How many seismographs do we need to record ground motion at a station?, Studia Geophys. et Geod

Seismograms represent basic data for structural studies and studies of earthquakes. There is a common opinion that with digital broad-band seismographs, seismologist’s life is relatively easy: “Just take your seismogram, apply the instrumental correction, and get the ground motion”. The author wants to show that this is nothing but a myth. Various limitations arise not only from specific instrumental characteristics (transfer functions), but also from instrumental noise. As a result, where a velocity recorder fails in a particular application, an acceleration recorder may be better, and vice versa. Moreover, there are also hidden features of instrument, not described by its transfer function at all (e.g. a spurious ringing). And, finally, seismographs are “living organisms”, changing their properties with time, often in a quite unexpected fashion, thus requiring extremely patient and careful treatment. All this will be demonstrated on a few examples from three sites, each one equipped by two collocated instruments (i.e. different instruments located very close from each other). As the author is not an instrumental specialist, he spent endless hours on recognizing some traps able to catch an unprepared data user (e.g. a modeler), or a student, when he wants to use seismograms in his work. Thus the main stimulation fo