Abstract. The formula of Burton et al.  provides a quick and simple means by which the strength of the ring current and the Dst index can be calculated based solely on upstream solar wind density, velocity, and the north/south component of the magnetic field. Solar wind data from ISEE-3, the Pioneer Venus Orbiter, and Helios-A are used to show how well Dst predictions based on the Burton et al. formula match Dst observations for upstream monitors at various heliospheric distances. It is shown that a solar wind monitor that provides a substantial geomagnetic forecast lead time, with usable predictions of the characteristics of the solar wind can be stationed ~0.7 AU if it is near the ecliptic plane and within 10o; East to 5o; West of the Earth-Sun line. It is also shown that the Burton et al. formula predicts equally well the Dst index resulting from the passage of disturbed solar wind associated with either CMEs or stream-interaction regions. A solar sail-type spacecraft is proposed as the ideal monitor for forecasting purposes.
For the purposes of forecasting geomagnetic storms, it is desirable to make predictions with the longest lead-time possible. Linear prediction filtering and recurrence/persistence techniques have been used for a forecast lead-time of ~27 days [Bartels, 1932; Clauer et al., 1981; Bargatze et al., 1985]. These techniques are useful during the declining phase of solar activity when the stream structure of the solar wind is only slowly evolving. Recurrence forecast methods do not account for newly developing streams or transient solar activity that may effect the Earth's geomagnetic environment. In practice, it is not possible to predict activity associated with coronal mass ejections (CMEs) that are responsible for the largest terrestrial disturbances [Lindsay et al., 1995]. To predict the potential effects of the most recent solar activity (e.g., CMEs), space weather forecasters presently observe the optical and magnetic characteristics of the solar surface in an attempt to forecast the state of the geomagnetic environment approximately 4 days later (where ~4 days is the average transit time of the solar wind between the Sun and 1 AU). The true forecast lead-time will be a function of the velocity of the solar wind, and the potential strength of the geomagnetic storm will depend upon the solar wind velocity and density as well as the strength of the southward component of the IMF [Akasofu, 1964; Siscoe, 1966; Hirshberg and Colburn, 1969; Russell and McPherron, 1974]. Predictions of solar wind conditions at 1.0 AU based on solar observations [Hoeksema, 1992; Marubashi, 1986] are affected by changes in solar wind conditions between the Sun and the Earth. The best way to forecast interplanetary conditions at the Earth is to have a solar wind monitor on the streamline that intersects the Earth. The closer the monitor is to the Earth and to the Earth-Sun line, the more accurate the prediction of geomagnetic activity will be. The closer the monitor to the Sun, the greater the lead-time of the prediction will be. It is the purpose of this paper to provide an assessment of the effect of changing distance on the quality of geomagnetic activity forecasts using presently available data.
Burton et al.,  provided a relatively simple formula that allowed the accurate prediction of ring current strength and the Dst index based on upstream solar wind conditions near 1 AU. The assumed sources of change in the strength of the ring current are, first, the amount of injection (F(E)=d(Ey-0.5), where d = -1.5x10-3 nT((mV/m-1)s)-1 and Ey is the y GSM component of the interplanetary electric field) and, second, the amount of decay which is proportional to the strength of the ring current. In their formula, dDsto/dt = F(E)-aDsto, where F(E) is non-zero only for southward solar wind magnetic fields (the half-wave rectifier assumption), and "a" is an empirically derived constant representing the fractional loss of ring current per unit time (3.6x10-5 s-1). When the IMF is southward and the solar wind convective electric field (VBz) is sufficiently strong such that F(E) exceeds aDsto, the ring current is energized. When the IMF turns northward, F(E) is zero and the ring current is no longer energized, decaying and becoming weaker. Dst is the perturbation of the horizontal component of the Earth's magnetic field as measured by mid-latitude ground stations in nanoteslas. It is the sum of the ring current and the magnetopause currents and equals zero when the magnetopause currents and ring currents have their quiet day values. Burton et al.  obtained their ring current value by subtracting from Dst the correction term b(V2)1/2-c. The constants "b" and "c," also empirically derived, represent the response to dynamic pressure changes in the solar wind (15.8 nT(nPa)-1/2) and the quiet day currents (20 nT), respectively. Large, negative Dst characteristic of geomagnetic storms occurs when the ring current is strong (i.e., during periods of large, duskward Ey). Positive Dst occurs when the magnetopause currents are strong and the ring current is weak (i.e., during periods of strong solar wind dynamic pressure and northward Bz).
Forecasting Dst, then, is essentially a matter of predicting the solar wind conditions at the front of the magnetosphere. Burton et al.,  tested their Dst prediction formula during geomagnetically active periods using Explorer 34 solar wind and IMF data observed at the nose of the magnetopause (~1.0 AU). However, it has not been tested over an extended period of time, nor has the observation site for the IMF been evaluated at any point other than just outside the magnetopause. Particularly, the ISEE-3 data at 0.98 AU have not been used in such a test. The capability of using data well inside 1 AU provides even greater lead-time. If the Burton et al.,  formula could be used on solar wind data observed near 0.3 AU, the forecast lead-time would increase to about 3 days for average solar wind and ~1.5 days for an 800 km/s solar wind.
Unlike any previous study, this study examines the range of applicability of the Burton et al. formula by applying it to solar wind observations from ISEE-3, Pioneer Venus Orbiter (PVO), and Helios-A, to evaluate the usefulness of the Burton et al. formula during both geomagnetic storm and geomagnetically quiet times and to determine the usefulness of the Burton et al. formula using solar wind observed inside 1 AU. It is found that the simple formula devised by Burton et al.,  is remarkably robust and predicts the Dst index well during both geomagnetic storm and quiet periods using measurements near the L1 libration point despite the occasional differences seen in the direction of the IMF between this site and the Earth [e.g. Russell et al., 1980; Crooker et al., 1982]. Moreover, geomagnetically useful predictions with 24-hour forecast lead times can be achieved with an upstream solar wind monitor as far inside 1.0 AU as 0.7 AU during both geomagnetic storm and geomagnetically quiet times, but the prediction is less accurate during all conditions for more distant monitors closer to the Sun. The apparent reason for this modicum of predictability despite the large distances involved is that it is the long scale sizes in the IMF that control the ring current energization. Finally, the Burton et al. formula is found to work well regardless of the source of the conditions in the solar wind (e.g., stream interactions or coronal mass ejections).
ISEE-3: 1.0 AU Monitor
The first two figures show comparisons between Dst predicted from the Burton et al. formula (red trace) using ISEE-3 solar wind data near 1.0 AU and that observed (black trace) by mid-latitude ground stations. The observed Dst (obtained from the NSSDC database) has a 1-hr resolution while the ISEE-3 predicted Dst has a 5-minute resolution. ISEE-3 provides about a 45-minute lead-time between Dst prediction and Dst observation at 1.0 AU. The predicted Dst shown in Figures 1 and 2 has not been delayed by the expected solar wind convection time between ISEE-3 and 1.0 AU. We have not attempted to predict this delay because the convection time depends on the location of ISEE-3 perpendicular to the streamline of the solar wind passing through the stagnation point and the orientation of the structure convected past ISEE-3. This orientation will be different for coronal mass ejections (CMEs) and stream interaction regions.
|Fig. 1.540-day comparison between observed Dst (black) and predicted Dst (red) obtained from 5-minute resolution ISEE-3 solar wind measurements 1 September, 1978 - 23 February, 1980. Each data strip is 27 days.|
|Fig. 2. 27-day interval (10 February - 9 March, 1978) extracted from the 540-day period in Figure 1.|
Figure 1 shows a 540-day comparison obtained from the ISEE-3 interplanetary data set for the mission period when the plasma analyzer was functioning (1 September, 1978 to 23 February, 1980). To highlight any recurrent features, the length of each data strip is 27 days (approximately one solar rotation). At this resolution, Dst predicted by the Burton et al. formula is in excellent agreement with the actual Dst observations. For most of the time, the predicted and observed Dst values differ only slightly (an average amount of only ~5 nT). However, there are isolated periods where the agreement in magnitude differs by an average of ~50 nT (29-30 September, 1978; 21-24 February, 1979; 24-28 April, 1979; 3-9 July, 1979; hereafter referred to as cases a, b, c, and d, respectively). In cases a and d, the observed Dst is less disturbed than predicted indicating that the ring current was less energized than predicted. This difference suggests that less reconnection occurred than assumed by the form of F(E) in the Burton et al. formula. In cases b and c, the observed Dst is more disturbed than predicted, indicating that more ring current energization occurred than predicted.
These occasional differences suggest that additional factors are affecting the F(E) in the Burton et al. formula. Two possible factors are the beta of the magnetosheath plasma and the dynamic pressure of the solar wind. High beta values in the magnetosheath have been found to reduce the rate of reconnection [Paschmann et al., 1986; Scurry and Russell, 1991; Scurry et al., 1994]. The beta in the magnetosheath is controlled principally by the solar wind magnetosonic Mach number and partially by solar wind beta. During each of these four periods the solar wind was characterized by extremely low solar wind beta (beta<0.4). Beta in this low regime is most often associated with the passage of a CME [Klein and Burlaga, 1982]. Indeed, during a, b, and c, CMEs passing by 1.0 AU were identified by Gosling et al., . The low beta solar wind associated with case d is due to the passage of the high speed portion of a fast/slow stream interaction region characterized by lower than average densities (~3/cm3) and higher than average magnetic fields (~12 nT). Solar wind beta does not appear to be a factor in any of our four cases.
The solar wind magnetosonic Mach number also helps determine the value of beta in the magnetosheath and it too varies from cases to case. In cases a and d, the magnetosonic Mach number was relatively low (~2-3) during a period when the observed Dst was less than predicted implying reconnection was weaker than expected. For cases b and c, the magnetosonic Mach number was near average to slightly higher than average (~4-6) when the observed Dst was greater than predicted, implying reconnection was stronger than expected. This is opposite the behavior expected from the study of Scurry and Russell, . Their results indicate that the efficiency of reconnection is nearly constant for Mach numbers less than about seven, and that the efficiency drops above a Mach number of seven when the beta in the magnetosheath becomes high. Again we draw the conclusion that magnetosheath beta variations do not affect our four cases.
Dynamic pressure has been found to be an important factor in enhancing geomagnetic activity and by inference, the rate of reconnection [Scurry and Russell, 1991]. Common between cases b and c, where more ring current energization than predicted was observed, is high solar wind dynamic pressure (~10-30 nPa) associated with large IMF and a high likelihood of Bz being southward. Common between cases a and d, where less ring current energization than predicted was observed, is average to low solar wind dynamic pressure (~1-3 nPa) associated with large IMF magnitudes and an equal likelihood of Bz being either northward or southward. Thus, varying solar wind dynamic pressure appears to be the cause of the varying efficiency of coupling in our four cases.
Figure 2 shows a particular 27-day interval extracted from the 540-day period in Figure 1. In this enlargement it is easier to see how well the variation and level of activity are predicted using the formula of Burton et al. with ISEE-3 upstream solar wind measurements. (As noted above, the one exception in this interval occurs 21-24 February, 1979, a period of high solar wind dynamic pressure.) During this time, the times of the increases and decreases in Dst are well predicted, but the predicted level of activity (magnitude) is less than that observed. Interplanetary shocks (black arrows) and coronal mass ejections (white arrows) as identified by Gosling et al.,  using bi-directional electron signatures are indicated. Each interplanetary shock produces a sharp increase in both predicted and observed Dst. The first CME produces a large decrease in observed Dst that is not predicted. The second and third CMEs produce decreases in Dst that are correctly predicted.
Pioneer Venus Orbiter: 0.7 AU Monitor
Figure 3 shows a 27 day comparison between predicted Dst using 10 minute resolution solar wind data measured at 0.7 AU with Pioneer Venus Orbiter (PVO) and observed Dst at 1.0 AU from 1-28 June, 1980. During this period, PVO is within ±10o of the Earth-Sun line, and close to the intersection of the Venus orbital plane with the ecliptic plane. It is assumed that solar wind velocity does not vary appreciably between 0.7 AU and 1.0 AU and that the values of and Bz vary as r-2 and r-1, respectively. In this case, the solar wind monitor is 0.3 AU inside 1.0 AU. At a typical solar wind velocity of about 450 km/s, the solar wind plasma would take about 24 hours to reach the Earth from 0.72 AU. We have not taken this delay into account in Figure 3. Thus the predicted trace shown should be ~1 day in advance of the observed Dst at 1.0 AU. Over the initial 10 days of this period, changes in the magnetic field have very similar magnitudes and occur nearly simultaneously. The recovery of the PVO predicted Dst occurred on Day 164, two days ahead of when it occurred at Earth. Thereafter Dst is predicted to be quiet and it is. So during this one period of co-alignment, measurements at 0.72 AU could be used to predict the general level of activity although not the specific timing to better than a day.
|Fig. 3. 27-day comparison between observed Dst (black) and predicted Dst (red) obtained from 10-minute resolution PVO solar wind measurements at 0.7 AU from 1 - 28 June, 1980.|
This period of good correspondence occurred even though there were several transient disturbances in the solar wind. During the period shown, two stream interactions (on 5 June and 20 June, 1988) and two CMEs (on 10 June and 26 June, 1980) are observed at 0.7 AU. The first stream interaction is associated with a predicted increase in Dst to positive values. The second stream interaction produces only a slight increase in Dst to ~0 nT. After both stream interactions, Dst decreases and becomes negative. The first stream interaction region produces a significant decrease in Dst, while only a slight decrease in Dst occurs after passage of the second stream interaction region. This behavior is predicted as well as observed. In both cases, the maximum magnitude of positive Dst is under-predicted. The first stream interaction appears to have been more conducive to ring current energization (indicated by Dst<0) due to more southward Bz. The second stream interaction is more conducive to sustained magnetospheric compression as suggested by the substantial period of Dst>0 resulting from its higher dynamic pressures.
Both CMEs are associated with predicted and observed periods of large, negative Dst. The CME observed during the first period has a peak velocity of ~600 km/s and southward Bz of ~20 nT. The second CME has a peak velocity of ~400 km/s and southward Bz of ~15 nT. The longer duration of negative Dst following the first CME probably occurred because Bz was southward for a longer period during the first CME than the second CME.
Helios A: 0.6 to 1AU Monitor
Figure 4 shows a 54-day comparison (17 October-30 November, 1975) between Dst predicted from the Burton et al. formula (red line) using 1-hour resolution solar wind data from Helios-A and observed Dst (black line). During this period Helios-A is between ~13o to ~5 o East of the Earth-Sun line, 9o to 3o North of the ecliptic, and varies in radial distance from ~0.6 to ~1.0 AU. The forecast lead-time changes from ~2 days to ~1 hr during the period shown. Solar wind velocity and Bz have been scaled to 1.0 AU before calculating Dst. Because of the long time that Helios-A spends near the Earth-Sun line and the variation in radial and angular separation between Helios-A and the Earth, this pass provides an opportunity to evaluate the merits of various locations of an upstream solar wind monitor.
|Fig. 4. 54-day comparison between observed Dst and predicted Dst using 1-hour resolution Helios-A solar wind measurements from 0.6-1.0 AU 17 October - 30 November, 1975.|
When Helios-A is at radial distances between ~0.6 and ~0.7 AU (17-27 October), it is ~8o north of the ecliptic and ~10o East of the Earth-Sun line. The Dst predictions and observations agree in the level of activity, but the correspondence in predicted and observed variation is not clear. From 0.7 to 0.8 AU (28 October - 9 November), Helios-A is ~6o north of the ecliptic and ~7o east of the Earth-Sun line. The agreement between predicted and observed Dst, the level of activity and the timing of changes is not as good as we saw above with PVO at a similar distance. PVO was very near the ecliptic during the period discussed whereas Helios-A is ~6o north of the ecliptic. Since the average solar wind is not latitudinally uniform [Woo, 1988], it is possible that the solar wind conditions sampled by Helios-A are not observed at Earth.
When Helios is between ~0.8-1.0 AU (10 November-30 November), it travels from ~6o north to ~3o north of the ecliptic. Helios-A is at its closest approach to the Earth-Sun line (~5o) on 9-13 November. During this period Helios predicts well the general level of activity and the timing of events although the magnitude of one event on November 23-24 is predicted to be much stronger than it was observed to be. Note that the interplanetary shock occurring 13:40 UT on 8 November, 1975 (identified by Volkmer and Neubauer, 1985) produces a predicted 40 nT increase in Dst. One day later, a ~30 nT increase is observed as the interplanetary shock passes 1.0 AU. Although the forecast lead-time varies from ~1 day (10 November) to ~6 hours (28 November) as the radial separation between Helios and Earth decreases, on 17 November the observations appear to precede the predictions. We attribute these differences to the rather short scale length for magnetic changes in the solar wind.
The explanation behind the observation at 1.0 AU that the stream interaction effects are seen prior to the time predicted may be found by considering the characteristics of the stream interface that passes the Helios-A spacecraft on 18 November, 1975. If the interface is perpendicular to the ecliptic plane (upstream normal to the interface lies in the ecliptic plane), as it co-rotates with the Sun, it will first pass the easternmost spacecraft (Helios-A in this case) then it will pass Earth. This will be the case even if the easternmost spacecraft does not lie in the ecliptic plane. However, stream interfaces are often at angles other than 90o with respect to the ecliptic [Pizzo, 1982]. If the interface is tilted beyond perpendicular (upstream normal points north of the ecliptic plane), it is possible that the Earth will detect the interface prior to the easternmost spacecraft (assuming this spacecraft is north of the ecliptic as is Helios-A). If the interface is tilted less than perpendicular (upstream normal points south of the ecliptic plane), it is possible that the Earth will detect the interface after the easternmost spacecraft. To be seen simultaneously at Helios-A and Earth, an interface observed on 18 November would need to be tilted such that the normal is ~51o north of the ecliptic given the position of Helios-A with respect to the Earth. However, comparison of predicted and observed Dst indicates that Helios-A observes the interface ~6 hours after Earth observation. Therefore, the stream interface must be tilted such that the normal is ~71o north of the ecliptic. Examination of the flow deflections observed by Helios-A during this period show that the solar wind velocity has a large northward component prior to the interface, consistent with a tilt greater than perpendicular. Finally, we note that at the time of the large discrepancy on November 23 (Day 327), there were significant data gaps in the Helios 1 data so that the predictions are less reliable at this time.
The Burton et al. formula is simple in concept. This simplicity makes it very easy to use. The algorithm requires no large, long-term databases, very little computer space, and takes virtually no time to execute. The forecast lead-time is determined mainly by the time it takes the solar wind to travel from the solar wind monitor to the Earth and its accuracy depends principally on the evolution or variation of the solar wind and IMF from the point of observations and the nose of the magnetosphere. This model could be used in conjunction with a model that forecasts the solar wind properties at 1 AU if a sufficiently accurate model existed. Despite the existence of ongoing research in this area, no such model has yet been developed.
Figures 1 through 4 demonstrate that despite the simplicity and inherent limitations in the Burton et al. formula, it is possible to make useful predictions of Dst using this formula and solar wind measurements made between ~0.7 and 1.0 AU. However, the accuracy of this extrapolation decreases as the distance between the solar wind monitor and Earth increases. Obviously, the best prediction capability is provided by a solar wind monitor that samples solar wind conditions representative of those later seen at 1.0 AU. In evaluating possible positions of solar wind monitors, we have assumed that the differences between predicted and observed Dst resulting from limitations in the Burton et al. formula are negligible compared to those resulting from variations in spatial orientation. This is reasonable given the good long-term comparison shown in Figure 1.
It is preferable to have a large forecast lead-time. This can only be provided by a monitor as far inside 1.0 AU as possible. The inherent characteristics of the solar wind cause two problems in stationing a monitor far from 1.0 AU. First, inside ~0.5 AU, the solar wind exhibits many small-scale features which are not observed at 1.0 AU [Schwenn, 1990]. These features are purported to be swept up, "entrained," as the solar wind travels outward and are not seen by the time the solar wind reaches 1.0 AU [Burlaga et al., 1985]. Also, characteristics in the solar wind that do not vary in a simple, radially dependent manner have a longer distance over which to change unpredictably. In contrast, the variations in the solar wind density and Bz between 0.7 and 1.0 AU are reasonably well predicted at 1.0 AU using the simple assumption of r-2 and r-1 variation. So, although the separation between Helios-A and the ecliptic plane may have biased the conclusion that monitor positions inside ~0.7 AU are not useful, the known properties of the solar wind lead us to conclude that a solar wind monitor should be placed at least beyond 0.5 AU.
Forecast lead-time is also dependent upon the type of large-scale solar wind phenomena being observed. This is because the transit time from the solar wind monitor to 1.0 AU varies according to both the speed of the solar wind and the geometry of the disturbance relative to the spacecraft and the Earth. When the solar wind monitor and Earth are in near alignment, the time delay between observations of a CME at the two locations is just that determined by the velocity of the CME which is presumed to travel radially outward. Typically in the data examined here, no east/west timing bias is seen with regard to CME observations. Stream interactions are co-rotational solar wind features, so that the simple time delay between observations at the monitor and 1.0 AU will depend upon the radial separation as well as the east/west separation angle between the monitor and Earth. For example, when Venus is eastward to just westward of the Earth-Sun line, PVO will observe the stream interaction first. The stream interaction will then be observed at Earth at a time determined by the angular separation between Earth and Venus and the co-rotational speed and archimedean spiral angle of the interaction region. As PVO travels from east to west of the Earth-Sun line, the time between observations at 0.7 AU and 1.0 AU will decrease. When PVO is ~18° west of the Earth-Sun line, the stream interaction may be detected nearly simultaneously at 0.7 AU and 1.0 AU. Thus, a useful solar wind monitor should be deployed no further west of the Earth-Sun line than ~18° at 0.7 AU. To provide the largest forecast lead time for both CMEs and stream interaction passages, the monitor should be placed east of the Earth-Sun line but not so far east that the characteristics of the solar wind are uncorrelated with those that intersect the Earth.
A Solar Wind Monitor
Solar wind monitor placement generates another problematic issue as a consequence of the fact that the most appropriate location for a monitor is near the ecliptic plane and near radial alignment with the Earth-Sun line during the entire year. This means that the monitor must have the same angular velocity as the Earth. In a Keplerian orbit, the angular velocity of a satellite orbiting the Sun (accounting only for the gravitational effects of the Sun) is determined by the balance between the centrifugal force of the spacecraft (Fc) and the gravitational force of the Sun (Fg).
|m2r = GMm/r2||(1)|
Here, r is the heliocentric distance of the spacecraft, m is the mass of the spacecraft, M is the mass of the Sun, is the angular velocity of the spacecraft, and G is the universal gravitational constant. A spacecraft orbiting inside the orbit of the Earth has an angular velocity greater than that of the Earth. Thus, to keep a solar wind monitor synchronous with the Earth about the Sun, its angular velocity about the Sun must be slowed. At the L1 libration point where SOHO and ACE are presently stationed their angular velocity is reduced by the gravitational pull of the Earth. At distances closer to the Sun than the L1 point, other means of reducing the inward pull of the Sun must be found in order to achieve Earth orbital synchronization.
Non-Keplerian orbits are possible using thrust vectoring for continuous station keeping. However, conventional technology for continuous station keeping is expensive because of the fuel required to counteract the solar-gravitational pull throughout the duration of a long mission and the additional launch costs exacted by the higher spacecraft weight. Innovative technologies such as solar wind sails could provide a practical alternative for this type of application. A solar sail provides a large area against which the Sun's radiation pressure acts. The solar photon force thus provides an outward force in addition to the centrifugal force. The orbital force balance of the spacecraft (1) then becomes
|m2r+ Ps = GMm/r2||(2)|
Ps is the radiation pressure of the Sun acting on the solar sail. Expressed in terms of distance, R, measured in AU, and the irradiance of the Sun at 1AU, I, equation 2 becomes
|m2R R31 AU + 2IA R21 AU /R2c = GMm/r2||(3)|
where A is the area of the sail in m2, I is 1.38x103 W/m2, c is the speed of light and R1AU is 1.496 X 1011 m. The solar pressure force acts in conjunction with the centrifugal force to balance solar gravity and slow the spacecraft, making it possible to have a solar wind monitor inside 1 AU, yet orbiting synchronously with the Earth at 1.991 x 10-7 radians/s.
Solar sail technology has the advantage over conventional technology in that there are no fuel requirements for this type of station keeping. Solar sails do have mass that must be accounted for in the spacecraft mass budget. As seen from (3), the closer a spacecraft gets to the Sun (r gets smaller), the larger the sail area (As) must become to maintain a constant orbital speed (e.g., to keep the monitor synchronous with the orbit of the Earth). For example, if a mass of 25 kg is the allowable mass of a solar sail, Joel Sercel of JPL (personal communication, 1994) has estimated that a solar wind monitor could be stationed at ~0.98 AU using a solar sail of ~2.3x103 m2 in area. Currently produced materials are available with area densities of ~10 g/m2. This produces a sail mass of ~23 kg. The present practical lower limit on surface density appears to be about 3 g/m2 [Harris, 1995]. A monitor at ~0.98 AU provides ~2.5 hours of forecast lead-time on average (~1.25 hours for high speed CMEs). A solar wind monitor at ~0.75 AU would provide a more desirable forecast lead-time of ~25 hours on average (~12 hours for high speed CMEs). To station a monitor at ~0.75 AU, a sail with an area of ~1.8x104 m2 is required. This implies a maximum allowable area density of 1.3 g/m2. Sail materials of this low area density are not currently available, but are forecast to be available in the near future.
Marubashi  notes that the keys to the establishment of space weather forecasting systems are: (1) the development of an efficient algorithm for predictions of critical solar and geophysical phenomena and (2) secure, continuous streams of real-time data required for the prediction algorithm. In its current form, the Burton et al. formula provides the first key. It is simple, fast and, based solely upon solar wind measurements, it generally predicts Dst to within a few nanoteslas of the observed Dst. Thus, a "magnetospheric" storm can presently be readily predicted. Deployment of an upstream solar wind monitor that would remain substantially inside 1.0 AU (0.7 to 0.95 AU), near the ecliptic plane, and orbit about the Sun synchronously with the Earth providing an ongoing capability to make these predictions would be the second key towards establishing a true space weather forecast system. Such a monitor will provide useful predictions of the energization of the magnetosphere by large-scale interplanetary structures.
Acknowledgments. This work was supported by the National Aeronautics and Space Administration under grant NAGW3492-PVO and by the Los Alamos National Laboratory/Institute of Geophysics and Planetary Physics under grant 4-443869-JL-69895-03.
Akasofu, S. -I., A source of the energy for geomagnetic storms and auroras, Planet. Space Sci., 12, 801, 1964.
Bargatze, L. F., D. N. Baker, R. L. McPherron, and E. W. Hones Jr., Magnetospheric impulse response for many levels of geomagnetic activity, J. Geophys. Res., 90, 6387-6394, 1985.
Bartels, Terrestrial-magnetic activity and its relations to solar phenomena, J. Geophys. Res., 37, 1-52, 1932.
Burlaga, L. F., V. J. Pizzo, A. Lazarus, and P. Gazis, Stream dynamics between 1 AU and 2 AU: a comparison of observations and theory, J. Geophys. Res., 90, 7377-7388, 1985.
Burton, R. K., R. L. McPherron, and C. T. Russell, An empirical relationship between interplanetary conditions and Dst, J. Geophys. Res., 87, 4204-4214, 1975.
Clauer, C. R., R. L. McPherron, C. Searls, and M. G. Kivelson, Solar wind control of auroral zone geomagnetic activity, Geophys. Res. Lett., 8, 915-918, 1981.
Crooker, N.U., G. L. Siscoe, C. T. Russell and E. J. Smith, Factors controlling degree of correlation between ISEE-1 and ISEE-3 interplanetary magnetic field measurements, J. Geophys. Res., 87, 2224-2230, 1982.
Gosling, J. T., D. N. Baker, S. J. Bame, W. C. Feldman, R. D. Zwickl, Bidirectional solar wind electron heat flux events, J. Geophys. Res., 92, 8519 - 8535, 1987.
Harris, H.M. (editor), Coronal transients and space weather prediction missions, JPL D-12611, Jet Propulsion Laboratory, Pasadena, 1995.
Hirshberg, J. and D. S. Colburn, Interplanetary field and geomagnetic variations - a unified view, Planet. Space Sci., 17, 1183-1206, 1968.
Hoeksema, J. T., Large-scale structure of the heliospheric magnetic field: 1976-1991in Solar Wind VII, edited by E. Marsch and R. Schwenn, Pergamon Press, Oxford, 191-196, 1992.
Klein, L. W. and L. F. Burlaga, Interplanetary magnetic clouds at 1AU, J. Geophys. Res., 87, 613-624, 1982.
Lindsay, G. M., C. T. Russell and J. G. Luhmann, Coronal mass ejection and stream interaction region characteristics and their potential geomagnetic effectiveness, J. Geophys. Res., 100, 16999-17013, 1995.
Marubashi, K., Structure of the interplanetary magnetic clouds and their solar origins, Adv. Space Res., 6, 335-338, 1986.
Marubashi, K., The Space Weather Forecast Program, Sp. Sci. Rev., 51, 197-214, 1989.
Paschmann, G., I. Papamastorakis, W. Baumjohann, N. Sckopke, C. W. Carlson, et al., The magnetopause for large magnetic shear: AMPTE/IRM observations, J. Geophys. Res., 91, 11009-11115, 1986.
Pizzo, V. J., A three dimensional model of corotating stream in the solar wind 3. Magnetohydrodynamic streams, J. Geophys. Res., 87, 4374-4394, 1982.
Russell, C. T., R. L. McPherron, and R. K. Burton, On the cause of geomagnetic storms, J. Geophys. Res., 79, 1105-1109, 1974.
Russell, C. T., G. L. Siscoe and E. J. Smith, Comparison of ISEE 1 and 3 interplanetary magnetic field observations, Geophys. Res. Lett., 7, 381-384, 1980.
Schwenn, R., Large-scale structure of the interplanetary medium in Physics of the Inner Heliosphere I, R. Schwenn, E. Marsch, Eds., Springer-Verlag, Berlin Heidelberg, 99-181, 1990.
Scurry, L. and C. T. Russell, Proxy studies of energy transfer to the magnetosphere, J. Geophys. Res., 96, 9541-9548, 1991.
Scurry, L., C. T. Russell, and J. T. Gosling, Geomagnetic activity and the beta dependence of the dayside reconnection rate, J. Geophys. Res., 99, 14811-14814, 1994.
Siscoe, G. L., A unified treatment of magnetospheric dynamics with applications to magnetic storms, Planet. Space Sci., 14, 947-967, 1966.
Volkmer, P. M. and F. M. Neubauer, Statistical properties of fast magnetoacoustic shock waves in the solar wind between 0.3 AU and 1 AU: Helios-1, 2 observations, Ann. Geophysicae , 3, 1985.
Woo, R., A synoptic study of Doppler scintillation transients in the solar wind, J. Geophys. Res., 93, 3919-3926, 1988.
G. M. Lindsay, HQ AFSPC/DRFS, 150 Vandenberg St., Suite 1105, Peterson AFB, CO 80914-4590.
J. G. Luhmann, Space Sciences Laboratory, University of California, Grizzly Peak Blvd. at Centennial Dr., Berkeley, CA 94720
C. T. Russell, Institute of Geophysics and Planetary Physics, University of California, 3845 Slichter Hall, Los Angeles, CA 90095-1567 (e-mail: email@example.com)