GNGTS 2022 - Atti del 40° Convegno Nazionale

GNGTS 2022 Sessione 1.2 97 LTA (top): the red bar consists of Trigger on and the Blue bar on Trigger off. These results are obtained by fixing a couple of sizes for STA and LTA. Note that the STA/LTA algorithm needs two input parameters to detect on and off triggers. In this work, we have considered these values as 4.0 and 1.0 respectively. Starting from a set of signals cut by the operator, a comparison was made between the start and end times of the EQ events triggered by STA / LTA on the same signals. For each considered raw signal, STA / LTA outputs a list of triggers that are compared individually with the ones cut by the operator. For the comparison, the absolute deviation in terms of temporal distance is calculated. If the deviation is at most 10 (seconds), both for the start and for the end of the event, then the cut is considered to be correct. Regardless of this check, the triggers produced by STA / LTA are counted and indicated as enq. Statistical measures and results. Two statistical measures were calculated to determine the effectiveness of the cut of the STA / LTA algorithm, i.e. the quality index  ( qi ) and the numerosity index  ( ni ) .  Let m  be the mean of all the deviation computed between the correct (below 10 secs) STA/LTA cuts and the operator cuts, the qi  is defined as qi = 1-(m/10) The ni , on the other hand, is defined to take into account the discrepancy among enq  (the correct number of cuts by STA / LTA) and the number of cuts by the operator, indicated as  tnq . In our data, the  tnq  is equal to 1506. The value of the ni  is set in this way: Finally, the product between the qi and the ni , named qni , is considered an overall measure to determine the effectiveness of the cut made by STA / LTA, and is defined as qni = qi * ni Fig. 3 - Results example for qni values.