The Causes and Effects of Adverse Space Weather.

Space weather refers to highly disturbed conditions on the sun, in the solar wind, magnetosphere, ionosphere, and thermosphere that can influence the performance and reliability of space-borne and ground-based technological systems and can endanger human life and health. Adverse changes in the near-Earth space environment can cause disruption of satellite operations, communications, navigation, and electric power distribution grids, leading to a variety of socioeconomic losses. This paper discusses some of the causes that lead to adverse space environment. The sources are believed to be on the sun. The propagation of these sources through the interplanetary space is reviewed. Finally, the interactions of the interplanetary disturbances with the earth’s magnetosphere that include bow shock, magnetopause, magnetosphere, and ionosphere are considered. The example of the June 24-28, 1999 event is given to demonstrate the solar/interplanetary/ magnetosphere inter-relationships. There is no doubt that the future COSMIC project will be important for the study of adverse space weather.


INTRODUCTION
In order to minimize the damage to technological systems that can result from severe geomagnetic disturbances, much attention has been paid to the prediction of storms and substorms (Joselyn, 1995). It is hoped that the solar eruptions can provide a key predictor, and the subsequent propagation of the solar generated disturbances to 1 AU that produce severe geomagnetic disturbances can be determined. At present the best understanding of the rela tionships between solar eruptions and resulting geoeffective solar wind events is statistical (e.g., Joselyn and Mcintosh, 1981;Hil dner, 1984, 1986;Gosling et al, 1991;Gos ling, 1993). For given solar wind parameters, such as the solar wind speed V, number density N, IMF B and possibly other parameters, the geomagnetic storms are modeled by estimation 1 Institute of Space Science, National Central University, Chung-Li, Taiwan, ROG 2Physics Department, National Cheng Kung University, Tainan, Taiwan, ROG 3National Space Program Office, Hsin-Chu, Taiwan, ROG *Corresponding author address: Dr. Jih-Kwin Chao, Institute of Space Science, National Central University, Chung-Li, Taiwan, ROG; E-mail:jkchao@jupiter.ss.ncu.edu.tw of the geoeffective parameters Dst and AE (e.g., Burton et al., 1975 ;Perreault and Akasofu, I 978; Akasofu and Chao, 1980;Sharma et al., 1993;Vassiliadis et al., 1995;Wu and Lundstedt, 1996;Chen et al, 1997).
On the other hand, the study of the solar source of geomagnetic storms has been continued for a long period (Dryer,1982 ;Gosling et al., 1991;Zhao and Hoeksema,1995 ;Hundhansen, 1993). The relation between solar flares and strong magnetic storms has long been recognized. Transient interplanetary (IP) shock waves have been associated with flares (Chao and Lepping, 1974;Hundhausen, 1972). However, many IP shocks are found no asso ciation with flares. Chao (1974) noted that the associations ofIP shocks with their flare origin are not totally satisfactory. The association of a shock wave at 1 AU with a particular flare is not always possible. Some shocks can be associated with large flares while some others can be attributed only to small ones. On the other hand, some large flares do not produce IP shocks near the earth. Later, Tang et al. (1989) showed that there is no correlation between the flare parameters and the strength of the IP shock at Earth. The sudden eruption of solar prominences has also been invoked as a source of geomagnetic perturbations (Joselyn and Mcintosh, 1981;Wright and McNamara, 1983). Their associations are not good (Bravo et al, 1999).
Coronal mass ejections (CMEs) were first observed in the 1970's as changes in coronal structure that occur on a time scale from a few minutes to several hours (Gosling, 1975 ;Dryer, 1982;Hundhausen, 1993). Observations of CMEs on the Sol wind coronagraph on board the P78l satellite have been compared with transient interplanetary shocks observed by the Helios 1 spacecraft from 1979 to 1983 by Sheeley et al. (1985). Virtually every shock ob served by Helios was preceded by a CME observed by Solwind. Since then, it has been widely accepted that CMEs are the pistons, which drive IP shocks ahead. When entering IP space, CMEs are often called interplanetary magnetic cloud (IMC). A high-density region between the preceding shock and the boundary of the IMC resembles the magnetosheath of the terres trial magnetosphere (Bravo et. al., 1999). Hence IMCs can be called as an interplanetary magnetosphere. IMCs contain coronal materials, which are much less dissipative than blast waves, and thus can propagate to a large distance in IP space. It is these IMCs, which often carry large southward IMF, and enhanced momentum flux due to compression by the pre ceded shock, that can generate adverse space weather (Gonzalez and Tsurutani, 1987).
During the IMC passage, the earth's bow shock and magnetopause will be compressed substantially. Often the magnetopause is pushed to the geosynchronous orbit (Shue et al., 1998). The positions of the bow shock can also fluctuate in large amplitude at short time intervals (Wu et. al. 2000, to be published). Under such interactions between the IMCs and the Earth's magnetosphere, energetic solar and magnetospheric charge particles (Baker et al., 1990), geomagnetic storms and magnetospheric substorms are also initiated. Adverse space weather is related to their occurrence. Within the magnetosphere, high-latitude convection pattern and the related electrodynamic parameters are changed under a direct result of solar wind/mag netosphere/ionosphere interactions (Richmond et al., 1998). Field-aligned currents and Alfven waves are also generated (Ma and Lee, 1999). It is anticipated that the whole ionosphere including the equatorial anomaly regions will be under the influence of adverse space weather.
In this paper, we will give the example of the June 24-29,1999 event to demonstrate a series of interactions starting from solar surface and ending on the ground. On the solar side, we use the data from SOHO's EIT and LASCO coronagraph data to identify the solar event. The source surface magnetic field data are obtained from Wilcox Observatory of Stanford University (Zhao and Hoeksema, 1995). A kinematic code (Hakamada and Akasofu, 1982) is used to calculate the propagation from the source surface to 1 AU. Interplanetary magnetic field and plasma data of WIND and Geotail are used as the upstream input parameters for predictions of the positions and shapes of the earth's bow shock and magnetopause. The IZMEM model (Papitashvili et al., 1999) is used to calculate the field-aligned currents in the polar region. Future calculations will use a more sophisticated AMIE code (Richmond and Kamide, 1988) for this purpose.
It is believed that the scheme we demonstrate is a useful one not only for understanding the physics of the couplings between different regions of the solar-terrestrial environment but also possibly for space weather prediction.

THE SOLAR SOURCE
The relation between solar flares, transient IP shocks and strong geomagnetic storms has long been recognized (Dryer, 1984). Therefore, solar flares were considered as the most likely solar cause for geomagnetic storms. However, more recent observations on board satellites from coronal and near-surface solar event measurements suggest that the source of the storms is coronal mass ejections (Sheeley et al;Harrison, 1994;Webb and Hundhausen, 1987). Re cently Brave et al (1999) found the percentage of solar associations of interplanetary magnetic clouds (IMCs) are 51 % for Ha. flare, 21 % for filament eruption, 7% for both of the previous two and 15% for neither of them. From all those studies, it is practically reas0nable to assume the solar source is the CMEs for space weather studies. In order to identify a solar source and use it for prediction purpose, we use a kinematic code. This code was designed by Hakamada and Akasofu (1982) and modified by Akasofu and Fry (1986) and Sun et al. (1985). This method combines the magnetic field frozen-in property and some observational property of the solar wind to construct a 3-D solar wind model. It is useful for the study of large structures in the solar wind particularly the large disturbances generated by IMCs. There are three impor tant assumptions made in the model: 1. The background solar wind speed variations are assumed to change with the solar magnetic latitude A as follows.
3. At 2.5 R Q (solar radii) from the sun, it is assumed that the solar wind flows with the frozen in magnetic field. 4. The source surface magnetic field at 2.5 Ro is routinely calculated from the Wilcox Solar Observatory of Stanford University. Therefore, the solar wind variations on the source surface are following the variations of the magnetic field. Solar disturbances caused by a CME or a filament eruption event (FD) are assumed to be spherical symmetric to the radial direction on the source surface and their intensities decrease from the center of the source following a Gaussian distribution The CMEs are from the coronagraph data of SOHO and FDs obtained from the Geophysi cal data published by World Data Center A. With the information of the initial disturbance and the ambient solar wind, simulation starts from the source surface at 2.5 R 0 from the Sun.

INTERPLANETARY SOURCE
Because of the rotation of the Sun, the solar wind and the disturbance entering the inter planetary space will interact with the ambient solar wind originating from different longitudes on the solar surface. This interaction can create additional source for geomagnetic storms.
Since the direct cause for storms is a large southward IMF Bz , we look for processes that can generate such a component.
A CME in general is composed of a bright loop, a dark region and a filament or promi nence close to the Sun (Hundhausen, 1993, Tsurutani andGonzalez, 1997;Tsurutani et al, 1999). When entering IP space, the material of the CME is called a driven gas (Bame et al., 1979;Hirshberg et al, 1970). Occasionally, magnetic fields of the given gas have the form of a magnetic cloud or giant flux rope ( Burlaga et al, 1987;Klein and Burlaga, 1982 ). This flux rope will have a Bz component. When the material carrying the magnetic cloud has a speed greater than the ambient solar wind by more than the ambient fast wave speed, fast shock wave will form. This MHD fast shock can compress the upstream magnetic field substantially. If a moderate southward Bz already exists upstream, a large southward Bz will be generated. When it reaches the magnetosphere, a large storm will be initiated. Fast MHD shock can generate the storm efficiently.
Interplanetary shock waves can be grouped into two types. The first type consists of corotating shocks, which are generated by interactions of solar wind streams. The lifetime of these streams may be longer or shorter than one solar rotation period. Hence, the corotating shocks do not necessarily have a recurrence tendency of 27 days (a solar rotational period).
The second type consists of transient shocks generated by IMCs. Non-linear large amplitude waves can steepen into fast shocks (Chao, 1973). Both these two types of shocks can amplify the ambient southward Bz to produce the interplanetary cause for geomagnetic storms. Nu merical and empirical models have been proposed for this generation mechanism.
The compressed region between the driver gas and the shock wave can be called the sheath region, which is generated in interplanetary space. In principle, the strength and the direction of this Bz can be predicted when the undisturbed source surface magnetic fields and solar wind speeds are known (Wu and Dryer, 1996). Large amplitude Alfven waves and turbulence when compressed by the shocks may also be the source for storms when large Bz' s are present. Tsurutani and Gonzalez (1997) have listed six types of possibilities of how large southward Bz are created: (1) shocked southward fields (Tsurutani et al., 1988), (2) bending of the heliospheric current sheets(HCS) (Tsurutani et al., 1984 ), (3) amplification of Alfven waves and turbulence (Tsurutani et al., 1995), (4) draped magnetic fields in the sheath region (Midgley and Davis, 1963; Zwan and Wolf, 1976;McComas et al., 1989), (5) equinoctial By effect (Russell and McPherron, 1973) and (6) fast stream-HCS interactions (Odstrcil and Pizzo, 1999). It is hoped that kinematic simulation can account for some of the above listed possibili ties.

MAGNETOSPHERIC EFFECTS
The supersonic solar wind impinges on the Earth' s magnetosphere generating the magne topause (MP) and bow shock (BS). Both MP and BS are never been found to disappear. During the recently observations by the ISTP satellites WIND, ACE, Geotail and IMP-8 on May 11, 1999, the number density of solar wind had dropped to below 1 per cubic cm for more than half a day. Both BS and MP have been found to cross some of these satellites at large distances from the Earth. On the other hand, under some extreme solar wind conditions when the high solar wind speed, number density and large southward Bz prevail, the MP and BS can be pushed much closer to the Earth. Sometimes the MP moves inside the geosynchronous orbit and some orbiting satellites may enter the magnetosheath and be exposed to the solar wind and fields. Some vulnerable satellites will have difficulties in coping with highly vari able fluctuations of the fields and energetic solar wind particles. Thus, forecasts of those geo synchronous MP crossings are very important to the safety of geosynchronous satellites.
The locations of the MP are not only important for modeling the magnetosphere but also essential in space weather forecasts. Models for the size and shape of the MP are plenty (Fairfield, 1971;Formisano et al., 1979;Russell, 1993,1996;Roelof and Sibeck, 1993;Shue et al.,1997Shue et al., , 1998. Only a few of them can be used for predictions. Shue et al. (2000) first compare two models (Petrinec and Russell, 1996;Shue et al., 1998) to test the capability of predictions of geosynchronous MP crossings by GOI;:Ss satellite using seven years of data. Yang et al. (2000) improve the prediction of Shue et al. (2000) by using a new model derived from a carefully selected database of MP crossings.
The models for the Earth's BS are also important for space weather studies (e.g., Fairfield, 1971;Formisano, 1979;Slavin and Holzer, 1981;Farris and Russell, 1994;Cairns and Lyon, 1995;Peredo et al., 1995;Bennett et al., 1997;Wu et al., 2000). Recently,  have selected a database of BS crossings from Geotail using only the mul tiple crossing events with quiet upstream conditions. The satellite WIND is used as a monitor to obtain the upstream parameters Dp, Bz, f3 and Mms, which are the solar wind dynamic pressure, IMF Bz, plasma beta and magnetosonic Mach number respectively. A model for the size and shape of the BS is thus derived. This model is able to predict the IMC induced BS crossings very accurately. As an example, the 26 Geotail's BS crossings, which are induced by the October 18-20, 1995 IMC event, are correctly predicted by this model except one. Solar wind disturbance induced BS and MP crossings for a selected event will be demon strated in the next section.
In Fig. 1, an example is given for the February 10 , 1969 event. It can be seen that the storm sudden commencement (ssc) starts at 20:2 1 UT and is followed immediately by intense substorm ac tivity as indicated by the AE index. The  I  3  8  9  It  IS  11  21  24  3  8  8  12  15  11  21  24 FEB. 10-11.1969 Fig.I. The interplanetary magnetic field (IMF) B, the three components (Bx, By, Bz), the solar wind speed V, the solar windmagnetosphere coupling function E, the magnetosphere substorm index AE and the ring current intensity index Ost for the February 10-11, 1969 storm (Akasofu and Chao, 1980) hours prior to the S.S.C. The geomagnetic storm starts after the S.S.C. as indicated by the Dst changes. The reasonably good correlation between the AE and the coupling function is apparantly noticed. This coupling function can be a good indicator for space weather predic tion.
It has been demonstrated in many studies that the large-scale ionospheric convection at high latitudes is primarily controlled by IMF B and solar wind dynamic pressure outside the magnetosphere. The couplings of solar-wind/ magnetosphere/ ionosphere determine the pat terns of high-latitude convection and related electrodynamic parameters in the ionosphere. Models of inner magnetospheric convection require knowledge of the electric potential distri bution around the polar cap boundary. Similarly, models of thermospheric dynamics need to know the plasma convection at high latitudes in order to model correctly the effects of ion drag and Joule heating. A model is designed for this kind of study, called AMIE (The Assimilative Mapping of Ionospheric Electrodynamics), which is used to synthesize collections of diverse data relating to high-latitude ionospheric electrodynamics into coherent patterns of conduc tivities, electric fields and currents, and related parameters (Richmond, 1992;Richmond et al., 1998). At present, AMIE is a specification model rather than a forecast model, although its mathematical structure could allow inclusion of time as an additional dimension, which would permit temporal extrapolation. Nonetheless, this specification model can be used to help ini tialize forecast models of thermospheric winds and composition, ionospheric electron density, and inner-magnetospheric particle populations. Another recent model designed for the study of ionospheric convection patterns is the IZMEM model (The IZMIRAN Electrodynamic Model). Both models can deduce the field-aligned current system in the polar cap. For a satellite at a typical altitude of 800 km the toroidal component of ionospheric current produces a relatively weak magnetic perturbation. By contrast, the field-aligned current system can produce relatively strong magnetic perturbations at satellite altitudes (Richmond and Kamide, 1988). A field-aligned current system calculated from the IZMEM model for the period June 24-29 is given in the next section. An origin of the field-aligned currents has been proposed by (Ma and Lee, 1999). They have carried out a three-dimensional compressible MHD simula tion to study the generation of field-aligned currents and Alfven waves by magnetic reconnection. The results indicate that the presence of IMF By leads to a shift of the reversal site between the downward and upward field-aligned currents that may contribute to the ob served region 1 field-aligned currents near noon in the polar ionosphere. This result can be incorporated into the AMIE model to study solar-wind/ magnetosphere coupling. THE JUNE 24-29, 1999 EVENT In this section, we present observations and analyses of a solar disturbance and the asso ciated IP disturbances that lasted for a little over two days and were observed at 1 AU by ISTP satellites. Such disturbances interact with Earth's bow shock and magnetopause causing their positions and shape to change. The interactions may also influence the polar as well as the equatorial ionosphere. The IPEI payload on ROCSAT-1 observes "bubbles" in the equatorial region of the ionosphere during the passage of this disturbance.

5.
(1) Identification of solar source A review of possible solar activities, which can be related to the interplanetary distur bance observed at lAU from 0200 UT of June 26 to 0300 UT of June 28, shows that two flares and one filament eruption (DSF) occurred at 1818 (N22E37),June 22, 0649(N23E42), June 23, and 1051 UT (N33E09), June 24, respectively, are the possible sources for the event. The observations provided by the LASCO and EIT on board SOHO also reveal solar disturbances ·at 1400 UT, June 24. Figure 2 shows the observations of the coronal disturbance and the flare activity by LASCO and EIT respectively. LASCO and EIT show a CME and a region of flare activity, respectively, at this time. This solar disturbance will be assumed as our solar source for this event.
(2) Interplanetary propagation These disturbances and solar wind will start from the source surface. The source surface of the magnetic field measurement is obtained from Wilcox Observatory of Stanford Univer sity. One Carrington Rotation (no. 1951) of the Solar Magnetic Field Synoptic Chart is shown in Fig. 3 where the projections of the locations of the Earth and the origin of the disturbances are indicated by a"*" and " 0 " respectively. With this information, the kinematic code is used to calculate the propagation of disturbances in 3-dimension interplanetary space. Solar wind is assumed to have a radial propagation from the source surface at 2.5 R0 from the Sun. The magnitude of the solar wind speed is assumed to be proportional to the magnetic field strength.
With the frozen-in condition assumed, the magnetic fields are carried to IP space. Without any disturbance on the source surface, the magnetic fields are assumed to be in the radial direc tions. When there is a disturbance added on the source surface with intensities of the velocities decreasing from the center of the source following a Gaussian distribution, the magnetic field will be stretched such that a non-radial component will be generated. This will be the source for Bz component. Figure 4 is a plot of the projected magnetic field line of force on the solar equatorial plane. Outward field is indicated by dash curves and inward field by solid curves. Compression and rarefaction of field lines can be easily noticed from the curves. The simula tion starts on 1818 UT, June 22 when the first disturbance is initiated. The circle is the posi tion of the Earth. The first plot shows the IP magnetic fields projected on the ecliptic plane at 0000 UT, June 24. The second one is for OOOOUT, June 25 when all the disturbances have already left the Sun. One can find that the first disturbance reaches 1 AU in late June 25. The kinematic code maps the source surface magnetic field structures in interplanetary space where the sectors of inward and outward magnetic fields are clearly seen. Fast and slow streams originating from different polarities of the solar surface can form the sector structures and interaction regions. Therefore, this code is also good for prediction of the arrival of corotation shocks due to fast-and slow-stream interactions. We would like to point out the discontinuity of field lines at longitude 0°or 360°. It is not real because the source surface given in Fig. 3 we use is not taken simultaneously. Since we are interested in the region far from this longitude, our results are not affected by this discontinuity. The simulated disturbance as seen at the Earth's position is shown in Fig. 5 where the solar wind radial velocity V, number density N, IMF B and its latitude 8 and longitude cI> are respectively shown from top to bottom. The disturbance arrives at the Earth in late June 25 and is so strongly compressed in its frontal part that a shock is formed at the leading edge. The whole event lasts until in early June 28. Smflll northward and then a little southward IMF Bz is observed inside the compressed sheath region    as can be noticed from the 8 changes. The cI> decreases first and then increases to 180°. IMF B is outward through out the disturbed period. Now, it is interesting to compare our simulations with the WIND and Geotail observa tions. Figure 6 shows the parameters of the solar wind and IMF B observed by WIND, which is about 220 Re upstream of the Earth. Interplanetary disturbance from 0200 UT, June 26 to 0700 UT, June 28 can be easily recognized from the magnitude of the IMF B measurements. (3) Earth's bow shock and magnetopause The disturbance in Fig. 6 lasts for two and half days. The parameters that control the size and shape of the BS and MP are Op, Bz, /3 and M . Before and after the disturbance, the solar ms wind plasmas and fields display much less fluctuation. Therefore, when the disturbance interacts with BS and MP, the positions of both BS and MP will change. The changes in position and shape of the BS are given in terms of two parameters y 0 and a where both are functions of the above four parameters. The y 0 and a determine the radial distance y of the BS by the following expression: y:::: : y 0( (1. +1])/(1. + 17cos8)) a , where ( y, 8) are the polar coordinates of the BS surface. The function form of y 0 and a is given by  where T] =1.03. These results can be used to predict the bow shock crossings for the Geotail, which is traveling in a region close to the BS and MP during this time. At this time, the ISTP satellites WIND and ACE are at the Lagrangian points upstream of the Earth. They should observe the solar wind disturbances 30-60 minutes before the Geotail. A magnetopause model  which has been tested to be quite successful for the predictions of the geosynchronous orbit satellite crossings (Yang et al., 2000) will be used to predict the Geotail's MP crossings caused by this disturbance where the upstream values observed by WIND are used. Now, the functional form for MP is similar to that of BS but y 0 and a are functions of Op and Bz only with TJ = 1.0. The predicted MP crossings are shown in the lower curve of Fig. 8 at 0100 and 0200 UT on June 29. Geotail also observed these two crossings as can be found in Fig. 7.

(4) Possible ionospheric and ground responses
This disturbance may cause the electrodynamic changes in the equatorial ionosphere.
The instrument IPEI on board ROCSA T-1 observed ionospheric bubbles during this period. A sudden increase in AE value at 0515 UT on June 28 is shown in Fig. 9 indicating the onset of a geomagnetic substorm. The other geomagnetic responses are found at Lunping station (Yumoto, 1995 ). The magnitude of B reaches a peak value at 0514 UT as shown in Fig. 10.
The power spectra for the time interval from 0510 to 0525 UT is shown in Fig. 11 indicating the presence of some low frequency wave activities. The polar ionosphere is strongly per turbed as the AE suddenly increases. The magnitude and the distribution of field-aligned cur rents are calculated by the IZMEN model (Papitashvili et al., 1999) and calibrated by the ion drift observations from DMSP satellites shown in Figs. 12(a) and 12(b) for the quiet and per turbed periods, respectively. The currents of both hemispheres in perturbed time increase to about twice that of the quiet time value. In the low latitude ionosphere, the ion density mea sured by IPEI is shown in Fig. 13 where each horizontal plot represents the measurement for one single orbit of ROCSAT-1. The horizontal axis shows the time for one orbit period of 97 minutes. The next curve from the bottom starts from the end of the previous one. Therefore, t h e vertical axis gives the numbers for the periods starting from the first one, which is in discrete numbers. These numbers can be converted to time such that the total time spans for the vertical axis is about two and half days. The density depletion regions in the plots show the presence of "bubbles" (Yeh et al., 1999). It is also interesting to note that the low latitude ionosphere shows less diurnal variation in number density of charged particles when the " bubbles " are present (please see Fig. 13). This may be caused by the compression of the magnetosphere during these periods, which might be related to the compression by the inter-   planetary disturbances and the increases in ionized oxygen in the night side region.

329
Research in solar-terrestrial physics has been conducted for many years. Only in the last few years have serious efforts been given to applying the findings for space weather predic tion. Since we still have many problems in each of the areas: the Sun, the interplanetary space and the magnetosphere as well as their couplings, the predictions we have attempted are very preliminary. Nevertheless, the scheme we outline above can offer useful results for space weather study.
On the solar side, the causes for adverse space weather need to be identified. Through  Fig. 12(b). Same plots as in Fig. 12(a) for the disturbed periods of June 28, 1999. The only input from interplanetary space is the IMF Bx and By. Note the enhancement of field-aligned currents during this period.
holes. These solar activities and the CME might be interrelated and the physics of their rela tionship is still not very clear. Therefore, for practical purposes, we take CMEs as the basis for the space weather prediction .. The ability to predict a CME release from the solar corona is still a long way off. The direct way to find the source is to observe the eruption of CMEs. The early satellites, like Skylab, SMM, and the recent SOHO can measure the CMEs seen only at the limb. The CMEs propagating earthward are observed as halo CMEs by SOHO. It is difficult to observe the head-on type CMEs. Hence, we take the next most likely source, a filament eruption as our disturbance. Once the source is selected, we have to describe how such a disturbance propa gates from the solar corona to the vicinity of the Earth.
Numerical simulations have been the most common practice for the propagation of solar interplanetary disturbances. Because of the non-uniform nature of the solar corona and inter planetary space, it is not easy to simulate such phenomena in 3-dimensional space with all the This kind of simulation can be performed even on a personal computer. Because it is ease to use and efficient, we hope to develop its capability for space weather prediction.
The BS and MP are the most important boundaries of the Earth 's magnetosphere. They protect us from the direct damage by the solar wind and some energetic particles. The predic tion of the positions and shape of the BS and MP is very essential for space weather prediction.
But, before a good prediction for the detailrd structures of the solar wind made from the solar source is available, predictions of changes of the locations and shape of BS and MP mainly rely on the upstream observations of the magnetosphere . Fortunately, the ISTP satellites, particularly the WIND and ACE are very useful for such purposes. Our models for the BS and MP respectively have demonstrated very accurate predictions for many events and hope they will be implemented for space weather prediction in the near future.
The responses inside the Earth 's magnetosphere due to the solar and interplanetary distur bances are under very active study particularly for space weather studies. The energy transfer function c needs further study so that the magnetosphere response can be more accurately calculated. The field-aligned currents due to interplanetary Alfven waves and rotational discontinuities need to be incorporated in the ionospheric circulation models such as the AMIE, IZMEN and . o thers. The global distribution of ionospheric election density and the polar region field-aligned currents obtained from the COSMIC project would provide valuable data for the study of ionospheric response to the adverse space weather.
In summary, we have demonstrated a scheme for modeling solar disturbances, which propagate through the interplanetary space and interact with the Earth 's magnetosphere caus ing changes of the BS, MP and the polar and equatorial ionospheres . Comparison with obser vations in these regions shows this prediction scheme warrents further development.