The response of the magnetotail to steady external driving is not well understood. The primary reason is that observationally such periods are rare and thus simulations and theories are not well bound by multi-spacecraft studies of such solar-wind-magnetosphere interactions. We have searched for extended (>6 hours) periods of southward IMF during the first two years of GEOTAIL operation. We have found one case during which GEOTAIL was near the midnight magnetotail at X=-96 Re and IMP8 was monitoring the solar wind. The IMF Bz was steady at ~-2 nT for 8 hours. During the first half of the interval the ground activity was very low, while the plasma sheet convection at GEOTAIL was Earthward, indicating that the distant tail neutral line was beyong the distance of GEOTAIL. In the middle of the interval a substorm onset took place while a tailward moving plasmoid was seen at GEOTAIL. Ground and geosynchornous observables are consistent with a classical substorm onset. There is no indication of a trigger in the IMP8 dataset or on the ground. These observations document a clear case of a spontaneous substorm onset. For 4 hours after onset the convection at GEOTAIL continued to be tailward. This suggests that the distant tail neutral line may have re-established itself closer to Earth than X=-96 Re after the spontaneous substorm.

Figure 1 shows the IMP8 data for the period under study.

Figure 1 / Simulation Input shows the same IMP8 data but cast at 1 min resolution for both plasma and magnetic field and deflagged. Deflagging was done via linear interpolation over gaps or last point extrapolation over interval edges. The one minute data that went into this figure are also available in ascci format. To get the data on the screen just clic with your mouse on the colored word "data". To get the data in a file at your directory clic on the colored word "data" while holding your SHIFT button.

Figure 2 shows the high latitude ground magnetic traces (H- or X- component) after a quiet day diurnal variation was substracted. Two clear onsets were seen at ~0415 and 0453 UT.

Figure 3a shows the H component traces from the west Greenland meridional chain of ground magnetometer stations. Data from SKT do exist and will become available shortly. The 0415 UT onset is clearly seen in NAQ and FHB. However, it is also obvious that higher latitude activations were taking place in the interval 00:00 to 04:00 UT. Figure 3b presents the Z component traces from the same stations. It is evident from the opposite sign Z perturbations in the interval 00-04 UT that these activations were centered between SKT and GHB. They correspond to activations near the high latitude boundary. This picture implies that a convection bay was established in this 4 hour interval.

Figure 4a shows the mid-latitude ground magnetic traces (H-component) after a quiet day diurnal variation was substracted.

Figure 4b shows the mid-latitude ground magnetic traces (D-component) after a quiet day diurnal variation was substracted. The first onset (0413 UT) was at dawn. The second (0453 UT) was at pre-midnight.

Figure 5 shows the response of the geosynchronous electrons during the interval. Two electon injections corresponding to the 0415 and 0453 UT onsets are evident. Also evident is the lack of an injection in the interval preceding the 0415 UT onset. The high latitude activations had no effect at geosynchronous.

Figure 6 shows the GOES 7 data. GOES 7 east geographic longitude is -112.3 degrees. The data are plotted in the MAG coordinate system in spherical coordinates. Superimposed on that is the magnetic field from the T89 model. There are four curves that represent model data corresponding to iopt=2,3,4 and 5 (Kp=+-1,1 which is the Kp at the time corresponds to iopt=2.). The iopt=2 curve is the one that has the least agreement with the data. This is because the field is much keep depressed than the model would predict for the Kp value at the time. Only Kp=7 tends to approach the field magnitude at GOES 7.

In an attempt to find the offset in the P component of the GOES 7 magnetometer we have computed the expected field in the dayside for several passes of GOES 7 in the month of January 1993 which were accompanied by solar wind data. The plots constructed along those lines include the input (Pdyn, Dst, AE, By,z gsm) that went into the T95_06 model and the GOES 7 P-component field (solid line) along with the T95 computed field in the direction of P (dashed line). Each day corresponds to one plot. In addition to the T95 model we have tried the T89 model for the kp index of the time. The results of the T89 model are plotted in the same figures with a dash-dotted line (the least variable line of the panel that shows the P-component). There are 6 plots: 930103, 930116, 930117, 930127, 930128, and 930129.

The above plots indicate first that the T89 and T95 models are in fair agreement with each other but often in dissagrement with the measured P. If the difference between model and measured field is to be attributed only to the offset then that offset has to vary with time. On January 03 it was small (varies on either side of data and variability is too large to compute safely) and it was also small (less than 5 nT) on January 16 and January 29. Conversely, on January 17, 27 and 28 it was between 7 and 13 nT depending on the model and the day. Since the day we are interested in is January 22, 1993, i.e., between January 17 and 27, we take the value of 10 nT for the offset. The actual day/model averages are:

Offset(930117,T95)=7.6 nT, Offset(930117,T89)=12.4 nT

Offset(930127,T95)=8.5 nT, Offset(930127,T89)=10.9 nT

Thus, we recompute Figure 6 including the P-component offset of 10 nT. The addition of the offset does not change our conclusion: Prior to the substorm the field was more compressed than the T89 model would predict. As the spacecraft was outside the current wedge, we do not expect dramatic dipolarization effects at onset and thus the smooth Blat variation is not unreasonable.

Note, however, that if you make the same comparison with T95 without offset and with offset you will get somewhat different results. In particular, even without the offset in P, the measured field is more dipol-like and stronger than the T95 model would predict. Thus, our results depend on which model is used. As the SW pressure is quite typical at the time, I suspect that T89 should do an adequate job; since T89 has also been subjected to the test of time, I tend to favor T89 for the analysis in this paper.

Figure 7 shows the GEOTAIL data. Bottom panel is computed Vxperp from Efield and Bfield. Notice plasmoid release at 04:30 UT in association with first substorm but not with second (localization of ground and tail activity).

Figure 8 shows an expanded plot of the GEOTAIL data at the time of plasmoid release. Bottom panel is computed Vxperp from Efield and Bfield. The bipolar signaturein Bz with the "core" field in By is apparrent. This is a "textbook" case of a plasmoid/flux rope that one could model quite successfully. Simple integration of the electric field gives a total magnetic flux transport of around 2x10^6 Wb/Re.

Conclusion #1. A clear example of a spontaneous substorm onset that is classical in many respects is presented here. Conclusion #2. Plasmoids are localized when ground activity is also localized.

A draft of the paper and figures are available for comments in postscript format at the anonymous ftp site: plasma1.ssl.berkeley.edu, in directory pub/vassilis/JGR930122EVENT. Both gziped and regular postscript files are available there.