The Dst index is normally used to study magnetic storms. This index is calculated by a complex procedure described in papers by Sugiura and colleagues.
Objections to the use of Dst have been raised by Campbell (1996) based on the time profile of Dst during a typical magnetic storm. Campbell noted that the temporal pattern of Dst approximates a log normal distribution. From this he concluded that it is a myth that Dst represents the ring current behavior since log normal distributions are produced by the superposition of many independent processes. It should be noted, however, that it is the probability dsensity function for Dst that should be log normal, not the waveform.
Kamide et al. (1998) summarize our more recent understanding of magnetic storms and their relation to substorms. The authors discuss many of the processes that contribute to the Dst index referencing the Campbell paper. Campbell did not care for the description of storms given by Kamide and in a comment [Campbell, 1999] enumerated even more processes that might contribute to the Dst index. In a reply [Kamide et al., 1999] point out that many of these processes have been recognized for decades and attempts are made to remove them both in the calculation of the index and in its analysis.
It is well known that magnetic activity is stronger at the two equinoxes than at other times of the year producing what is known as the semiannual variation of geomagnetic activity. The standard explanation for this variation is known as the Russell-McPherron effect. At equinox the Earth's dipole axis is tilted in a plane orthogonal to the Earth-Sun line as much as 35° relative to the ecliptic pole. At this time an IMF of the correct polarity ("Spring to and Fall away") lying in the solar equatorial plane can have a component antiparallel to the Earth's dipole field. This causes magnetic reconnection and geomagnetic activity. Recently Cliver et al. (2000) presented evidence that the R-M effect is not the primary cause of the semiannual variation in Dst and other indices. Cliver et al. suggest instead that the tilt of the Earth's dipole towards or away from orthogonal to the Earth-Sun line modulates reconnection efficiency in such a way as to cause most of the semiannual variation. The cause of this modulation is not know. Additional evidence supporting the importance of this equinoctial effect is presented in
Additional factors may affect the coupling of the solar wind to the ring current. For example, [Lopez et al., 2004] point out that the solar wind density influences reconnection through its effect on the compression ratio of the bow shock. Borovsky & Funsten 2003 demonstrate that the level of turbulence in the solar wind effects activity through the viscous interaction. Siscoe et al., (2002a) discuss how saturation of the polar cap potential can affect geomagnetic actiity. A physical explanation for this saturation in terms of Region 1 currents is presented in Siscoe et al., (2002b).
There is great interest in predicting geomagnetic activty as discussed by Joselyn (1995). Unfortunately only for what she calls "nowcasting" is it possible to make reasonably accurate predictions of indices. Such forecasts are based on linear and non-linear mappings of solar wind input series measured upstream of the Earth into a magnetic index. The coefficients describing the mapping are determined empirically from historical data. A very successful application of this method is described in Temerin and Li (2002). Medium-term forcasting (1-4 days) are so poor as to be of little value to most users. McPherron and Siscoe (2004) advocate a different approach which they refer to as "Probabilistic Forecasting by Air Mass Climatology". In this technique the past history of magnetic activity relative to some specific (and predictable) feature in the solar wind is quantified in terms of cumulative probability distributions. If it is known when this feature will arrive it is then possible to predict the probability that activity will exceed some threshold. The authors illustrate the technique using the stream interfaces imbedded within a corotating interaction region. Richardson (2001) discuss the relative importance of different structures in the generation of activity suggesting all known structures are about equally important in long-term averages.
LWS CDAW Working group 4 is concerned with topic of storm forecasting. The group leaders have identified a number of outstanding questions related to storms. For some of these it may be possible to make some progress during the workshop. The complete list is contained in Questions for Working Group 4 .