GNGTS 2013 - Atti del 32° Convegno Nazionale

where M LBB stands for the local magnitude simulated on the BB via the WA filter. The local magnitudes calculated by the BB seismometer are slightly higher than the actual MAW s. For the 833 events considered, on average, the M LBB overestimation is equal to 0.11. We compared the MAW s with the local magnitude provided by INGV M LINGV , the institute that is responsible for the official magnitude publication in Italy (Fig. 2b). There were considered two different data sets and thus two corresponding fitting curves: the data until April 16, 2005 (red empty diamonds in Fig. 2b) are taken from Italian Seismic Bulletin (INGV, 2010), the following data (orange filled diamonds in Fig. 2a) are taken from the Italian Seismological Instrumental and parametric database ISIDe (INGV, 2010). Unfortunately the two data set of magnitudes are not entirely compatible because a mix of duration magnitude MD and local magnitude M L is reported in the first data set while an ML simulated from BB recordings is available in the second data set. Considering only the earthquakes recorded after April 16, 2005 (538 events) the fit is: M LINGV = (0.930 ± 0.007) MAW + (0.356 ± 0.024) R 2 = 0.8874 (4) Taking into account also the events before April 16, 2005 (680 earthquakes), the fit is: M LINGV = (0.872 ± 0.006) MAW + (0.511 ± 0.021) R 2 = 0.8743 (5) The M LINGV are on average higher compared to the MAW. In particular the overestimation (on average equal to 0.17) is more accentuated for events with a magnitude smaller than 3, while for higher magnitudes the difference is slightly reduced (on average it is equal to 0.13.) As further analysis, we have compared MAW s and MD s provided by OGS in the time window October 22, 2004 to May 20, 2012, the day of the first strong event of the Emilia seismic sequence (Fig. 2c). The two catalogues have 187 events in common. A fixed uncertainty equal to 0.1 has been assumed on OGS MD s (Gentili et al. , 2011). The uncertainty on MAW was, however, obtained from the amplitude measurements as described in the next paragraph. The equation of the fit is: M D = (0.84 ± 0.01) MAW + (0.67 ± 0.03) R 2 = 0.8874 (6) It canbe seen that MD overestimates MAW for lowmagnitudes,while it tends tounderestimate it for high magnitudes. The result is qualitatively similar to that obtained in Gentili et al. (2011) comparing the local magnitude with that of duration comparing a set of local magnitude from Bragato and Tento (2005) and Garbin (2009) with the duration magnitude of the OGS bulletin. Considerations on the WA magnification factor. The WA transfer function, determined empirically, is equivalent to an inertial pendulum with a free period of 0.8 s and damping of 0.8. Regarding the magnification, Anderson and Wood (1925) proposed a static magnification of 2800, which was commonly used since then. Uhrhammer and Collins (1990) and Uhrhammer et al. (1996) report a static magnification equal to 2080. According to Uhrhammer and Collins (2011), the difference derives from the wrong assumption by Anderson and Wood (1925) that the wire stretched in suspension used in the sensor WA does not deviate from a straight line. The deformation is actually sufficient to increase the moment of inertia and reduce the static magnification of about 30%. The difference in estimated magnification does not affect the measure of the amplitudes recorded by the original WA sensors, but it becomes crucial when synthetic seismograms are simulated. Using 2800 instead of 2080 in MAW estimation may cause an increase of magnitude of 0.129 (Uhrhammer et al. , 2011). We tried to verify the static magnification factor of our WA with two different methods. 120 GNGTS 2013 S essione 1.1