This study focuses on the foreshocks of earthquake sequences recorded by the Italian national seismological network between 1975 and 1991. We consider as foreshocks events exceeding a given threshold magnitude that follow a period of quiescence and come before a mainshock occurring within a given time-distance range. The total number of potential foreshocks identified by the algorithm depends on the size and the duration of the predefined quiescence area, whereas the number of mainshocks preceded by foreshocks depends also on the size of the area in which the event is expected and the duration of the alarm period. Our aim was to optimize the choice of the parameters defining potential foreshocks as precursors of stronger earthquakes, in order to improve their validity and reliability.
A preliminary investigation carried over the whole Italian territory allowed the definition of zones where foreshocks-mainshocks series are likely to occur. One of these zones is an area of about 32,000 km2 in Central Italy including a total of 2671 events of M ≧ 2.5, where there is a significant tendency of moderate shocks to be followed at short distance and in short time by larger shocks. Our method would lead to issuing 63 alarms with the optimal definition of foreshocks as events of magnitude exceeding 3.0 after at least 80 days of quiescence in an area of 140 km radius. The rate of occurrence for mainshocks of magnitude exceeding 4.0 during the following 48 hours, in the specified area of 30 km radius, is 6 of 23 events. The corresponding probability gain is about 150. For comparison, another larger zone (Po Valley) with 453 events of M ≧ 2.5 exhibits only a rate of 1 of 82 precursors. In this area most events are single shocks. Our analysis shows that the foreshock activity can be regarded as a significant earthquake precursor, so suggesting that real time observation of seismicity can be used to improve the quality of an earthquake prediction system.