is calculated from the hydrostatic equilibrium equation (3). As for the N2 concentration, the barometric law for the molecular nitrogen density is used to calculate it above the turbopause level instead of calculating it as difference between the total mass density and mass density of molecular and atomic oxygen (equation (5)) because of great errors which arise when this difference is small. The equations (2) are used to calculate the horizontal meridional (Vx) and zonal (Vy) components of the neutral wind velocity. To obtain the vertical wind velocity we use the continuity equation (4) because the vertical component of the momentum equation for the neutral gas is reduced to (3) which does not contain the vertical component of the neutral wind velocity. At last, the heat balance equation (7) is used to compute the neutral temperatureT.
The system of equations (1)-(7) is completed by initial and boundary conditions. At the upper boundary (h = 520 km) we assume that
dV / dr = dT / dr = 0 ,
and all neutral components are in the diffusion equilibrium there. At the lower boundary (h = 60 km) the wind velocity is taken according to the geostrophical approximation (or tides can be taken) and the temperature and concentrations of the neutral components are taken from the MSISE-90 empirical thermospheric model (Hedin, 1991). We also use this model to obtain the initial spatial distribution of the neutral concentrations and temperature. As for wind, we use the zero velocity as the initial condition. To obtain the stationary solution we need to integrate the modelling equation system until the results of integration do not differ under continuating of integration. Usually to reach it, several days (geophysical but not computing time) of integration are required.
2.1.2. Lower ionosphere (D, E and F1 ionospheric regions)
In this section the following parameters of D, E and F1 ionospheric regions are calculated: the total concentration of the molecular ions n(XY+) = n(NO+) + n(O2+) + n(N2+), ion and electron temperatures Tt and Te and molecular ion velocity V(XY+) for the height range from 60 km to 175 km (for Tt and Te) or 520 km (for
n(XY+) and V(XY+)). The following equations are solved:
d n(XY+) / dt = Q(XY+) - L(XY+) , (8)
(3 n,k/2 ) dT,/dt = PQ + PT + P,T„ , (9)
( 3 „e k/2 ) dTe/dt = PeQP + PeQc + PeT + PeT„ , (10)
nmg - V(n,kT,) - Ea„ v,„n (V - V) + en, (E +V,xB) = 0, (11)
„
V ( ne kTe ) + ene (E +VexB) = 0 , (12)
ne = nt = n (XY+) . (13)
In these equations Q(XY+), L(XY+) are the production and loss rates of the molecular ions taking into account ionization by solar EUV direct and scattered radiation, ionization by precipitating electrons, ion-molecular reactions and dissociative recombination; Pt q J is the rate of the Joule heating of the ion gas; PiTe, PiT„ are the rates of the heat exchange between ion and electron and neutral gases; ne is the electron concentration; PeQ , PeQc are the rates of heating of the electron gas by photoelectrons and by precipitating magnetospheric electrons; PeT' = - PiTe; PeT„ is the rate of the elastic and inelastic exchange of heat between electron and neutral gases; m, is ion mass; g is the vector of gravity acceleration; v,^, = v„n„; e is the electron charge; E, B are the electric and magnetic fields. More information about terms of the equations (8)-(12) can be found in Namgaladze et al. (1988) and Brunelli and Namgaladze (1988).
As one can see from (8)-(10) we neglect the heat and particle transport processes in the D, E and F1 ionospheric regions due to dominating of the photochemical and local heating and heat exchange processes in these ionospheric regions. The equations (11)-(12) are used to obtain components of the ion velocity vector which are needed to calculate thermospheric winds and temperature with taking into account ion drag and Joule heating.
2.2. Ionospheric F2 region and protonosphere block
In this block, parameters of the ionospheric F2 region and protonosphere are calculated, namely: the atomic oxygen and hydrogen ion number densities n(O+) and n(H+) as well as ion and electron temperatures Tt and Te and ion velocities V(O+) and V(H+) for the height range from 175 km to the radial distance of 15RE. We consider that all charged components of the upper atmosphere are magnetized fully at the heights above 175 km because vin « a>i at these heights where coi is ion gyrofrequency so the geomagnetic field has a very strong effect on the behaviour of the ion and electron gases. That is why we use the magnetic dipole coordinate system in this block and perform the integration of modelling equations along dipole geomagnetic field lines simultaneously taking into account electromagnetic plasma drift perpendicular to geomagnetic field lines. The following continuity, momentum and heat balance equations are solved in this block:
Dn,/Dt + Vpar(n,V,par) = Q- Lt - nt Vppe Ver, (14)
2 mn{QxVi)par = m,ngpar - Vpar(n,kT, ) - (n/n) Vpar(nekTe) -
- Z MmVnn, (V/ar-Vpar) -1 v-n, (V/ar-V/ar), (15)
n i
V/er = Veper = E x B/B2 , (16)
Veper = I n Vtper/ne, (17)
i
(3 n,k/2 ) (DTi/Dt + V/ar Vpa% ) + ( n,kTi ) VV1par - Vpar(^IVparTI ) =
= P,qj + PiTe + PT + P,t" , (18)
(3 nek/2 ) (DTe/Dt + VepaVparTe ) + ( nekTe ) Wepar - V^ *;V parTe ) =
= PeQp + PeQC + Pet' + PeTJ + PeT" . (19)
In these equations the subscripts i, j and e refer to ions O+ and H+, and electrons, respectively. The symbols par and per refer to the directions parallel and perpendicular to the geomagnetic field. The operator D/Dt = 8/8t +(Vper,V) gives the Langrangian temporal derivatives along the electromagnetic drift trajectory determined by the equation (16). Q, Li are the production and loss rates of O+ and H+ions which take account of photo- and corpuscular ionization, ion-molecular reactions between O+ and O2 and N2, charge exchange processes between O+ and H, and between H+ and O; gpar is a geomagnetic field aligned component of the sum of gravity and centrifugal accelerations; Pqj is the rate of the Joule heating of the ion gas; PiTe, P.T, Pt" are the ion heat exchange rates; PeQp, PeQc are the rates of local and non-local heating of the electron gas by photoelectrons and by precipitating magnetospheric electrons; PeT', PeT, PeTn are the electron heat exchange rates.
For the densities of neutral hydrogen we use the barometric law with a boundary condition at 500 km altitude from the neutral atmosphere model of Jacchia (1977). The detailed description of the terms of the equations (14)-(19) has been given by Namgaladze et al. (1988) and by Brunelli and Namgaladze (1988).
The integration of the equations (14)-(19) is done along dipole geomagnetic field lines. The boundary conditions are given near the bases of the field lines in the northern and southern hemispheres at height of 175 km. The atomic ion concentrations at this boundary are obtained from photochemical equilibrium conditions. The values of the ion and electron temperatures at this boundary are calculated from the equations (9)-(10) of heat balance. We assume that geomagnetic field lines with L > 15 (L parameter of Mcllwain) are open and ion concentrations and heat fluxes are set equal to zero at r = 15 RE.
Zero ion concentrations, and ion and electron temperatures equal to the temperature of the neutral gas or the results of preceding calculations of the modelled parameters may be chosen as initial conditions.
2.3. Electric field computation block
The next block of our model is the electric field computation block. The equation for the potential (p of the electric field E = -V^ is solved numerically in this block taking into account the dynamo-action of the thermospheric winds:
V[<r (Vp- VxB ) -jm ] = 0 , (20)
where cr is the ionospheric conductivity tensor and jm is the magnetospheric current density. After integrating the equation (20) over the height of the current-carrying layer with neglect of the height dependence of the electric field components in this layer, the problem to define the electric potential becomes two-dimensional and is solved by an iterative technique in the geomagnetic coordinate system. The ionospheric conductivities needed to solve (20) are calculated using the standard formulae with values of parameters of the ionosphere and thermosphere from the thermospheric and ionospheric-protonospheric blocks of the model.
2.4. Magnetospheric block
The magnetospheric block (Volkov et al., 1996a) contains the following equations for the magnetospheric plasma:
Sn,/ Dt + V( nV ) = 0 , (21)
n,e( E + V,xB) = Vp , (22)
d (pV ) / d t = 0, y = 5/3 , (23)
l
jn = ez[VV,Vp,] /B , V = B iB4 dz , (24)
0
where jn is the field-aligned current density, E and B are the electric and geomagnetic fields, ez is the field-aligned (along B) unit vector, V is a half-volume of the geomagnetic field tube, z is a distance along the geomagnetic field line, l is the value of z at the top of the geomagnetic field line, pi is the magnetospheric ion gas pressure considered isotropic and constant along the geomagnetic field line, ni is the magnetospheric ion concentration, Vi is the magnetospheric ion drift velocity, e is the electron charge. The magnetospheric electrons are considered cold and their pressure is neglected in comparison with that of the magnetospheric ions. The geomagnetic field is considered as a dipole one at latitudes equatorward from the polar cap boundary (0 =75°) and having the field lines opened inside the polar caps.
3. Inputs of the model
The input parameters of the model are 1) solar UV and EUV spectra; 2) precipitating particle fluxes; 3) field-aligned currents connecting the ionosphere with the magnetosphere and/or the electric field potential distribution at the polar cap boundaries. For the solar UV and EUV fluxes and their dependencies on solar activity we use the data from Ivanov-Kholodny and Nusinov (1987). Intensities of night sky scattered radiation are chosen equal to 5kR for X = 121.6 nm and 5R for each of other emission lines ( XX = 102.6 nm, 58.4 nm, 30.4 nm).
Spatial distributions of the precipitating electron fluxes are taken at the upper boundary of the thermosphere (h = 520 km) in a simple form:
I( 0,A,E ) = Im (E) exp [- (0- 0m(E))2/(A0(E))2 - (A - Am(E))2/(AA(E))2 ] , (25)
®m = ( ®md+ 0m„) / 2 + (cos^ ) (0md- 0m„) / 2 , (26)
where 0, A are geomagnetic latitude and longitude, A = 0 corresponds to the midday magnetic meridian, Im(E) is the maximum intensity of the precipitating electron flux, E is the energy of the precipitating electrons, 0md, 0m„ are the geomagnetic latitudes of the maximum precipitation at the midday and midnight magnetic meridians. All precipitation parameters (Im, 0md, 0m„, A0, Am, A4)can vary depending on geophysical conditions.
The magnetospheric sources of the electric field are field-aligned currents in zones 1 and 2 and in the cusp region (Iijima and Potemra, 1976). The first zone of field-aligned currents, flowing into the ionosphere on the dawn side and out on the dusk one, is at the polar cap boundary (±75° magnetic latitude). The second zone of the field-aligned currents flowing opposite to the zone 1 currents is located equatorward from the zone 1.
Spatial distributions of the current intensities in all zones vary depending on geophysical conditions. If we do not use the magnetospheric block, we take either all FAC systems as inputs of the model or we take the electric potential distribution at the polar cap boundary and the cusp and zone 2 FACs as inputs. When we use the magnetospheric block, we take the electric potential distribution at the polar cap boundary and the cusp FACs as inputs. In this case, the zone 2 FACs are calculated in the magnetospheric block, and the precipitating electron fluxes are taken in proportion to the magnetospheric plasma-sheet ion number density, being normalized to the empirical data by Hardy et al. (1985).
4. Numerical grids
In the neutral atmosphere and lower ionosphere block the equations (1)-(13) are solved by finite-difference numerical methods in a spherical geomagnetic coordinate system. In the preceding calculations (Namgaladze et al., 1988, 1990, 1991, 1994) the steps of the numerical integration were 10° in geomagnetic latitude, 15° in geomagnetic longitude, variable in altitude (3 km near the low boundary (h = 80 km), 5 km near h = 115 km, 15 km near h = 220 km, 25 km near h = 330 km, 40 km near h = 500 km, 100 km near h = 1000 km, etc., having 30 levels in the altitude range from 80 to 520 km) and 5 min in time.
The neutral atmosphere parameters calculated in the spherical geomagnetic coordinate system are interpolated to the nodes of the finite-difference magnetic dipole coordinate grid to calculate the parameters of the ionospheric F2 region and protonosphere. Earlier we used the following steps of this grid: 5° or 8° in latitude near the bases of geomagnetic field lines at the 175 km altitude, 15° in longitude and variable in a distance along geomagnetic field line. The number of the nodes of the grid along B varies from 9 on the lowest equatorial field line to maximum value 140 on the field line with L = 15. In turn, the necessary parameters of the ion and electron gases are put into the neutral atmosphere block from the ionospheric F2 region and protonosphere block which uses electric field from the electric field computation block. In this latter block the two-dimensional grid was used with the 5° latitude and 15° longitude steps.
Now, the following main alterations have been done (Namgaladze et al., 1995b, 1996b). In all blocks we have replaced the constant latitudinal steps of the numerical integration of the modelling equations by the differences - &k which are not constant but depend on latitude, i.e. they depend on numbers of i-th and k-th latitudinal nodes of the grid which, in turn, depend on latitude. So, in each block the kind of the grid is determined by its own law of the dependence of the latitudinal step on latitude.
—i-1-1-1-1-1-1-1-1
0 10 20 30 40 50 60 70 80 90 Magnetic latitude, deg.
Fig.2. The dependencies of the latitudinal integration steps on geomagnetic latitude for variable grids used in the neutral atmosphere and lower ionosphere block (dashed curve), and in the ionospheric F2 region and protonosphere and electric field computation blocks (solid curve).
To investigate the effects of the choice of the latitudinal integration step on the results of numerical solution of the modelling equations we have run the following test calculations (Namgaladze et al., 1996b). The system of the modelling equations (1)-(20) has been solved numerically. Started with the same initial conditions we have performed three variants of the modelling calculations of all parameters. These three variants differ only in latitudinal steps of integration of the modelling equations. The following grids have been used:
1) So-called "rough" grid. It has the constant latitudinal steps of 10° in the neutral atmosphere and lower ionosphere block, and 5° in the ionospheric F2 region and protonosphere block and in the electric field computation block.
2) So-called "fine"grid. It has the constant latitudinal step of 2° in all blocks of the model.
3) Variable grid. It has the variable latitudinal steps which vary from 10° at the magnetic equator to 2° at the auroral zones in the neutral atmosphere and lower ionosphere block, and from 5° at the magnetic equator to 2° at the auroral zones in the F2 region and protonosphere and electric field computation block. The
10
8
6
4
2
dependencies of the latitudinal integration steps on latitude used in this variant of calculations are shown in Fig.2. As we can see from this figure the minimum of the latitudinal step is at 70° geomagnetic latitude.
As for other integration steps they are the same in all variants of the calculations (15° in longitude and variable in altitude as mentioned above, and 2 min in time). Figs.3, 4 show the nodes of the height-latitude grids used in these calculations.
variable grid
I«
fine grid
Ittllltlllllltlttltlttlllltllllllllllllll
rough grid
variable grid
fine grid
rough grid
Magnetic latitude (deg.)
Fig.3. Nodes of numerical integration grids for the thermosphere and lower ionosphere parameters at heights 80-155 km (bottom) and for the F2 region parameters at heights 175650 km (top).
fine
grid 10
variable ■ grid 10
llÊë:
rough grid 10
Ml
Wi - y '
o | pFit'i'f-i
1 2 0 Geocentric distance, RE
Fig.4. Nodes of numerical integration grids for the protonosphere parameters at geocentric distances 1-2.5 RE (left) and 1-15 Re (right).
The calculations have been made for the time interval 0000-0624 UT on 24 March 1987 using the same input parameters as in the paper by Namgaladze et al. (1996a) for quiet conditions. The results of these calculations show (Namgaladze et al., 1996a), that the use of the "rough" grid can lead to significant errors in the calculated ionospheric variations at the polar latitudes but at middle latitudes the numerical results obtained by the use of the "rough" and "fine" grids are rather similar and the use of the "rough" grid is quite justified for the midlatitude phenomena studies. As concerns the variable grid which is close to the "fine" grid at latitudes between 60 and 80 and to the "rough" grid at latitudes equatorward from 30 as Fig.2 shows, the use of this variable grid gives the results close to those obtained by the use of the "fine" grid not only at the polar latitudes but over the whole globe for most of the calculated parameters. It is important that the use of the variable grid economizes the computer resources without significant damage to the accuracy of the computations and permits us to use personal computers with RAM 8-16 Mb for the modelling calculations. 5. Comparison with empirical models
It is interesting now to compare the results of our modelling calculations made by the use of the variable latitudinal integration grid with the data of empirical thermospheric and ionospheric models. Such a comparison has been done with the data of the MSIS-86 thermospheric model (Hedin, 1987) and empirical ionospheric model by Chusovitin et al. (1987) and the results are shown in Figs.5-8.
The calculations have been made for the same quiet conditions as in the paper Namgaladze et al. (1996b), but the input data for the field-aligned currents and precipitating electron fluxes have been corrected to improve general agreement with the empirical data. We have taken the field-aligned current zone 1 at 76°
15 -
600
2
450
300
5
150
0
2
5
t 125
0
15
2
75
5
75
0
0
75
latitude with the maximum current density 0.16 |iA/m2 and zone 2 at 72° latitude with the maximum field-aligned current density 0.06 |iA/m2. With such field-aligned currents the electric field potential drop across the polar cap is 21 kV.
As for precipitating electrons we divided them accordingly to Hardy and Gussenhoven (1985) into two groupes with the characteristic energies 3 keV (hot electrons) and 0.2 keV (cold electrons). The precipitation parameters for them in the expressions (25), (26) have been taken as following. For 3 keV electrons &md = 78°, 0mn = 73°, A0 = 2.5°, Am = 187°, AA = 360°, Im = 4.2x108cm-2s-1. For 0.2 keV electrons ®md = 78°, 0mn = 68°, A0= 2.5°, I = I1 +12, Am1 = 303°, AA1 = 90°, Im1 = 1.9x109cm-2s-1, Am2 = 125°, AA2 = 35°, Im2 = 3x109cm-2 s-1. These parameters characterize the electron precipitations in the auroral zone. Besides, we add electron precipitations over the all polar cap with the maximum intensity 108cm-2 s-1 for 3 keV electrons and 109cm-2 s-1 for 0.2 keV electrons over the magnetic pole and A@ = 15°.
The calculated height-latitude variations of some modelled ionospheric and thermospheric parameters (electron and neutral atomic oxygen concentrations, electron and neutral temperatures) at the northern hemisphere along the midnight and midday magnetic meridians at 2400 UT are shown in Figs.5-8 (bottom panels) together with the corresponding data of the empirical ionospheric and thermospheric models of Chusovitin et al. (1987) and Hedin (1987) (top panels). Fig.9 shows the calculated local time variations of the meridional (positive northward) and zonal (positive eastward) thermospheric wind velocity at the height 319 km along the 70°N geographic latitude for various UT together with the empirical seasonal average HWM90 meridional and zonal winds for the same latitude and solar activity (Hedin et o/., 1991).
Fig.5. The calculated electron concentration (lg Ne) at heights 100-520 km over the northern geomagnetic pole along the midday (left parts of the plots) and midnight (right parts of the plots) magnetic meridians at 2400 UT on 24 March 1987 (bottom) and corresponding data of the empirical ionospheric model of Chusovitin et al. (1987) (top).
ELECTRON TEMPERATURE, K
400 1000 1600 2200 2800
Magnetic latitude, deg.
Fig.6. The same as in Fig.5, but for the electron temperature.
LG ATOMIC OXYGEN DENSITY, m 3
12.00 13.30 1 4.60 15.90 1 7.20
100-1-'-■-1-■-'-1-■-■-1-'-■-1-■-'-1-■-'--
45 60 75 90 75 60 45
Magnetic latitude, deg.
Fig.7. The calculated neutral atomic oxygen concentration at heights 100-520 km over the northern geomagnetic pole along the midday (left parts of the plots) and midnight (right parts of the plots) magnetic meridians at 2400 UT on 24 March 1987 (bottom) and corresponding data of the empirical thermospheric MSIS-86 model of Hedin (1987) (top).
NEUTRAL TEMPERATURE, K
160 360 560 760 960
500400-
Fig.9. The calculated meridional (positive northward) and zonal (positive eastward) thermospheric winds at the height 319 km along the 70oN geographic latitude together with seasonal average HWM90 meridional and zonal winds for the same latitude and solar activity (Hedin et al, 1991).
As one can see in these figures, the general agreement between the calculated and empirical data is rather good although some details are different. The empirical electron concentration and temperature data are smoother than theoretical ones and do not reveal such noticeable ionospheric F2 region trough and related electron temperature enhancement as in the theoretically calculated data. Perhaps it is due to variability of these phenomena connected with the electric field and precipitating particle flux variations. 6. Numerical modelling of the behaviour of the Earth's upper atmosphere during substorms
6.1. Ionospheric disturbance over EISCAT on 25 March 1987
The aim of this section of the paper is to demonstrate the capability of the global numerical model to describe quantitatively the disturbed behaviour of the polar ionosphere over EISCAT and at the neighbouring areas during a specific event, and to investigate by the use of this model how the observed decrease of the F2 region electron concentration is connected with the main ionospheric trough dynamics, and what are the physical mechanisms of the observed ionospheric variations (Namgaladze et al., 1996a).
The event of 24-25 March 1987 considered by us for the modelling was described by Collis and Haggstrom (1989, 1991). They presented the results of the EISCAT observations obtained with high temporal resolution during this period when the quiet day of 24 March was followed by the disturbed day of 25 March with SC starting about 1540 UT. The main features of these data are the following.
The quiet day (24 March 1987) reveals rather regular, solar controlled ionospheric and geomagnetic variations, but the next day the geomagnetic records from Kiruna show a positive geomagnetic bay in the magnetic X component starting at 1543 UT and the EISCAT measurements show simultaneous increases of the northward electric field (E n), F2 region ion (T) and electron (Te) temperatures, and a sharp decrease of the F2 region electron concentration (Ne). All these disturbances reach their maxima at about 1700 UT and then the geomagnetic and ionospheric parameters return to their quiet levels for two or three hours after 1700 UT. The E region electron concentration reveals an increase starting later than the other disturbances, at 1630 UT, and remains increased untill at least 2230 UT. A very similar behaviour of the ionosphere over EISCAT was observed on 21 October 1987 (Collis and Haggstrom, 1989), so we may consider it as rather typical for substorm conditions.
To simulate numerically the behaviour of the ionosphere over EISCAT observed on 24-25 March 1987 we selected the model input data for the auroral precipitating particle fluxes and the field-aligned currents in order to have an acceptable agreement in the first place between calculated and observed variations of the electric field and electron concentration in the E region. Then the other calculated and observed parameters such
as ion and electron temperatures, and electron concentration in the F2 region, were compared and the differences between the results of the calculations and observations were analyzed and discussed.
We present here, in this section the numerical results obtained by Namgaladze et al. (1996a) using the MSIS-86 thermospheric model (Hedin, 1987). The fully self-consistent solutions are presented in the sections 6.2 and 6.3. In the simulation we have used so-called "fine" latitudinal grid. The steps of the numerical integration are 2° in geomagnetic latitude, 15° in geomagnetic longitude and 5 min in time.
For the quiet conditions of 24 March 1987 we have taken the following magnetospheric parameters as inputs. For the auroral electrons having the exponential spectrum with the characteristic energy of 5 keV, 0m = 73°, A0 = 4°, Am corresponds to the midnight geomagnetic longitude, AAO= 48°, Im = 2x108cm-2s-1. For the soft 0.2 keV electrons 0m = 74°, A0O= 5°, AAE^- » (circle zone), Im = 4x108cm-2s-1.
The field-aligned currents of the zone 1 flowing into the ionosphere on the dawn side and flowing out on the dusk side (Iijima and Potemra, 1976) have been taken with the maximum density of 0.04 |iA/m2 at 76° geomagnetic latitude and the currents of the zone 2 flowing opposite to the zone 1 currents, have been taken with the maximum density of 0.015 |iA/m2 at 70° geomagnetic latitude.
For the disturbed day of 25 March 1987, during the time interval from 1540 UT to 1630 UT the precipitating electron fluxes are the same as on the previous quiet day of 24 March, but the maximum field-aligned current densities increase linearly from their quiet values at 1540 UT up to 1.04 |iA/m2 for zone 1 and 0.24 |iA/m2 for zone 2 (which is shifted to 66° geomagnetic latitude) at 1630 UT. During the time interval from 1630 UT to 1700 UT the maximum field-aligned current densities increase also linearly but more quickly up to 2.24 |iA/m2 at zone 1 and 1.26 |iA/m2 at zone 2 at 1700 UT. Besides, we have introduced an additional system of substorm field-aligned currents. They flow into the ionosphere along 76° geomagnetic latitude in the interval of 2100 - 0500 MLT and flow out of the ionosphere along 70° geomagnetic latitude in the interval of 1800 -2400 MLT with the maximum density of 2.2 |iA/m2 at 1700 UT in this sector. The total current flowing out is equal to the total current flowing into the ionosphere. Similar currents were observed by Lopez et al. (1991) in the midnight sector during the expansion phase of the substorm.
The maximum intensity Im of the 5 keV auroral electron fluxes increases linearly from the quiet level at 1630 UT to 34x108 cm-2s-1 at 1700 UT. Besides, the position Am of the maximum is shifted by 20° eastward from midnight and the longitudinal extension AA of the precipitation region increases from 48° to 110°. The maximum intensity of the soft 0.2 keV precipitating electron fluxes is not changed but their latitudinal width A0 decreases from 5° to 3° (Makita et al., 1985).
After reaching the peak intensity at 1700 UT the disturbances of the field-aligned currents return smoothly to their quiet levels by 1900 UT, while those of the precipitating electron fluxes decrease more slowly.
UT
Fig.10. The observed by EISCAT (dashed curves labelled by the letter E) and modelled (solid curves labelled by the letter M) variations of the northward electric field (the bottom panel), E region electron concentration at 111 km altitude, F2 region ion and electron temperature and electron concentration at 279 km altitude for the quiet day of 24 March (labelled by 24) and disturbed day of 25 March 1987 (labelled by 25).
In Fig.10 the modelled (solid curves labelled by the letter M) and observed by EISCAT (dashed curves labelled by the letter E) variations of the northward electric field, E region electron concentration at 111 km altitude and F2 region ion and electron temperatures and electron concentration at 279 km altitude are shown for the quiet (labelled by 24) and disturbed (labelled by 25) days of 24 and 25 March 1987. We can see that the agreement between the observed and calculated variations of the northward electric field, E region electron concentration and F2 region ion temperature is not bad but it should be noted that the calculated En is slightly overestimated while the calculated T is underestimated in comparison with the observations. The observed F2 region electron temperature variations are more intensive and irregular than the calculated ones.
1540 UT 1700 UT
ELECTRON DENSITY, 1011 m":
ION TEMPERATURE , K ION TEMPERATURE , K
ELECTRIC POTENTIAL , kV ELECTRIC POTENTIAL , kV
Fig.11. The calculated F2 region electron concentration (the top panel), ion temperature (the middle panel) at 279 km altitude, and electric field potential (the bottom panel) in the polar geomagnetic coordinate system for the beginning of the disturbance at 1540 UT on 25 March 1987 (the left plots) and for the time of the maximum disturbance at 1700 UT on 25 March 1987 (the right plots). The position of EISCAT is marked by the symbol +.
The calculated and observed variations of the F2 region electron concentration at 279 km altitude are rather similar. But the calculated variation for the quiet day is underestimated while the calculated drop of the electron concentration after 1600 UT on the disturbed day of 25 March 1987 is not as sharp as the observed one. To find out how this drop is related to the main ionospheric trough dynamics we can look at the calculated spatial distributions of the F2 region electron concentration at 279 km altitude shown in Fig.11 (the top panel) in the polar geomagnetic coordinate system for the beginning of the disturbance at 1540 UT on 25 March (the left plot) and for the time of the maximum disturbance at 1700 UT on 25 March (the right plot). Here we can see that during the disturbance the calculated trough moves equatorward in the midnight sector but its edges move westward and eastward as well, so that EISCAT (0 = 66.9° N, A= 117.2°) lies in the trough after 1600 UT on
the disturbed day. These west and east edges of the trough are related to the so-called "hot spots" in the ion temperature distribution, as can be seen from comparison of the top panel with the middle one of Fig.11, where the calculated spatial distributions of the F2 region ion temperature are shown for the quiet and disturbed conditions.
The calculated spatial distributions of the electric field potential for the quiet and disturbed conditions are shown in Fig.11 (the bottom panel). We can see that the potential drop across the north polar cap increases from approximately 10 kV at the beginning of the disturbance to about 90 kV at 1700 UT. Besides, the Harang discontinuity appears due to the additional field-aligned currents.
Unfortunately, we had no possibility to compare our model calculation results and EISCAT data at the altitudes other than 111 and 279 km. That is why in Fig. 12 we showonly calculated height profiles of the modelled parameters (the ion concentrations, ion and electron temperatures, field-aligned ion fluxes, neutral wind velocities) for the same UT moments as in Fig.11 (the time of the beginning of the disturbance and the time of the maximum of the disturbance).
550 450 350 250 150
-1-1-1-1-1
9.0 9.5 10.0 10.5 11.0 11.5 Log Ni ( m-3 )
1 1 1111 1 1111 1 1111 1 1111 1 I
0 500 1000 1500 2000 2500 Temperature, K
550
450
m
350-
(-
ig
ei I 250
150-
50
550 450 350 250 150
I 1 I 1 I 1 I 1 I 1 I 1 I 1 I 1
1 2 3 4 5 6 7 8 Ion flux, 1012 m-2 s-1
TH
9 10
T
-50 0 50 100 Wind velocity, m s-1
Fig.12. The calculated height profiles of the ion concentrations (the top plot in the left panel), ion and electron temperatures (the top plot in the right panel), field-aligned ion fluxes (the bottom plot in the left panel), and zonal (VnE , positive eastwards) and meridional (VnS, positive southwards) neutral wind velocities (the bottom plot in the right panel) for the time of the beginning of the disturbance (1540 UT, dashed curves) and the time of the maximum of the disturbance (1700 UT, solid curves).
We can see in Fig.12 (the top plot in the left panel) that ion composition calculations using the neutral composition from the thermospheric MSIS-86 model show a significant increase of the molecular ion abundance at heights below 200 km during the disturbance. But near the maximum of the F2-layer the O+-ions remain dominant. Their relative content decreases, but not so much as Haggstrom and Collis (1990) have estimated for this case from the analysis of the ion temperature results. The shape of the O+-ion density height profile does not change, in agreement with the results of Collis and Haggstrom (1991) who reported that altitude profiles of electron density from EISCAT show that the whole F region became depleted, i.e. this was not an effect of vertical redistribution of plasma above the radar.
Fig. 12 (the top plot in the right panel) shows the calculated height profiles of the ion and electron temperature. They reveal a significant increase of Ti at the heights from 150 to 450 km during the disturbance and practically no change of Te below 350 km during this period. At the top of the F2 layer both Tt and Te decrease so the shape of their height profiles changes and it may be a cause of an increase of the O+-ion upflow which appears at these heights as seen in Fig.12 (the bottom plot in the left panel) where the calculated height profiles of the field-aligned (along the geomagnetic field line) ion fluxes are shown.
50
50
The calculated height profiles of the meridional (positive southwards) and zonal (positive eastwards) neutral wind velocities are shown in Fig.12 (the bottom plot in the right panel). As we can see here, the zonal neutral wind changes more drastically (from +50 m/s to - 80 m/s at heights near the maximum of the F2 layer) during the disturbance due to the ion drag than the meridional one. This result differs from the conclusion of Haggstrom and Collis (1990) who considered that the zonal wind velocity remains stable and has the constant value of +50 m/s during the disturbance, while the zonal ion drift velocity reaches the values more than 1200 m/s.
Thus, as Fig.10 shows, the agreement between the calculated and observed ionospheric variations over EISCAT during the disturbance of 25 March 1987 may be considered as being satisfactory in general, although some details are different. We consider that the most important differences are those connected with the F2 region ion temperature variation. While the calculated electric field variation is rather close to the observed one and even slightly exceeds it, the calculated ion temperature variation at 279 km altitude is smaller (by about 150300 K near the maximum of the disturbance) than the observed one. In turn, this leads to a less sharp drop of the calculated F2 region electron concentration after 1600 UT in comparison with the observations.
Our tentative calculations show that when we take the disturbed ion temperature variation more close to that observed we obtain a calculated disturbed F2 region electron concentration which is closer to that observed as well. This means that the observed drop of the F2 region electron concentration is really connected with the increase of the ion temperature due to Joule heating and caused by the corresponding increase of the O+- ion loss rate (Schunk et al., 1976; Schunk and Sojka, 1982; Evans et al., 1983).
Joule heating, which is proportional to (Vi -Vn)2, where Vi and Vn are the ion and neutral gas velocity vectors, respectively, is underestimated in our calculations perhaps due to the use of the MSIS-themospheric model for the calculation of the thermospheric wind velocities. It is not quite correct because the disturbance of the thermospheric gas by Joule heating is not taken into account in this case. Besides, the changes of the neutral composition due to the disturbances of the thermospheric temperature and wind can alter the ion composition and temperature as well as the F2 region electron concentration.
As concerns the F2 region electron temperature, its observed disturbed variation is larger and more irregular than that calculated. We assume that heating by the enhanced soft electron precipitation may be responsible for the observed variation of the electron temperature. It is possible that the difference between the calculated and observed variations of the F2 region electron concentration on the quiet day of 24 March 1987 is caused by the soft electron precipitation as well.
The main conclusion of our investigation is that present understanding of the ionospheric processes permits us to simulate them numerically and to describe the observed behaviour of the ionosphere over EISCAT not only qualitatively but to some extent quantitatively, at least for the events such as that of 25 March 1987. Our calculations have helped us to divide the contributions of the ion heating and plasma transport into the main ionospheric trough dynamics during the disturbance. As Fig. 11 shows, the equatorward movement of the midnight part of the trough is connected with the enhanced plasma transport, while the apparent westward and eastward movements of the evening and morning edges of the trough are connected with the "hot spots" caused by Joule heating of the ion gas. Of course, the fully self-consistent solution of the modelling equations including the heat balance and continuity equations for the main thermospheric components is necessary to investigate the global effects of the polar ionosphere disturbances, including internal gravity wave generation and propagation.
6.2. Substorm current wedge modelling
In the section 6.1, the model input data for field-aligned currents and precipitating electron fluxes have been selected to obtain an acceptable agreement between variations observed by EISCAT (Collis and Hdggst^ m, 1989, 1991) and modelled ionospheric variations for the quiet day of 24 March 1987 and disturbed day of 25 March. The best agreement has been achieved when the field-aligned currents of the substorm current wedge were added to the region 1 and 2 field-aligned currents during an active phase of a substorm.
The three-dimensional current system named, the substorm current wedge has been suggested and discussed by many authors (e.g., Bonnevier et al., 1970; McPherron et al., 1973; Kamide et al., 1976; Rothwell et al., 1984; Rostoker and Eastman, 1987; Kan et al., 1988, 1992; Kan, 1993). It consists of the field-aligned currents flowing out of the ionosphere at the pre-midnight sector and flowing in at the post-midnight sector closed by the horizontal ionospheric currents and magnetospheric currents perpendicular to the geomagnetic field. In the paper by Namgaladze et al. (1996a) the spatial and time variations of such a field-aligned current system were not calculated but selected to obtain the best fit to the electric field variations observed by EISCAT.
In the papers by Volkov and Namgaladze (1996a, b) an attempt is made to calculate the field-aligned currents of the region 2 and those of the substorm current wedge for the same event of 25 March 1987 rather than selecting them. In this case the field-aligned currents are transferred from the category of input parameters
for the model to the category of calculated parameters. Simultaneously, new input parameters namely, electric-field potential at the polar-cap boundaries and equivalent magnetospheric conductivity, are introduced and variations of these parameters are selected to obtain the best agreement between EISCAT-observed and modelled ionospheric variations. In this way we attempt to answer the question: what magnetospheric conductivity variations can be responsible for the formation at the substorm current wedge and for the corresponding ionospheric variations of the electric field and electron and ion temperature and concentration? As it will be shown, the development of the current wedge during the substorm active phase can be related with the region of the decreased cold magnetospheric electron content travelling westward with a velocity about 1 km/s at ionospheric levels.
The field-aligned currents have been calculated by solving the time-dependent continuity equation for the cold manetospheric plasma-sheet electrons integrated along the closed geomagnetic field lines:
jp = - B,dZm/8t - Re2sin-10 [ (dZm/d&) (d^O/OdA) - (d£m/8A) (dpC/60) ], (27)
where jp is the field-aligned current density defined to be positive for currents flowing out the ionosphere, Bi is the geomagnetic field at the base of the field line in the ionosphere (h = 175 km), Em=eN / B, is the integrated pseudo-Hall magnetospheric conductivity, e is the electron charge, N = B, j(n/B)dl, where the integration is carried out along the geomagnetic field lines from h = 175 km up to the top of the field line, n is the concentration of the magnetospheric plasma-sheet electrons, and so N is a half of the total plasma sheet electron content in the field line tube, t is time, RE is the Earth's radius, 0 is the geomagnetic colatitude, A is the geomagnetic longitude measured from the magnetic midnight to east, (p is the electric field potential determined from the continuity equation for the ionospheric currents (20).
The equation (27) is obtained from the time-dependent continuity equation for the magnetospheric plasma-sheet electrons under the following assumptions. The field-aligned currents are carried by the electrons. The magnetospheric plasma-sheet electrons are cold, i.e. their gradient and other drifts are neglected in comparison with the electromagnetic one. The electric field is potential. The geomagnetic field lines are electrically equipotential; n is constant along the geomagnetic field lines. For the stationary case the equation (27) has been obtained by Vasyliunas (1972) and Maltsev (1974). In these calculations the field-aligned currents are assumed to be flowing only at the closed dipole geomagnetic field lines up to the polar cap boundary, along which the distribution of the electric field potential is taken in the following form:
(Pb = [^c (t) /2] sin A, (28)
where (pc is the potential drop across the polar cap. Its variation during the growth phase of the substorm (1540 -1640 UT) is taken as the linear increase from 20 to 80 kV. The variations of the precipitating electron fluxes are taken in the same form as in the paper by Namgaladze et al. (1996a).
The integrated magnetospheric conductivity distribution during the quiet conditions till up to the moment of the beginning of the growth phase (1540 UT) is taken in the form:
Zm0 = exp[-(0- 0b)2/(A0)2] , 0> 0b = 16o, 29)
where 0b is the geomagnetic colatitude of the polar cap boundary, A0O= 10°, 20 =D100DSm. During the growth phase of the substorm (1540 - 1640 UT), the polar cap boundary moves linearly 4° equatorwards.
The expansion phase of the substorm continues for 20 min (1640 - 1700 UT). During this phase the integrated magnetospheric conductivity varies by the following means:
2m = Zmo {l-0.3exp[- (A - Ao(t))2/(AA)2] }, (30)
where A4 = 23° is the longitudinal half-width of the region of the decreased plasma sheet electron content centered at the longitude A0 . The centre of the region is moving westward with a speed of 1.2 km/s at the ionosphere level. The maximum decrease of the plasma-sheet electron content and correspondingly of the integrated magnetospheric conductivity is 30%. Fig.13 shows the northern polar geomagnetic (0, A) plots of the integrated magnetospheric conductivity at 1550 UT (the left plot) and 1650 UT (the right plot).
During the recovery phase of the substorm (after 1700 UT), the distributions of the magnetospheric conductivity and electric field potential at the polar cap boundary recover to the undisturbed state exponentially, with the characteristic time of 1.5 hour.
The equations (27) and (20) were solved numerically together with all other equations of the model, namely the continuity, momentum and heat balance equations for the main neutral gases (N2, O2, O), molecular (O2+ and NO+ ) and atomic (O + and H + ) ions, and electrons for the height range from 80 km up to 15 RE geocentric distance. The variable latitudinal steps of numerical integration have been used. They vary from 10° for the thermospheric parameters and 5° for the ionospheric F2 region and protonosphere parameters at the equator to 2 ° at the auroral zones for all parameters. In the papers by Namgaladze et al. (1996a) and Volkov and Namgaladze (1996a) the empirical MSIS-86 (Hedin, 1987) thermospheric model has been used to calculate the temperature and composition of the thermosphere. In the present section all calculations are self-consistent. It means that the full system of the modelling equations for the neutral and charged particles is solved (Volkov and Namgaladze, 1996b). The differences between the self-consistent solutions and those obtained by the use of the MSIS-86 model were discussed by Namgaladze et al. (1995) (see the next section). These differences are significant for the calculated thermospheric wind disturbances but they are not important for the calculated field-aligned current and electric field variations discussed here.
The calculated electric field potential and field-aligned currents at the northern high-latitude ionosphere for the different phases of the substorm are shown in Figs.14 and 15. The field-aligned currents of the region 1 are distributed along the polar cap boundary and are not shown in Fig. 15. During the quiet conditions and growth phase of substorm the calculated electric field potential and field-aligned currents are consistent with the average statistical picture of these parameters for the weakly disturbed geomagnetic conditions (Heppner and Maynard, 1987; Iijima and Potemra, 1978). During the expansion phase of the substorm a pair of the field-aligned currents flowing out of and into the ionosphere (the substorm current wedge) appears at the midnight sector. It is produced by the westward-travelling region of the decreased magnetospheric conductivity. The out flowing current is westward from that in flowing. The maximum density of the out flowing current is about 1 A/m2 . In the case when the region of the decreased magnetospheric conductivity is not travelling, the current-wedge field-aligned currents are generated by -VE' VN in the magnetospheric plasma, where VE is ExB plasma drift. Therefore, they are generated at the eastern and western edges of the region of decreased plasma content as far as VE is eastward in the region of decreased plasma content. We can see that it is really so in the midnight sector in our calculations when comparing Figs. 13, 14 and 15 (right panels).
1550 UT
1650 UT
o
□
270
ISO
ISO
0
25
50
75
100
INTEGRATED MAGNETOSPHERIC CONDUCTIVITY, Sm
Fig.13. The northern geomagnetic (0, A) polar plots of the integrated magnetospheric conductivity at the growth phase (left) and at the expansion phase (right) of the substorm. The sun position is at
the top of the figure.
-16
I
16 -40
ELECTRIC POTENTIAL, kV
Fig.14. The geomagnetic (0, A) polar plots of the calculated electric field potential (kV) in the northern high latitude ionosphere at the growth phase (left) and at the expansion phase (right) of the substorm. The sun position is at the top of the figure.
1550 UT
1650 UT
ISO
-0.2 -0.1 0
I
FIELD-ALIGNED CURRENT DENSITY, A km"'
Fig.15. The geomagnetic (0, A) polar plots of the calculated field-aligned current density (A km-2) in the northern high latitude ionosphere at the growth phase (left) and at the expansion phase (right) of the substorm. The sun position is at the top of the figure.
1 -
E < 0
1200
1620
1 1 1 /\ / 2 \ I $ = 68°
\2/ V1
0000 LT
0740
1200
Fig.16. The calculated longitudinal variations of the field-aligned current density along the 68° geomagnetic latitude in the end of the substorm expansion phase for the cases: 1) the region of decreased magnetospheric conductivity is travelling westward and 2) it is at rest.
An influence of the westward travelling of the region of the decreased magnetospheric conductivity on the calculated field-aligned current density is illustrated in Fig.16. It shows the longitudinal variations of the field-aligned current density along the geomagnetic latitude 68° calculated for the cases when the region of the decreased magnetospheric conductivity is travelling westward and when it is motionless. In the last case the field-aligned current density is noticeably less than in the case of the travelling region. Correspondingly, the electric field in the midnight sector is decreased. The influence of the travelling speed Vt r of the decreased plasma content region on the current wedge field-aligned current generation is opposite to that of VE : the westward travelling of the "hole" acts as the eastward electromagnetic plasma drift.
UT
Fig.17. The calculated (solid curves and black circles) and observed by EISCAT (dashed curves) variations of the northward electric field (the bottom panel), F2 region ion and electron temperature at 279 km altitude, E region electron concentration at 111 km altitude and F2 region electron concentration at 279 km altitude for the substorm event on 25 March 1987. The solid curves correspond to the self-consistent calculations and the black circles correspond to the calculations using the empirical MSIS-86 thermosphere model.
The calculated time variations of the northward electric field component, electron concentration at the heights 111 and 279 km, and electron and ion temperature at the 279 km altitude over the EISCAT transmitter position are shown in Fig.17 together with the variations of these parameters observed by EISCAT on 25 March 1987 (Collis and Hdggst^m, 1989, 1991). As can be seen in this figure an agreement between the observed and calculated variations is quite satisfactory.
The presented results show that the behaviour of the electric field, electron concentration, electron and ion temperature observed by EISCAT during the isolated substorm on 25 March 1987 can be satisfactory simulated in the numerical model calculations assuming the appearence of the westward-travelling region of the
decreased plasma-sheet electron content during the expansion phase of the substorm. Due to the appearence of this region the substorm current wedge is formed in accordance with the ideas of Bonnevier et al. (1970), McPherron et al. (1973), Kamide et al. (1976) and others, and with the observations of the field-aligned currents during the substorm expansion phase (Opgenoorth et al., 1983; Lopez et al., 1991; Hoffman et al., 1994).
Opgenoorth et al. (1983) presented the results of the observations of the field-aligned currents at the western and eastern edges of the auroral surge. The field-aligned current flowing out of the ionosphere is connected with the westward-travelling bend of auroras at the western edge of the auroral surge. The speed of the travelling is about 1-2 km/s at the ionosphere level. Baumjohann et al. (1991) presented the satellite observation data about plasma sheet variations during the substorm expansion phase for 39 substorm events. In all these cases decreases of the central-plasma-sheet ion concentration of about 50% were observed simultaneously with temperature increases. These variations are apparently related with the geomagnetic field line reconnection processes leading to the plasma heating and the pushing out of the region of the heating. The conductivity of this region is decreased and a part of the transverse magnetospheric current is closed through the ionosphere forming the current wedge. The observations of the decreased magnetospheric plasma concentration by Baumjohann et al. (1991) are not direct evidence of the decreased plasma tube content. Nevertheless, we consider it is reasonable to assume that the magnetospheric plasma concentration variations during the substorm expansion phase reflect the plasma tube content variations. Due to the dipolarization of the geomagnetic field inside the current wedge (Kan et al., 1992) the geomagnetic field tube volume should be decreased, and if the total plasma tube content is not decreased, the plasma concentration should be increased, in contradiction with the observations by Baumjohann et al. (1991).
So, we have obtained the magnetospheric conductivity variations permitted us to simulate numerically the behaviour of the field-aligned currents, electric fields and high-latitude ionosphere parameters in agreement with the observations. During quiet geomagnetic conditions and at the substorm growth phase, the distribution of the field-aligned currents and electric field potential in the high-latitude ionosphere corresponds to the magnetospheric conductivity model which is uniform in longitude and drops exponentially with latitude equatorward from the polar cap boundary with the characteristic latitude scale of about the auroral zone width.
During the substorm expansion phase a region of decreased (about 30% in comparison with a ground state) plasma-sheet electron content in the geomagnetic field tube appears at the midnight sector and travels westwards with a speed of about 1 km/s at the ionosphere level, forming the substorm current wedge. An appearence of such a region of decreased magnetospheric conductivity agrees with the central-plasma-sheet ion concentration decreases observed simultaneously with increases of their temperature during the substorm expansion phase (Baumjohann et al., 1991).
6.3. Thermospheric disturbance during the substorm of 25 March 1987
The responses of the ionosphere and thermosphere to time-dependent magnetospheric forcing have been modelled by many authors (Richmond and Matsushita, 1975; Fuller-Rowell and Rees, 1984; Sojka and Schunk, 1983, 1984; Roble et al., 1987; Forbes et al., 1987; Maeda et al., 1989; Millward et al., 1993a,b; Sojka et al., 1994). Some of them considered specific events, simulating either thermospheric (Roble et al., 1987; Forbes et al, 1987; Maeda et al., 1989) or ionospheric (Sojka et al, 1994) variations, but none of them calculated the magnetospheric electric field variations together with those of the ionosphere and thermosphere parameters. The behaviour of the ionosphere over EISCAT on 24-25 March 1987 when the quiet day of 24 March was followed by the disturbed day of 25 March with a geomagnetic sudden commencement (SC) starting about 1540 UT was described by Collis and Hdggs^m (1989, 1991) and simulated numerically by Namgaladze et al. (1996a) using a global numerical model of the Earth's upper atmosphere including the calculations of the electric field variations both of magnetospheric and thermospheric dynamo origin. The model input data for the auroral precipitating particle fluxes and the field-aligned currents were selected to get an acceptable agreement between the calculated and observed ionospheric variations. A rather good similarity between the calculated and observed variations of the electric field, E region electron concentration, F2 region ion temperature and electron concentration was found. Spatial distributions of the electric field potential and F2 region electron concentration and ion temperature were obtained and the picture of the main ionospheric trough dynamics during the disturbed period was investigated. It was found that the equatorward movement of the midnight part of the trough was connected with enhanced plasma transport while the apparent westward and eastward movements of the evening and morning edges of the trough were connected with the "hot spots" caused by Joule heating of the ion gas (section 6.1).
Those calculations were performed using the empirical thermospheric model MSIS-86 (Hedin, 1987) to calculate the thermospheric temperature and main neutral gas component concentrations. It is well known that
these thermospheric parameters may vary significantly during geomagnetic disturbances especially during magnetic storms (Roble et al., 1987; Forbes et al., 1987; Burns et al., 1991). During isolated substorms, the effects depend on the substorm intensity and duration and for events of about 3 hours in duration or less they are not reproduced by the MSIS-86 model which operates with the ap index of geomagnetic activity. To check how the substorm thermospheric variations may influence the ionospheric behaviour over EISCAT, new calculations have been conducted by Namgaladze et al. (1995c) and Volkov and Namgaladze (1996b). Numerical integration of the continuity, momentum and heat balance equations for neutral, ion and electron gases as well as the equation for the electric field potential was performed to calculate the time-dependent three-dimensional distributions of electric fields both of the thermospheric dynamo and magnetospheric origin, thermospheric winds, electron, ion and neutral gas temperatures and concentrations for the substorm event of 25 March 1987. By this means the self-consistent solutions of the hydrodynamical and electrodynamical equations were obtained and they were compared with EISCAT data and the results obtained with using the MSIS-86 model.
The precipitating electron flux and field-aligned current variations in these calculations were the same as in the paper by Volkov and Namgaladze (1996b). They are very close to those used by Namgaladze et al. (1996a). The main difference is that in the paper by Volkov and Namgaladze (1996b) the field-aligned currents needed to simulate the substorm variations of the electric field and other parameters observed by EISCAT were not selected but calculated. The calculations of the field-aligned currents have been performed by means of numerical integration of the time-dependent continuity equation for the cold magnetospheric electrons. By this means it has been found that during the active phase of the substorm a current wedge is formed. It is connected with the region of the decreased cold magnetospheric electron content travelling westward with a velocity about 1 km/s at ionospheric levels (Volkov and Namgaladze, 1996b).
The calculated ionospheric variations are shown in Fig.17 together with the EISCAT data observed on 25 March 1987. The black circles in this figure correspond to the calculations using the MSIS model and the solid curves correspond to the self-consistent ionosphere-thermosphere calculations. Both calculation variants use the same initial conditions at the time of the substorm onset at 1540 UT on 25 March taken from the calculations for the quiet day of 24 March performed using the MSIS model. As we can see in this figure, the ionospheric variations calculated in both variants are practically the same (except for the electron temperature being by 40-180 K lower in the self-consistent variant, but this is not related with substorm effect) but the structure of the thermospheric wind disturbances shown in Fig. 18 differs appreciably. After the time of maximum of the geomagnetic substorm disturbance (1700 UT) the self-consistently calculated wind velocity disturbance (left plots in Fig.18) reveals all typical features of the atmospheric internal gravity wave (Hines, 1960; Richmond and Matsushita, 1975). These waves are the low-frequency equivalents of acoustic waves. They are "internal" (not "surface") because they can support a substantial vertical component of phase propagation. One of the peculiarities of the internal atmospheric gravity waves is that they propagate energy upwards when phase progression is downwards, although the horizontal components of energy and phase velocity are directed in the same sense.
The wind disturbance intensity increases with height approximately in proportion to exp( h / 2H) where H is the barometric scale height of the neutral atmosphere, until the dissipative processes due to molecular viscosity and thermal conductivity of the neutral gas stop this increase. The horizontal speed of the propagation of the disturbance also increases with height due to the growth of the neutral temperature with height in the thermosphere. It causes the apparent front of the wave propagation in the meridional (height-latitude) plane to be oblique with variable inclination of the front. The apparent horizontal speed of the propagation may be roughly estimated as the speed of the horizontal movement of the surface on which the wind velocity is of a constant value, for example, 200 m/s. It is of about 590 m/s at heights of about 400 km as estimated for the 50-70° latitude range, being directed equatorwards at these latitudes. This speed is comparable with values of about 400-718 m/s estimated from the observations of atmospheric gravity waves (with periods of about 60 min) during the October 1985 WAGS campaign by Rice et al. (1988) and Williams et al. (1988) and with value 550 m/s obtained by Sheen and Liu (1988) in their modelling calculations for the event on October 18, 1985 observed by the incoherent scatter radar at Sondrestrom.
Fig.18. The northward wind velocity variations along the EISCAT 117° geomagnetic meridian during the substorm event of 25 March 1987 calculated self-consistently (left plots) and those calculated using the empirical MSIS-86 thermosphere model (right plots) for various times of UT from 1700 to 1800.
It can be seen well in the left plots in Fig.18 that firstly (from 1700 to 1720 UT) the wind velocity disturbance moves both equatorwards and polewards from the auroral zone. But after 1720 UT we can see the reversal of the movement of the disturbance at the polar cap, where interaction takes place between the wave disturbances coming from the opposite sides of the auroral zone, and the movement becomes equatorward -from the pole to the evening auroral zone - due to arrival of more intensive disturbance from the opposite side of the auroral zone (where the neutral temperature disturbance is maximal, see Fig.19). This disturbance is seen well in the top right corner in the left plot at 1750 UT in Fig.18.
In the MSIS-86 variant the wind disturbance (right plots in Fig.18) is more localized and propagates much slower because of the thermosphere temperature and density being practically fixed during the event. Over the source region (in the vicinity of 70° geomagnetic latitude), the horizontal winds calculated in both variants are similar. The difference between the variants of the calculations is that in the variant using the empirical MSIS-86 model the heating of the thermosphere due to the electron precipitation and Joule heating has not been taken into consideration, and only ion drag has been allowed for. The fully self-consistent theoretical calculations take into account all these effects. The corresponding disturbed neutral temperature distribution at 1800 UT at 300 km altitude is shown in Fig.19 (bottom plot in the left panel) together with the quiet initial
distribution at 1540 UT (top plot) and the MSIS-86 model neutral temperature distribution for 1800 UT (middle plot in the left panel).
NEUTRAL TEMPERATURE (K) WIND VELOCITY (m/s)
800 850 900 950 1000 640 4801 320 160
I I I I l 560 400 240 80
Fig.19. The geomagnetic polar plots of the neutral temperature and wind velocity vectors at 300 km altitude in the northern polar thermosphere. The sun position is at the top of each plot. The top panel shows the initial quiet distributions at 1540 UT. Other panels show the temperature and wind vectors at 1800 UT calculated using the MSIS-86 model (middle panel) and self-consistently (bottom panel).
The disturbed neutral temperature and corresponding thermospheric wind (bottom right plot in Fig.19) distributions, obtained in the self-consistent calculations, are rather similar to those obtained by Roble et al. (1987) for 1230 UT in their model simulations of the thermospheric dynamics during the March 22, 1979, magnetic storm. Although this was another event under higher solar activity level, the first substorm observed at 1000-1230 UT on March 22, 1979, was rather similar to that observed on March 25, 1987. The neutral temperature disturbance determined as a difference between disturbed and quiet values of the temperature has a maximum of about 200 K in comparison with 250 K obtained by Roble et al. (1987) both being positioned in the morning sector of the auroral zone. The thermospheric wind disturbances are maximal in the morning sector as well, reaching velocity values of about 640 m/s in comparison with 500-600 m/s obtained by Roble et al. (1987). All this means that patterns of the neutral temperature and wind disturbances obtained in both cases are typical
for the thermospheric response to a substorm. Note that Roble et al. (1987) have used an empirical model for the ion convection whereas the latter has been calculated theoretically in the present work and presented in the paper by Volkov and Namgaladze (1996b).
The differences between the MSIS-86 and self-consistently calculated winds (middle and bottom right plots in Fig.19) are practically absent in the near-midnight sector where the temperature disturbance is minimal. On the dayside an additional meridional wind component of about 100 m/s appears at latitudes lower than approximately 73o in the self-consistent variant of calculations leading to the equatorward turning of the wind vector and to the equatorward expansion of the whole disturbance picture. This turning almost completely compensates for the antisunward flow due to solar heating.
As for neutral composition variations due to the thermosphere heating, a decrease in the O/N2 concentration ratio of the order of about 40% occurs at 300 km altitude near the neutral temperature maximum position with a smaller decrease in the vicinity of the EISCAT transmitter position. Ionospheric variations over EISCAT are not significantly influenced.
Thus, a comparison of the results obtained in the self-consistent variant of calculations and in that using MSIS-86 shows the significant role of thermal sources acting together with the processes of momentum exchange between ions and neutrals and affecting the propagation character of thermospheric disturbances. But calculated substorm variations of the ionospheric parameters over EISCAT do not depend very much on the thermospheric model used (at least in the evening sector) because of the small influence of thermospheric winds on the high latitude ionosphere due to a large geomagnetic field inclination. It means that the previously published results by Namgaladze et al. (1996a) are not superseded by the present ones.
Thus, the behaviour of the polar ionosphere and thermosphere during the substorm event of 25 March 1987 has been calculated in two variants of the calculations. One of them uses the empirical MSIS-86 thermospheric model to calculate the temperature and concentrations of the neutral gases, the other calculates all thermospheric parameters self-consistently by solving the corresponding heat balance and continuity equations. A comparison of the results obtained in these variants shows considerable difference between the calculated thermospheric winds at ionospheric F2 region heights. The self-consistent calculated wind velocity disturbance reveals all typical features of the internal atmospheric gravity wave. It propagates equatorward from the auroral zone with an oblique front and at a speed of about 590 m/s at heights of about 400 km. This speed is close to values estimated from the observations of AGW during the October 1985 WAGS campaign. At the polar cap, an interaction takes place between the wave disturbances coming from the opposite sides of the auroral zone. In the evening sector, the wind disturbance is moving firstly polewards from the auroral zone, but afterwards the movement becomes equatorward - from the pole to the evening auroral zone - due to arrival of more intensive disturbance from the opposite side of the auroral zone where the neutral temperature disturbance is maximal. In the MSIS-86 variant, the wind disturbance is more localized and propagates much slower because of the thermosphere temperature and density being practically fixed during the event. However, this difference does not have much influence on the calculated substorm variations of the ionospheric parameters over EISCAT because of the small influence of the thermospheric winds on the high latitude ionosphere due to a large geomagnetic field inclination. The similarity of the results presented with those obtained by Roble et al. (1987) for the higher solar activity level, allows to conclude that patterns of the neutral temperature and wind disturbances obtained in both cases are typical for the thermospheric response to a substorm.
6.4. Numerical modelling of the thermospheric winds over Loparskaya
The ionosphere-thermosphere interaction via the momentum and energy interchange between neutral and charged particles is an effective way to transport the energy, entering the high latitude upper atmosphere from the magnetosphere, to the middle and low latitudes by means of the global thermospheric circulation and its disturbances. As a result the high latitude momentum and energy sources affect the global thermosphere and ionosphere dynamics. Owing to the three-dimensional character of the ionosphere-thermosphere coupling its study demands using both observations done in a wide range of latitudes and time-dependent three-dimensional self-consistent ionosphere-thermosphere models. Applied to the global effects in the F layer of the ionosphere such a synthetic approach was fulfilled in the Dynamics Explorer satellite programme (Killen and Roble, 1988) using the global ionosphere-thermosphere models by Fuller-Rowell et al. (1987) and Roble et al. (1988).
At the heights of the E layer of the ionosphere (90-150 km) the situation is much more complicated especially at the high latitudes. On the one hand, the nature of the ion-neutral interaction changes from the almost full dragging of ions by neutrals in the lower E region to the almost total "frozen-in" state in the upper part of it where ions are fully magnetized. On the other hand, it is in the E layer where the rate of the ionization by the precipitating auroral (keV) electrons has a maximum, that the ionospheric currents flow, the electric fields are generated by the ionospheric dynamo effect and the tidal activity is important. Moreover, there are no
satellites at these altitudes. All this causes particular interest to the studies of ionosphere-thermosphere coupling at these heights.
Modelling of the neutral winds in the auroral E region was done by Rees and Fuller-Rowell (1990) using the global ionosphere-thermosphere model by Fuller-Rowell et al. (1987). In this work particular attention has been paid to influence of the enhanced ionospheric plasma convection on the neutral wind spatial distribution. It was proved that in the E region this influence is similar to that in the F region although being 2-3 times weaker. The calculations, however, were done for the rarely existing stationary convection conditions and have not been compared with any wind observations in the E region. The National Center for Atmospheric Research Thermospheric General Circulation Model - NCAR TGCM (Dickinson et al, 1981) has been used by Johnson et al. (1987) to simulate the Chatanika radar observations of the high latitude lower-thermospheric winds. The model results are generally in good agreement with the observations at Chatanika, but they differ in the dusk sector, however. Enhanced equatorward flows were observed at Chatanika at active intervals, while model results predict an increased poleward flow in all LT zones other than the early morning sector. Inaccurate magnetospheric input for the model has been suggested as a possible cause of this disagreement.
Loparskaya, Emission 557.7 nm, 26.01.93
kR
19 20 21 22 23
Fig.20. The time variations of the meridional wind velocity (positive being directed to the south) observed to the pole (solid line) and to the equator (dashed line) of Loparskaya (bottom panel) and the auroral situation over the region observed with the meridional scanning photometer (top panel). From Leontyev et al. (1997).
In the paper by Leontyev et al. (1997) an attempt has been made to interpret the observed dynamics of the meridional thermosphere winds in the pre-midnight E region at latitudes poleward and equatorward of the precipitation zone. The wind observations were done with the Fabry-Perot interferometer (FPI) at Loparskaya (68° N, 33° E). The modification by Namgaladze et al. (1996b) of the global thermosphere-ionosphere-protonosphere model by Namgaladze et al. (1988, 1990, 1991, 1994) was used to simulate the observed meridional thermospheric wind behaviour and explain the distinct decrease in the mean wind velocity, when
passing across the auroral zone from the pole to the equator, and also a near-counterphase character of the wind velocity fluctuations on the both sides of the precipitation zone.
The time variation of the meridional wind velocity (positive southward) measured poleward and equatorward of Loparskaya is shown in Fig.20 (the bottom panel) by the solid and dashed lines correspondingly. The top panel in Fig.20 shows the auroral situation according to the meridional scanning photometer data. The zenith of Loparskaya corresponds to 90° (the angle is indicated on the left side of this panel); the wind measurements taken poleward of Loparskaya and shown by solid lines in the top panel in Fig.20 correspond to the angle of 150° and the measurements taken equatorward of Loparskaya correspond to 30°. As it is seen from the top panel in Fig.20 the auroral zone is located mostly between the regions from which the emission is measured, and there are several auroral intensifications during the observation time interval.
The main distinct features of the observed behaviour of the meridional thermospheric winds in the evening E layer poleward and equatorward of the precipitation zone are the following:
1) The mean meridional wind velocity measured 2° of latitude to the pole of Loparskaya is about 100 m/s being directed southward whereas that measured 2° to the equator of Loparskaya is about zero, i.e. the southward velocity drops by about 100 m/s when passing through the precipitation zone from the pole to the equator.
2) The meridional wind velocity fluctuations of about 100 m/s with quasi-periods of about 20-60 min occur on the both sides of the precipitation zone being of "quasi-counterphase" character.
We will try to explain these features of the observed meridional wind behaviour using numerical model calculations.
In this study the model calculations were performed using both the completely self-consistent, on thermospheric parameters, variant of the model and the variant using the thermosphere temperature and density distributions given by the empirical MSIS-86 model (Hedin, 1987) to calculate the winds and ionospheric parameters. The calculations were done for the conditions of the quiet 3-keV electron precipitation and its steplike 50 times enhancement of one hour in duration.
Spatial distribution of the precipitating electron intensity at the upper boundary of the thermosphere (h = 520 km) is set in the model correspondingly to (25-26) with the following values of the precipitation parameters:
"cold" electrons (with the characteristic energy E = 0.2 keV): @mn = 68°, A<t>D= 2.5°, I = 1+-12, Am1 = 125°, AA1 = 35°, 1m1 = 6.0x109cm-2 s-1, Am2 = 165°, AA2 = 20°, 1m2 = 1.5x109cm-2s-1;
"hot" electrons (with the characteristic energy E = 3 keV): 1m = 4.2x108cm-2s-\ @mn = 73°, Am = 187°, AAD= 60°.
These parameters characterize the quiet electron precipitations in the auroral zone. The "cold" electron precipitation with the maximum intensity of 3x109 cm-2s-1 and "hot" electron precipitation with maximum intensity of 0.3x 109 cm-2 s-1 has been added over the polar cap with 0m = 90° and A@ = 15° for the best fit of the calculated electron concentration to the empirical ionospheric model data by Chusovitin et al. (1987).
The maximum current density in the zone 1 is assumed to be 0.32 |iA/m2 and 0.12 |iA/m2 in the zone 2. With such field-aligned currents the electric potential drop across the polar cap is 32 kV.
It is worth noticing that we use the zero wind velocity values at the lower boundary (h = 80 km). It means that no upwards propagating tides, connected with insolation absorption by H2O and O3 in the lower atmosphere, have been included in the self-consistent variant of the model apart from those in situ generated due to UV and EUV absorption in the thermosphere.
The model calculation results revealed the following features in the behaviour of the meridional thermospheric winds under quiet conditions. In the pre-midnight sector of the north polar thermosphere the calculated meridional winds at the altitudes above 105 km are directed equatorward in both variants of the calculations. Fig.21 shows the calculated latitudinal variations of the meridional wind velocities at 0000 UT along the 2200 MLT geomagnetic meridian (solid curves). The variations are shown for the altitudes of 105, 115, 125, 150 and 200 km. Data given in the left column represent the wind variations calculated by means of the MSIS-86 model, i.e. using the neutral temperature, gas concentrations and, therefore, pressure distribution from the MSIS-86 model not taking into account the effects of the thermosphere heating by the precipitating electrons and Joule heating. The results of the fully self-consistent theoretical calculations taking into account along with others also the effects of the thermosphere heating by the precipitation and current dissipation are shown in the right column.
As it is seen from Fig.21 when moving across the precipitation zone (i.e. through the geomagnetic latitudes of 70-75° in these calculations) from higher latitudes to lower ones the 2-3 times decrease in the meridional wind velocity occurs at the heights above 105 km. The latitudinal width of this decrease area is about 5-10° broadening in the F region where the latitudinal dependence of the meridional wind velocity is smoother
than it is in the E region. To the pole of the precipitation zone the meridional wind velocities are about 30-110 m/s at the altitudes of 115-150 km in the self-consistent variant of the calculations and about 90-160 m/s in the MSIS-86 variant; to the equator of the precipitation zone the meridional wind velocities are 12-45 m/s and 30-85 m/s respectively. Thus, both variants of the model calculations show in accordance with the observations the drop of the equatorward wind velocity when crossing the precipitation zone from the pole to the equator. The only difference between the calculation variants is that the velocity magnitudes in the E region are smaller in the self-consistent variant than in that of MSIS-86 being similar in the F region. Perhaps, it is due to neglecting tides propagating upwards from the lower atmosphere in the self-consistent variant of the calculations.
MSIS-86 THEORETICAL MODEL
Fig.21. The calculated latitudinal variations of the meridional wind velocity (positive being directed to the south) along the 22.00 MLT geomagnetic meridian. The left column represents the wind variations calculated using the MSIS-86 model. The results of fully self-consistent theoretical calculations are shown in the right column. Solid lines correspond to the results obtained for 0000 UT using the normal quiet field-aligned currents. The results for 0130 UT are shown with dashed lines for zero field-aligned currents and with black circle lines for the tripled ones as compared with the normal quiet field-aligned currents. From Leontyev et al. (1997).
The physical nature of the discussed latitudinal drop of the meridional wind velocity is the following. With no significant ionosphere-thermosphere interaction the latitudinal profile of the meridional wind velocity could be of smooth character corresponding to the thermospheric gas flowing over the poles from the sun-heated dayside into the cold nightside when the diurnal oscillation dominates, or be more complicated, but also smooth in the presence of the semi-diurnal tidal wind component. Existence of the auroral zone with higher concentration of ions, being partly or fully "frozen-in" and braking (in the case of their immobility) the movement of neutrals, prevents thermospheric gas flowing. Therefore, it causes decrease in the meridional neutral gas velocity when crossing the auroral zone. At the same time the auroral zone is an area of higher heating of the thermosphere by the precipitating electrons and by Joule heating, and hence it is a region of higher thermospheric gas pressure. The force of the gas pressure gradient accelerates the neutral gas equatorward and poleward of the auroral zone "quenching" the velocity drop caused by ion braking. As it follows from Fig.21 the influence of the heating effects on the meridional winds in the E region is small under
the quiet conditions as shown by the insignificant differences between the results when allowing and not allowing for the local auroral heating.
The reasoning behind the role of ion braking is valid on the supposition of immobility of the ions. If the ions are driven by the electric field, for example, they can accelerate as well as brake the neutral gas. For example, the antisolar ion convection at the polar cap accelerates the neutral gas flowing over the polar cap from the dayside to the nightside. Inversion of the ion convection in the nightside auroral zone brakes equatorward flowing of the neutrals.
To estimate the value of this effect the calculations similar to above have been done for different magnitudes of the input field-aligned currents responsible for the electric field distribution in the ionosphere. Two additional variants of the calculations have been performed using zero input field-aligned currents and tripled as compared with those taken above. The calculations started at 0000 UT. The initial conditions for them were taken from the results presented by solid curves in Fig.21. The results for 0130 UT are shown in Fig.21 by dashed lines for zero field-aligned currents and by black circle lines for the tripled ones as compared with those corresponding to the results presented by solid lines.
MSIS-86 THEORETICAL MODEL
h=125 km h=125 km
Fig.22. Calculated latitudinal variations of the meridional wind velocity (positive being directed to the south) at the 125 km altitude for the different moments of UT after the sudden jump of the precipitating flux intensity at 0000 UT. As in Fig.21 the left column corresponds to the calculations done using the MSIS-86 model, and the right one corresponds to the self-consistent calculations. From Leontyev et al. (1997).
Comparison of the results presented in Fig.21 by solid, dashed and black circle lines shows that with out field-aligned currents the latitudinal variation of the meridional wind velocity has not changed significantly being slightly smoothed in the MSIS-86 variant of the calculations. Tripling of the field-aligned currents increased considerably the value of the meridional velocity drop at all altitudes above 105 km in both MSIS-86 and self-consistent variants of the calculations. Thus, under quiet conditions with the electric potential drop across the polar cap less than 30 kV the influence of the ion movements, i.e. of the electric fields, on the thermospheric winds is small; nevertheless, the meridional wind velocity drop occurs when crossing the auroral zone, and it is due to braking of the neutral gas by the almost stationary ions. The ion convection enhancement
increases considerably the value of the meridional velocity drop which may be greater than 100 m/s in the upper E region.
Usually the energetic particle precipitations from the magnetosphere are of non-stationary character with sharp increases and decreases in the precipitating particle fluxes. It may lead to the atmospheric gravity wave generation. Let us consider the influence of the sudden precipitation enhancement on the meridional wind behaviour. The following model was used: the flux intensity of the precipitating 3 keV electrons increased 50 times suddenly at 0000 UT and returned abruptly to the quiet level at 0100 UT, the field-aligned currents did not change at this time and corresponded to the quiet conditions. The calculated latitudinal variations of the meridional wind velocity at the height of 125 km are shown in Fig.22 for the different moments of UT. As in Fig.21 the left column corresponds to the calculations done with using the MSIS-86 model, and the right one corresponds to the self-consistent calculations.
As it is seen from Fig.22 in both variants of the calculations the meridional wind velocity drops polewards of the precipitation zone and increases equatorwards of it after the sudden precipitation enhancement. The amplitude of the meridional wind velocity disturbance is of the same order as the value of the latitudinal drop of the meridional wind velocity under the quiet conditions. These features of the meridional wind velocity disturbances correspond completely to the observed wind velocity fluctuations shown in Fig.20. In the self-consistent variant of the calculations the wind velocity disturbance is of the inner gravity wave character and the non-disturbed state restores itself in an hour. In the MSIS-86 variant the recovery of the non-disturbed state takes much more time because of the thermospheric temperature and density being practically fixed during the disturbance.
Thus, the comparison between FPI-observations of the meridional thermospheric winds at the E layer heights poleward and equatorward of the auroral zone and model calculations are in the quite good agreement. In the calculations the drop of the average meridional wind velocity when crossing the precipitation zone from the pole to the equator in the pre-midnight sector is displayed together with the intensive (of the order of the mean velocity) quasi-counterphase fluctuations at the latitudes to the pole and to the equator of the precipitation zone in accordance with the observations.
Comparison of the results obtained in the self-consistent variant of the calculations and in that using the MSIS-86 model shows the insignificant role of local pressure gradient forcing due to the auroral heating of the thermosphere in comparison with the processes of the momentum exchange between ions and neutrals in forming the meridional wind velocity variations in the vicinity of the precipitation zone under the quiet conditions. Under non-stationary conditions when the precipitation intensity changes sharply, the role of the thermal processes becomes very important affecting the propagation character of the disturbances. This conclusion was made also by Namgaladze et al. (1995) from the results of the numerical simulation of the upper atmosphere behaviour over EISCAT during a substorm.
According to the calculation results, the value of the meridional velocity latitudinal drop and amplitude of the velocity fluctuations depends mostly upon two factors: the electric field intensity and the height where the wind velocity is measured. The latter is determined by the maximum luminosity height, i.e. by the energy of the precipitating electrons exciting the 557.7 nm emission. The variations of the precipitating electron energy may cause additional fluctuations in the observed wind velocity because of the height dependence of the wind velocity (see, for example, Fig.21). Within the uncertainty of the above factors the presented observed thermospheric wind data and their model calculations are consistent not only qualitatively, but quantitatively as well.
7. Numerical modelling of the ionosphere-thermosphere responses to the precipitation and field-aligned current variations in the cusp region
7.1. Equinoctial conditions
The cusp is a region where magnetosheath solar-wind particles have direct access to the magnetosphere and some of them may precipitate into the ionosphere. The local ionospheric effects of the soft electron precipitation in the cusp, such as increases of the F2 region electron concentration and temperature, are well known (Shepherd, 1979). As far as the thermospheric effects of the soft electron precipitation are concerned, they are non-local, due to the internal atmospheric gravity waves propagating from the region of the abrupt electron precipitation. The same can be said of the ionospheric and thermospheric effects of the field-aligned current variations in the cusp because they influence the whole pattern of the polar ionosphere convection and related thermospheric disturbances. This means that these effects should be modelled by the use of the three-dimensional time-dependent self-consistent ionospheric-thermospheric model including the electric field calculations.
In recent years, the two- and three-dimensional thermospheric models have been used to study the response of the thermosphere to the model electric field and auroral particle precipitation variations (Richmond and Matsushita, 1975; Fuller-Rowell and Rees, 1981, 1984; Fuller-Rowell, 1984; Roble et al., 1987; Maeda et al., 1989; Burns et al., 1991; Fuller-Rowell et al., 1991). A modelling study of the effect of a short-lived, localized enhancement in the high-latitude dawn side convection electric field has been made using the coupled ionosphere/thermosphere model (Millward et al., 1993a). However, the magnetospheric convection electric field variations were not calculated but taken as inputs in all these model simulations, and the cusp region was not considered as a separate high latitude source of the thermospheric and ionospheric disturbances. This source has some specific features in comparison with other auroral sources being more localized in longitude and having lower characteristic energies of the precipitating electrons.
The main goal of this study is to investigate the thermospheric and ionospheric effects of the soft electron precipitation and field-aligned current variations in the cusp, of the order of an hour in duration, using a new version of the global numerical model of the Earth's upper atmosphere developed for studies of polar phenomena (Namgaladze et al., 1996b, c). The questions wanted to be answered in our investigation are the following. How far from the cusp can the thermospheric and ionospheric effects of the precipitation be seen? How does the spatial distribution of the electric field potential react to the variations of the field-aligned currents in and near the cusp? How do these electric field changes influence the disturbances of the thermospheric temperature and circulation and ionospheric parameters? What is the relative role of the electric field penetration at remote distances from the cusp and the atmospheric gravity wave propagation there?
The answers to these questions are of interest because of the location of the EISCAT Svalbard Radar in the cusp region and the associated observations at lower latitudes that will be possible using the existing EISCAT UHF and VHF radars. This study makes predictions for both these regions and these predictions will be tested by joint observations by ESR, EISCAT UHF/VHF and other ground-based ionosphere/thermosphere observations.
The effects of the soft electron precipitation in the cusp were modelled by the following means (Namgaladze et al., 1996c). The precipitating 0.23 keV electron flux (a Maxwellian with characteristic energy of 0.23 keV) intensity in the cusp region Im and the geomagnetic latitude of the precipitation maximum Fm have been used as the variable inputs of the model; Fm varies between 78° and 73° geomagnetic latitude; DF = 3.5°, Lm corresponds to the local midday; DL = 45°, i.e. the cusp region extends approximately from 0900 to 1500 MLT. The undisturbed value of Im has been chosen equal to 1.9x10 9 cm-2s-1.
In the first variant of the calculations this flux was increased suddenly by a factor of 10 at 0000 UT and maintained at such a level for 30 min and then returned suddenly to the initial level. In the second variant of the calculations the precipitating electron flux intensity was increased linearly with time over 30 min from 0000 UT to 0030 UT. Simultaneously, the position of the intensity maximum moved from 78° to 73° geomagnetic latitude. During the next 30 min both maximum intensity and its position returned linearly to their initial levels. Such movements of the cusp have been observed, for example, by Sandholt et al. (1994). The time-integrated peak flux of the precipitating electrons was the same in both variants of the calculations. The background quiet precipitations outside of the cusp region were the same as in previous papers for the quiet equinoctial conditions under low solar activity (for example, Namgaladze et al., 1996a) being close to those given by Hardy et al. (1985). They were not varied in these calculations. The magnetospheric sources of the electric field for the undisturbed conditions are field-aligned currents in zones 1 and 2 (Iijima and Potemra, 1976). To investigate the effects of the disturbed field-aligned current variations in the cusp for IMF By < 0 we have used the following model input variations of the field-aligned currents based on the data by Taguchi et al. (1993), Yamauchi et al. (1993), Ohtani et al. (1995). We have added the field-aligned current flowing into (out of) the ionosphere at the northern (southern) hemisphere along the 80° geomagnetic latitude at the 1130-1400 MLT sector and flowing out at the 1000-1130 MLT sector. These currents are closed by the additional zone-1 currents. The time variation of all these additional field-aligned currents has the following form. Their density increases linearly from 0 to the maximum value during the first 30 min (0000 - 0030 UT) and then recovers to 0 during the next 30 min (0030 - 0100 UT). The maximum density of the field-aligned current flowing into the ionosphere at 80° geomagnetic latitude is 1.6 A km-2 which is ten times larger than the quiet zone-1 field-aligned current density. The corresponding magnetic disturbance in the cusp region is estimated approximately as 600 nT. Such a disturbance was observed in the cusp region when By component of IMF was equal - 9 nT (Taguchi et al., 1993). It means that the modelled situation corresponds to the case when By changes from 0 to - 9 nT and back to 0 over a 1-h period.
Four variants of the calculations have been performed: 1) the cusp position is fixed and only the sudden precipitation of 0.23 keV electrons takes place over 30 min (from 0000 UT to 0030 UT); 2) the cusp is moving and the precipitation is linearly increased and then decreased over 1 h; 3) the same as in variant 2, but the
additional field-aligned currents are included being linearly increased and then decreased over 1 h; 4) the same as in variant 3, but the additional field-aligned currents keep their maximum values after 0030 UT.
The magnetosphere may generate these four cases by the following means. Variant 1 employs a square wave pulse of enhanced electron precipitation flux in the cusp region, which maintains a fixed position in the ionosphere. This could well be the result of a corresponding pulse in the density of the solar wind impinging on the magnetosphere. Such a pulse would compress the dayside magnetosphere but would not move the latitude of the dayside cusp in the ionosphere (because the ionosphere is largely incompressible, in the sense that the magnetic field there is almost constant).
Variant 2 has a triangular pulse in cusp electron flux, which again could be caused by a similar variation in the solar wind density. The cusp migrates equatorward, as it is often seen in observations. This would be expected if the magnetopause compression were to be accompanied by a proportionally enhanced rate of magnetopause reconnection, eroding the dayside magnetopause and bringing the cusp to lower latitudes. However, we would expect (after about 10-15 min delay) this to cause a rise in the field-aligned currents and associated ionospheric convection (see, for example, Cowley and Lockwood, 1992). Thus, variant 2 is unlikely to be observed. Nevertheless, it is useful to model variant 2, as it helps to distinguish the effects of the precipitation from those of the electrodynamics. We consider variant 3, in which the erosion is accompanied by field-aligned currents in synchronization with erosion, to be more realistic than variant 2.
Electron density, m-3 Electron temperature, K
0000 0030 0100 UT 0000 0030 0100 UT
1115 1145 1215 MLT 1115 1145 1215 MLT
Fig.23. Time variations of the calculated electron concentration (left plots) and electron temperature (right plots) at h = 300 km, A = 240°, at various geomagnetic latitudes for variants 1 (dashed curves), 2 (solid curves), 3 (black circle curves) and 4 (open circles) of the calculations (see text for explanation).
Lastly, variant 4 does not ramp down the field-aligned currents after the rise; this means that the cusp region currents and associated flows persist after the enhanced cusp precipitation decays away. This would
happen if the dayside reconnection persists after the decay of the solar wind pressure pulse. This situation would thus apply to a southward turning of the IMF, occuring at the time of the solar wind pressure pulse.
Figs.23-25 show the calculated time variations of the ionospheric and thermospheric parameters at various northern geomagnetic latitudes in the range 60-85° for the daytime 240° geomagnetic meridian at the height 300 km. Variants 1, 2, 3 and 4 of the calculations are presented in these figures by the dashed, solid, black circle curves and open circles, correspondingly.
Let us consider firstly the results of variants 1 and 2 of the calculations when only precipitation acts as a source of the disturbances. As we can see in Figs.23 and 24, at the geomagnetic latitudes equatorward from about 75°, the disturbances of the electron concentration and temperature, as well as of the ion and neutral temperature are minimal in case of the fixed cusp position (variant 1 of the calculations). The exception for this is the initial electron temperature burst near the cusp region in the beginning of the abrupt precipitation. This burst takes place when the ion concentration is not yet high enough to cool the electron gas effectively.
Neutral temperature, K
1000 л 85°_______
900 1100 -
1000 -
900
1100 -л
1000 -
900 1100 -,
1000 -
900
1000
900 1000 -1
900
900 -850
900 -850
Ion temperature, K
1100 n 85
900
-I-I-I-I-J-I-I-I-I-J-I-I-I-I-J
-1 -.-3 1400 -,
4
1200 -
1000 1400 -I
1200 -
1000 1300 -I
1100 -
900
1100 -
900 1100 -]
1000 ^ 70° 900
1000 -900
65°
I I I I' I 11 I I I I I l'^'l1
1—I—I—I—I—I—I—I—I—I—I—I—I—I—I
0000 0030 0100 UT 1115 1145 1215 MLT
1000 -900
60°
"I I I iiT^]
0000 0030 0100 UT 1115 1145 1215 MLT
Fig.24. The same as in Fig.23 but for the neutral (left plots) and ion (right plots) temperature.
At the end of the precipitation burst (0030 UT) in variant 1 of the calculations, electron concentration Ne at h = 300 km (left plots in Fig.23) reaches its maximum value of about 8.4x10n m-3 at 78° geomagnetic latitude in comparison with the initial value of about 2.9x10n m-3 at 0000 UT. This then decreases to the quiet level during next 30 min. In the case of the moving cusp (variant 2 of the calculations) Ne (300 km) reaches the maximum value of 6.7x10n m-3 at 0030 UT at 73° geomagnetic latitude, in comparison with the initial value of about 1.2X1011 m-3 and recovers to the quiet level at about 0130 UT. These Ne enhancements are caused by the precipitating electron impact ionization, although there is also an increase of the ion O+ loss rate due to the neutral composition and ion temperature disturbances. At 70° geomagnetic latitude, the Ne enhancement is still rather high in variant 2 of the calculations, whereas in variant 1 the disturbance magnitude drops very significantly. At lower latitudes only weak positive disturbances of Ne caused by the disturbed thermospheric
wind action propagate equatorward with the average speed of about 540 m/s in both variants of the calculations as estimated for the 65°- 60° latitude range.
The electron temperature disturbances are shown in Fig.23 (right plots). They are positive in the cusp region during the majority of the time when the heating of the electron gas by the precipitating electrons is active but becomes negative after the ending of the precipitation. That is because the remaining increased ion concentration acts to cool the electron gas.
The ion temperature variations shown in Fig.24 (right plots) are very similar to those of the neutral temperature (left plots in Fig.24) being more intensive in case of the moving cusp (variant 2) at the geomagnetic latitudes lower than about 78°. Both ion and neutral temperature disturbances are positive due to the heating of the ion and neutral gases by electrons. They propagate away from the cusp region as large-scale gravity waves with the average speed of about 690 m/s, estimated for the 70°- 60° latitude range.
-1 -1 Northward wind, m s Eastward wind, m s
Fig.25. The same as in Fig.23 but for the meridional, positive northward (left plots) and zonal, positive eastward (right plots) thermospheric wind velocity.
The meridional thermospheric wind disturbances (left plots in Fig.25), driven by the pressure gradient forcing from the heated cusp region, reveal an analogous character of the propagation being of about 140, 90 and 46 m/s in magnitude at the geomagnetic latitudes 70°, 65°, 60° respectively in case of the moving cusp. The corresponding values are about 90, 50 and 26 m/s at the same latitudes in case of the fixed cusp position. The zonal wind disturbances (right plots in Fig.25) are insignificant in the midday sector for cases when only enhanced precipitation takes place. The neutral composition disturbances (not shown) are virtually confined to the cusp region where the concentration O/N2 ratio diminishes by about 25% at 75° geomagnetic latitude in case of the moving cusp and 24% at 80° in case of the fixed cusp position.
Now let us consider the results of the calculations in variants 3 and 4, where the additional field-aligned currents in the cusp region are included. These have largest current densities at 0030 UT and subsequently either return to the quiet level at 0100 UT (variant 3) or remain fixed at their maximum values (variant 4).
A comparison of the results shown in the left and right columns in Figs.23-25, as well as of the results presented by solid curves with and without the black circles in these figures, demonstrates the differences between the effects caused by the joint action of the precipitation and field-aligned current variations, and those caused by the precipitation only. It can be seen from these figures that the main differences are in the thermospheric wind and ion temperature disturbances in the cusp region. The latter are much more intensive (up to about 600 K) at geomagnetic latitudes 75°-80° in case of the joint precipitation and field-aligned current action due to Joule heating of the ion gas in the cusp region, whereas the neutral temperature does not increase so greatly (Fig.24).
The most significant changes are in the zonal thermospheric wind variations due to the ion drag. Eastward wind disturbances of about 140-200 m/s appear at geomagnetic latitudes 75°- 80° in the midday sector and of about 200-300 m/s in the afternoon sector. The meridional wind disturbances are about 90 m/s at geomagnetic latitudes 80°- 85° in the midday sector and of about 180 m/s in the afternoon sector. They are also caused by the ion drag which acts in the opposite direction, in comparison with the pressure gradient forcing in the midday sector.
The response of electron density to the field-aligned current disturbances (Fig.23) is very localized near the 80° geomagnetic latitude, being negative because of increase in ion temperature due to Joule heating of the ion gas. The electron temperature change is insignificant. The differences between the results of the calculations in variants 3 and 4 are seen in Figs.24 and 25 to be largely in the ion temperature and the zonal and meridional thermospheric wind variations at the geomagnetic latitudes 75°- 85°.
Thus, the main prominent feature of the results of our calculations is that the thermospheric disturbances outside the cusp are generated mainly by the thermospheric heating due to the soft electron precipitation. They reveal appreciable magnitudes at significant distances from the cusp region being noticeably larger in case of the moving region of the precipitation. For example, the meridional wind velocity disturbance at 65° geomagnetic latitude is of the same order as the background wind due to the solar heating, but is oppositely directed. We can conclude from these calculations that the most distinguishable disturbances outside of the cusp are those of the thermospheric wind. It means that Fabri-Perot interferometer observations outside of the cusp could be used as a means of remote investigation of the cusp dynamics.
The thermospheric disturbances propagate from the cusp to lower latitudes as large scale atmospheric gravity waves with the mean horizontal velocity of about 690 m/s. This speed is comparable with values of about 400-718 m/s estimated from the observations of AGW (with periods of about 60 min) during the October 1985 WAGS campaign (Rice et al., 1988; Williams et al., 1988). It is worth noting that at the geomagnetic latitudes equatorward from 70°, all disturbances have a similar form of time variation, almost independent of the time variation of the cusp disturbance. This arises because of the attenuation of the higher frequency harmonics.
The ionospheric disturbances have appreciable magnitudes at the geomagnetic latitudes 70°- 85°. The electron concentration and temperature disturbances are caused mainly by the ionization and heating processes due to precipitation. On the other hand, the ion temperature disturbances are influenced strongly by Joule heating of the ion gas due to the field-aligned currents and associated electric field disturbances in the cusp. The latter strongly influence the meridional and, in particular, the zonal wind disturbances via ion drag so these disturbances can reach values of about 200-300 m/s in the afternoon sector at 75°- 85° geomagnetic latitudes.
7.2. Seasonal effects
There have been many reported ground-based and satellite investigations of the thermospheric and ionospheric responses to the soft electron precipitation and field-aligned current variations in the cusp region (e.g., Shepherd, 1979; Kelly and Vickrey, 1984; Kofman and Wickwar, 1984; McCormac and Smith, 1984; Oliver et al., 1984; Robinson et al., 1984; Smith, 1984; Vennerstrom et al., 1984; Wickwar, 1984; Thayer et al., 1987; Sandholt et al, 1994; Wu et al., 1996) but there has not been any reported systematic observational picture of the seasonal behaviour of the thermospheric and ionospheric disturbances in the cusp region, partly due to the suppresive influence of the solar emission on the ionization and heating processes in the summer polar upper atmosphere.
In the present investigation we have studied mainly the seasonal effects in the thermospheric and ionospheric responses to the soft electron precipitation and field-aligned current variations, of the order of an hour in duration, in the summer and winter cusp regions simultaneously. It should be expected that seasonal effects in the thermospheric and ionospheric disturbances may be rather significant due to at least two factors: (1) seasonal variations in the background state of the undisturbed thermosphere and ionosphere, i.e. of the neutral, ion and electron densities and temperatures and electric conductivities (Fuller-Rowell et al, 1988; Sojka
and Schunk, 1989; Kirkwood, 1996); and (2) seasonal variations of the "input" parameters such as the precipitating particle fluxes and field-aligned current densities (Iijima and Potemra, 1976; Bythrow et al, 1982; Fujii and Iijima, 1987; Newell and Meng, 1988; Yamauchi and Araki, 1989; Lu et al., 1994, 1995). Correspondingly, two variants of the calculations have been performed both for the IMF By < 0.
In the first variant, we have performed the model calculations for the solstice conditions of 22 June 1987 (low solar activity). In these calculations, the model input data for the summer and winter precipitating fluxes and field-aligned currents have been taken as geomagnetically symmetric (i.e. symmetric relatively to the geomagnetic equator) and equal to those used earlier in variant 3 of our calculations for the equinoctial conditions (Namgaladze et al., 1996c, see section 7.1) to investigate only the effects related with the background state of the ionosphere and thermosphere.
In the second variant, the calculations have been performed for the events of 28-29 January 1992. In reality, not only the background state of the ionosphere and thermosphere is different in the summer and winter but the precipitating fluxes and FAC's may be different in the summer and winter hemispheres as well (see, e.g., Lu et al., 1995, and references therein). That is why we performed the second variant of the calculations where geomagnetically asymmetric input data for the summer and winter precipitating fluxes and field-aligned currents have been taken from the patterns derived by Lu et al. (1995) by combining data obtained from the satellite, radar and ground magnetometer observations for these events when precipitations were weaker but the magnetospheric convection was stronger than in the first variant. Lu et al. (1995) used the assimilative mapping of ionospheric electrodynamics (AMIE) technique, derived by Richmond and Kamide (1988), to estimate global "snapshot" distributions of high-latitude convection and field-aligned current by combining data obtained nearly simultaneously both ground and from space. The results of the model calculations will be compared with those obtained in the first variant and with the observations and the possible physical causes of the predicted seasonal ionospheric and thermospheric effects in the cusp regions will be discussed.
Thus, the first variant of the calculations takes into account only the seasonal effects due to the seasonal variations in the background state of the undisturbed thermosphere (higher neutral temperature and density in summer) and ionosphere (higher electron concentration and conductivity in summer) which influence very strongly the ionization and electron, ion and neutral heating rates as well as the ion-neutral momentum exchange. It is very difficult, if possible, to understand and predict these seasonal effects only on the qualitative assessment basis, without any numerical model calculations because different processes may influence oppositely the upper atmosphere parameters. For example, the soft electron precipitation increases the electron concentration in the F2 region whereas the enhanced ion temperature due to the enhanced Joule heating increases the ion loss rate and correspondingly decreases the electron concentration, both effects depend strongly on the background neutral gas temperature and density, etc. The merit of the numerical calculation is that it permits us to take into account many coupled physical processes in the ionosphere and thermosphere simultaneously.
The second variant of the calculations is more realistic because it takes into account not only the background seasonal variation of the ionospheric and thermospheric parameters but also includes the seasonal effects in the magnetospheric input parameters such as geometry and FAC intensity in accordance with data by Lu et al. (1995). As it has been shown below in the paper, these seasonal effects in the input parameters do not influence significantly the ionospheric and thermospheric responses to the precipitation and FAC variations in the cusp region; the effects of the background state, as well as the relation between the precipitation and FAC intensity, are much more important.
Figs.26 and 27 show the ionospheric convection and field-aligned current patterns derived by Lu et al. (1995) at 0155 UT on January 29, 1992, in the northern hemisphere (Fig.26) and at 0011 UT on January 28, 1992, in the southern hemisphere (Fig.27). By comparing these patterns with the corresponding spectrograms of precipitating particles, the following signatures have been identified by Lu et al. (1995):
1) For the cases studied, which all had an IMF with both By and Bz < 0 for more than one hour prior to the time when the patterns were derived, the cusp precipitation was encountered by the DMSP satellites in the postnoon sector in the northern hemisphere and in the prenoon sector in the southern hemisphere.
2) The pair of field-aligned currents near local noon, i.e., the cusp/mantle currents, are coincident with the cusp or mantle particle precipitation. Thus, these currents are generated on open field lines. In distinction, the FACs on the dawnside and duskside, i.e., the region 1 currents, are usually associated with the plasma sheet precipitation and, therefore, they are generated mainly on closed field lines.
3) Topologically, the cusp/mantle currents appear as an expansion of the region 1 currents from the dawnside and duskside and they overlap near local noon. When By < 0, in the northern hemisphere the downward FAC is located poleward of the upward current; whereas in the southern hemisphere the upward current is located poleward of the downward current.
1982 JAN 29 01:55 UT 12 ELECTRIC 1992 JA,N 29 01:55 UT 13 DOWNWARD
Fig.26. (a) The ionospheric convection pattern derived at 0155 UT on January 29, 1992, in the northern hemisphere. The pattern has a contour interval of 10 kV. The satellite trajectories which have been converted to apex coordinates are indicated as either dots (if the observations were made prior to 0155 UT) or plus signs (if they were made after 0155 UT). The solid arrows show the direction of the satellite motion. (b) Distribution of the field-aligned current density, with solid lines representing the downward current and dashed lines the upward current. The contour interval is 0.3 |i A/m2 , starting ± 0.1 |i A/m2. The total downward current integrated over the area poleward of 50° latitude is given at the upper right. The different magnetospheric plasma regimes are indicated by the different shadings. From Lu et al. (1995).
Fig.27. Patterns of the (a) ionospheric convection and (b) field-aligned current derived at 0011 UT on January 28, 1992, in the southern hemisphere. The contour interval for the field-aligned current is 0.4 |iA/m2, starting ±0.2 |iA/m2. From Lu et al. (1995).
To reproduce these peculiarities of the FACs and precipitation, we have constructed more or less regular approximations of the field-aligned current density and soft electron (with a Maxwellian characteristic energy of 50 eV) precipitating flux data shown in Fig.28 (two upper panels) and used them as inputs for the second variant of the model calculations. These approximations take into account such differences between the hemispheres as the shift of the precipitation maximum to the postnoon sector in the northern hemisphere and to the prenoon sector in the southern hemisphere, more intensive FACs in the southern hemisphere (by a factor of about 1.5 in comparison with those in the northern hemisphere), opposite polarities of FACs in the cusp region and more poleward position of the cusp in the southern hemisphere in accordance with the data by Lu et al. (1995).
225 225
FIELD-ALIGNED CURRENT DENSITY, A km"2
0.0 1.0 2.0 3.0 4 0
i i i i i i
50 eV ELECTRON FLUX, 109crrr2s-1
45 45
-60 -35 -10 15 40
ELECTRIC POTENTIAL, kV Fig.28. Approximations of the field-aligned current density and soft electron (with characteristic energy of 50 eV) precipitating flux data (two upper panels) used as inputs for the second variant of the model calculations. The bottom panel shows the calculated patterns of the ionospheric convection. The left plots correspond to the northern (winter) hemisphere, the right plots correspond to the southern (summer) hemisphere. The sun position is at the top of the figure.
So, the main differences between the inputs in variants 1 and 2 are the following. In variant 2, the solar activity is high, the input precipitating electron fluxes and FACs are geomagnetically asymmetric and stable during two hours, the precipitating electron fluxes are weaker and FACs are stronger than in variant 1.
7.2.1. Results of the model calculations in variant 1 (geomagnetically symmetric inputs, enhanced precipitation, enhanced FACs in the cusp region, quiet zone 1 FACs)
Using geomagnetically symmetric inputs, the same as in the calculations made by Namgaladze et al. (1996c) for the equinoctial conditions of 22 March 1987 (section 7.1), we have performed the model calculations for the solstice conditions of 22 June 1987. Electric field is calculated for given FAC (both electric field and FAC inter to the ionosphere together, but this effect is not considered). The bottom panel in Fig.28 shows the calculated electric field potential patterns at 0030 UT in the summer (left plot) and winter (right plot) polar regions. The maximum electric field intensities are of about 30 mV/m in the summer cusp region and 70 mV/m in the winter cusp region in comparison with 30 mV/m obtained by Namgaladze et al. (1996c) for the equinoctial conditions. This difference is due to the different background conductivities in the summer and winter polar caps not connected with each other by the geomagnetic field lines because they are open there (the closed geomagnetic field lines equalize the electric potential; the boundary between the open and closed geomagnetic field lines is at 76° geomagnetic latitude in the model used). It is the main cause of the seasonal
effects in the ionospheric and thermospheric responses to the precipitation and field-aligned current variations in the summer and winter cusp regions. It is interesting that the potential pattern which consists of three convection cells in equinox (Namgaladze et al, 1996c), transforms to the four-cell pattern in solstice in both summer and winter hemispheres.
^ ill ELECTRON CONCENTRATION DISTURBANCE, Д Lg N
ELECTRON TEMPERATURE DISTURBANCE, К
0 150 300 450 500
ION TEMPERATURE DISTURBANCE, K
Fig.29. Geomagnetic polar plots (latitudes 60°-90°) of the calculated ionospheric disturbances, i.e. the differences between disturbed and undisturbed values of the calculated ionospheric parameters, at h = 300 km in the northern (summer, left plot) and southern (winter, right plot) hemispheres at 0030 UT which is the time of maximum of the ionospheric disturbances. The sun position is at the top of the figure.
Figs.29 and 30 show the calculated ionospheric and thermospheric disturbances, i.e. the differences between disturbed and undisturbed values of the calculated electron concentration, electron, ion and neutral temperature, meridional (positive northwards) and zonal (positive eastwards) wind velocity at 300 km altitude in the summer (left plots) and winter (right plots) polar regions at the times of maximal disturbances (0030 UT for the ionospheric disturbances and 0040 UT for the thermospheric ones). A comparison of the results shown in the left and right columns in these figures demonstrates the seasonal differences between the effects caused by the joint action of the precipitation and field-aligned current variations in the summer and winter cusp regions.
The electron concentration disturbances are caused mainly by the precipitating electron impact ionization although there is also an increase of the ion O+ loss rate due to ion temperature and neutral composition disturbances, as well as ion transportation effects due to electromagnetic drifts. The maximum positive disturbance of lgNe due to the precipitation in the cusp is 0.7 in the summer cusp region and 0.9 in the winter one in comparison with 0.8 in equinox. The ion drift effects are seen well in the winter cusp region at 8085° geomagnetic latitudes ("hole" and "tongue" in the right top plot in Fig.29) but they are very weak in the summer cusp region.
The electron and ion temperature disturbances (middle and bottom panels in Fig.29) both are larger in the winter cusp region in comparison with the summer one and this seasonal effect is more distinct in the ion temperature because of Joule heating of ion gas which depends on the electric field intensity. Max ATe is 760 K in the summer and 1070 K in the winter cusp region in comparison with 950 K in equinox. Max ATi is 390 K in the summer and 610 K in the winter cusp region in comparison with 600 K in equinox.
The neutral temperature disturbance is larger in the summer cusp region in absolute values (top panel in Fig.30) but relatively to the quiet background temperature (about 670 K in the winter cusp region and 1000 K in the summer one) it is noticeably larger in the winter cusp region being of about 30% in comparison with 20% in the summer cusp region and in equinox. Correspondingly, the neutral composition disturbances (not shown) are larger in the winter cusp region where the O/N2 concentration ratio diminishes by about 45% at 75° geomagnetic latitude in comparison with 15% in the summer cusp region and 25% in equinox.
0 55 110 165 220
1 i i i i
NEUTRAL TEMPERATURE DISTURBANCE, K
EASTWARD WIND DISTURBANCE, ms*1
-200 -60 80 220 360
^ l l l l
NORTHWARD WIND DISTURBANCE, ms"1
Fig.30. Geomagnetic polar plots (latitudes 60°- 90°) of the calculated thermospheric disturbances, i.e. the differences between disturbed and undisturbed values of the calculated thermospheric parameters, at h = 300 km in the northern (summer, left plot) and southern (winter, right plot) hemispheres at 0040 UT which is the time of maximum of the thermospheric disturbances. The sun position is at the top of the figure.
The zonal thermospheric wind disturbances due to the ion drag are largest in the winter cusp region (southern hemisphere) where they reach values of about 260 m/s being directed westwards in the 1200-1400 MLT sector in comparison with 140 m/s eastward wind disturbances in this sector of the summer cusp region (middle panel in Fig.30). In equinox, the maximum zonal wind disturbances were of about 300 m/s being eastward in the norhern hemisphere and westward in the southern hemisphere.
The meridional wind disturbances are caused mainly by the pressure gradient forcing in the midday sector. They reach the largest values of about 320 m/s in the winter cusp region being directed equatorwards in comparison with 120 m/s in the summer cusp region (bottom panel in Fig.30) and 180 m/s in equinox.
So, all ionospheric and thermospheric disturbances caused by the precipitation and field-aligned current variations in the cusp are more intensive in the winter cusp region in comparison with the summer one.
m
70° S
E. 120 -o
Summer
Winter
65° S
-v
60° S
o .o
0:00 11:15
0:30 11:45
1:00 12:15
1:30 UT 12:45 MLT
Fig.31. The time variations of the meridional wind velocity disturbance (positive northward) at height 300 km at latitudes 70, 65 and 60° in the summer (solid lines) and winter (dashed lines) hemispheres (variant 1 of the calculations).
As for the speed of the horizontal propagation of the disturbances, it is apparently higher at the high summer latitudes in comparison with the winter ones as it is seen in Fig.31 which shows the time variations of the meridional wind velocity disturbance at height 300 km at latitudes 70, 65 and 60° in the summer (solid lines) and winter (dashed lines) hemispheres. Being estimated at the 70 - 65° latitude interval, the horizontal propagation speed is about 770 m/s in the summer hemisphere and about 680 m/s in the winter one. Evidently, this difference is due to the seasonal difference in the background temperature of the thermosphere and it depends on latitude, decreasing equatorwards.
7.2.2. Results of the model calculations in variant 2 (geomagnetically asymmetric inputs corresponding to the events of 28-29 January 1992, quiet precipitation, enhanced FACs in the cusp and zone 1 regions)
The second variant of the calculations (geomagnetically asymmetric inputs, enhanced convection, quiet precipitation) was performed by the following means. Starting from the "quiet" (i.e. undisturbed by the FACs in the cusp region) conditions at 0000 UT on 28 January 1992, we inserted the "disturbed" input values for the FACs in the cusp region in accordance with the data shown in Fig.28 and ran the model till 0200 UT. The cusp position and precipitation and FAC intensities were stable during this time period. Then we repeated the calculations from 0000 to 0200 UT without the FACs in the cusp region but with the same precipitating electron fluxes shown in Fig.28 as in the previous case. The difference between the results of these calculations made with and without the FACs in the cusp region we name as the "disturbance due to the FACs in the cusp region".
The bottom panel in Fig.28 shows the calculated patterns of the electric field potential at 0200 UT for the northern (winter; left plot) and southern (summer; right plot) hemispheres. A comparison of these patterns with those obtained by Lu et al. (1995) (Figs.26 and 27) shows a good agreement between them. It is not so trivial that our model result is in a good agreement with Lu's convection, because the «external» part (FACs) is not absolutely the same, and beside of that we used our own ionospheric conductivity calculated simultaneously with convection using precipitating electron fluxes. The agreement means that our approximation of Lu's FACs and our conductivity model permits us to obtain quantitatively a correct distribution of the electric potential both in the summer and winter hemispheres. It is worth pointing out that the data of Lu et al. (1995) correspond to the different but adjacent days of 28 and 29 January 1992 when both IMF By and Bz components were negative, whereas our model calculations correspond to the single day of 28 January 1992 with the different FACs and precipitations in the northern and southern hemispheres. Nevertheless, both observed and calculated patterns reveal the similar differences between the hemispheres: (1) the electric fields are more intensive in the winter cusp region whereas FACs are larger in the summer one and (2) the zonal component of the ion flow at the geomagnetic latitudes > 72° is eastward in the winter cusp region and westward in the summer one.
These peculiarities of the ion flow are reflected very well in the calculated patterns of the horizontal thermospheric wind shown in Fig.32. The top panel in this figure shows the calculated patterns of the horizontal thermospheric wind velocity at height 300 km for 28 January 1992 in the northern (winter; left plots) and southern (summer; right plots) hemispheres. The bottom panel in Fig.32 shows the calculated patterns of the
wind disturbance, i.e. the difference between the wind velocities calculated with and without taking into account the FACs in the cusp region. These patterns demonstrate an appearance of the eastward wind disturbances of about 200 m/s in the afternoon cusp region in the northern (winter) hemisphere and the westward wind disturbance of about 100 m/s in the cusp region in the southern (summer) hemisphere created by ion drag due to the FACs related with the IMF By < 0. The wind disturbances are located more poleward in the summer hemisphere due to more poleward FACs in summer. These wind disturbaces lead to the total horizontal wind patterns shown in the top panel in Fig.32 with the oppositely directed zonal winds in the afternoon cusp region: eastward in the northern (winter) and westward in the southern (summer) hemisphere.
Fig.33 shows the calculated disturbances of the electron number density, electron, ion and neutral temperature due to the FACs in the cusp region at height 300 km along the meridian of 1319 MLT at 0200 UT on 28 January 1992 for the northern (winter; solid lines) and southern (summer; dashed lines) hemispheres. We can see from this figure that the summer thermospheric and ionospheric temperatures and densities react very weakly on the FACs in the cusp region whereas in the winter hemisphere there are noticeable disturbances especially in the ion temperature due to Joule heating. The electron number density disturbance is not large being negative due to the enhanced ion loss rate caused by the enhanced ion temperature (Schunk et al., 1976).
Now we can compare the results of the model calculations in variant 1 and 2 between themselves and with the observations and discuss the physical mechanisms of the seasonal effects in the response of the thermosphere and ionosphere to the FAC and precipitation variations in the cusp region.
28.01.92 0200 UT h = 300 km
HORIZONTAL WIND, m s"1
0 140 280 420 560
-60F 00 MLT NORTH (WINTER) 12 MLT
00 MLT SOUTH (SUMMER) 12 MLT
C
X
x
WIND DISTURBANCE, in s 1
100
x
X
X
X
□
Fig.32. The calculated patterns of the horizontal thermospheric wind velocity (top panel) and wind disturbance (bottom panel) at height 300 km for 28 January 1992 in the northern (winter; left plots) and southern (summer, right plots) hemispheres.
28.01.92 0200 UT 1319 MLT h=300 km
60 70 80 90 N
-60 -70 -80 -90 S
Magnetic latitude
Fig.33. The calculated disturbances of the electron number density, electron, ion and neutral temperature due to the FACs in the cusp region at height 300 km along the meridian of 1319 MLT at 0200 UT on 28 January 1992 for the northern (winter; solid lines) and southern (summer; dashed lines) hemispheres.
7.2.3. Comparison between the model results in variants 1 and 2
In variant 1 of the calculations, ionospheric convection is weak everywhere except of the confined cusp region where the winter electric fields are more than twice larger than summer ones (the maximum values are 70 and 30 mV/m) due to the low ionospheric conductivities in the winter polar cap. In variant 2 of the calculations, ionospheric convection is stronger because of the larger input FACs in the cusp and zone 1 regions. Again, electric fields are larger in the winter cusp region in comparison with the summer one (the maximum values are 100 and 50 mV/m) despite of that FACs are larger in the summer hemisphere. This seasonal difference in the electric fields plays an important role in forming of the seasonal effects in the response of the thermospheric circulation, ion temperature and electron number density to the FACs in the cusp region.
When comparing the thermospheric wind disturbances calculated in variant 1 and 2 shown in Figs.30 and 32, we can see well that in variant 1 both zonal and meridional components of the wind velocity are of the same order of the magnitude (the eastward wind velocity disturbanes are between -260 and +160 m/s and the northward ones are between -200 and +360 m/s) both being larger in the winter hemisphere whereas in variant 2 of the calculations the zonal wind disturbance dominates having maximum values of about 200 m/s in the winter hemisphere and 100 m/s in the summer one. This difference between the variants has been caused by the difference in the thermospheric temperature disturbances which are significant in variant 1 due to the enhanced soft electron precipitation and insignificant in variant 2 (cf. Figs.30 (top panel) and 33) when precipitation is the same as under the quiet conditions. Correspondingly, in variant 2 the wind disturbances are driven mainly by the ion drag, whereas in variant 1 both ion drag and pressure gradient forcings have an important influence on the wind pattern. Neutral gas pressure gradient is larger in the winter cusp region (see top panel in Fig.30) as well as electric fields so the total wind disturbances are larger in the winter hemisphere in both variants of the calculations.
Ionospheric F2 region electron number density and electron temperature disturbances both are positive due to the enhanced precipitation and significantly more intensive in variant 1 of the calculations in comparison with the variant 2 (cf. Figs.29 and 33) where the electron concentration disturbances are negative due to the enhanced ion loss rate. The ion temperature disturbances, in contrast, are larger in variant 2 of the calculations in comparison with the variant 1 due to the larger electric fields in variant 1. They are larger in the winter cusp region again due to the larger electric fields in the winter polar cap in comparison with the summer one.
In all parameters, the effects of the input source asymmetry are not so appreciable as the seasonal effects due to the seasonal variation of the background state of the ionosphere and thermosphere.
7.2.4. Comparison with the observations
The most extensive data set about the behaviour of the thermosphere and ionosphere in the northern and southern polar caps has been obtained by the Dynamics Explorer satellites (Killeen andRoble, 1988). These
data have been analyzed by many authors, among them by Rees et al. (1986) and Roble et al. (1987, 1988a) using their thermospheric general circulation models and a good general agreement has been found between TGCM-predicted neutral winds and DE-2 observations showing the dominant influence of magnetospheric convection on the high-latitude circulation.
Fig.34 shows the geomagnetic polar plots of the mean thermospheric circulation measured on the DE-2 satellite (Thayer et al., 1987) between November and January in the years 1981-1982 in the northern (left plot) and southern (right plot) hemispheres for the IMF By < 0. A comparison of this figure with the top panel in Fig.32 shows that calculated patterns of the thermospheric circulation are in a good agreement with the average circulation for the southern (summer) hemisphere obtained from DE-2 data, but for the northern (winter) hemisphere there is some disagreement at high latitudes in the afternoon sector. The most distinctive feature of the calculated pattern in the northern cusp region is the eastward flow in the afternoon sector of the cusp region, but it is absent in the average DE-2 data.
The DE-2 neutral wind vectors presented by Thayer et al. (1987) were averaged for two 3-month periods into bins of 5° magnetic latitude and 1-h magnetic local time, whereas our calculations have been made for the specific event and UT moment and have a more high spatial resolution. To contribute to the average pattern significantly, there should be sufficient amount of the events with IMF By < 0 during one or more hours (to influence on the winds via ion drag) when satellite orbit intersects (passes through) the afternoon sector of the cusp region. The eastward ion flow in the afternoon cusp region seen in the convection patterns for the northern hemisphere in Figs.26 and 28 is not an unusual phenomenon. It is present in all empirical models of the ionospheric convection for IMF By < 0, (e.g., Reiff and Burch, 1985; Heppner and Maynard, 1987; Weimer, 1995) so it should have a corresponding reflection in the thermospheric circulation during the period with the stable IMF By < 0 as it has been predicted by TGCMs (Rees et al., 1986) and as it was observed by the Fabry-Perot interferometer at Longyearbyen, Spitsbergen (78.2° N, 15.6° E, 75° mag. lat.) by McCormac and Smith (1984). They obtained that the zonal neutral winds averaged over 12 days near winter solstice in the period 1979 to 1983 when the IMF By was negative were eastward in the afternoon sector of the cusp region.
NORTH POLE 12
^ 500 m/s
18
.. AV
W'
60
£
00 00
Fig.34. The geomagnetic polar plots of the mean thermospheric circulation measured on the DE-2 satellite (Thayer et al., 1987) between November and January in the years 1981-1982 in the northern (left plot) and southern (right plot) hemispheres for the IMF By < 0.
At last it is interesting to compare our results with those obtained by Wu et al. (1996) which presented two detailed case studies of the ionospheric and thermospheric response to soft particle precipitation in the cusp/cleft region using multi-instrument observations from the DE-2 satellite during orbits 688 and 748, together with supporting model calculations. They used one-dimensional hybrid satellite track model (Deng et al., 1995) to calculate thermospheric and ionospheric structures below the satellite altitude employing various DE-2 measurements as inputs and upper boundary conditions. In both cases the IMF By was negative during several hours and the zonal winds were eastward everywhere along the tracks in the polar cap (orbit 688 passed through the cusp region in the prenoon MLT sector and orbit 748 did it in the afternoon MLT sector) in contradiction with the average pattern by Thayer et al. (1987) for IMF By < 0 in the northern hemisphere (Fig.34) but in
agreement with the results by McCormac and Smith (1984) and with our results (Fig.32). It is remarkably as well that their two cases reveal opposite behaviour of the electron density in the cusp region. The electron density was enhanced during orbit 688 when the precipitation flux, 630 nm volume emission rate, electron and neutral temperature all were significanly enhanced, and the electron density was decreased in the cusp region during orbit 748 when the precipitation was weak, the 630 nm emission and electron temperature disturbances were relatively small, the neutral temperature was undisturbed and only ion temperature was distinctively increased in the cusp region. All these differences between the cases correspond quite well to the differences between our variants 1 and 2 of the model calculations.
7.2.5. Summary and conclusions
The seasonal effects in the thermosphere and ionosphere responses to the precipitating electron flux and field-aligned current variations, of the order of an hour in duration, in the summer and winter cusp regions have been investigated using the global numerical model of the Earth's upper atmosphere. Two variants of the calculations have been performed both for the IMF By < 0.
In the first variant, the model input data for the summer and winter precipitating fluxes and field-aligned currents have been taken as geomagnetically symmetric and equal to those used earlier in our calculations for the equinoctial conditions. The soft electron precipitation has been increased ten times in comparison with background state in this variant as well as FACs in the cusp region whereas the FACs in zone 1 have been weak. It has been found that both ionospheric and thermospheric disturbances are more intensive in the winter cusp region due to the lower conductivity of the winter polar cap ionosphere and correspondingly larger electric field variations leading to the larger Joule heating effects in the ion and neutral gas temperature, ion drag effects in the thermospheric winds and ion drift effects in the F2 region electron concentration.
In the second variant, the calculations have been performed for the events of 28-29 January 1992 when precipitations were weaker but the magnetospheric convection was stronger than in the first variant. Geomagnetically asymmetric input data for the summer and winter precipitating fluxes and field-aligned currents have been taken from the patterns derived by Lu et al. (1995) by combining data obtained from the satellite, radar and ground magnetometer observations for these events. Calculated patterns of the ionospheric convection and thermospheric circulation have been compared with observations and it has been established that calculated patterns of the ionospheric convection for both winter and summer hemispheres are in a good agreement with the results by Lu et al. (1995). Calculated patterns of the thermospheric circulation are in a good agreement with the average circulation for the southern (summer) hemisphere obtained from DE-2 data for IMF By < 0 (Thayer et al, 1987) but for the northern (winter) hemisphere there is a disagreement at high latitudes in the afternoon sector of the cusp region. At the same time, the model results for this sector agree with the DE-2 data analyzed by Wu et al. (1996) and with the ground-based FPI data by McCormac and Smith (1984). This contradiction is a question to be tested by the future observations in the cusp region such as EISCAT Svalbard Radar and optical measurements. All ionospheric and thermospheric disturbances in the second variant of the calculations are more intensive in the winter cusp region in comparison with the summer one and this seasonal difference is larger than in the first variant of the calculations, especially in the electron density and all temperature variations. This means that the seasonal effects in the cusp region are stronger in the thermospheric and ionospheric responses to the FAC variations than to the precipitation disrurbances.
8. Numerical modelling of the behaviour of the Earth's upper atmosphere during geomagnetic storms
In contrast with magnetospheric substorms, magnetic storms are more long-time phenomena, from several hours to several days in duration. To model them, it requires to perform long model runs such as it has been done, for example, by Volkov et al. (1996b) in their numerical simulation of the field-aligned current and electric field variations and corresponding ionospheric and thermospheric effects using the coupled ionosphere-thermosphere-magnetosphere model. In this work, the zone-2 field-aligned current formation and corresponding electric field variation and their ionospheric and thermospheric effects have been calculated using the global numerical model of the Earth's upper atmosphere (Namgaladze et al., 1995b) supplemented with a magnetospheric block (Volkov et al., 1996a) containing the magnetohydrodynamic continuity, momentum and energy balance equations for the magnetospheric plasma. These equations have been added to the modelling equation system for the ion, electron and neutral gasses of the ionosphere, protonosphere and thermosphere including the equation for the electric field potential to be solved jointly. The geomagnetic field is considered as a dipole one at latitudes equatorward from the polar cap boundary and having the field lines opened inside the polar cap.
The magnetosphere is empty at the closed field lines at the initial time moment when the potenial drop across the polar cap appears suddenly together with the plasma source at the polar cap boundary. The initial
values of p and n have been taken as 2x 10- Pa and 0.2 cm- everywhere in the magnetosphere at the initial time moment (0000 UT) when the potential drop of 100 kV across the polar cap appears. After that this potential drop has been kept constant as well as the magnetospheric plasma pressure and ion density inside of the polar caps acting as electric field and plasma sources. As a sequence of this the redistribution of the plasma pressure takes place at the closed geomagnetic field lines leading to the plasma sheet and zone 2 field-aligned current formation. This formation has been calculated as well as the corresponding ionospheric and thermospheric effects for moderate disturbed conditions. Calculated spatial distributions of the electric field potential, plasma sheet pressure, field-aligned currents, ionospheric F2 region electron density, ion temperature as well as thermospheric temperature, composition and winds at various time moments are presented in Figs.35, 36 and 37.
ELECTRIC POTENTIAL, kV 0:30 UT
PLASMA PRESSURE, Pa d-nn I IT
7:00 UT
7:00 UT
В
ED . 0 37 . 5 25.0 12 . 5 0.00 -12.5 -25 . 0 -37.5 -50 . 0
»10
0.3000
I
0.1500 0.1125 0.0750 0.0375
1 . 2 0 . 0
I
Fig.35. Northern geomagnetic polar plots of the calculated electric field potential, plasma-sheet pressure, and zone 2 field-aligned currents at various UT moments. The starting time is 0000 UT.
The results of the calculations shown in Fig.35 demonstrate a good accordance with the well known experimental and theoretical data concerning a development of the magnetospheric electric field potential, plasma sheet and zone 2 field-aligned currents leading to the shielding of the inner magnetosphere from the penetration of the magnetospheric electric field (Iijima and Potemra, 1978; Harel et al, 1981; Lyatsky and Maltsev, 1983; Pudovkin and Zakharov, 1984; Heppner and Maynard, 1987; Peymirat and Fontaine, 1994).
NEUTRAL TEMPERATURE, К 225 0:00 UT
7:00 UT
LOG [ N (0) ] ¿0:00 UT
LOG[N(N2)]
0:00 UT
7:00 UT
7:00 UT
1160
1110
1060
1010
960
910
860
810
760
8 . 600 8.512 8 . 425 8 . 337 8 . 250 8 . 162 8 . 075 7 . 987 7 . 900
8 . 3 8 . 2 8 . 1 8 . 0 7 . 9 7 . 8 7 . 7 7 . 6 7 . 5
height 300 km
Fig.36. Northern geomagnetic polar plots of the calculated neutral temperature, atomic oxygen and molecular nitrogen number density at height 300 km at 0000 UT and 0700 UT. The starting time is 0000 UT for which thermospheric parameters are taken from the MSIS-86 model (Hedin, 1987) for quiet magnetic conditions.
Figs.36 and 37 illustrate the main well known features of the thermospheric and ionospheric responses to the long-time increase of the high-latitude electric field during a geomagnetic storm (Appleton and Ingram, 1935; Appleton andPiggot, 1952; Seaton, 1956; Duncan, 1969; Matuura, 1972; Mayr and Volland, 1973; Hays et al., 1973; Pr^ss, 1980; Chandra and Spencer, 1981; Brunelli and Namgaladze, 1988; Fuller-Rowell et al., 1994, 1996; Burns et al., 1995; Mikhailov et al., 1995, and references therein):
1) the enhancement of the neutral and ion temperatures due to Joule heating;
2) the decrease of the atomic oxygen number density and increase of the molecular nitrogen number density due to the upward and equatorward motions of the neutral gas;
3) the appearance of the vortexes in the thermospheric circulation due to the ion drag;
4) the large scale depletion in the electron number density at polar and subauroral latitudes (known as negative ionospheric storm) due to the neutral composition changes, and
5) the redistribution of the ionospheric plasma at polar and subauroral latitudes due to the electromagnetic plasma drifts.
HORIZONTAL WIND, m/s
LOG [Ne]
330 275 220 165 110
5 . 50 Б . 25 5.00 4 . 75 4 . 50 4 . 25 4 . 00 3.75
ION TEMPERATURE, К
0:00 UT
height 300 km
7:00 UT
Fig.37. North geomagnetic polar plots of the calculated horizontal thermospheric wind velocity and electron number density and ion temperature at height 300 km at 0000 UT and 0700 UT. The starting time is 0000 UT for which thermospheric winds and ionospheric parameters are calculated using the MSIS-86 model (Hedin, 1987) for quiet magnetic conditions.
Whereas the ionospheric storms are usually negative (electron density is decreased near the F2 maximum) at high and middle latitudes, they are positive (electron density is increased near the F2 maximum) at low latitudes (Matuura, 1972; Pr^ss, 1980; Brunelli and Namgaladze, 1988). Negative ionospheric storms are explained practically unanimously as consequence of the decrease of the concentration ratio O/N2 during magnetic storms resulting in a ion loss rate enhancement. The situation with positive ionospheric storms is more complicated. The positive phase of F2 region ionospheric storm at middle and low latitudes may be caused by winds and (or) electric fields which are generated during magnetic storms and move the ionospheric plasma
upward where the ion loss rates are slower (Matuura, 1972; Prylss, 1980; Brunelli and Namgaladze, 1988; Mikhailov et al., 1995). However, some authors believe that the main cause of the positive phase is the neutral composition changes in the low-latitude thermosphere, namely the increase of the concentration ratio O/N2 (Fuller-Rowell et al., 1994; Burns et al, 1995). To test these hypotheses, we performed numerical simulations for the magnetic storm of 24-27 January 1974 to obtain the global pattern of the thermospheric and ionospheric effects of this magnetic storm.
25 JANUARY 1974
Fig.38. Potential drop across the polar cap (kV) during 25 January 1974.
The global self-consistent numerical model of the thermosphere-ionosphere-protonosphere system including the magnetospheric block (Volkov et al., 1996a) has been used in this investigation. The storm input has been controlled by the electric potential drop across the polar cap (Fig.38) which has been derived from the variation of the planetary index of magnetic activity Kp during the magnetic storm of 25 January 1974 by the use of the emprirical equation: A^(kV) = 20 + 13Kp.
The precipitation parameters have been taken as following. The so-called "hot" electrons with the characteristic energy E = 3 keV have been taken in proportion to the ion density in the plasma sheet and normalized to the precipitation pattern by Hardy and Gussenhoven (1985). For the so-called "cold" electrons, the same precipitation parameters have been taken as for the quiet conditions: = 78°, 0mn = 68°, A0= 3.5°, E1 = 0.2 keV, E2 = 0.17 keV, Am1 = 330°, AA = 90°, Im1 = 3.8x109cm-2s-1, Am2 = 112°, A42 = 45°, Im2 = 2.9x109cm-2s-1. These parameters characterize the electron precipitations in the auroral zone.
Fig.39 (top panels) demonstrates the calculated changes in the neutral composition at the height 250 km caused by the magnetic storm, namely it shows the calculated latitudinal variations of the ratio R determined as
R = ( [N2] / [O]storm) / ( [N2] / [O^et),
where [N2] / [O]storm and [N2] / [O]quiet refer to the ratio of the molecular nitrogen number density to the atomic oxygen number density in storm and quiet conditions respectively.
The left and right plots in Fig.39 refer to 1250 UT and 2133 UT (the time of storm maximum) respectively. We can see that the results of the numerical simulation of the neutral composition changes agree in general with the observations obtained by AE-C satellite and with the MSISE-90 model. Also, we can see that although ratio [N2] / [O] in storm maximum increases more than by factor 6 in comparison with the quiet level at high latitudes, at low latitudes it decreases below the quiet level neither in the numerical simulation nor in the satellite data nor in the empirical model. Meanwhile, as the bottom panels in Fig.39 show, the ionospheric storm is positive at low latitudes. Thus, we can conclude that the positive phase of the ionospheric storm is created by thermospheric winds (Fig.40) which cause upwelling of the ionospheric F2 region plasma along the geomagnetic field lines to the heights where the ion loss rate is lower and at low latitudes this enhanced plasma is moved equatorwards by the oppositely directed thermospheric winds blowing from the northern and southern high latitudes.
It is interesting to notice that in the auroral zones, the electron density in the F2 region is lower than in the E region during the storm as it is shown in Fig.41, because the precipitations increase greatly the electron concentration in the E region whereas the neutral composition changes decrease the electron concentration in the F2 region.
_ 8
tr
g 7 С
z
Ж 5 I 4 ° 3
С
2
с
T 1 0
1
tr
-«г 0
T
25.01.1974
1250 UT
-i—i—i—i—i
model
h = 250 km
Theoi MSIS AE-C
J_i_L
Data
J_i_L
2133 UT
1 1 1 1 1 ■ i i i i 1
L ?'/ ° -
:
..... .....
-60 -30 0 30 60 Geodetic latitude 19E (14.1 LT)
90
-60 -30 0 30 60 Geodetic latitude 273E (15.8 LT)
90
Fig.39. Latitudinal variations of the neutral composition at the altitude 250 km (top panels) during the magnetic storm of 25 January 1974 obtained from the numerical simulation (solid line), AE-C satellite data (Mikhailov et al., 1995; dashed line) and the empirical model MSISE-90 (Hedin, 1991; open circles). The bottom panels show the calculated latitudinal variations of the disturbance of f0 F2, determined as Af0 = f0 F2]storm - [f0F2]quiet , where subscripts "storm" and "quiet" refer to storm and quiet conditions respectively.
-90 -60 -30 0 30 60 90 -90 -60 -30 0 30 60 90 Geodetic latitude Geodetic latitude
19 E (14.1 LT) 273 E (15.8 LT)
Fig.40. The calculated meridional thermospheric winds (positive northward) at the height 250 km along the same meridians as in Fig.39. Solid lines correspond to the storm conditions, dashed lines correspond to the quiet conditions.
ELECTRON DENSITY, 1011m3
4 5 90
Magnetic latitude, (leg.
Fig.41. The height-latitude plots of the electron concentration in the quiet (left plot ) and storm (right plot)
conditions.
9. Numerical modelling of the turbulent energy dissipation in the high-latitude mesosphere
Until recently, the model was mainly used above 100 km, but recent ground-based measurements of turbulent intensities using the EISCAT VHF system have prompted an investigation of the feasibility of extending into the mesosphere. Hall et al. (1997b) reported the first results delivered by the model for predicting turbulent intensities in the height regime 80 to 90 km. According to them, the turbulent energy dissipation rate s is simply related to the eddy diffusion coefficient via:
K = CS/Wb2 ,
where ob is the Brunt-Vaisala frequency and C is a constant for which they use the value 0.81 as suggested by Weinstock (1982). Hall et al. (1997a,b) employ the maximum value, Km, and its height, zm, as functions of season and latitude from Danilov and Kalgin (1996). Linear interpolation was employed to obtain Km and zm at all latitudes. The seasonal variation is taken as being sinusoidal, with Danilov's and Kalgin's solstice values representing minima and maxima. The form of the profiles given by Danilov and Kalgin (1996), however, are unconvincing when compared with the wealth of profiles given by Hocking (1990) and, therefore, the shape of the profiles are derived from Shimazaki (1971):
K(z) = Km exp [ S (z - zm)2] , z > zm , (31)
K(z) =(Km - Ko) exp [ -S2 (z - zm)2 ] + Kq exp [ -S3 (z - zm)] , z < zm , (32)
where s} = s2 = 0.05 km-2, s3 = 0.07 km-2 and K0/Km = 1/5.
Turbulent energy dissipation rate, mW/kg
Fig.42. Estimates of the turbulent energy dissipation rate, s, as a function of height for June (left plot) and December (right plot) for EISCAT site 69° N, 19° E. The solid line shows the radar observation; the dotted line shows the model profile using the temperatures given by Lbbken and von Zahn (1991); the dashed line shows the model profile using the temperatures given by MSISE-90 (Hedin, 1991).
Fig.42 shows the results for June (left plot) and December (right plot). In each case the model has computed profiles of s using both the MSISE-90 and Lbbken and von Zahn (1991) temperature profiles as starting points. Considering the uncertainty in K, and also the C factor, the differences in the resulting profiles are negligible, although Lbbken and von Zahn (1991) temperatures do tend to increase the second derivative. The results of Hall (1997) using EISCAT measurements are shown for comparison. Nowhere are model and measurement more than a factor 2 apart. Since the two input temperature profiles give such similar results, Hall et al. (1997b) restricted themselves to the MSISE-90 temperatures.
Month
Fig.43. Seasonal variation of the turbulent energy dissipation rate, s (mW/ kg), as a function of height: top - as indicated by the model profile using the temperatures given by MSISE-90 (Hedin, 1991), bottom - as determined by the EISCAT VHF radar (Hall, 1997).
Fig.43 shows s from model (top) and EISCAT (bottom) as functions of height and month. Note that data were not available for the months of April and November from EISCAT, and interpolation is used to fill the gaps. Moreover, EISCAT experiments are run for 1 to 2 days at a time and on average each month is represented by 2 such experiments so that the monthly values are not necessarily representative for the whole month. The same general variation is seen in each case with the exception of the very top of the height regime. Above 86 km, the EISCAT results indicate a winter maximum in excess of 24 mW/kg, whereas the model indicates only just over 15 mW/kg. Below 84 km the agreement is very satisfactory and in both cases one can discern a tendency for the lowest turbulence intensities to be just on the spring side of the solstice.
Finally, the left plot of Fig.44 shows the data averaged from Fig.43 to obtain average energy dissipation rates for the height regime 80-90 km as functions of season. The EISCAT results show a higher variability, but general form of the variation is reproduced in each case. As shown in the right plot of Fig.44, the difference is considerably reduced if we take the average for the 80-85 km interval, although in this case, the summer minimum indicated by the model is centered on the solstice. At 90 km there is a summer minimum in s according to EISCAT, whereas the opposite is indicated by the model. In situ measurements from northern Scandinavia, however, support the model: higher s values are evident for July and August than winter months (Hall et al, 1997a; Lbbken et al, 1993).
Fig.44. Comparison of average turbulent energy dissipation rates, s, from EISCAT (Hall, 1997) (solid line) and the model using MSISE-90 (Hedin, 1991) temperatures (dashed line) for the height regime 80 to 90 km (the left plot) and 80 to 85 km (the right plot) as a function of season.
Thus, recent estimates of the turbulent energy dissipation rate, using the EISCAT VHF radar (Hall, 1997; Hall and Hoppe, 1997; Hall et al., 1997a) have been compared to the global model calculation results. The agreements between those and EISCAT for summer and winter solstice mesospheres are excellent when temperature is close to that given by the MSISE-90 model and eddy diffusion coefficient has the maximum value, Km, and its height, zm, as functions of season and latitude taken from Danilov and Kalgin (1996) combining with the shape of the profiles from Shimazaki (1971). The general seasonal variation has been investigated, again showing good agreement with the EISCAT results, although when examining the average energy dissipation in the 80-90 km height regime, the model shows less variability.
10. Summary and conclusions
The global numerical model of the Earth's upper atmosphere constructed at the Kaliningrad Observatory of IZMIRAN on the basis of the previous numerical models of the mid-latitude ionosphere, equatorial ionosphere, protonosphere and thermosphere was modified in last years at the Polar Geophysical Institute and Murmansk State Technical University for the studies of the high-latitude phenomena. The spatial and time resolution of the model was significantly enhanced by the use of the variable latitudinal steps of numerical integration. The latitudinal steps can be taken as small as 1 degree or even less at high latitudes instead of 5-10 degrees used in previous versions of the model, and the longitudinal steps can be decreased from 15 to 4 degrees. A new MHD magnetospheric block was incorporated in the model to calculate the zone 2 field-aligned currents instead of using them as input of the model.
As a next step of development of the model, the mesosphere has been included in consideration to use the model for interpretation of the mesosphere and lower thermosphere observational data. Previously, the
empirical model of the thermosphere MSIS-86 (Hedin, 1987) was used as initial and lower boundary conditions to calculate neutral gas temperature and density, but the lower limit of this model is 85 km and to extend the model to 80 km an extrapolation was used that was a rather rough approximation. Now, the MSISE-90 (Hedin, 1991) extending MSIS-86 to lower atmosphere is incorporated in the model to use it, first, as initial and lower boundary conditions (now lower boundary can be taken at any height between 60 and 80 km) and, second, for global calculations of the neutral gas temperature, density and composition at any height in parallels with (or instead of) self-consistent theoretical calculations of these parameters to compare the results obtained by the use of theoretical and empirical models of the neutral atmosphere. The possibility of using the empirical model atmosphere of Lbbken and von Zahn (1990) is also envisaged.
Thus, the high-latitude version of the global numerical model of the Earth's upper atmosphere describes the mesosphere, thermosphere, ionosphere, protonosphere and inner magnetosphere of the Earth as a single system by means of numerical integration of the corresponding time-dependent three-dimensional continuity, momentum and heat balance equations for neutral, ion and electron gases as well as the equation for the potential of the electric field both of magnetospheric and thermospheric (dynamo) origin.
The results of the model calculations for the quiet magnetic conditions have been compared with the data of the empirical ionospheric and thermospheric models as well as with the EISCAT data and, in general, reasonable agreement between theoretical and empirical data has been found. The new high-latitude version of the model has been applied as well to the investigations of the disturbed behaviour of the Earth's upper atmosphere during geomagnetic substorms and storms and during disturbances in the cusp region. The physical mechanisms of the upper-atmosphere responses to the solar wind and magnetospheric forcings have been understood using the model in the several case studies.
The numerical modelling of the behaviour of the ionospheric E and F-regions over EISCAT during the quiet day of 24 March 1987 and disturbed day of 25 March 1987 has shown that the present understanding of the ionospheric processes permits us to simulate them numerically and to describe the observed behaviour of the ionosphere over EISCAT not only qualitatively but to some extent quantitatively. The calculations have helped to divide the contributions of the ion heating and plasma transport into the main ionospheric trough dynamics during the disturbance. It has been found the equatorward movement of the midnight part of the trough is connected with the enhanced plasma transport while the apparent westward and eastward movements of the evening and morning edges of the trough are connected with "hot spots" caused by Joule heating of the ion gas.
The variations of the magnetospheric conductivity, needed to simulate numerically the behaviour of the field-aligned currents, electric fields and high-latitude ionosphere parameters in agreement with the observations have been found. During quiet geomagnetic conditions and at the substorm growth phase the distribution of the field-aligned currents and electric field potential in the high latitude ionosphere corresponds to the magnetospheric conductivity model which is uniform in longitude and drops exponentially with latitude equatorward from the polar cap boundary with the characteristic latitude scale of about the auroral zone width. During the substorm expansion phase a region of decreased (on about 30% in comparison with a ground state) plasma sheet electron content in the geomagnetic field tube appears at the midnight sector and travels westward with the speed of about 1 km/s at the ionosphere level forming the substorm current wedge. An appearence of such a region of decreased magnetospheric conductivity agrees with the central plasma sheet ion concentration decreases observed simultaneously with increases of their temperature during substorm expansion phase.
A comparison of the results obtained in the self-consistent variant of the calculations and in that using the MSIS-86 model shows considerable difference between the calculated thermospheric winds at ionospheric F2 region heights. The self-consistent calculated wind velocity disturbance reveals all typical features of the internal atmospheric gravity wave. It propagates equatorward from the auroral zone with an oblique front and at a speed of about 590 m/s at heights of about 400 km. This speed is close to values estimated from the observations of AGW during the October 1985 WAGS campaign. At the polar cap, an interaction takes place between the wave disturbances coming from the opposite sides of the auroral zone. In the evening sector, the wind disturbance is moving firstly polewards from the auroral zone, but afterwards the movement becomes equatorward - from the pole to the evening auroral zone - due to arrival of more intensive disturbance from the opposite side of the auroral zone where the neutral temperature disturbance is maximal. In the MSIS-86 variant, the wind disturbance is more localized and propagates much slower because of the thermosphere temperature and density being practically fixed during the event. However, this difference does not have much influence on the calculated substorm variations of the ionospheric parameters over EISCAT because of the small influence of the thermospheric winds on the high latitude ionosphere due to a large geomagnetic field inclination.
The results of the FPI-observations of the meridional thermospheric winds at the E layer heights poleward and equatorward of the auroral zone and their model calculations are in quite good agreement. In the calculations the drop of the average meridional wind velocity when crossing the precipitation zone from the pole
to the equator in the pre-midnight sector is displayed together with the intensive (of the order of the mean velocity) quasi-counterphase fluctuations at the latitudes to the pole and to the equator of the precipitation zone in accordance with the observations. Comparison of the results obtained in the self-consistent variant of the calculations and in that using the MSIS-86 model shows the insignificant role of local pressure gradient forcing due to the auroral heating of the thermosphere in comparison with the processes of the momentum exchange between ions and neutrals in forming the meridional wind velocity variations in the vicinity of the precipitation zone under the quiet conditions. Under non-stationary conditions when the precipitation intensity changes sharply, the role of the thermal processes becomes very important affecting the propagation character of the disturbances.
The numerical modelling of the responses of the ionosphere and thermosphere to the precipitation and field-aligned current variations in the cusp region has shown that the thermospheric disturbances outside the cusp are generated mainly by the thermospheric heating due to the soft electron precipitation. They reveal appreciable magnitudes at significant distances from the cusp region being noticeably larger in case of the moving region of the precipitation. For example, the meridional wind velocity disturbance at 65° geomagnetic latitude is of the same order as the background wind due to the solar heating but is oppositely directed. We can conclude from these calculations that the most distinguishable disturbances outside of the cusp are those of the thermospheric wind. It means that Fabri-Perot interferometer observations outside of the cusp could be used as a means of remote investigation of the cusp dynamics.
The thermospheric disturbances propagate from the cusp to lower latitudes as large scale atmospheric gravity waves with the mean horizontal velocity of about 690 m/s. This speed is comparable with values of about 400 - 718 m/s estimated from the observations of AGW (with periods of about 60 min) during the October 1985 WAGS campaign. It is worth noting that at the geomagnetic latitudes equatorward from 70°, all disturbances have a similar form of time variation, almost independent of the time variation of the cusp disturbance. This arises because of the attenuation of the higher frequency harmonics.
The ionospheric disturbances have appreciable magnitudes at the geomagnetic latitudes 70°- 85°. The electron concentration and temperature disturbances are caused mainly by the ionization and heating processes due to precipitation. On the other hand, the ion temperature disturbances are influenced strongly by Joule heating of the ion gas due to the field-aligned currents and associated electric field disturbances in the cusp. The latter strongly influence the meridional and, in particular, the zonal wind disturbances via ion drag so these disturbances can reach values of about 200-300 m/s in the afternoon sector at 75°- 85° geomagnetic latitude.
To investigate the seasonal effects in the thermosphere and ionosphere responses to the precipitating electron flux and field-aligned current variations, of the order of an hour in duration, in the summer and winter cusp regions, two variants of the calculations have been performed both for the IMF By < 0. In the first variant, the model input data for the summer and winter precipitating fluxes and field-aligned currents have been taken as geomagnetically symmetric and equal to those used earlier in our calculations for the equinoctial conditions. It has been found that both ionospheric and thermospheric disturbances are more intensive in the winter cusp region due to the lower conductivity of the winter polar cap ionosphere and correspondingly larger electric field variations leading to the larger Joule heating effects in the ion and neutral gas temperature, ion drag effects in the thermospheric winds and ion drift effects in the F2 region electron concentration.
In the second variant, the calculations have been performed for the events of 28-29 January 1992 when precipitations were weaker but the magnetospheric convection was stronger than in the first variant. Geomagnetically asymmetric input data for the summer and winter precipitating fluxes and field-aligned currents have been taken from the patterns derived by Lu et al. (1995) by combining data obtained from the satellite, radar and ground magnetometer observations for these events. Calculated patterns of the ionospheric convection and thermospheric circulation have been compared with observations and it has been established that calculated patterns of the ionospheric convection for both winter and summer hemispheres are in a good agreement with the results by Lu et al. (1995). Calculated patterns of the thermospheric circulation are in a good agreement with the average circulation for the southern (summer) hemisphere obtained from DE-2 data for IMF By < 0 (Thayer et al, 1987), but for the northern (winter) hemisphere there is a disagreement at high latitudes in the afternoon sector of the cusp region. At the same time, the model results for this sector agree with the DE-2 data analyzed by Wu et al. (1996) and with the ground-based FPI data by McCormac and Smith (1984). This contradiction is a question to be tested by the future observations in the cusp region such as EISCAT Svalbard Radar and optical measurements. All ionospheric and thermospheric disturbances in the second variant of the calculations are more intensive in the winter cusp region in comparison with the summer one and this seasonal difference is larger than in the first variant of the calculations, especially in the electron density and all temperature variations. This means that the seasonal effects in the cusp region are stronger in the thermospheric and ionospheric responses to the FAC variations than to the precipitation disturbances.
The results of the numerical modelling of the behaviour of the Earth's upper atmosphere during geomagnetic storms demonstrate a good accordance with the well known experimental and theoretical data concerning a development of the magnetospheric electric field potential, plasma sheet and zone 2 field-aligned currents leading to the shielding of the inner magnetosphere from the penetration of the magnetospheric electric field as well as the main well known features of the thermospheric and ionospheric responses to the long-time increase of the high-latitude electric field during a geomagnetic storm: 1) the enhancement of the neutral and ion temperatures due to Joule heating; 2) the decrease of the atomic oxygen number density and increase of the molecular nitrogen number density due to the upward and equatorward motions of the neutral gas; 3) the appearance of the vortexes in the thermospheric circulation due to the ion drag; 4) the large scale depletion in the electron number density at polar and subauroral latitudes (known as negative ionospheric storm) due to the neutral composition changes and 5) the redistribution of the ionospheric plasma at polar and subauroral latitudes due to the electromagnetic plasma drifts.
The results of the numerical simulation of the neutral composition changes agree in general with the observations obtained by AE-C satellite and with the MSISE-90 model. Although ratio [N2]/ [O] in storm maximum increases more than by factor 6 in comparison with the quiet level at high latitudes, at low latitudes it decreases below the quiet level neither in the numerical simulation nor in the satellite data nor in the empirical model. Meanwhile, the ionospheric storm is positive at low latitudes. Thus, we can conclude that the positive phase of the ionospheric storm is created by thermospheric winds which cause upwelling of the ionospheric F2 region plasma along the geomagnetic field lines to the heights where the ion loss rate is lower.
At last, recent estimates of the turbulent energy dissipation rate, using the EISCAT VHF radar have been compared to the global model calculation results. The agreements between those and EISCAT for summer and winter solstice mesospheres are excellent when temperature is close to that given by the MSISE-90 model and eddy diffusion coefficient has the maximum value, Km, and its height, zm, as functions of season and latitude taken from Danilov and Kalgin (1996) combining with the shape of the profiles from Shimazaki (1971). The general seasonal variation has been investigated, again showing good agreement with the EISCAT results, although when examining the average energy dissipation in the 80-90 km height regime, the model shows less variability.
Acknowledgments. This work has been supported by the grants No.RLX000 from the International Science Foundation, No.RLX300 from the ISF and Russian Government and NN.94-05-17321, 95-05-14505 from the Russian Foundation of Fundamental Investigations, from the SCOSTEP Bureau (1995), "Barents project" 020/94 from the Foreign Ministry of Norway and the grant 110192/730 from the Norwegian Research Council. We appreciate very much fruitful cooperation and discussions with our colleagues from the Kaliningrad Observatory of IZMIRAN Drs Yu.N.Korenkov, V.V.Klimenko, V.A.Surotkin, I.V.Karpov, F.S.Bessarab and V.M.Smertin having constructed many of the numerical algorithmes used in the model.
References
Appleton E.V. and Ingram L.J. Magnetic storms and upper-atmospheric ionization. Nature, v.136, p.548-549, 1935.
Appleton E.V. and Piggot W.R. The morphology of storms in the F2 layer of the ionosphere, 1, Some
statistical relationships. J.Atmos.Terr.Phys., v.2, p.36-252, 1952. Baumjohann W. Determination of the polytropic index in the plasma sheet. Geophys.Res.Lett, v.16, p.295-298, 1989.
Baumjohann W., Paschmann G., Nagai T. and Luhr H. Superposed epoch analysis of the substorm plasma
sheet. J.Geophys.Res, v.96, p.11,605-11,608, 1991. Bogdanov N.N., Galakhov V.A. and Leontyev S.V. Measurements of the temperature and wind velocity of
neutral gas at the high latitude thermosphere. Geomagn. Aeronomy, v.30, p.1016-1021, 1990. Bogdanov N.N. and Leontyev S.V. Meridional winds in the evening-midnight sector of the auroral oval.
Geomagn. Aeronomy, v.35, p.158-161, 1995. Bonnevier B.R., Bostrom R. and Rostoker G. A three-dimensional model current system for polar magnetic
substorms. J.Geophys.Res, v.75, p.107-122, 1970. Bostrom R. Mechanisms for driving Birkeland currents. Physics of the Hot Plasma in the Magnetosphere,
edited by B.Hultqvist, Plenum, New York, p.314-365, 1975. Brunelli B.E. and Namgaladze A.A. Physics of the ionosphere. (Monograph), Moscow, Publ.House "Nauka", p.527, 1988.
Burns A.G., Killeen T.L. and Roble R.G. A theoretical study of thermospheric composition perturbations during an impulsive geomagnetic storm. J.Geophys.Res, v.96, p.14,153-14,167, 1991.
Burns A.G., Killeen T.L., Deng W., Carignan G.R. and Roble R.G. Geomagnetic storm effects in the low- to middle-latitude upper thermosphere. J.Geophys.Res, v.100, No.A8, p.14,673-14,691, 1995.
Bythrow P.F., Potemra T.A. and Hoffman R.A. Observations of field-aligned currents, particles and plasma drift in the polar cusps near solstice. J.Geophys.Res, v.87, No.A7, p.5131-5139, 1982.
Chandra S. and Spencer N.W. Thermospheric storms and related ionospheric effects. J.Geophys.Res, v.81, p.5018-5026, 1981.
Chusovitin Yu.K., Shirochkov A.V., Besprozvannaya A.S., Gulyaeva T.L., Denisenko P.F., Armenskaya O.A., Ivanova S.E., Kashirin A.I., Klueva N.M., Koryakina E.A., Mironova L.S., Sykilinda T.N., Shushkova V.B., Vodolazkin V.I., Sotsky V.V. and Sheidakov N.E. An empirical model for the global distributions of density, temperature and effective collision frequency of electrons in the ionosphere. Adv.Space Res., v.7, p.49-52, 1987.
Collis P.N. and Häggström I. High resolution measurements of the main ionospheric trough using EISCAT. Adv. Space Res, v.9, p.(5)45-(5)48, 1989.
Collis P.N. and Häggström I. High-latitude ionospheric response to a geomagnetic sudden commencement. J.Atmos.Terr.Phys, v.53, p.241-248, 1991.
Cowley S.W.H. and Lockwood M. Excitation and decay of solar wind driven flows in the magnetosphere-ionosphere system. Ann.Geophys., v.10, p.103-115, 1992.
Deng W., Killeen T.L., Burns A.G., Johnson R.W., Emery B.A., Roble R.G., Winningham J.D. and Cary J.B. One-dimensional hybrid satellite track model for the Dynamics Explorer 2 (DE 2) satellite. J.Geophys.Res, v.100, p.1611-1624, 1995.
Danilov A.D. and Kalgin Y.A. Eddy diffusion studies in the lower thermosphere. Adv. Space Res., v. 17, p.(11)17-(11)24, 1996.
Dickinson R.E., Ridley E.C. and Roble R.G. A three-dimensional general circulation model of the thermosphere. J.Geophys.Res, v.86, p.1499-1512, 1981.
Dickinson R.E., Ridley E.C. and Roble R.G. Thermospheric general circulation with coupled dynamics and composition. J.Atmosph.Sci, v.41, p.205-219, 1984.
Duncan R.A. F-region seasonal and magnetic storm behaviour. JAtmos.Terr.Phys, v.31, p.59-70, 1969.
Evans J.V., Holt J.M., Oliver W.L. and Wand R.H. The fossil theory of nighttime high latitude F-region troughs. J.Geophys.Res.A, v.68, p.7769-7782, 1983.
Forbes J.M., Roble R.G. and Marcos F.A. Thermospheric dynamics during the March 22, 1979, magnetic storm, 2, Comparisons of model predictions with observations. J.Geophys.Res, v.92, p.6069-6081, 1987.
Fujii R. and Iijima T. Control of the ionospheric conductivities on large-scale Birkeland current intensities under geomagnetic quiet conditions. J.Geophys.Res, v.92, No.A5, p.4505-4513, 1987.
Fuller-Rowell T.J. A two-dimensional, high resolution, nested-grid model of the thermosphere 1. Neutral response to an electric field "spike". J.Geophys.Res., v.89, p.2971-2990, 1984.
Fuller-Rowell T.J., Codrescu M.V., Moffet R.J. and Quegan S. Response of the thermosphere and ionosphere to geomagnetic storms. J.Geophys.Res., v.99, p.3893-3914, 1994.
Fuller-Rowell T.J. and Rees D. A three-dimensional, time-dependent global model of the thermosphere. J.Atmosph.Sci, v.37, p.2545-2567, 1980.
Fuller-Rowell T.J. and Rees D. A three-dimensional, time-dependent simulation of the global dynamical response of the thermosphere to a geomagnetic substorm. J.Atmosph.Terr.Phys., v.43, p.701-721, 1981.
Fuller-Rowell T.J. and Rees D. Derivation of a conservation equation for mean molecular weight for a two-component gas within a three-dimensional, time-dependent model of the thermosphere. Planet.Space Sci., v.31, p.1209-1222, 1983.
Fuller-Rowell T.J. and Rees D. Interpretation of anticipated long-lived vortex in the lower thermosphere following simulation of an isolated substorm. Planet.Space Sci., v.32, p.69-85, 1984.
Fuller-Rowell T.J., Rees D., Quegan S., Bailey G.J. and Moffett R.J. The effect of realistic conductivities on the high-latitude neutral thermospheric circulation. Planet.Space Sci., v.32, p.469-480, 1984.
Fuller-Rowell T.J., Rees D., Quegan S., Bailey G.J. and Moffett R.J. Interactions between neutral thermospheric composition and the polar ionosphere using a coupled ionosphere-thermosphere model. J.Geophys.Res, v.92, p.7744-7748, 1987.
Fuller-Rowell T.J., Rees D., Quegan S., Moffett R.J. and Bailey G.J. Simulations of the seasonal and universal time variations of the high-latitude thermosphere and ionosphere using a coupled, three-dimensional model. Pure and Applied Geophysics, v.127, p.189-217, 1988.
Fuller-Rowell T.J., Rees D., Rishbeth H., Burns A.G., Killeen T.L. and Roble R.G. Modelling of composition changes during F-region storms: a reassessment. J.Atmosph.Terr.Phys, v.53, p.541-550, 1991.
Fuller-Rowell T.J., Codresku M.V., Moffett R.J. and Quegan S. Response of the thermosphere and ionosphere to geomagnetic storms. J.Geophys.Res, v.99, p.3893-3914, 1994.
Fuller-Rowell T.J., Codresku M.V., Rishbeth H., Moffett R.J. and Quegan S. On the seasonal response of the thermosphere and ionosphere to geomagnetic storms. J.Geophys.Res, v.101, p.2343-2353, 1996.
Hall C.M. Kilometer scale kinetic energy perturbations in the mesosphere derived from EISCAT velocity data. Radio Sci, v.32, p.93-101, 1997.
Hall C.M. and Hoppe U.-P. Characteristic vertical wavenumbers for the polar mesosphere. Submitted to Geophys.Res.Lett, 1997.
Hall C.M., Blix T.A., Thrane E.V. and Lübken F.-J. Seasonal variation of mesospheric turbulent energy dissipation rates over northern Scandinavia. Submitted to Geophys.Res.Lett, 1997a.
Hall C.M., Brekke A., Martynenko O.V. and Namgaladze A.A. EISCAT measurements and model calculations of the turbulent energy dissipation in the high-latitude mesosphere. VIII International EISCAT Workshop, Leicester, Abstracts, p.6, 1997b.
Hardy D.A., Gussenhoven M.S. and Holeman E. A statistical model of auroral electron precipitation. J.Geophys.Res, v.90, p.4229-4248, 1985.
Harel M., Wolf R.A., Reiff P.H., Spiro R.W. and Chen C.K. Quantitative simulation of a magnetospheric substorm, 2, Comparison with observations. J.Geophys.Res, v.86, p.2242-2260, 1981.
Hays P.B., Jones R.A. and Rees M.H. Auroral heating and the composition of the neutral atmosphere. Planet. Space Sci, v.21, p.559-573, 1973.
Haggstrum I. and Collis P.N. Ion composition changes during F-region depletions in the presence of electric field at auroral latitudes. J.Atmosph.Terr.Phys., v.52, p.519-530, 1990.
Hedin A.E. MSIS-86 thermospheric model. J.Geophys.Res, v.92, p.4649-4662, 1987.
Hedin A.E. Extension of the MSIS thermosphere model into the middle and lower atmosphere. J.Geophys.Res, v.96, p.1159-1172, 1991.
Hedin A.E., Biondi M.A., Burnside R.G., Hernandez G., Johnson R.M., Killeen T.L., Mazaudier C., Meriwether J.W., Salah J.E., Sica R.J., Smith R.W., Spencer N.W., Wickwar V.B. and Virdi T.S.
Revised global model of thermospheric winds using satellite and ground-based observations. J.Geophys.Res, v.96, p.7657-7688, 1991.
Heppner J.P. and Maynard N.C. Empirical high latitude electric field models. J.Geophys.Res, v.92, p.4467-4489, 1987.
Hines C.O. Internal atmospheric gravity waves at ionospheric heights. Can.J.Phys, v.38, No.11, p.1441-1481, 1960.
Hocking W.K. Turbulence in the region 80-120 km. Adv.Space Res, v.10, p.(12)153-(12)161, 1990.
Hoffman R.A., Fujii R. and Sugiura M. Characteristics of the field-aligned current system in the night-time sector during auroral substorms. J.Geophys.Res, v.99, p.21,303-21,325, 1994.
Iijima T. and Potemra T.A. The amplitude distribution of field-aligned currents at northern high latitudes observed by Triad. J.Geophys.Res, v.81, p.2165-2174, 1976.
Iijima T. and Potemra T.A. Large-scale characteristics of field-aligned currents at northern high latitudes observed by Triad. J.Geophys.Res, v.83, p.599-615, 1978.
Ivanov V.E., Kirillov A.S., Sergienko T.I. and Steen A. Modelling of the altitude distribution of green line (5577 Ц ) luminosity in aurora. Airglow and aurora, Ed.Leontiev, Proc SPIE 2050, p.105-107, 1993.
Ivanov-Kholodny G.S. and Nusinov A.A. Korotkovolnovoe izluchenie Solntza i ego vozdeistvie na verchnyuu atmospheru i ionospheru. Issledovaniya cosmicheskogo prostranstva (Investigations of Outer Space), VINITIPress, Moscow, v.26, p.80-154, 1987.
Jacchia L.G. Thermospheric temperature, density and composition: New models. SAO Special Report, No.375, p.106, 1977.
Kamide Y., Yasuhara F. and Akasofu S.-I. A model current system for the magnetospheric substorms. Planet.Space Sci, v.24, p.215-222, 1976.
Kan J.R., Zhu L. and Akasofu S.-I. A theory of substorm: Onset and subsidence. J.Geophys.Res, v.93, p.5624-5640, 1988.
Kan J.R., Sun W., Akasofu S.-I., Iijima T. and Potemra T.A. A global model of the substorm current system. Geophys.Res.Lett, v.19, p.969-972, 1992.
Kan J.R. A global magnetosphere-ionosphere coupling model of substorms. J.Geophys.Res, v.98, p.17,263-17,275, 1993.
Karpov I.V. and Namgaladze A.A. On causes of gas composition changes in the thermosphere. Geomagn. Aeronomy, v.28, p.246-250, 1988.
Kelly J.D. and Vickrey J.F. F-region ionospheric structure associated with antisunward flow near the dayside polar cusp. Geophys.Res.Lett, v.1, p.907-910, 1984.
Killeen T.L. and Roble R.G. Thermospheric dynamics: contributions from the first 5 years of the Dynamics Explorer. Rev.Geophys., v.26, p.329-367, 1988.
Kirkwood S. Lower thermosphere mean temperatures, densities, and winds measured by EISCAT: Seasonal and solar cycle effects. J.Geophys.Res, v.101, No.A3, p.5133-5148, 1996.
Klimenko V.V. and Namgaladze A.A. On the role of the convection in the formation of the trough and plasmapause. Geomagn. Aeronomy, v.20, p.946-950, 1980.
Kofman W. and Wickwar V.B. Very high electron temperatures in the daytime F-region at Sondrestrom. Geophys.Res.Lett, v.1, p.919-922, 1984.
Leontyev S.V., Namgaladze A.A., Namgaladze A.N. and Bogdanov N.N. Thermospheric meridional winds in the vicinity of the auroral zone: observation and modelling. Submitted to JASTP, 1997.
Lopez R.E., Spence H.E. and Meng C.-I. DMSP F7 observations of a substorm field-aligned current. J.Geophys.Res.A., v.96, p.19,409-19,415, 1991.
Lu G., Richmond A.D., Emery B.A., Reiff P.H., O.de la Beaujardiere, Rich F.J., Denig W.F., Kroehl H.W., Lyons L.R., Ruohoniemi J.M., Friis-Christensen E., Opgenoorth H., Persson M.A.L., Lepping R.P., Rodger A.S., Hughes T., McEwin A., Dennis S., Morris R., Burns G. and Tomlinson L. Interhemispheric asymmetry of the high-latitude ionospheric convection pattern. J.Geophys.Res, v.99, No.A4, p.6491-6510, 1994.
Lu G., Lyons L.R., Reiff P.H., Denig W.F., O.de la Beaujardiere, Kroehl H.W., Newell P.T., Rich F.J., Opgenoorth H., Persson M.A.L., Ruohoniemi J.M., Friss-Christensen E., Tomlinson L., Morris R., Burns G. and McEwin A. Characteristics of ionospheric convection and field-aligned current in the dayside cusp region. J.Geophys.Res, v.100, p.11,845-11,861, 1995.
Lbbken F.-J. and U. von Zahn. Thermal structure of the polar mesopause region at polar latitudes. J.Geophys.Res, v.96, p.20841-20857, 1991.
Lbbken F.-J., Hillert W., Lemacher G. and U. von Zahn. Experiments revealing small impact of turbulence on the energy budget of the mesosphere and lower thermosphere. J.Geophys.Res, v.98, p.20,369-20,384, 1993.
Lyatsky W.B. and Maltsev Yu.P. The magnetosphere-ionosphere interaction. Moskow, Nauka, 192p., 1983.
Makita K., Meng C.-I. and Akasofu S.-I. Temporal and spatial variations of the polar cap dimension inferred from the precipitation boundaries. J.Geophys.Res, v.90, p.2744-2752, 1985.
Maltsev Yu.P. An influence of the ionospheric conductivity on the convection system in the magnetosphere. Geomagn. Aeronomy, v.14, p.154-155, 1974.
McCormac F.C. and Smith R.W. The influence of the interplanetary magnetic field Y component on ion and neutral motions in the polar thermosphere. Geophys.Res.Lett, v.11, p.935-938, 1984.
McPherron R.L., Russel C.T. and Aubry M.P. Satellite studies of magnetospheric substorms on August 15, 1968. Phenomenological model for substorms. J.Geophys.Res., v.78, p.3131-3149, 1973.
Maeda S., Fuller-Rowell T.J. and Evans D.S. Zonally averaged dynamical and compositional response of the thermosphere to auroral activity during September 18-24, 1984. J.Geophys.Res, v.94, p.16,869-16,883, 1989.
Matuura N. Theoretical models of ionospheric storms. Space Sci.Rev., v.13, p.124-189, 1972.
Mayr H.G. and Volland H. Magnetic storm characteristics of the thermosphere. J.Geophys.Res, v.78, p.2251-2264, 1973.
Mikhailov A.V., Skoblin M.G. and Förster M. Daytime F2-layer positive storm effect at middle and lower latitudes. Ann.Geophys., v.13, p.532-540, 1995.
Millward G.H., Quegan S., Moffett R.J., Fuller-Rowell T.J. and Rees D. A modelling study of the coupled ionospheric and thermospheric response to an enhanced high-latitude electric field event. Planet.Space Sci, v.41, p.45-56, 1993a.
Millward G.H., Moffett R.J., Quegan S. and Fuller-Rowell T.J. Effects of an atmospheric gravity wave on the midlatitude ionospheric F- layer. J.Geophys.Res, v.98, p.19,173-19,179, 1993b.
Namgaladze A.A., Latishev K.S., Korenkov Yu.N. and Zakharov L.P. A dynamical model of the midlatitude ionosphere for the height range from 100 to 1000 km. Acta Geophysica Polonica, v.25, No.3, p.173-182, 1977.
Namgaladze A.A., Korenkov Yu.N., Klimenko V.V., Karpov I.V., Bessarab F.S., Surotkin V.A., Glushchenko T.A. and Naumova N.M. Global model of the thermosphere-ionosphere-protonosphere system. Pure and Applied Geophysics, v.127, p.219-254, 1988.
Namgaladze A.A., Korenkov Yu.N., Klimenko V.V., Karpov I.V., Bessarab F.S., Surotkin V.A., Glushchenko T.A. and Naumova N.M. Global numerical model of the thermosphere, ionosphere, and protonosphere of the Earth. Geomagn. Aeronomy (in Russian), v.30, p.612-619, 1990.
Namgaladze A.A., Korenkov Yu.N., Klimenko V.V., Karpov I.V., Surotkin V.A. and Naumova N.M. Numerical modelling of the thermosphere-ionosphere-protonosphere system. JAtmos.Terr.Phys, v.53, p.1113-1124, 1991.
Namgaladze A.A., Korenkov Yu.N., Klimenko V.V., Karpov I.V., Surotkin V.A., Bessarab F.S. and Smertin V.M. Numerical modelling of the global coupling processes in the near-earth space environment. COSPAR Coll.Ser., v.5, p.807-811, 1994.
Namgaladze A.A., Martynenko O.V. and Namgaladze A.N. Numerical simulation of substorm ionospheric variations observed by EISCAT by the use of the self-consistent ionosphere-thermosphere model including electric field calculations. VII International EISCAT Workshop, Cargese, Abstracts, p. 114, 1995.
Namgaladze A.A., Martynenko O.V., Namgaladze A.N., Volkov M.A., Korenkov Yu.N., Klimenko V.V., Karpov I.V. and Bessarab F.S. Numerical simulation of the ionospheric disturbance over EISCAT by the use of the global ionospheric model. JAtmos.Terr.Phys, v.58, p.297-306, 1996a.
Namgaladze A.A., Martynenko O.V. and Namgaladze A.N. Global model of the upper atmosphere with variable latitudinal steps of numerical integration. IUGG XXI General Assemblly, Boulder, 1995, Abstracts, GAB41F-6, B150, 1995b, and (in Russian) Geomagn. Aeronomy, v.36, p.89-95, 1996b.
Namgaladze A.A., Namgaladze A.N. and Volkov M.A. Numerical modelling of the thermospheric and ionospheric effects of magnetospheric processes in the cusp region. Ann. Geophysicae, v. 14, p. 13431355, 1996c.
Newell P.T. and Meng C.-I. Hemispherical asymmetry in cusp precipitation near solstices. J.Geophys.Res, v.93, No.A4, p.2643-2648, 1988.
Ohtani S., Potemra T.A., Newell P.T., Zanetti L.J., Iijima T., Watanabe M., Blomberg L.G., Elphinstone R.D., Murphree J.S., Yamauchi M. and Woch J.G. Four large-scale field-aligned current systems in the dayside high-latitude region. J.Geophys.Res, v.100, p.137-153, 1995.
Oliver W.L., Foster J.C., Holt J.M., Loriot G.B., Wickwar V.B., Kelly J.D., O. de la Beaujardiere, Bythrow P.F., Meng C.-I., Rich F.J. and Huffman R.E. Initial Millstone Hill, Sondrestrom, and HILAT observations of thermospheric temperatures and frictional heating. Geophys.Res.Lett., v.1, No.9, p.911-914, 1984.
Opgenoorth H.J., Pellinen R.J., Baumjohann W., Nielsen E., Marklund G. and Eliasson L. Three-dimensional current flow and particle precipitation in a westward travelling surge. J.Geophys.Res, v.88, p.3138-3152, 1983.
Peymirat C. and Fontaine D. Numerical simulation of magnetospheric convection including the effect of field-aligned currents and electron precipitation. J.Geophys.Res, v.99, p.11,155-11,176, 1994.
Prölss G.W. Magnetic storm associated perturbations of the upper atmosphere: Recent resultes obtained by satellite-borne gas analyzers. Rev.Geophys., v.18, p.183-202, 1980.
Pudovkin M.I. and Zakharov V.E. An investigation of the dynamic processes in the magnetospheric plasma. Magnetosph. Investigations, Sov.Radio, Moscow, No.3, p.67-85, 1984.
Rees D., Fuller-Rowell T.J., Gordon R., Heppner J.P., Maynard N.C., Spencer N.W., Wharton L.E., Hays P.B. and Killeen T.L. A theoretical and empirical study of the response of the high-latitude thermosphere to the sense of the "Y" component of the interplanetary magnetic field. Planet.Space Sci., v.34, No.1, p.1-40, 1986.
Reiff P.H. and Burch J.L. IMF By-dependent plasma flow and Birkeland currents in the dayside magneto sphere. 2. A global model for northward and southward IMF. J.Geophys.Res, v.90, p. 15951609, 1985.
Rice D.D., Hunsucker R.D., Lanzerotti L.J., Growley G., Williams P.J.S., Graven J.D. and Frank L. An
observation of atmospheric gravity wave cause and effect during the October 1985 WAGS campaign.
Radio Science, v.23, p.919-938, 1988.
Richmond A.D. and Kamide Y. Mapping electrodynamic features of the high-latitude ionosphere from localized observations, Technique. J.Geophys.Res, v.93, p.5741-5759, 1988.
Richmond A.D. and Matsushita S. Thermospheric response to a magnetic substorm. J.Geophys.Res, v.80, p.2839-2850, 1975.
Richmond A.D., Ridley E.C. and Roble R.G. A thermosphere-ionosphere general circulation model with coupled electrodynamics. Geophys.Res.Lett, v.19, p.601-604, 1992.
Robinson R.M., Evans D.S., Potemra T.A. and Kelly J.D. Radar and satellite measurements of an F-region ionization enhancement in the post-noon sector. Geophys.Res.Lett, v.1, No.9, p.899-902, 1984.
Roble R.G. and Ridley E.C. A thermosphere-ionosphere-mesosphere-electrodynamics general circulation model (TIME-GCM): equinox solar cycle minimum conditions (30-500 km). Geophys.Res.Lett, v.21, p.417-420, 1994.
Roble R.G., Forbes J.M. and Marcos F.A. Thermospheric dynamics during the March 22, 1979, magnetic storm, 1, Model simulations. J.Geophys. Res, v.92, p.6045-6068, 1987.
Roble R.G., Killeen T.L., Spencer N.W., Heelis R.A., Reiff P.H. and Winningham J.D. Thermospheric dynamics during November 21-22, 1981: Dynamics Explorer measurements and thermospheric general circulation model predictions. J.Geophys.Res, v.93, p.209-225, 1988a.
Roble R.G., Ridley E.C., Richmond A.D. and Dickinson R.E. A coupled thermosphere/ionosphere general circulation model. Geophys.Res.Lett, v.15, p.1325-1328, 1988b.
Rostoker G. and Eastman T. A boundary layer model for magnetospheric substorms. J.Geophys.Res, v.92, p.12,187-12,201, 1987.
Rothwell P.L., Silevitch M.B. and Block L.P. A model for the propagation of the westward travelling surge. J.Geophys.Res, v.84, p.8941-8948, 1984.
Sandholt P.E., Farrugia C.J., Burlaga L.F., Holtet J.A., Moen J., Lybekk B., Jacobsen B., Opsvik D., Egeland A., Lepping R., Lazarus A.J., Hansen T., Brekke A. and Friis-Christensen E. Cusp/cleft auroral activity in relation to solar wind dynamic pressure, interplanetary magnetic field Bz and By. J.Geophys.Res, v.99, p.17,323-17,342, 1994.
Schunk R.W. A mathematical model of the middle and high latitude ionosphere. Pure and Appl.Geophys, v.127, p.255-303, 1988.
Schunk R.W., Banks P.M. and Raitt W.J. Effects of electric fields and other processes upon the nighttime high-latitude F-layer. J.Geophys.Res, v.81, p.3271-3282, 1976.
Schunk R.W. and Sojka J.J. Ionospheric hot spots at high latitudes. Geophys.Res.Lett, v.9, p. 1045-1048, 1982.
Seaton M.J. A possible explanation of the drop in F-region critical densities accompanying major ionospheric storms. J.Atmos.Terr.Phys., v.8, p.122-124, 1956.
Sheen D.R. and Liu C.H. Modelling F-region gravity waves observed during the WAGS campaign 1. Special event. Radio Sci, v.23, p.879-904, 1988.
Shepherd G.S. Dayside cleft aurora and its ionospheric effects. Rev.Geophys.Space Phys, v.17, No.8, p.2017-2033, 1979.
Shimazaki T. Effective eddy diffusion coefficient and atmospheric composition in the lower thermosphere. J.Atmos.Terr.Phys., v.33, p.1383-1401, 1971.
Smith R.W. Interferometer observations of ion and neutral dynamics. The Polar Cusp, edited by J.H.Holtet and A.Egeland, D.Reidel, Norwell, Mass., p.243-260, 1984.
Sojka J.J. Global scale physical models of the F-region ionosphere. Rev.Geophys, v.27, p.371-403, 1989.
Sojka J.J. and Schunk R.W. A theoretical study of the high latitude F region's response to magnetospheric storm inputs. J.Geophys.Res, v.88, p.2112-2122, 1983.
Sojka J.J. and Schunk R.W. A theoretical F-region study of ion composition and temperature variations in response to magnetospheric storm inputs. J.Geophys.Res, v.89, p.2348-2358, 1984.
Sojka J.J. and Schunk R.W. A model study of how electric field structures affect the polar cap F-region. J.Geophys.Res, v.93, p.884-896, 1988.
Sojka J.J. and Schunk R.W. Theoretical study of the seasonal behaviour of the global ionosphere at solar maximum. J.Geophys.Res., v.94, p.6739-6744, 1989.
Sojka J.J., Schunk R.W. and Denig W.F. Ionospheric response to the sustained high geomagnetic activity during the March '89 great storm. J.Geophys.Res., v.99, p.21,341-21,352, 1994.
Surotkin V.A., Klimenko V.V. and Namgaladze A.A. A numerical model of the equatorial ionosphere. Investigation of the ionospheric dynamics, Moscow, IZMIRAN, p.58-68, 1979.
Taguchi S., Sugiura M., Winningham J.D. and Slavin J.A. Characterization of the IMF By-dependent field-aligned currents in the cleft region based on DE 2 observations. J.Geophys.Res., v.98, p. 1393-1407, 1993.
Thayer J.P., Killeen T.L., McCormac F.G., Tschan C.R., Ponthieu J.-J. and Spencer N.W. Thermospheric neutral wind signatures dependent on the east-west component of the interplanetary magnetic field for Northern and Southern Hemispheres as measured from Dynamics Explorer-2. Ann.Geophys., v.5, p.363-368, 1987.
Twerskoy B.A. About field-aligned currents in the magnetosphere. Geomagn. Aeronomy, v.22, p.991-995, 1982.
Vasyliunas V.M. The interrelationship of magnetospheric processes. "Earth's Magnetospheric Processes", edited by M.McCormac; D.Reidel, Norwell, Mass, p.29-39, 1972.
Vennerstrom S., Friis-Christensen E., Jorgensen T.S., Rasmussen O., Clauer C.R. and Wickwar V.B. Ionospheric currents and F-region plasma boundaries near the dayside cusp. Geophys.Res.Lett., v.1, p.903-906, 1984.
Volkov M.A. and Namgaladze A.A. Model calculations of the field-aligned currents and electric fields during the substorm expansion phase. Geomagn. Aeronomy (in Russian), v.36, p.(4)38-(4)44, 1996a.
Volkov M.A. and Namgaladze A.A. Models of field-aligned currents needful to simulate the substorm variations of the electric field and other parameters observed by EISCAT. Ann. Geophysicae, v.14, p.1356-1361, 1996b.
Volkov M.A., Martynenko O.V. and Namgaladze A.A. MHD-magnetospheric block for the global numerical model of the Earth's upper atmosphere. "Physics of Auroral Phenomena", Proc. XIX Annual Seminar, Apatity, p.101-102, 1996a.
Volkov M.A., Martynenko O.V. and Namgaladze A.A. Numerical simulation of the field-aligned current and electric field variations and their ionospheric and thermospheric effects using a joint ionosphere-thermosphere-magnetosphere model. International Conference on Problems of Geocosmos. Book of Abstracts. St.-Petersburg, p.114-115, 1996b.
Weimer D.R. Models of high-latitude electric potentials derived with a least error fit of spherical harmonic coefficients. J.Geophys.Res, v.100, p.19,595-19,607, 1995.
Weinstock J. Vertical turbulent diffusion in a stably stratified fluid. J.Atmos.Sci., v.35, p.1022-1027, 1978.
Wickwar V.B. Thermospheric neutral wind at 39° azimuth during the daytime sector at Sondrestrom. Geophys.Res.Lett, v.1, p.927-930, 1984.
Williams P.J.S., Growley G., Schlegel K., Virdi T.S., McCrea I., Watkins G., Wade N., Hargreaves J.K., Lachlan-Cope T., Muller H., Baldwin J.E., Warner P., A.P. van Eyken, Hapgood M.A. and Rodger A.S. The generation and propagation of atmospheric gravity waves observed during the Worldwide Atmospheric Gravity-Wave Study (WAGS). J.Atmosph.Terr.Phys., v.50, p.323-338, 1988.
Wu Q., Killeen T.L., Deng W., Burns A.G., Winningham J.D., Spencer N.W., Heelis R.A. and Hanson W.B. Dynamics Explorer 2 satellite observations and satellite track model calculations in the cusp/cleft region. J.Geophys.Res, v.101, p.5329-5342, 1996.
Yamauchi M. and Araki T. The interplanetary magnetic field By-dependent field-aligned current in the dayside polar cap under quiet conditions. J.Geophys.Res, v.94, No.A3, p.2684-2690, 1989.
Yamauchi M., Lundin R. and Woch J. The interplanetary magnetic field By effects on large scale field-aligned currents near local noon: contributions from cusp part and non-cusp part. J.Geophys.Res, v.98, p.5761-5767, 1993.
