Metadata evalutation using normal modes.
The occurence of several very large earthquakes since 2004 provides a special opportunity to validate the accuracy of sensor sensitivities reported for the GSN. A goal of the GSN is to publish instrument responses to an accuracy of 1% in amplitude and 1 degree in phase. These earthquakes excited long period free oscillations with low decay rates to amplitudes that could be observed above ambient noise for many weeks after the event. By examining the subset of modes least sensitive to short wavelength structure, one may test the reliability of published station response information. We investigated (see citations below) the amplitude of the radial mode 0S0, which to first order is predicted to be uniform globally, and found that there was a broader distribution in observed excitation amplitudes than expected from the measurement methods. We concluded that the distribution was widened due to errors in the instrument responses for the stations used in these studies.
Between the time of the initial study in 2005 and the present, a campaign to improve instrument response information was started ( see RECALIBRATION in sidebar ). Based on the 2012 paper, 0S0 observations from the later events clearly benefited from this work: as one would expect from theory, they cluster much closer to the same unique amplitude for each earthquake.
Davis, P., M. Ishii, and G. Masters (2005). An assessment of the accuracy of GSN sensor response information, Seismol. Res. Lett. 76, 678-683.
Davis, P. and J. Berger (2012). Initial impact of the Global Seismographic Network quality initiative on metadata accuracy, Seismol. Res. Lett. 83, 697-703.