Slow Earthquakes Linked Along Dip in the Nankai Subduction Zone

See allHide authors and affiliations

Science  10 Dec 2010:
Vol. 330, Issue 6010, pp. 1502
DOI: 10.1126/science.1197102


We identified a strong temporal correlation between three distinct types of slow earthquakes distributed over 100 kilometers along the dip of the subducting oceanic plate at the western margin of the Nankai megathrust rupture zone, southwest Japan. In 2003 and 2010, shallow very-low-frequency earthquakes near the Nankai trough as well as nonvolcanic tremor at depths of 30 to 40 kilometers were triggered by the acceleration of a long-term slow slip event in between. This correlation suggests that the slow slip might extend along-dip between the source areas of deeper and shallower slow earthquakes and thus could modulate the stress buildup on the adjacent megathrust rupture zone.

Slow earthquakes, as defined by anomalously low rupture velocity, reflect a different slip regime than fast faulting of regular earthquakes. In the Cascadia and Nankai subduction zones, episodic tremor and slip (ETS) occur in the transition zone between locked and steady sliding zones along the plate interface (13). ETS, composed of concurrent nonvolcanic tremor and a short-term slow slip event (SSE), is characterized by migration of the sources together along the strike direction of the subducting slab for several days to weeks (24). In the Bungo channel region at the western margin of the Nankai megathrust rupture zone (5), long-term SSEs, which last for several months, recur just updip from the ETS zone about every 6 years (6, 7) and activate ETS (7, 8). In addition, shallow very-low-frequency earthquakes (VLFEs) occur in some areas along the Nankai trough (9). However, the connection between shallow VLFEs and other slow earthquakes remains unknown. We present new observations of temporally correlated occurrences of a long-term SSE, shallow VLFEs, and tremor.

Shallow VLFEs were identified from July 2003, and tremor was recorded from January 2001 through the recurrence of the long-term SSEs in 2003 and 2009–2010 (Fig. 1A) (10). The Global Positioning System (GPS) displacement time series of each SSE is characterized by a slow onset and an acceleration phase in the middle of an episode. The updip tremor activity in the Bungo channel (Fig. 1B) shows a rapid increase in tremor count that corresponds to the acceleration phase in the GPS record during the two SSEs (Fig. 1A), and the tremor activity gradually decreases as the SSE decelerates. In contrast, the downdip tremor (Fig. 1B) exhibits steady activity with an almost constant rate throughout the observation period (8). At the same time, the shallow VLFE activity to the south (Fig. 1B) abruptly begins almost simultaneously with the acceleration phase of the SSEs (Fig. 1A). The VLFE activity is of shorter duration than the SSEs and the updip tremor activity, becoming quiet well before the end of the SSEs.

Fig. 1

(A) Time series of the cumulative number of shallow VLFEs (pink line) to the south, off Cape Ashizuri [pink circles in (B)], cumulative number of tremor sources in the downdip (blue line) and updip (red line) regions in the Bungo channel [blue and red dots in (B)], and (black dots) detrended GPS displacement record (east component) at Ohtsuki [green square in (B)] with respect to Kamitsushima [red square in (B) inset]. The shaded regions indicate the approximate durations of the two long-term SSEs that occurred in 2003 and 2009–2010. The dashed lines denote the beginning of the VLFE activity off Cape Ashizuri. (B) Map of tremor epicenters (orange, blue, and red dots), VLFE epicenters (gray and pink circles), and the 2010 Bungo channel long-term SSE fault (pink rectangle) in southwest Japan. Purple contours show the slip distribution of the 1946 Nankai earthquake (5). Gray contours show ocean depths.

The epicenters of the updip tremor are covered by the slip area of the SSE, whereas most of the downdip tremor is located outside of the slip area (Fig. 1B). This indicates that the contrast in the tremor activities is controlled by the extent of the SSE slip area. A similar relation between SSE slip and shallow VLFEs is suggested; that is, the SSE slip area may extend through the region of no measurable slip to the shallower VLFE source area and activate the VLFEs.

If slow slip connects the deep ETS zone and the shallow VLFE area in the dip direction, then the source area of slow earthquakes might act as a barrier to nearby megathrust rupturing because the slow slip area repeatedly releases the accumulated strain in an interseismic period of the megathrust events. Moreover, the slip area adjoins the megathrust rupture zone, suggesting that the repeating aseismic slip can modulate the stress buildup on the rupture zone. This indicates the importance of monitoring slow earthquakes as proxies for the stress modulation process.

Supporting Online Material

Materials and Methods


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. GEONET, operated by the Geospatial Information Authority of Japan, is gratefully acknowledged.

Stay Connected to Science

Navigate This Article