[1] We present the results of an investigation of the sequence of events from the Sun to the Earth that ultimately led to the 88 major geomagnetic storms (defined by minimum Dst À100 nT) that occurred during 1996-2005. The results are achieved through cooperative efforts that originated at the Living with a Star (LWS) Coordinated DataAnalysis Workshop (CDAW) held at George Mason University in March 2005. On the basis of careful examination of the complete array of solar and in situ solar wind observations, we have identified and characterized, for each major geomagnetic storm, the overall solar-interplanetary (solar-IP) source type, the time, velocity, and angular width of the source coronal mass ejection (CME), the type and heliographic location of the solar source region, the structure of the transient solar wind flow with the storm-driving component specified, the arrival time of shock/disturbance, and the start and ending times of the corresponding IP CME (ICME). The storm-driving component, which possesses a prolonged and enhanced southward magnetic field (B s ), may be an ICME, the sheath of shocked plasma (SH) upstream of an ICME, a corotating interaction region (CIR), or a combination of these structures. We classify the Solar-IP sources into three broad types: (1) S-type, in which the storm is associated with a single ICME and a single CME at the Sun; (2) M-type, in which the storm is associated with a complex solar wind flow produced by multiple interacting ICMEs arising from multiple halo CMEs launched from the Sun in a short period; (3) C-type, in which the storm is associated with a CIR formed at the leading edge of a high-speed stream originating from a solar coronal hole (CH). For the 88 major storms, the S-type, M-type, and C-type events number 53 (60%), 24 (27%), and 11 (13%), respectively. For the 85 events for which the surface source regions could be investigated, 54 (63%) of the storms originated in solar active regions, 11 (13%) in quiet Sun regions associated with quiescent filaments or filament channels, and 11 (13%) were associated with coronal holes. Remarkably, nine (11%) CME-driven events showed no sign of eruptive features on the surface or in the low corona (e.g., no flare, no coronal dimming, and no loop arcade, etc.), even though all the available solar observations in a suitable time period were carefully examined. Thus while it is generally true that a major geomagnetic storm is more likely to be driven by a frontside fast halo CME associated with a major flare, our study indicates a broad distribution of source properties. The implications of the results for space weather forecasting are briefly discussed.

We report the first detection at long radio wavelengths of interaction between coronal mass ejections (CMEs) in the interplanetary medium. The radio signature is in the form of intense continuum-like radio emission following an interplanetary type II burst. At the time of the radio enhancement, coronagraphic images show a fast CME overtaking a slow CME. We interpret the radio enhancement as a consequence of shock strengthening when the shock ahead of the fast CME plows through the core of the preceding slow CME. The duration of the radio enhancement is consistent with the transit time of the CME-driven shock through the core of the slow CME. As a consequence of the interaction, the core of the slow CME changed its trajectory significantly. Based on the emission characteristics of the radio enhancement, we estimate the density of the core of the slow CME to be ∼ cm Ϫ3 . The CME interaction has important implications for space weather prediction based on halo 4 4 # 10 CMEs: some of the false alarms could be accounted for by CME interactions. The observed CME interaction could also explain some of the complex ejecta at 1 AU, which have unusual composition.

[1] We studied the geoeffectiveness, speed, solar source, and flare association of a set of 378 halo coronal mass ejections (CMEs) of cycle 23 (1996-2005, inclusive). We compiled the minimum Dst values occurring within 1-5 days after the CME onset. We compared the distributions of such Dst values for the following subsets of halo CMEs: disk halos (within 45 deg from disk center), limb halos (beyond 45 degrees but within 90 deg from disk center), and backside halo CMEs. Defining that a halo CME is geoeffective if it is followed by Dst À50 nT, moderately geoeffective if À50 nT < Dst < À100 nT, and strongly geoeffective if Dst À100 nT, we find that the disk halos are followed by strong storms, limb halos are followed by moderate storms, and backside halos are not followed by significant storms. The Dst distribution for a random sample is nearly identical to the case of backside halos. About 71% of all frontside halos are geoeffective, supporting the high rate of geoeffectiveness of halo CMEs. A larger fraction (75%) of disk halos are geoeffective. Intense storms are generally due to disk halos and the few intense storms from limb halos occur only in the maximum and declining phases. Most intense storms occur when there are successive CMEs. The delay time between CME onset and minimum Dst value is the smallest for limb halos, suggesting that the sheath is geoeffective in these cases. The geoeffectiveness rate has prominent dips in 1999 and 2002 (the beginning and end years of the solar maximum phase). The numbers of all frontside and geoeffective frontside halos show a triple peak structure similar to the number of intense geomagnetic storms. The difference in flare sizes among geoeffective and nongeoeffective halos is not significant. The nongeoeffective CMEs are generally slower and have more easterly or limbward solar sources compared to the geoeffective ones; source location and speed are the most important parameters for geoeffectiveness.

Abstract. We investigated the characteristics of' coronal mass ejections (CMEs) associated with long-wavelength type II radio bursts in tim near-Sun interplanetary medium. Type II radio bursts in the decameter-hectometric (DH) wavelengths indicate powerful MHD shocks leaving the inner solar corona and entering the interplanetary medium. Almost all of these bursts are associated with wider and faster than average CMEs. A large fraction of these radio-rich CMEs were fbund to decelerate in the coronagraph field of view, in contrast to the prevailing view that most CMEs display either constant acceleration or constant speed. We fbund a similar deceleration for the fast CMEs (speed > 900 km s -1) in general. We suggest that the coronal drag could be responsible fbr the deceleration, based on the result that the deceleration has a quadratic dependence on the CME speed.About 60% of the fast CMEs were not associated with DH type II bursts, suggesting that some additional condition needs to be satisfied to be radio-rich. The average ß width (66 ø) of the radio-poor, fast CMEs is much smaller than that (102 ø) of the radio-rich CMEs, suggesting that the CME width plays an important role. The special characteristics of the radio-rich CMEs suggest that the detection of DH radio bursts may provide a useful tool in identif•/ing the population of geoeffective CMEs.