The duration of a volcanic earthquake swarm was found to be the most commonly reported parameter. The database contains 385 swarm duration records with high quality grades. The distribution of swarm durations is shown in Figure 5. The durations vary from very short, intense swarms lasting only a few hours, such as those reported at Piton de la Fournaise and Kilauea, to swarms lasting a few years, such as the activity recorded at Long Valley Caldera and Usu (Usu at 1682 days is the longest swarm in the database). The distribution of durations is nearly log-normal with a geometric mean of 5.5 days, a median of 7 days and a mode of 2 days. The log10 transformed duration distribution was tested for normality using the Kolmogorov-Smirnov test (Rock, 1988). The test showed that the distribution was not normal. The duration distribution is skewed to the left. In other words, there is an excess of shorter swarms.
Swarm durations were then separated based on their relationship to eruptive activity: those which preceded eruptions (Type I) and those not associated with eruptions (Type III). The durations of Type I swarms tended to be longer than Type III swarms (fig. 6). The geometric mean durations were 8 and 3.5 days, respectively. The means of the duration distributions were found to be significantly different from one another, in other words the durations of each swarm type are drawn from different parent populations. The log10 transformed duration distributions for the two swarm types were again tested for normality. We found that the Type I swarm duration distribution is again not log-normal, while the Type III distribution is log-normal. As with the combined distributions, the Type I duration distribution is skewed towards shorter durations.
Apart from swarm duration the next most commonly and reliably reported parameter is the magnitude of the largest shock (Mmax). The database contains 113 magnitude records with high quality grades. The Mmax values range from 0.5 to 6.2. The largest events in the database, Ms6.2 at Miyake-jima in 1983 and M6.0 at Oshima in 1986, occurred during large fissure eruptions. A M5.6 event at Soputan in 1985 is the largest event preceding an eruption in the database. The largest events occurring without a following eruption are a MJMA5.7 at Unzen in 1984 and a M6.2 in Long Valley Caldera in 1980.
Although many magnitude scales are used in reporting these events, only the Russian energy class measurements were converted for comparison. The Russian energy scale Ks was converted to M using: M = (Ks -4.6)/1.5; where M is the magnitude determined from surface waves; Ks is the mean energy class determined as the arithmetical mean from short-period S-waves of several stations (Gorelchik, 1989). The mean of the Mmax distribution for precursory swarms (Type I) is not significantly different from the mean for swarms not associated with eruptions (Type III) (fig. 7). The mean Mmax is 3.2 for both precursory or Type I swarms and for non-precursory or Type III swarms. The Type I distribution is slightly more spread out than the Type III. The standard deviations of the distributions are 1.3 and 1.0 for Type I and Type III swarm respectively.
The eruptive character is used to divide Type I swarms into two groups. The shaded bars of the top histogram of figure 7 show events that preceded central vent eruptions, while the open bars are from swarms that preceded eccentric (flank) or radial fissure eruptions. With two exceptions, the eccentric and radial fissure eruptions are preceded by larger shocks than central vent eruptions.
Move on to the [NEXT] page.
Return to the GVESD or to the AVO home page.