localized to DRAO itself are not the source of
the transient reversals. The 14.5-month aver-
age recurrence is similar to the period of theChandler resonance (8). Nonetheless, thisresonance cannot explain the observed base-
line changes or the sawtooth pattern in thetime series.
M. Meghan Miller,* Tim Melbourne, Daniel J. Johnson,
earthquake that began on 7 February 2002 at
Friday Harbor, Washington (Fig. 1; inset). Discrete offsets in the north and east compo-nents fit the pattern observed at nearby sta-
fected region may last for 8 weeks. The slow
tions in previous events and are visible at this
earthquakes occur on average every 14.5 Ϯ 1
new station in the daily positions. The offset
months over the past 10 years and typically
and partial recovery in the north component
amounts of strain energy without detectable
begin in the Puget Lowlands near the arch in
reflect propagation of a slow earthquake un-
seismic shaking, and they are thus known as
the downgoing Juan de Fuca plate. Observa-
slow or silent earthquakes. Because subduc-
tion zones generate the largest earthquakes,
quakes invites an assessment of the role of
seismic hazard assessment relies on under-
transient, aseismic creep events in the de-
forming Earth. Over the time scale of re-
quakes reduce the energy released by infre-
currence intervals for the plate interface,
quent large earthquakes. We present data that
indicate the periodicity of slow earthquakes
in Cascadia and the current onset of such an
field determined over the past 10 years by
GPS contains a rich record comprising per-
manent deformation overlain by an elastic
occurred over its 1500-km length from cen-
strain field, which is ultimately recovered
tral British Columbia to northern California
as the Juan de Fuca plate subducts beneath
the North American plate (1– 4 ). Since 1992,
regular and cyclical nature of the transient
continuous global positioning system (GPS)
events indicates that they are a fundamen-
monitoring of changes in station position
have been used to monitor surface deforma-
tion associated with tectonic events (5, 6 ). Inthe fore arc that lies between the offshore
References and Notes
plate boundary and the Cascade volcanoes,
1. J. Adams, Tectonics 9, 569 (1990).
2. K. Satake, K. Shimizaki, Y. Tsuji, K. Ueda, Nature 379,
current interseismic deformation is dominat-
ed by northeast-directed contraction, the re-
3. B. F. Atwater, E. Hemphill-Haley, Recurrence intervals
sult of shallow convergence of the Juan de
for great earthquakes of the past 3,500 years atFig. 1. Time series for GPS stations along the northeastern Willapa Bay, U.S. Geol. Surv. Prof. Pap.
Fuca plate beneath the North American plate.
Cascadia margin. Baseline lengths are from
108 (1997).
During 1999, a sudden shift from fore arc
4. D. K. Yamaguchi, B. F. Atwater, D. E. Bunker, B. E.
weekly station positions with respect to Pen-
contraction to short-lived extension led to
Benson, M. S. Reid, Nature 389, 922 (1997).
ticton (DRAO). (Inset) Time series shows daily
5. H. Dragert, R. D. Hyndman, Geophys. Res. Lett. 22,
positions for Friday Harbor (SC02) for 2002 in
the global reference, illustrating the onset of
6. M. M. Miller et al., Tectonics 20, 161 (2001).
7. H. Dragert, K. Wang, T. S. James, Science 292, 1525
8. L. Xie, S. R. Dickman, Geophys. J. Intl. 121, 117
indicates that eight slow earthquakes have
occurred since 1992. Both fore arc station
for each event. Yet the last three events have
9. This material is based on work supported by the
NSF under grants EAR-0002066, EAR-000003483,
positions in the global reference frame and
well-mapped displacement fields that differ
EAR-9616540, and EAR-9973191 and through sup-
precise baselines relative to Penticton, British
across the network, implying differences in
port of the University NAVSTAR Consortium Boul-
the fault area that failed. Modeling suggests
der Facility; by NASA contract NAG5-7672; and by
extension events superimposed on interseis-
creep of a few cm along the plate interface at
the U.S. Geological Survey through annual awards
under the National Earthquake Hazard Reduction
mic crustal shortening (Fig. 1). DRAO lies
Program to member institutions of the PANGA
inland and is largely unaffected by conver-
release approximates a M ϭ 6.7 earthquake,
consortium. The Western Canada Deformation Ar-
gence-related contraction. The transient re-
consistent with previous analysis (7 ).
ray operated by the Geological Survey of Canada is
versals from secular contraction are consis-
No corresponding periodicity characteriz-
tent with southwest-directed fore arc motion
es the DRAO time series in the global refer-
that is both spatially and temporally coherent.
Department of Geological Sciences, Central Washing-
Each event starts over a 3-week period and
other PANGA back arc stations, those that lie
ton University, Ellensburg, WA 98926, USA.
lasts 2 to 4 weeks at any one station. Propa-
east of the volcanic front. Thus, tectonic,
*To whom correspondence should be addressed. E-
gation of the slow earthquake across the af-
tropospheric, or monument instability signals
www.sciencemag.org SCIENCE VOL 295 29 MARCH 2002
Hardware removal Surgery PoSt-oPerative inStructionS Bleeding through the dressing is normal. Expect some reddish staining of the splint near the surgical site. An ACE wrap with additional gauze pads may be used to cover this up and reinforce the dressing/splint. You maybe bear weight on the leg or arm as tolerated unless instructed otherwise. Elevate the extremity on 2-3 pillows as much as