Recorded (or raw) and processed pre/post-stack seismic data (relative amplitudes) usually display a standard normal distribution, albeit with a very small amount of skewness.
What would be possible causes and/or significance if amplitudes did not retain a mean value of zero (or very nearly zero)?
I'm sure there are several possible reasons.
I'm not a seismologist, this is an amateur answer only.
Local acceleration has very small long-term drifts due to tidal forces from the Sun and the Moon. These will have periods of roughly 1 day as the Earth rotates. Most seismometer data will have been run through a high-pass filter with a much higher cutoff frequency than this (of course along with other filtering). This might be implemented with some baseline restoring function other than a simple filter.
Prompt gravity signals -- changes in local g as mass distribution in the Earth changes -- have been detected, and this kind of measurement may become more widely used in the future. Of course a typical seismometer will not be sensitive to pick this up, unless of course your seismometer is sitting atop a substantial geological event!
See:
