Introduction
Seismic events with weak P phases but well-developed monochromatic shear-wave trains are often recorded on the University of Alaska’s high-gain seismic station SCM in southern Alaska (Fig. 1). Occasional large events are also recorded on some of the National Oceanographic and Atmospheric Administration’s Palmer Seismological Observatory stations. These additional data permit determination of their source areas. What may be similar events have been reported from south-east Alaska and British Columbia (G. C. Rogers, Victoria Geophysical Observatory; personal communication from J. Davies) and apparently from Antarctica (Reference AdamsAdams, 1971). However, we know of no published study in which sufficient data were accumulated to allow accurate source locations.
Event Description
The events considered in this study typically show a weakly developed P phase and an almost monochromatic, non-dispersive, well-developed wave train propagating at shear-wave velocities (Fig. 2). The period of this distinctive wave train is normally about 0.6 s and the ground motion at recording stations seems to exclude their interpretation as surface waves. Some of the larger events have been recorded as far away as station PJD, 420 km from the source, and show on dispersion. Smaller events with the same characteristic monochromatic wave train and a P phase so weak as to be undetectable are often recorded at station SCM. The source of these weak events is probably the same as that for the larger events which were recorded well enough to allow location.
Frequency of Occurrence
Although complete tallies of the events have not been kept, there are several time periods during which useful data on their occurrence were obtained as an adjunct to other studies. Daily counts for two of these time periods are shown in Figure 3. Station SCM is closer to the source area and of higher gain than the Palmer stations, so one should not compare the level of activity of the two time periods. Although the smaller events recorded at SCM are often difficult to distinguish from “background noise”, they often occur in large numbers (Fig. 4), as many as 60 per day.
No data have been compiled to investigate seasonal variations in frequency of occurrence. However, these events have been noticed throughout the year.
Location of Events
To determine the source area of these unusual events, all of the larger ones which occurred during the time periods shown in Figure 3 were scaled from both the Palmer Observatory and the Geophysical Institute records. Due to the very weak nature of the P phases, only 12 of the events were recorded well enough at nearby stations to allow epicenter determination. A notable event which occurred in 1968 (Fig. 2) was also included and these 13 events were processed with the Geophysical Institute’s routine earthquake epicenter-location computer program which is based on P-phase transit times. The locations of the events are shown in Figure 5 and other parameters are given in Table I. Most of the events are clustered on or near Harvard Glacier at the head of College Fiord and all but two of the events yielded a surface source.
Event 1, located north-west of the cluster at a depth of 15 km, has a signature similar to events located on the glacier and may be slightly in error. Event 3 is probably not from Harvard Glacier as the signal character (Fig. 6) and the order of P-phase arrivals at the stations indicates a different source. Event 7 is definitely from a source near the mouth of College Fiord and, unlike events from Harvard Glacier, was first recorded at station PMS. Other events too small to locate but with similar signature and arrival times have been observed, although the source area for event 7 is not as prolific as the Harvard Glacier area. Event 9 probably represents another source for these events, although no others have been located there and none has been noted on the records to have similar character. Event 10 has first P-phase arrival at station PMR and a character different from the events near Harvard Glacier. However, the calculated depth of 250 km and the rather poor quality of the P phase indicates possible error in location and it seems prudent to disregard event 10.
Figure 6 contrasts typical events 2 and 5 from the group on Harvard Glacier with the anomalous events 3, 7, 9 and 10 as recorded at SCM. The differences in signature characteristics are readily apparent and, in addition, event 7 was noted to have a very unusual signature at PMS, the station closest to that event.
Similarities of signature for the many small events recorded at SCM show that the great majority of them originate in the Harvard Glacier area, with a few coming from the area near the mouth of the fiord.
A few rare events similar in appearance to those from College Fiord have been recorded at stations BLR and PAX. The S minus P times of these small events and the order of arrival at the two stations dictate a source within the Alaska Range, although none of the events has been large enough to locate. We speculate that the common signal forms of these events indicate a mechanism similar to the source in the College Fiord area.
The uncommon monochromatic shear-wave train is a source characteristic since it propagates in different azimuths and to a distance of 400 km (for the larger events) conserving its signature. The peculiarities of this predominantly shear-type source will be further investigated.
Magnitude and Energy
The Richter magnitudes computed for the event of 20 May 1968 at stations SCM, BLR and PJD were 2.3, 2.0 and 2.1, respectively. For an earthquake of magnitude 2.1, the energy, E (in ergs), calculated from the empirical formula (Reference RichterRichter, 1958, p. 366)
is 1014.5 erg (107.5 J). If one assumes that these events are produced by glacier motion, it is instructive to calculate the amount of motion required to provide this much seismic energy. Applying this formula to the events in this study will give only a rough estimate of the total energy involved because little is known of the conversion efficiency of glacier potential energy to seismic energy.
If an area A of the glacier bed is stressed by the yield stress Y of the ice, the displacement on release of the stress is d and, if the seismic efficiency is e, then the energy available is
Specifying reasonable upper limits for unknowns, A and e, will give a lower limit for d. Thus, assuming a yield strength of 1 bar, Y = 105 N m−2, and using the calculated energy E = 107.5 J, we have Ade ≃ 300 m3, so that if we suppose A not more than 1 km2 and e not more than 10%, we find that d is not less than 300 × 10−5 m or d ≃ 3 mm. Although these figures are based on some assumptions that may not be valid, they appear reasonable and are at least within a few orders of magnitude of reality and indicate that measurement of seismic energy release for a substantial time period could give an indication of the total mass movement on a glacier.
We hypothesize that these unusual seismic events are due to glacier movement and that the movements associated with some events may be quite large. Although all the events we have been able to locate have been near glaciers in marine environments, we do not feel that they are produced by calving because the locations are well inland with the exception of event 7. Also the weak events of similar character from the Alaska Range exclude calving as an essential source.
Harvard Glacier, where most of the events are produced, is the most active glacier in the Prince William Sound area (personal communication from Larry Mayo, U.S. Geological Survey).
The very nature of these events excludes several sources. For example, earthquakes at comparable distances give records which have a much higher frequency content and are not at all monochromatic. The very weak P phases and pronounced S phases of these events are unlike those of any other event observed by the authors.
Conclusions
There is every indication that the unusual seismic events generated in the College Fiord-Harvard Glacier area are associated with glaciers. The energy of the largest events and the frequency of occurrence of the smaller events indicates that jerky, short-lived rapid motion could significantly contribute to the total movement of the glacier.
Acknowledgements
The authors are indebted to Mr Howell Butier of the Palmer Observatory for access to data from stations PMR, PMS, PWA and TOA. These data were critical to this study. Mr John Davies graciously contributed the data in the upper part of Figure 3. The continued interest and enthusiasm of Dr William Harrison is sincerely appreciated. We extend our thanks to the anonymous referee who made valuable suggestions for the section on magnitude and energy. The study was financed by various grants and contracts from the Air Force Office of Scientific Research, the National Science Foundation and the State of Alaska.