For each volcano, one or more earthquake swarm records are linked to the volcano table’s records. The swarm records are linked through the Volcanoes of the World catalog-number. Each swarm record is composed of: a header of key fields; a body of swarm, instrumental, and other geophysical parameters; a variable length section containing report excerpts; and references. Three fields are used to ensure that every earthquake swarm record is unique. These fields are called the key fields, and are the volcano catalog-number, the swarm start date, and the swarm type. Every swarm record has a unique value of these three combined fields.
The start date of a swarm in most cases originates directly from the BVE reports. Typically the beginning of a swarm is described as an increase in the number earthquakes reported per day (see swarm duration definition below). Gradual increases in seismicity, problems in network coverage, a high detection threshold, and the lack of a clear definition of when a swarm begins (or ends) are all problems with determining the start time and duration of a swarm. Often these difficulties lead to reports that describe the onset of a swarm in imprecise terms. In order to track these problems a field was added to capture the uncertainties in these measurements. A typical swarm report may read: “Seismicity increased in the middle of November to about 60 events per day. However, there was a decline to 5-20 per day in late December” (Bagana volcano, BVE, 1985, no. 25, p. 20). This swarm was entered into the GVESD as beginning on 85/11/15 + 5 days, with a duration of 40 + 10 days. Table 6 describes the uncertainty values used in several common situations.
Some reports in the BVE include vague descriptions or occurrences of seismicity that cannot be easily classified. We use one place holding record per year to incorporate this information into the GVESD. Place holder records are delimited using the start date, the uncertainty, and the key phrase fields (see below). The start date is ‘year’/7/1 with an uncertainty of + 183 days for a place holding record. The key phrase field also contains the text; “place holder for ‘year’...” to set these records apart. These are examples of our solutions to the problem of coding highly variable reports into quantitative form.
Earthquake swarms are generally defined as a sequence of events closely clustered in time and space without a single outstanding shock (Mogi, 1963). Our working definition follows Mogi’s outline and also requires a significant increase in the rate of local volcanic earthquakes above the background rate. We take volcanic earthquakes to be of any type, for example A, B, (Minakami, 1960), high frequency, low frequency, short period, long period, (Koyanagi, and others, 1987), volcano tectonic, (Latter, 1981), explosion events, etc., but they must occur within an arbitrary near distance to the volcano (typically < 15 km). We do not identify a significant increase over the background rate in a strict statistical sense, but accept the experience and point of view of each reporter. In other words, if the reporter feels that an increase in seismicity is significant enough to report, then we include that report as a swarm record along with a quality modifier. We also do not consider seismic crises (peak seismicity rates within a swarm), obvious mainshock-aftershock sequences, and tremor episodes as swarms. These "non-swarm" seismic sequences are recorded in the GVESD and are delimited in a separate field (QC field explained below). This working definition was developed through the systematic examination of over 600 swarms. One single fixed definition or algorithm might be preferable, but is not feasible due to the widely varying qualities and formats of the data. Future studies would greatly benefit from standardized reporting and the strict application of an algorithm to distinguish the starts, ends and durations of swarms.
We grouped volcanic earthquake swarms according to their temporal relationship to eruptive activity. The swarm types are schematically summarized in figure 3. The main categories are: swarms that precede (Type I), or accompany (Type II) eruptive activity, and those not associated with eruptive activity (Type III). There are a few reported cases of eruptions occurring without a detectable increase in seismicity. These eruptions are included in the database and are identified as Type IV. Roman numerals are used throughout the discussion of swarm type, while Arabic numerals are used in the database for compactness.
Type I, or precursory swarms (46% of the GVESD records), were further divided into 4 sub-types (I a, I b, I c, and I d) according to when the swarm ends in relation to the eruptive activity. Type I a are swarms that begin and end before the eruption commences (for example, 1989 precursory swarm at Izu-Tobu). Type I b are swarms that begin before the eruption and end coincident with the start of the eruption (for example Asama, 1983). Type I c are swarms that begin before the eruption, continue through the duration of the eruption, and end as the eruption ends (for example Oshima, 1987). Type I d swarms begin before the eruption and end after the eruption has ceased (for example Soufriere de Guadeloupe, 1976).
Type II swarms, those accompanying eruptions (15% of the GVESD records), are separated into three sub-types (II a, II b, II c). Type II a swarms begin and end with the eruption. Type II b swarms begin with the eruption and then continue after the cessation of the eruption. Type II c is reserved for swarms that occur during an extended eruption (e.g., the continuing eruption of Kilauea).
Type III swarms are not associated with eruptions (39% of the GVESD records). To separate this category from swarms of Type I a, the time period between the end of the swarm and the next eruption was measured. This quiescent duration is generally less than 10 days with no cases greater than 3 months. Using this observation 100 days is used as a cut-off to separate Type III from Type I a. Post-eruption swarms are also included in Type III category.
In order to further describe the nature of the seismicity that makes up a swarm we added an event type(s) field. Table 7 shows a list of the events types found in the GVESD. Within any swarm there maybe one or more types of seismic event recorded. The event type field attempts to reflect this complexity by listing (in the order of occurrence, if reported) all the event types that occurred during the swarm. Some swarms are defined and reported by event type. At Kilauea, for example, swarms are reported by location and event type. We separated swarms reported at this level of detail into individual swarm records. Most reports do not provide this level of detail. Therefore, most swarm records contain many different event types. Table 7 lists the event types we have defined, with their abbreviations and the numbers of swarm records in which each was used.
We assign an overall quality grade (QC) to each swarm record. The quality grade is intended to be a qualitative statement of the reliability of the report and the swarm record. We assigned quality grades of A through C to each swarm record. The first two grade levels, A and B primarily reflect the report data source. A QC grade of A is given to swarm records that are taken from the primary reviewed literature or from data to which we have primary access. We assume that swarm records derived from these sources are the most dependable. QC grades of B are assigned to swarm records extracted from reports in the BVE. This QC grade level makes up a majority of the records in the GVESD. The C grade is not a reflection of the data source, but is given to records where there is some question about whether the seismicity constitutes a swarm. Mainshock-aftershock sequences, seismic crises, and vague reports of seismic activity are given a QC grade of C. A parallel grading system is used for tremor episodes. Tremor episodes are delimited from true earthquake swarms by using a lower case QC grade (e.g., a, b, and c). Approximately 15% (93 records) of the swarm records in the GVESD are derived from the reviewed literature or locally available data (A-QC). Over half (327 records) of the records were drawn from the BVE (B-QC). A quarter (148 records) of the records are questionable swarms and therefore given a C-QC grade. Tremor episodes comprise about 8% (50 records) of the GVESD.
The maximum magnitude field contains the magnitude of the largest shock within each swarm. Over one fourth (168 cases) of the swarm records contain the magnitude of the largest shock in the swarm. We added a magnitude scale field to qualify the type of magnitude reported (e.g., ML, mb, MJMA, etc.). A specific magnitude scale is reported with the maximum magnitude in only 6% of the swarm records.
The references used to compile the swarm records are listed at the bottom of each record. The first reference in the list is the primary data source, unless otherwise noted. The other references of seismological interest are included with the BVE reports. The comment field is above the reference field within the swarm record. This variable length field contains text excerpted from the original reports. If the report includes pertinent figures a short note is added in the comment field. The short comment field is a one or two line summary of the comment field and gives the essence of the report from which the numerical data were derived.
Move on to the [NEXT] page.
Return to the GVESD or to the AVO home page.