On May 4, 1998, the Wind and ACE satellites observed a period of extremely strong southward interplanetary magnetic field (IMF) followed by a period of highly variable IMF. The magnetopause moved inside the geosynchronous orbit, exposing Los Alamos National Laboratory (LANL) geosynchronous satellites and Polar to the magnetosheath. In a simple model the magnetopause erosion may be linearly proportional to the magnitude of southward IMF. A careful comparison of the predictions derived from various magnetopause location models to the in situ observations indicates a nonlinear dependence of the magnetopause erosion on IMF Bz. The observations from the Wind and ACE satellites are significantly different in the period of highly variable IMF. The predictions based on the ACE solar wind data are more consistent with the LANL observations than those based on the Wind solar wind data. The positions of the satellites show that ACE and LANL were mainly on the same side of the Sun-Earth meridian plane. However, for the Polar observations the predictions based on the Wind data are better than those based on the ACE data. Wind and Polar were on the same side of the equatorial plane. These results suggest that we choose a solar wind monitor that is on the same side of the meridian or equatorial plane as the magnetospheric satellite for better predictions.