It was pointed out to me by Terrence, that the quantity that I was using from the geophones was the velocity rather than the displacement. Therefore, I turned the integrator on and measured the transfer functions again.

The coincidence between the measurement made with the geophones and LVDTs is good, so I guess the inter calibration was done already. However, there are small bumps around 1 Hz, that may suggest that the calibration filters of the geophones are inaccurate.