L R Lyons (Atmospheric Sciences Department, UCLA, Los Angeles, CA 90095-1565); 310-206-7876; e-mail: larry@atmos.ucla.edu); G T Blanchard (The Aerospace Corp, PO Box 92957, Los Angeles, CA 90009); J. C. Samson (Department of Physics, University of Alberta, Edmonton, Alberta, T6G 2E9); R Lepping (NASA/Goddard, Greenbelt, MD 20771); T Yamamoto (Inst. of Space and Astronautical Science, Sagamihara, Kanagawa 229, Japan); T Moretto (Danish Meteor. Institute, Lyngbyvej 100, Dk-2100, Copenhagen 0 Denmark)

A prerequisite for understanding substorms is a determination of the extent to which onsets are the result of an internal magnetospheric instability versus the result of changes in the interplanetary medium. Despite its importance, a definitive answer to this question has not yet been obtained. To address this question, we have examined all the photometer data from the CANOPUS meridian scanning photometers from 1 October 1995 to 15 May 1996. We have identified 20 isolated substorms with well-defined onsets, having the poleward propagating expansion-phase aurora associated with classical substorms, and having data on the interplanetary magnetic field (IMF) available. Precise onset times have been obtained with ground magnetometer data. A clear association with expected IMF substorm triggers (northward turnings, large reductions in |By|) is found for 70% of the substorms. A positive association between substorms and triggers is investigated by comparing this figure to the occurrence rate of trigger-like IMF changes in random intervals of the same length as our timing uncertainty in associating triggers with substorm onsets. We find that the probability of observing an IMF trigger depends strongly upon the radial distance r of IMF observing satellites from the Earth-sun line, triggering being seen is our data set 100% of the time when r < 20 RE, but only 50% of the time when 56 RE > r > 40 RE. IMF data from more than one spacecraft is available for several of the substorms, and in a few cases an IMF trigger is seen in the data from only one satellite, When this occurs, it is always the satellite with the smallest r that observes the trigger. A dependence on the upstream distance of an observation from the magnetosphere is not seen. We thus conclude that the onset of isolated classical substorms is primarily the result of IMF triggering. Cases where a trigger is not seen appear to be the result of spatial structure of the IMF causing observing satellites to measure an IMF that is not representative of that which actually impinges upon the magnetosphere.