Substorm probabilities are best predicted from solar wind speed

• Substorm frequency is best predicted by v.
• This explains radiation belt dependency.
• Frontside merging controls auroral power but v controls burstiness.

Most measures of magnetospheric activity – including auroral power (AP), magnetotail stretching, and ring current intensity – are best predicted by solar wind-magnetosphere coupling functions which approximate the frontside magnetopause merging rate. However radiation belt fluxes are best predicted by a simpler function, namely the solar wind speed, v. Since most theories of how these high energy electrons arise are associated with repeated rapid dipolarizations such as associated with substorms, this apparent discrepancy could be reconciled under the hypothesis that the frequency of substorms tracks v rather than the merging rate – despite the necessity of magnetotail flux loading prior to substorms. Here we investigate this conjecture about v and substorm probability. Specifically, a continuous list of substorm onsets compiled from SuperMAG covering January 1, 1997 through December 31, 2007 are studied. The continuity of SuperMAG data and near continuity of solar wind measurements minimize selection bias. In fact v is a much better predictor of onset probability than is the overall merging rate, with substorm odds rising sharply with v. Some loading by merging is necessary, and frontside merging does increase substorm probability, but nearly as strongly as does v taken alone. Likewise, the effects of dynamic pressure, p, are smaller than simply v taken by itself. Changes in the solar wind matter, albeit modestly. For a given level of v (or Bz), a change in v (or Bz) will increase the odds of a substorm for at least 2 h following the change. A decrease in driving elevates substorm probabilities to a greater extent than does an increase, partially supporting external triggering. Yet current v is the best single predictor of subsequently observing a substorm. These results explain why geomagnetically quiet years and active years are better characterized by low or high v (respectively) than by the distribution of merging estimators. It appears that the flow of energy through the magnetosphere is determined by frontside merging, but the burstiness of energy dissipation depends primarily on v.