Slow Slip Events Following the 2002 Mw 7.1 Hualien Offshore Earthquake Afterslip

The recurrence intervals of slow slip events may increase gradually after a large earthquake during the afterslip. Stress perturbations during coseismic and postseismic periods may result in such an increase of intervals. However, the increasing recurrence intervals of slow slip events are rarely observed during an afterslip. The evolution process along with the afterslip remains unclear. We report an observation of slow slip events following the 2002 M w 7.1 Hualien offshore earthquake afterslip in the southernmost Ryukyu subduction zone. Slow slip events in 2005, 2009, and 2015 are adjacent to the M w 7.1 earthquake hypocenter. An increasing slow-slip interval of 3.1, 4.2, and 6.2 years has been observed after the earthquake. We calculated coseismic and postseismic slips from the M w 7.1 earthquake and then estimated the Coulomb stress changes in the slow slip region. The M w 7.1 earthquake has contributed positive Coulomb stresses to both the 2005 slow-slip region and 2009/2015 repeating slow-slip region. The coseismic and postseismic Coulomb stress change on the 2005 slow-slip region is approximately 0.05 MPa and 0.035 MPa, respectively. However, both Coulomb stress changes on the 2009/2015 repeating slow-slip region are not over 0.03 MPa. The ongoing afterslip following the M w 7.1 earthquake last for at least ve years, evolving with a decaying stress rate with time. The long-term stress perturbations may be able to trigger the 2005 slow slip event during the afterslip. The 2009 slow slip event seems to be inuenced by the afterslip as well. Postseismic stress evolution and frictional and stressed conditions of the slow-slip region can be a reason to affect the evolution process of slow slip events intervals. perturbations the intervals? To this end, we calculated the coseismic and postseismic slips from the M w 7.1 earthquake in the subduction zone. Then Coulomb stress changes are estimated based on both the slips on the SSEs region. The Coulomb stress changes were examined to identify whether they were sucient for an increase of SSE intervals. We nd that SSE regions were very likely overlapped by the afterslip region of the M w 7.1 earthquake. The 2005 SSE region has been imposed by higher positive Coulomb stresses than the 2009/2015 SSEs region. The afterslip lasted from 2002 April to at least early 2007 with a time-decaying stress rate. The continuous positive stress loads may be sucient for the triggering of 2005 SSE and affect the 2009 SSE. Our study provides an observation of SSEs sequence following afterslip that can support the Boso SSEs case.

Hualien offshore earthquake was situated at the western end of the ISZ and above its interface in the overriding plate (Fig. 1a). The M w 7.1 earthquake was less studied due to the lack of near-eld observations, only for the coseismic displacement regarded as repeating SSEs. The three SSEs lasted from 2 to 4 months with a potential maximum size, M w 6.4 to 6.6. The SSEs were likely originated from a high V P /V S ratio zone on the subduction interface (Huang et al. 2014) and are accompanied by overriding plate seismicity with maximal magnitudes greater than M w 5.0. The peak slip of the three SSEs is adjacent to a high b-value region in the northeastern Taiwan orogen . A state of low differential stress may thus appear in that region (Scholz 2015). These observations agree with a broad consequence of SSEs that usually occur in the state of rich high-pressure uids, low effective stress, and transitional friction (e.g., Bürgmann 2018; Saffer and Wallace 2015; Schwartz and Rokosky 2007). The close distances from the SSEs to the M w 7.1 earthquake allow us to investigate the relation between afterslip and SSE intervals.

Data And Methods
To answer the questions raised in this study, we calculated Coulomb stress changes from the 2002 M w 7.1 Hualien offshore earthquake coseismic and postseismic slips. Coseismic and postseismic Coulomb stress change can commonly explain the triggering of spatiotemporally neighboring earthquakes around a mainshock (e.g., King et al. 1994;Stein 1999). However, there were no near-eld GNSS observations around the M w 7.1 earthquake (Fig. 1a). We made assumptions for the calculations of the coseismic and postseismic slips. First, the coseismic slip was calculated on a single fault in the subduction zone by fareld GNSS coseismic displacements (Chen et al. 2004). Strike and dip of the fault are 277° and 44° (northward), respectively, constrained by earthquake focal mechanism in a relocated earthquake catalog (Wu et al. 2008). An elastic half-space dislocation model (Okada 1992) was used to invert the coseismic displacements with the rigidity of 30 GPa. The fault dimension was initially presumed to be 50 x 30 km (Lee et al. 2009) and then optimized by grid search for the geometrical parameters. This method minimized the residuals between observed GNSS and modeled displacements. The two components of the fault dislocation vector were estimated by the least-squares. Second, the postseismic displacements following the M w 7.1 earthquake were determined from the same far-eld GNSS observations. The data are derived from the northeastern Taiwan region and Yonaguni Island (Fig. 1a) (Fig. 4a). The peak-slip location can correspond to the largest coseismic displacements that appear to the northeastern Taiwan region. The coseismic slip over 1.5 m covers almost half of the fault plane and is commonly less than 1.0 m in the updip region. The postseismic displacements likely lasted from 2002 April to early 2007 in the nearest GNSS station to the epicenter (Fig. 3). The duration of afterslip is at least ve years, consistent with a previous study using another GNSS station on Yonaguni Island (Nakamura 2009). Analyzing postseismic displacements from continuous GNSS records, most of the afterslip were surrounding the region of coseismic peak slip (Fig. 4b). The peak afterslip appears at the northwestern downdip end of the fault plane with approximately 2.6 m. The peak-slip location seems closer to the 2009/2015 repeating SSEs region than that of the coseismic peak slip. The amount of the overall afterslip is 60 to 70 % percent of the coseismic slip. It is reasonable because the postseismic displacements were less than the coseismic displacements by similar amounts. . The rst and second Boso SSEs showed much more shortening than the SSEs in our study area. The amount of shortening SSE intervals along with afterslip may be controlled by earthquake size and corresponding stress perturbations (Luo and Liu 2019). The larger the earthquake size is, the greater the amount of shortening SSE intervals will be. It might explain a smaller amount of the rst SSE shortening interval from this study (Fig. 1a). It might also explain the shortening SSE intervals decrease slowly with time in the M w 7.1 case. A weaker afterslip following the M w 7.1 earthquake size than the M w 9.0 one could be a reason. Note that the increasing recurrence intervals of the Boso SSEs in Japan recurred on the same fault patch. This patch is not overlapped by the afterslip region of the M w 9.0 Tohoku earthquake but is close enough. Our results indicate that the afterslip region of the 2002 M w 7.1 Hualien offshore earthquake very likely overlapped the SSE regions. The close relation makes the in uence of afterslip on SSE sequence more reasonable observed with positive coseismic and postseismic stress loads. Thus, SSEs at the edge or even within the afterslip region may be triggered when the SSE region is approaching a triggering stress level. This interpretation does not limit the observations to the repeating SSEs on the same fault patch. Continuous monitoring of future SSEs in our study region and the Boso region in Japan would understand the discrepancy.

Conclusions
We report an increase of SSEs intervals after a large earthquake in the southernmost Ryukyu subduction The GNSS data are available at http://gps.earth.sinica.edu.tw/. The Coulomb stress calculations were performed using Coulomb 3 software, available from https://www.usgs.gov/software/coulomb-3.

Competing interests
The authors declare that they have no competing interests.      The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.