Solar wind driving and substorm triggering

Abstract

[1] We compare solar wind driving and its changes for three data sets: (1) 4861 identifications of substorm onsets from satellite global imagers (Polar UVI and IMAGE FUV); (2) a similar number of otherwise random times chosen with a similar solar wind distribution (slightly elevated driving); (3) completely random times. Multiple measures of solar wind driving were used, including interplanetary magnetic field (IMF) Bz, the Kan-Lee electric field, the Borovsky function, and dΦMP/dt (all of which estimate dayside merging). Superposed epoch analysis verifies that the mean Bz has a northward turning (or at least averages less southward) starting 20 min before onset. We argue that the delay between IMF impact on the magnetopause and tail effects appearing in the ionosphere is about that long. The northward turning is not the effect of a few extreme events. The median field shows the same result, as do all other measures of solar wind driving. We compare the rate of northward turning to that observed after random times with slightly elevated driving. The subsequent reversion to mean is essentially the same between random elevations and substorms. To further verify this, we consider in detail the distribution of changes from the statistical peak (20 min prior to onset) to onset. For Bz, the mean change after onset is +0.14 nT (i.e., IMF becomes more northward), but the standard deviation is σ = 2.8 nT. Thus large changes in either direction are common. For EKL, the change is −15 nT km/s ± 830 nT km/s. Thus either a hypothesis predicting northward turnings or one predicting southward turnings would find abundant yet random confirming examples. Indeed, applying the Lyons et al. (1997) trigger criteria (excluding only the prior requirement of 22/30 min Bz < 0, which is often not valid for actual substorms) to these three sets of data shows that “northward turning triggers” occur in 23% of the random data, 24% of the actual substorms, and after 27% of the random elevations. These results strongly support the idea of Morley and Freeman (2007), that substorms require initial elevated solar wind driving, but that there is no evidence for external triggering. Finally dynamic pressure, p, and velocity, v, show no meaningful variation around onset (although p averages 10% above an 11 year mean).