What's that !?

Onsala, Sep. 12, 2011
An earthquake of magnitude (mb) 4.6 was reported with an epicentre near Emmerich, Nordrhein-Westfalen, Germany.
(USGS annotated it as in The Netherlands, never mind. Accept "Emmerich" for the scope of this page.)
Location N51.78, E5.96, Time 2011-09-08 19:02:48.6 UTC (GfZ)

Comparing our Superconducting gravimeter record of Sep. 08, 2011, 19:00-- UTC with the
Güralp 120s Seismograph (both instruments are located a few metres from each other), this came as a surprise:

The oscillation in the centre of the diagram has a period of 10-11 seconds and an amplitude of 20 nm/s2.
If that would be due to a change of sea level in Kattegatt, roughly 1m amplitude would be required!
No, no, no!!! Not this size, not this frequency.

In order to compare similar parameters, I have low-pass filtered the seismograph record and converted it to
1-second acceleration (very roughly by simply differencing the time series: (1, -z))
Here's the "accelerogram":

The low-pass filter spectrum.

Here is the question:

What on Earth is the Gravimeter showing that the Seismometer does not sense?
There is no other earthquake sufficiently in magnitude and distance to produce the signature.
If it's from the Emmerich event, what kind of wave would travel the 610 km to Onsala with a speed of 2.0 km/s?
(There's nothing particular on our barometer)

It is no surprise that the body waves don't show up; as the accelerogram shows, their signal is found above the gravimeter's Nyquist of 0.5 Hz.
However, the 3-times-noise-level-sized-signal that starts near 19:07 is not picked up by the seismometer, in particular not in velocity.
How come?

Yours curiously,
Hans-Georg Scherneck


1 - GfZ


3 The whole day's gravity series

Link to the ASCII time-series for the entire day 2011-09-08, sampled at 1s
Reduction model: Measurement - tides(empirical model) - air pressure (-3.286 nm/s2/hPa)
Units: nm/s2