Negative Magnetic susceptibilities

In running VOXI my models end up with some negative magnetic susceptibilities.  As was mentioned previously if the RMI has any negative values then negative susceptibilities will follow.  I would assume this is due to the background level shift or using the wrong value.  When making diurnal, etc. corrections to a base station it is standard to remove the base station values and then add a somewhat arbitrary value to the data, so that the values are in the typical field range...52,000 nT or whatever.  Then I remove the IGRF and that becomes RMI until I decide to remove a baselevel or linear or quadratic function regionals.  To avoid negative susceptibilities caused by negative RMI why don't I simply determine the minimum value of the RMI and then add that amount back into the field.  I assume that would solve the problem except for negatives caused by remnant polarization.  Does this make any sense and does anyone have experience with this?