Earthquakes at Loihi Submarine Volcano and the Hawaiian Hot Spot

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 87, NO. B9, PAGES , SEPTEMBER 10, 1982 Earthquakes at Loihi Submarine Volcano and the Hawaiian Hot Spot FRED W. KLEIN 1 U.S. Geological Survey, Hawaiian Volcano Observatory, Hawaii National Park, Hawaii Loihi is an active submarine volcano located 35 km south of the island of Hawaii and may eventually grow to be the next and southernmost island in the Hawaiian chain. The Hawaiian Volcano Observatory recorded two major earthquake swarms located there in and 1975 which were probably associated with submarine eruptions or intrusions. The swarms were located very close to Loihi's bathymetric summit, except for earthquakes during the second stage of the swarm, which occurred well onto Loihi's southwest flank. The flank earthquakes appear to have been triggered by the preceding activity and possible rifting along Loihi's long axis, similar to the rift-flank relationship at Kilauea volcano. Other changes accompanied the shift in locations from Loihi's summit to its flank, including a shift from burst to continuous seismicity, a rise in maximum magnitude, a change from small earthquake clusters to a larger elongated zone, a drop in b value, and a presumed shift from concentrated volcanic stresses to a more diffuse tectonic stress on Loihi's flank. The swarm began at depths of km about 1 month before the shallow swarm started and suggests an upward migration of seismicity as the first stage of the 13-month swarm. The seismic 'root' of Kilauea volcano is well defined by earthquakes which plunge to the south and southwest to a depth of km, terminating in a region in which earthquakes are associated with deep harmonic tremor and magma. The zone of these deeper earthquakes and tremor lies between Kilauea, Loihi, and Mauna Loa volcanoes, may feed magma to all three volcanoes, and probably locates the Hawaiian hot spot. INTRODUCTION Loihi is an active submarine volcano lying at the southeastern end of the Hawaiian-Emperor island and seamount chain, about 35 km offshore from the island of Hawaii. Loihi thus presents a unique opportunity for observation of an active submarine volcano. The youth and activity of Loihi are now clearly established by dredge hauls of fresh basalts, pillow structures, active hydrothermal fields, and bathymetric fault scarps [Moore et al., 1982; Malahoff et al., 1981; Malahoff, 1981]. Observations of shallow earthquake swarms in and 1975 also indicate contemporary volcanism, and perhaps even submarine eruptions during those years. A poorly located swarm in 1952 [Macdonald, 1955] probably occurred at Loihi also. As the Pacific plate is carried to the northwest over the Hawaiian hot spot in he mantle, new centers of volcanism result in new islands to the southeast [Dalrymple et al., 1973]. Loihi may thus be the next island volcano in the Hawaiian chain if its growth continues. Loihi joins the four other active volcanoes in the Hawaiian islands. Loihi's membership in the volcanic family is emphasized by its morphological similarities to the adjacent subaerial volcanoes. Loihi possesses a summit caldera and two radial rift zones [Malahoffet al., 1981; Malahoff, 1981], as do Kilauea and Mauna Loa volcanoes. The seismic and morphologic similarities between Loihi, Kilauea, and Mauna Loa make comparisons particularly [Klein, 1978]. The crustal velocity model used in the locafruitful in interpreting Loihi's seismicity. The basic model of volcanic structure and the eruptive cycle of Kilauea, formulated by Eaton and Murata [1960] and Eaton [1962], has been refined only in detail by later research [cf. Dzurisin et al., 1980]. A key element of this model of Kilauea is a magma tions was derived from a large set of earthquake arrival times that emphasized travel paths on the south side of the island of Hawaii [Klein, 1981]. Earthquake travel times in other parts.of the network also seem to be well modeled. The model uses a velocity of 1.9 km/s at the surface, 6.5 km/s at reservoir located 2 or 3 km beneath the summit caldera. 4.6 km, 6.9 km/s at 15 km, and 8.3 km/s at 16.5 km and Between eruptions, this reservoir refills with magma from a below. Linear velocity gradients are interpolated between Now at U.S. Geological Survey, Menlo Park, California This paper is not subject to U.S. copyright. Published in 1982 by the American Geophysical Union. Paper number 2B source region below about 60 km. Eruptions are generally accompanied by partial emptying of the magma reservoir, deflation of the summit area, and earthquake swarms. Many lines of physical and petrologic evidence demonstrate that most eruptions are fed directly or indirectly from the summit reservoir. Earthquakes and ground deformation observations reveal magma movement through the rift zone to the eruptive vent. Kilauea grows both by surface lava flows originating at the summit and rifts and by intrusion of magma to form dike complexes within the rift zones. In this paper I will first put the earthquakes located at Loihi into the context of seismic observations of Kilauea and the island of Hawaii. Next, the major earthquake swarms of and 1975 will be discussed. Last, the pattern of deeper earthquakes will be examined for their relationship to Loihi and the Hawaiian hot spot. HAWAIIAN EARTHQUAKE DATA Earthquake locations and errors. Between 1970 and 1980 the seismic network operated by the Hawaiian Volcano Observatory grew from about 24 station sites to 47. Increasingly sensitive and more widely distributed stations have improved the quality of earthquake locations from the Loihi area. The seismic network is fully described by Klein and Koyanagi [1980] and Nakata et al. [1980]. All earthquakes were located using the computer program HYPOINVERSE the above points. No single velocity model is totally adequate, since lateral velocity variations may be of the order of 20-30% or larger [Ellsworth and Koyanagi, 1977; Broyles et al., 1979; Zucca and Hill, 1980]. In light of these uncertainties, and the reasonable agreement with the model of Zucca

2 7720 KLEIN' LOIHI EARTHQUAKES TABLE 1. The Distribution of Horizontal Errors for the Loihi Earthquakes With One or Horizontal More S With No S Error Readings Readings and Hill [1980] derived from an unreversed refraction profile south of Hawaii, the above linear gradient model appears to be adequate for locating Loihi earthquakes. The standard errors of epicenter locations calculated by the HYPOINVERSE program [Klein, 1978] indicate that most Loihi earthquakes locate with errors of less than a few kilometers. The program calculates the magnitudes and orientations of the three principal standard errors defining the one standard deviation error ellipsoid. The horizontal error, as used here, is defined as the largest of the three principal errors when projected on a horizontal plane. The distribution of horizontal errors for the Loihi earthquakes appears in Table 1. It is apparent that location errors improve when at least one S phase is used because of the improved distance control for earthquakes outside the network of seismic stations. The larger-magnitud earthquakes (M > 3.0) are all well located, but the location errors do not depend on magnitude for the majority of earthquakes (2.0 -< M < 3.0). Only earthquakes with horizontal errors less than 5 km will appear on epicenter plots which include 96% of events with S readings and 74% of those without S. Epicenter patterns can thus be considered as resolved to better than 5 km in the direction of worst position control (NNW) and somewhat better in other directions. Unmodeled crustal or station biases could introduce systematic location errors of perhaps a few kilometers or less. Depth errors are generally comparable to or larger than horizontal errors but are subject to larger systematic errors arising from uncertainties in the crustal model. General seismicity. Figure 1 summarizes the seismicity of the south side of Hawaii Island. During the 12 years for which earthquakes are plotted, Kilauea was very active with several eruptions and earthquake swarms and one magnitude 7.2 earthquake with a 60-km-long aftershock zone [Tilling et al., 1976; Ando, 1979]. At depths mainly between 2 and 10 km, most of Hawaii' s earthquakes are at Kilauea caldera, its east and southwest rift zones, and the seaward volcanic flank south of the rift zones (Figure l a). These Kilauea earthquakes include (1) shallow (2-4 km) and episodic swarms associated with eruptions and intrusions on the rifts or at the summit caldera and (2) midcrustal earthquakes between 5- and 10-km depths, primarily beneath the Kilauea flank south of the rift zones and in the Kaoiki fault zone between Kilauea and Mauna Loa calderas. These 5- to 10- km earthquakes are tectonic in nature: they occur nearly continually, often in mainshock-aftershock sequences rather than in swarms, and also seem to result from lateral compressional stresses generated by intrusions within the rift zones [Koyanagi et al., 1972]. Mauna Loa volcano was relatively inactive during the period and only erupted once at its summit in a short, relatively low volume eruption in July Rift zones extend to the northeast and southwest from Mauna Loa's summit, and a few earthquakes along these rifts can be seen in Figure l a. The deeper earthquakes (Figure lb), which form a different pattern than the shallow earthquakes, are not closely related to surficial geologic features. Kilauea has a welldefined seismic 'root' which includes a major concentration of earthquakes between 25- and 35-km depth centered just south of the summit caldera. Mauna Loa and Loihi also have seismic roots extending into the upper mantle. The deeper earthquakes are generally scattered beneath and between the three southernmost active volcanoes and around the imme- diate offshore area. Notable clustering of the deepest (35-55 km) earthquakes occurs about 30 km southwest of Kilauea ß.MAUNA LOA:....,... kll' AUEA ' '..'.' '.;" ',¾; ; LOIHI 19 ø 19 ø ß :';.s.'..., km ß. " km depths Fig. la. ß O.....2,0km 20-60[ km depths..., i,,,,,,,, !... Fig. lb. 155 ø Fig. 1. Epicenters of earthquakes during located by the Hawaiian Volcano Observatory on the south side of Hawaii Island. Maximum horizontalocation errors are less than 5 km. (a) Epicenters with depths shallower than 20 km. Most earthquakes are within or adjacen to Kilauea caldera and its two rift zones. (b) Epicenters with depths deeper than 20 km. Most events occur below or to the south of Kilauea caldera.

3 KLEIN' LOIHI EARTHQUAKES ,,, 100 O E 0 m T ' I ' I' ' Ill I' rl' ' I' ' lid ;' 1'' I I I I ' I! I ' I I' ISLE' $EP OCT NOV DEC [19?2FEB R RPR HR¾ JUN JUL RUG P I Seismicity of h... a... y u,,,c u, events detected on the closest coastal station (top) and by the magnitudes of the earthquakes large enough to be located (bottom). These profiles are typical of an earthquake swarm. Magnitudes of zero were assigned to many earthquakes too small to register on the Wood-Anderson seismographs used to compute most magnitudes. The time periods A-F corresponding to frames in Figure 4 were chosen to emphasize changing patterns of earthquakes during the swarm. T caldera. Magma is apparently present in this zone: earthquakes there are often associated with harmonic tremor which is deep but otherwise is similar to shallow tremor seen during eruptions [Aki and Koyanagi, 1981]. The association of some deeper earthquakes with tremor or a root zone beneath Kilauea implies that deeper magma conduits cause some of these earthquakes. Focal mechanism solutions of the deeper earthquakes (E. T. Endo, personal communication, 1979] display radially symmetrical compression axes which point to the center of the island. The island-wide occurrence of deeper earthquakes and orientations of fault plane solutions imply that many deeper earthquakes result from lithospheric flexure caused by the growing load of the island's volcanic pile. Thus the deeper earthquakes reveal both volcanic conduits and recent island growth by lava accumulation. EARTHQUAKES AT LOIHI Swarms. A major earthquake swarm occurred south of Kilauea during March and April 1952 [Macdonald, 1955]..,, liil,, I,i i,, ll ll II Irll i Ill lil Ill l' I'' 200,', cc 0 z 1975FEB I I I I I I HRR I I I R ;I R I HR¾ I I I J ;I N I JUL I I I R lu; IslE[; V D C 1976FEB Fig. 3. Daily counts (top) and magnitudes of located earthquakes (bottom), plotted as in Figure 2, for the Loihi swarm of Note that the scales are identical to those in Figure 2. The time periods G and H correspond to the frames in Figure 4.

4 7722 KLEIN: LOIHI EARTHQUAKES 10' ',./ i; 7:'.-"/ /.- : ',... I f - 19ø 19o i! : / : I...,,i / / 91117'1 ' ß ' [ -10/1/71 30' 20' 10' 15 o 30' 20' 10' 15 10' 10' 19o 19,, IIIIIIIIId,/42112/71-1/1/72 30' 20' 10' 15 30' 20' 10' 15 Fig. 4. Earthquakes during successive time periods A-H at Loihi. The periods are labeled on the time axes of Figures 2 and 3. All plotted earthquakes are shallower than 20 km and have maximum horizontal location errors less than 5 km. The bathymetric contour interval is 500 m. Dashed contours [from Chase eta!., 1980] are derived from bathymetric data readily available. Solid contours are from a more precise and detailed survey [Moore eta!., 1982]. The epicenter locations were derived from S-P times at three or four stations, all at least 55 km north of Loihi, and are poor by today's standards. The earthquakes scatter in an E- W band, and this scatter probably results from the one-sided station distribution. When located using the current crustal model, the 1952 epicenters fall about 10 km north of Loihi's summit. Thus the 1952 swarm could have occurred on Loihi or its north flank. As Figure 1 shows, earthquakes at Loihi are both shallow and deep. The shallow events concentrate near Loihi's edifice, while the pattern of deeper earthquakes suggests a large but diffuse active zone including Loihi, Kilauea, Mauna Loa, and the south side of the island. Nearly all of the shallow and most of the deeper earthquakes plotted at Loihi occurred during two major swarm episodes in and Both the and the 1975 earthquake sequences are apparently volcanic in nature. Figures 2 and 3 depict

5 KLEIN: LOIHI E^RTHQU^ICES 7723 HAHAll ½ : 10' - '// /.// / C /,. / / /.--J /, 19o : :., x,,\ ;H t / " / J -9/11/72: / 30' 20' 10' 15 30' 20' 10' 155o 10' i / i i l, ' /I-I:.....,a / ///// / o/7s ' ß [ -" ' ' 20' 10' 15 Fig. 4. (continued) :30' 20' 10' 15 profiles of the Loihi swarms, both as daily counts of earthquakes mostly too small to locate and as magnitudes of earthquakes large enough to locate. Both sequences are clearly swarms because neither begins with a dominating mainshock. Swarms are typical of stress concentrations such as volcanic intrusions or eruptions [Mogi, 1963]. No harmonic tremor was observed from Loihi during the swarms. However, the tremor typically seen during Kilauea eruptions is only recorded on nearby stations, so the shallow tremor could not be detected from Loihi. On Kilauea and Mauna Loa, earthquake swarms generally accompany the earlier phases of eruptions and intrusions. Eruption seismicity at these volcanoes often subsides after a few days, either because the eruption has ended or because magma flow becomes established through conduits. The opening of new conduits, with its forceful intrusion of magma and earthquakes, is not a required process in an eruption's later stages. The durations of several months of the Loihi swarms are therefore long by Hawaiian standards and may be associated with infrequent eruptions. A low eruption rate for Loihi is suggested by analogy with Kilauea's rift zones as follows. During the last few decades an

6 7724 KLEIN: LOIHI EARTHQUAKES w z w I00 SEPT 17, 1971 AUG 20, : ) YEAR Fig. 5. Cumulative numbers of located Loihi earthquakes separated into deep and shallow categories. The major swarms of and 1975 account for most events. The swarm began at depth about 1 month before shallow earthquakes started, implying an upward migration. The apparent increase in rate beginning in is a result of expanded station coverage and improved recording and analysis procedures. eruption or intrusion occurred on Kilauea's east rift once a year or so, and swarms typically lasted only a few days. By contrast, an eruption or intrusion occurred on Kilauea's southwest rift much less frequently, and the only prolonged Kilauea swarm in recent years occurred on the southwest rift between January and June The infrequent eruptions may mean that the southwest rift cools and hardens between the infrequent eruptions and accommodates new intrusions only with extensive and prolonged seismicity. A second major southwest rift intrusion occurred soon after the first in August It lasted less than 1 week, presumably because the earlier intrusion opened and heated the riff so that later magma found an easy pathway. The Loihi swarms thus suggesthat prolonged and probably infrequent eruptions or intrusions occur at Loihi. Detailed earthquake locations. Epicenters shallower than 20-km depth from the and 1975 swarms are plotted in Figure 4 along with bathymetric contours. Dividing the swarms into discrete time periods aids in relating the seismicity to Loihi's morphology in addition to revealing the changes in earthquake patterns as the swarms developed. Figure 4a shows the first burst of swarm activity during September 1971; this activity is located 1 or 2 km west of Loihi's summit. The scatter of epicenters during this first period is relatively small and has the size and shape of the calculated ellipse of error of any one earthquake. The observed pattern in Figure 4a could thus have come from superimposing random errors on several earthquakes from precisely the same location. The apparent NNW elongation of this earthquake cluster is a result of scattering preferen- tially in the direction of least control. Earthquakes in Figure 4b, which shows the next burst of earthquakes during October 1971, are located in a distinct group about 2 or 3 km east of Loihi's summit. The following four swarm bursts (combined in Figure 4c) are located just southwest of the summit. The first phase of the swarm is thus characterized by temporally and spatially distinct small clusters of earthquakes. These bursts of earthquakes appear to accompany the first stages of intrusive or eruptive episodes near Loihi's summit. Earthquakes during the most intense part of the swarm, plotted in Figures 4d and 4e, represent a marked change in location. The earthquakes of Figure 4d define a NNW trending lineation on Loihi's southwest flank. The seismic lineation is too long (20 km) to result solely from the random scatter of epicenters, although the direction of greatest location error is parallel to the lineation. Both Figures 4d and 4e show that earthquakes fanned out into Loihi's southwest flank following the initial intrusions near the summit. The flank pattern was nearly unchanged during the period of Figure 4e. This behavior is closely analogous to patterns often seen at Kilauea's east rift, where an eruption or intrusion produces seismicity directly under the riff zone followed by earthquakes under the adjacent south flank [Koyanagi et al., 1972]. The summit-flank pattern at Loihi is clearer than at Kilauea because Kilauea is far more active and its flanks are continually active and mobile. The flank earthquakes at Loihi are thus caused by rifting at the summit, and they suggest that Loihi has an active, mobile flank to the southwest. Landslide features seen on Loihi's southwest flank (J. G. Moore, personal communication, 1981) similarly attest to the flank's activity, and some landsliding may be triggered by the M - 4 earthquakes which occur there. The last burst of earthquakes in the swarm (Figure 4f) is located in a small cluster near Loihi's summit, where the swarm began. Earthquakes during the last part of the swarm thus marked a return to the pattern of occurrences in small regions and in short bursts. The concentration of the four bursts illustrated (Figures 4a-4f) around Loihi's summit suggests that any systematic earthquake mislocation is probably 3 km or less. Values of b. A change in b value (slope of the log earthquake frequency versus magnitude distribution) accompanied the shift in earthquakes from summit to flank during the swarm. All other variables being equal, differences in b values can be interpreted in terms of differences in stress. Laboratory data, theoretical considerations, and some earthquake data imply that b is inversely related to effective stress [Scholz, 1968; Wyss, 1973]. Higher b values are also associated with heterogenous and more concentrat ed sources of stress such as volcanoes [Mogi, 1962]. All b values in this paper were computed using the mean magnitude method [Aki, 1965]. Only earthquakes of magnitude 2.5 and larger were used because the frequency-magnitude distribution reveals that the set of earthquakes is incomplete below this magnitude. The b value of the earthquake clusters near Loihi's summit (Figures 4a-4f) is 2.13 _ 0.23 using 89 events. The flank earthquakes (Figures 4d and 4e) display a significantly lower b value of using 239 events. The difference in b value corresponds to a shift in earthquake locations from summit and riff to flank. The changes in b value and location

7 ß ß ß ß KLEIN: LOIHI EARTHQUAKES o 60 South 0 LOIHI.:;.'".':.. ':..... ß (km) ol ß ".o;% ß... ß 11 ß ß ß "ø i ß ø ß :.{? t ' ß '.. '" '! ß ß ß.. ß. ;.. ß KILAUEA ,,,.,,i,,,,,,,... '... North Fig. 6. North-south cross section of south Hawaii earthquakes. The section passes through Kilauea caldera and Loihi seamount and includes the earthquakes of Figures la and lb within 20 km of the section plane. The seismic 'root' of Kilauea can be traced to a depth of about km near the bottom center of the figure. Maximum location errors are 5 km both horizontally and vertically. also correspond to a shift from short and distinct bursts of relatively smaller earthquakes to a more continuous mode of earthquake occurrence where all of the largest (M 3.5) events take place. All these characteristic changes of rift shifting to flank seismicity are seen at Kilauea, where nearly continually occurring flank earthquakes result from intrusions and short-lived bursts of earthquakes within the rift zones. At Kilauea, swarms located under the rift zones which accompany eruptions or intrusions generally show higher b values and maximum magnitudes of about 3.5 compared to lower b values and frequent large earthquakes in the adjacent volcano flanks (F. W. Klein, unpublished data, 1981). Therefore Loihi appears to be a miniature version of Kilauea. Kilauea has two general types of seismicity: First, rift earthquakes accompany intrusions or eruptions, occur in short bursts, are concentrated spatially, are limited in size, have higher b values, and are inferred to result from low but concentrated sources of stress. Second, flank earthquakes increase following rift activity, occur continually, are diffuse spatially, are often large in magnitude, have low b values, and are deduced to result from higher and more diffuse tectonic stress. The words 'volcanic' and 'tectonic' aptly label these two classes of seismicity, respectively. The 1975 swarm was confined to Loihi's rift axis and summit region (Figures 4g and 4h). The swarm was smaller and of shorter duration than the one in and did not trigger any earthquakes in Loihi's flank. A northward migration of epicenters occurred, perhaps triggered by magma movement or migration of stress during a rifting episode. The b value of the 1975 sequence was using 64 events, and seismicity was nearly continuous without isolated bursts. These characteristics are similar to those of the flank events during the earlier swarm and suggest that the 1975 swarm resulted from relatively higher and less concentrated stresses than the earlier swarm. DISCUSSION Loihi and the Hawaiian hot spot. An important question is whether any earthquake evidence exists to define the deep structure of Loihi and relate it to Kilauea and Mauna Loa volcanoes. The large depth errors of Loihi earthquakes prevent a finely detailed look at the vertical structure, although enough resolution exists to contrast deep and shallow events. Earthquakes initiating the major swarm reveal an upward migration. Figure 5 plots the cumulative numbers of deep and shallow events. The swarm clearly contributed the majority of earthquakes and marked the only significant increase of deep events. The swarm began at depth about 1 month before the shallow earthquakes occurred, implying an upward migration of earthquakes and probably magma movement at a speed of very roughly 1 km/d. Depth errors do not permit a closer look at this upward migration. The migration speed is comparable to the lateral migration rates of earthquakes in Kilauea' s southwest rift zone during slow intrusions but is about an order of magnitude slower than similar intrusions in the more active (and presumably hotter) east rift. Upward migration of earthquakes beneath Kilauea is not clearly evident because earthquakes and eruptions are frequent and an obscuring level of background seismicity is usually present. The deeper earthquake structure of Loihi and Kilauea is best shown by the north-south cross section of Figure 6. The seismic root of Kilauea is clearly seen extending below the dense earthquake cloud marking Kilauea's rift zones and south flank. Though the seismic root is somewhat uneven, it bends southward with depth. The diffuse cloud of earthquakes near the bottom left center of Figure 6 is located below Kilauea's lower SW rift zone (see Figure lb), and earthquakes there are often associated with bursts of deep harmonic tremor which signify the presence of magma flow [Aki and Koyanagi, 1981]. Earthquakes seldom occur below 60 km, implying that the asthenosphere cannot sustain brittle fracture even though stresses associated with a deep magma conduit are certainly present. The reason for the major concentration of hypocenters at 30 km below Kilauea is not clear, though the feature was relatively less active before 1968 and after 1975 and the pattern is not permanent. Hypocenters thus define a magma conduit zone to a depth of about 60 km below Kilauea. A very diffuse band of hypocenters links Loihi with the same deep source below Kilauea's lower SW rift. There seems to be a greater density of hypocenters linking Loihi with this deep source than occur directly beneath Loihi. The deep source of tremor beneath Kilauea's lower SW rift happens to be nearly equidistant between Kilauea, Loihi, and Mauna Loa, and this source may represent a common zone through which magma is supplied to the three active volcanoes. Kilauea's seismic root is the best developed of the three volcanoes because it has erupted frequently during the period plotted in Figures lb and 6. Mauna Loa erupted only once (in 1975) during the period, which accounts for the relatively sparse seismicity there in Figures 1 and 6. Even though earthquakes do not reveal much fine structure, the seismic roots of the most active volcanoes appear to coalesce into the same deep source. The 60-km-diameter region of deeper earthquakes in Figures lb and 6 thus may define the location and maximum size of the present Hawaiian hot spot at a depth of km. Future possibilities. The two earthquake swarms at Loihi in the past decade are significant seismic events which are probably associated with magma intrusions or submarine

8 7726 KLEIN: LOIHI EARTHQUAKES eruptions. As at Kilauea, summit and flank earthquakes at Loihi are distinct. When compared to flank earthquakes, summit events occur first and in shorter bursts, are more concentrated spatially, have smaller maximum magnitudes, and show larger b values. The background level of seismicity at Loihi is low, so that future swarms should be easily recognizable. The probable duration of a few months means that there is a chance of diverting an oceanographic ship to study a possible submarine eruption. Techniques such as measuring water temperature and locating earthquakes by using sonobuoys or ocean bottom seismographs (OBS) could be used to locate an eruptive vent for study by precise bathymetry, dredging, water chemistry, and bottom photographs. Future earthquake locations on Loihi will improve somewhat in accuracy as more seismic stations are deployed on land, including horizontal seismometers to improve $ wave timing. The greatest return of information on Loihi earthquakes, however, could come from OBS. A detailed microearthquake study by temporary OBS may not be practical owing to the low background seismicity. A temporary deployment during a swarm would be very useful, although OBS with triggered recording may be filled with many earthquakes in a short time. The most desirable OBS system would be a small array of perhaps four stations tethered to a radio buoy on the surface and recorded at the Hawaiian Volcano Observatory along with the rest of the seismic network. Loihi is a unique opportunity to study an active submarine volcano with a nearby seismic network. Acknowledgments. This study would have been impossible without the quality effort spent in data gathering and seismogram reading by the staff of the Hawaiian Volcano Observatory and especially Robert Koyanagi. I am also grateful to Jim Moore and Alex Malahoff for encouragement and useful discussions in this work. REFERENCES Aki, K., Maximum likelihood estimate of b in the formula log N = a- bm and its confidence limits, Bull. Earthquake Res. Inst. Tokyo Univ., 43, , Aki, K., and R. Y. Koyanagi, Deep volcanic tremors and magma ascent mechanism under Kilauea, Hawaii, J. Geophys. Res., 86, , Ando, M., The Hawaii earthquake of November 29, 1975: Low dip angle faulting due to forceful injection of magma, J. Geophys. Res., 84, , Broyles, M. L., W. Suyenaga, and A. S. Furumoto, Structure of the lower east rift zone of Kilauea volcano, Hawaii, from seismic and gravity data, J. Volcanol. Geotherm. Res., 5, , Chase, T. E., C. P. Miller, B. A. Seekins, W. R. Normark, C. E. Gutmacher, P. Wilde, and J. D. Young, Topography of the southern Hawaiian Islands, U.S. Geol. Surv. Open File Map, , Dalrymple, G. B., E. I. Silver, and E. D. Jackson, Origin of the Hawaiian Islands, Am. Sci., 61, , Dzurisin, D., L. A. Anderson, G. P. Eaton, R. Y. Koyanagi, P. W. Lipman, J.P. Lockwood, R. T. Okamura, G. S. Puniwai, M. K. Sako, and K. M. Yamashita, Temporal variations in gravity on Kilauea volcano, Hawaii, 2, Implications for the magma budget, November 1975-September 1977, J. Volcanol. Geotherm. Res., 7, , Eaton, J.P., Crustal structure and volcanism in Hawaii, in The Crust of the Pacific Basin, Geophys. Monogr. Ser., vol. 6, edited by G. A. Macdonald and H. Kuno, pp , AGU, Washington, D.C., Eaton, J.P., and K. J. Murata, How volcanoes grow, Science, 132, 925, Ellsworth, W. L., and R. Y. Koyanagi, Three-dimensional crust and mantle structure of Kilauea volcano, Hawaii, J. Geophys. Res., 82, , Klein, F. W., Hypocenter location program HYPOINVERSE, U.S. Geol. $urv. Open File Rep., , Klein, F. W., A linear gradient crustal model for south Hawaii, Bull. Seismol. Soc. Am., 71, , Klein, F. W., and R. Y. Koyanagi, Hawaiian Volcano Observatory seismic network history , U.S. Geol. $urv. Open File Rep., , Koyanagi, R. Y., D. A. Swanson, and E. T. Endo, Distribution of earthquakes related to mobility of the south flank of Kilauea volcano, Hawaii, U.S. Geol. $urv. Prof. Pap., 800D, D89-D97, Macdonald, G. A., Hawaiian volcanoes during 1952, U.S. Geol. $urv. Bull., 1021-B, 35-44, Malahoff, A., Contemporary volcanic and hydrothermal activity on Loihi seamount (abstract), Eos Trans. AGU, 62, 1082, Malahoff, A., S. Hammond, and D. Fornari, Loihi submarine volcano: An emerging Hawaiian island? (abstract), Eos Trans. AGU, 62, 431, Mogi, K., Study of elastic shocks caused by the fracture of heterogenous materials and its relations to earthquake phenomena, Bull. Earthquake Res. Inst. Tokyo Univ., 40, , Mogi, K., Some discussions on aftershocks, foreshocks, and earthquake swarms- The fracture of a semi-infinite body caused by an inner stress origin and its relation to the earthquake phenomena, 3, Bull. Earthquake Res. Inst. Tokyo Univ., 41, , Moore, J. G., D. A. Clague, and W. R. Normark, Diverse basalt types from Loihi seamount, Hawaii, Geology, 10, 88-92, Nakata, J. S., W. R. Tanigawa, F. W. Klein, and D. Dzurisin, Hawaiian Volcano Observatory, summary 79 part 1, seismic data, January to December 1979, U.S. Geol. Surv. Hawaiian Volcano Observ., Hawaii National Park, Scholz, C. H., The frequency-magnitude relation of microfracturing in rock and its relation to earthquakes, Bull. Seismol. Soc. Am., 58, , Tilling, R. I., R. Y. Koyanagi, P. W. Lipman, J.P. Lockwood, J. G. Moore, and D. A. Swanson, Earthquake and related catastrophic events, island of Hawaii, November 29, 1975: A preliminary report, U.S. Geol. Surv. Circ., 740, Wyss, M., Towards a physical understanding of the earthquake frequency distribution, Geophys. J. R. Astron. Soc., 31, 341, Zucca, J. J., and D. P. Hill, Crustal structure of the southeast flank of Kilauea volcano, Hawaii, from seismic refraction measurements, Bull. Seismol. Soc. Am., 70, , (Received January 5, 1982; revised June 25, 1982; accepted June 28, 1982.)