Reply to comment by Harald U. Frey on “Substorm triggering by new plasma intrusion: THEMIS all‐sky imager observations”

[1] The sequence of events leading to substorm onset has been a long outstanding issue in magnetosphere‐ionosphere coupling physics. Nishimura et al. [2010a, hereafter N2010] proposed a potential resolution to this problem based on observations from the all‐sky imager (ASI) array [Mende et al., 2008] of the THEMIS project [Angelopoulos, 2008], which provides high spatial and temporal resolution auroral images with broad latitudinal and longitudinal coverage. Usingthecapabilityoftheimagerarray,wefoundarepetitive precursor auroral sequence prior to substorm auroral onset. The preonset auroral sequence reported by N2010 is initiated by a poleward boundary intensification (PBI), which is followed by an approximately north‐south (N‐S) oriented arc (also referred to as an auroral streamer) moving equatorward toward the onset latitude and sometimes turning into an east‐ west (E‐W) oriented luminosity enhancement propagating azimuthally. It finally leads to onset instability in the near‐ Earth plasma sheet and is observed as auroral onset. Because of the linkage of fast magnetotail flows to PBIs and to N‐S auroras, this sequence gives strong support to the idea that onset instability develops following enhanced plasma flows fromtheopen‐closedboundarytowardthenear‐Earthplasma sheet [Lyons et al., 2010a, 2010b]. [2] Frey [2010, hereafter F2010] has commented on our study of the auroral sequence leading to substorm onset. The main issues raised by F2010 are as follows. [3] 1. Time differences of auroral intensifications less than 30 min are too short to be called two separated onsets. Thus some of the events considered by N2010 are not onsets but are just intensifications of earlier substorms, and inclusion of such intensifications might affect our statistical results. [4] 2. A large number of substorms were missed in the N2010 analysis, based on a comparison to the event list by F2010. [5] In this reply, we first show that the events separated by short time intervals are indeed auroral onsets and that the preonset sequence found by N2010 is commonly seen in both first and subsequent onsets occurring within ∼30 min. Then we show that over half of the events in the F2010 list not included in the N2010 study are not substorms but other types of auroral phenomena, and that the majority of the remainder had onsets that were not within the field of views (FOVs) of available imagers. We further demonstrate that the N2010 event list covers most of the F2010 substorm onsets. Finally, we show statistical results using only isolated events and provide evidence that the preonset sequence found by N2010 is common for isolated substorms, and has essentially the same high occurrence probability as for all events.