In the thermal catastrophe model of substorms, onset is attributed to an explosive heating of the plasma sheet boundary layer (PSBL) due to the resonant absorption of Alfv¿n waves. These waves are assumed to have their origins at the magnetopause where solar-wind-driven Kelvin-Helmholtz surface waves launch magnetoacoustic waves through the lobes toward the PSBL which are then mode-converted to kinetic Alfv¿n waves in the PSBL. On the basis of this wave source mechanism, the thermal catastrophe model predicts the locations in the tail where catastrophe and hence onset is most likely to occur. In this paper, the model predictions are extended to include the effects of a realistic tail field (represented by the Tsyganenko 1987 long model) and are compared to observations of substorm onset as determined from Viking auroral images. The use of a realistic tail field causes a deflection of the catastrophe domain towards the dusk (dawn) flanks of the magnetosphere for gardenhose (orthogardenhose) interplanetary magnetic field (IMF) conditions, and it is also found that this region extends into the deep tail only for higher solar wind speeds, thereby confining catastrophe to the relatively near-Earth portions of the tail for typical conditions.

Most of the other results obtained with the inclusion of a realistic tail field produce trends similar to the original model predictions including the results that onset should get closer to midnight for increasing solar wind speed and that onset should occur in the premidnight (postmidnight) sector for gardenhose (orthogardenhose) IMF conditions. Although a weak consistency with measured onset locations is observed, many more events than are contained in the Viking database are required in order to provide a firm conclusion. However, the reasonably well established result that onset occurs very close to the Earth on closed field lines is inconsistent with the predictions of the model which places onset on PSBL field lines. Thus the thermal catastrophe model is not a likely mechanism for typical substorm onsets. Nevertheless, evidence is shown that the mechanism may still operate in the magnetosphere to produce a different type of auroral intensification. ¿ American Geophysical Union 1995