GNGTS 2023 - Atti del 41° Convegno Nazionale

Session 2.1 GNGTS 2023 Fig. 2. The scenario to verify the EQ forecasting algorithm for East Pacific given the EB detections by NOAA satellites. EQ epicenters that contributed to the correlation were reported in the scenario by red dots. The alarm volume VA in red was generated by EB detections in yellow in the precursor volume VP in cyan. A satellite semi-orbit is indicated by a dashed line. Earthquakes indicated by red stars were failures if occurred outside of VA, and were successes if occurred inside VA. Using 16.5 years of NOAA-15 data, the VT was obtained by multiplying the East Pacific area by the total number of hours which were 36,135 for 4 h intervals. The total number of 199 EQs occurred in this volume with M ≥ 6 and depths ≤ 200 km, resulting in a P(EQ) = 0.0055. The probability gain calculated for this case was 2, which means that if an EB was observed onboard NOAA in the specific region VP, the EQ probability doubled (Fidani, 2022). However, the total number of EBs was high and equal to 3371, which are alarms with a P(EB) = 0.093. The total number of successes was 37 CEs. Thus, the false alarm rate was also high 1 − (CEs/EBs) = 0.989, very much more alarms than those for the West Pacific EQs (Fidani, 2021). Although the geohazard risk reduction can gain valuable preparation time by adopting a probabilistic short-term warning a few hours prior, especially for tsunamis (Zollo et al., 2009), a low number of false alarms is essential to have a usable scenario. From all this comes the need to reduce the number of false alarms, i.e. the