During the summer of 1993, residents of the Norris Lake community in the Piedmont Province of Georgia, USA, experienced a swarm of shallow (<1.5 km) earthquakes. The swarm contained five events in the magnitude range of 2.2-2.7 (mb). In this epicentral area no earthquakes were previously reported and no significant events have occurred since, and the swarm is considered complete and isolated. The swarm began June 8 with a short, 5-day burst of activity. The number of events per day and cumulative strain release increased exponentially to mid-August, when more than 200 events occurred per day. After the mid-August peak, the number of events per day decreased at a rate consistent with an aftershock sequence. The largest event (mb = 2.7) occurred late in the sequence on September 23, 1993.
Times of occurrence were read and magnitudes computed from visual records covering the time period of June 13, through December, 1993 for over 4,800 events with mb magnitude greater than -2i149 The correlation dimension of occurrence times suggests that the swarm may be divided into constant-valued segments that start and end at a large event. The peak of the swarm was characterized by a high dimension, suggesting a uniform rate of occurrence of earthquakes, whereas the beginning and end of the swarm were characterized by low dimensions typical of clusters of events. The average b-value in the Gutenberg-Richter recursion relation for the complete swarm was 0.8. The b-value for shorter time segments of the swarm increased gradually from 0.4 in July to greater than 1.0 in September, including the period of the highest rate of seismicity in August. The b-value returned to its average, 0.8, at the end of the swarm. The b-values were constant or increased gradually before the four largest events, changed at the time of three of these events, and on average tended to decrease with time afterward.
We speculate that the systematic increase and subsequent decrease in b-value and correlation dimension are characteristics of earthquake swarms triggered during a transient decrease in strength. Furthermore, we speculate that the decrease in strength may be caused by increases in fluid pressure and that the increase in fluid pressure may have been augmented by the conversion of elastic stress to hydrostatic stress during the occurrence of earthquakes in the swarm.