GNGTS 2013 - Atti del 32° Convegno Nazionale

The first method involves a direct action on the instrument. According to Wood and Anderson (1925), GS is determined by: (7) where A is the seismogram trace amplitude, a is the amplitude of the ground motion component normal to the equilibrium plane, l is the mass swinging center distance from the rotation axis, L is the optical lever length, g is the gravity acceleration (981 cm/s 2 ), acceleration, b is the instrument tilt angle (in radiants), and T 0 is its period of oscillation (0.8 s). Tilting the instruments of a known angle b and measuring the output voltage from the PSD, which is proportional to A , and applying Eq. (7) we can calculate GS (Tab. 2). Tab. 2 – Parameters used for the WA GS computation. I = WA seismograph component; O = PSD controller output (V); A = equivalent trace amplitude on paper (mm). I O (V) A (mm) GS N-S 2.00±0.07 45.8 ± 1.6 2092 ± 73 E-W 2.31±0.07 52.9 ± 1.8 2339 ± 82 We must emphasize that the measure made on the N-S component of the instrument is more reliable than the E-W one because the latter was damaged. The moving mirror was partially detached and it was repaired at best with the tools and skills of the OGS technical staff. The total error associated to the estimate is evaluated as an amplifier error, equal to 1%, on the linearity of the response, plus the uncertainty on the voltmeter, equal to 0.05 V. In order to assess the actual WA GS , the second method is based upon a comparative analysis of the data, in particular on the maximum amplitudes (peak to peak) of the seismograms traces, recorded by the two instruments placed side by side. On the BB seismometer we fixed GS equal to 2800. Sliding a window of 50 WA samples on the values of the amplitudes recorded by the BB seismometer, the GS values have been calculated as the weighted average of the corresponding ratios (i.e. WA/BB×2800; see Fig. 3a). The uncertainty on the individual measurements was obtained by perturbing the error on the gain of the instrument. We simulated a series of test measurements moving the needle between two notches on the instrument from time to time, which correspond to one theoretical shift of 10 cm on the paper. We therefore measured the average number of counts of the test measures: the gain is the ratio between mm and counts. The standard deviation of the distribution of the test measurements is the error on the gain due to the imprecision in making the movement of the needle. To this, we added a further 1% error on the mean value of the measures of the test due to the amplifier. The GS values decrease with increasing amplitude (Fig. 3a), reaching a value approximately constant in correspondence of 0.2 mm and close to that determined by Uhrhammer and Collins (1990) and equal to 2080. For amplitude values in the range 0.05-0.07 mm it is close to the original 2800 value. The asymptotic values in the two cases are very similar to those obtained with the first method (see Tab. 2). The magnitude estimation is slightly but clearly affected by adopting the theoretical magnification value equal to 2800 (Fig. 3b). If we simulate the WA through a BB seismometer fixing the magnification factor equal to 2800, instead of the real values of Fig. 3 A and B, we introduce an error that depends on the amplitude measured by the instrument, ranging between 0 and 0.13. Conclusions. The Trieste Wood Anderson seismograph, officially discontinued in 1992, was recovered, modernized and after a decade of interruption, it continues presently to record earthquakes. We recovered the amplitudes of the two components of the past events (319 121 GNGTS 2013 S essione 1.1