Earth and Planetary Science Letters 223 (2004) 203 – 212 www.elsevier.com/locate/epsl
Inter-plate coupling in the Nicoya Peninsula, Costa Rica, as deduced from a trans-peninsula GPS experiment Takeshi Iinuma a,*, Marino Protti b,1, Koichiro Obana c,2, Victor Gonza´lez b,1, Rodolfo Van der Laat b,1, Teruyuki Kato a,3, Shin’ichi Miyazaki a,4, Yoshiyuki Kaneda c,5, Enrique Herna´ndez b,1 a
Earthquake Research Institute, the University of Tokyo, 1-1-1, Yayoi, Bunkyo Tokyo, Japan Observatorio Vulcanolo´gico y Sismolo´gico de Costa Rica, Universidad Nacional, Apartado 2346-3000, Heredia, Costa Rica c Institute for Frontier Research on Earth Evolution, Japan Marine Science and Technology Center, 3173-25, Showa-machi, Kanazawa-ku, Yokohama 236-0001 Japan
b
Received 21 November 2003; received in revised form 2 April 2004; accepted 12 April 2004
Abstract We investigated the state of plate coupling at the subduction zone beneath the Nicoya Peninsula, northwestern Costa Rica, from 1.5 years of trans-peninsula GPS campaign experiments. This area is recognized as a seismic gap located between the rupture areas of two M z 7 earthquakes that occurred in the early 1990s, and it has not ruptured since 1950. We carried out campaign GPS observations beginning in autumn 2001 to obtain the profile of site velocities across the peninsula for the interseismic period. The obtained velocity field indicates a strong coupling at the plate interface. However, velocity directions deflect counterclockwise, which suggests a trench-parallel forearc sliver motion of Nicoya Peninsula at f 8.5 mm/year. An inversion analysis to infer interplate coupling at Nicoya seismic gap shows that the strongly coupled zone agrees well with seismogenic zone inferred from seismicity. Cumulative slip deficits since 1950 at the Nicoya seismic gap would produce a thrust earthquake as large as MW z 7.5. D 2004 Elsevier B.V. All rights reserved. Keywords: Costa Rica; Nicoya Seismic Gap; inter-plate coupling; forearc sliver
1. Introduction * Corresponding author. Tel.: +81-3-5841-5736; fax: +81-35689-7234. E-mail addresses:
[email protected] (T. Iinuma),
[email protected] (M. Protti),
[email protected] (K. Obana),
[email protected] (V. Gonza´lez),
[email protected] (R. Van der Laat),
[email protected] (T. Kato),
[email protected] (S. Miyazaki),
[email protected] (Y. Kaneda),
[email protected] (E. Herna´ndez). 1 Tel.: +506-261-0781; fax: +506-261-0303. 2 Tel.: +81-45-778-5436; fax: +81-45-778-5439. 3 Tel.: +81-3-5841-5730; fax: +81-3-5689-7234. 4 Tel.: +81-3-5841-5684; fax: +81-3-5689-7234. 5 Tel.: +81-45-778-5398; fax: +81-45-778-5439. 0012-821X/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2004.04.016
Costa Rica is located in the western margin of the Caribbean plate, where the Cocos plate is subducting beneath the Caribbean plate from the Middle American Trench (Fig. 1). The convergent rate between these two plates is about 90 mm/year in this area [1]. Because of this rapid convergence, large earthquakes have frequently occurred beneath the western coast of Costa Rica. The Nicoya subduction segment in northwestern Costa Rica ruptured with large earthquakes in 1853, 1900, and 1950.
204
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
Fig. 1. Tectonic map of Costa Rica [4]. Solid circles and solid triangles represent locations of M>7 earthquakes of 20th century with their year and active volcanoes, respectively. Dashed lines perpendicular to the Middle American Trench delineate thrust segments. Dotted lines shows the aftershock zones of the earthquakes occurred in 1992 (at Nicaragua segment), 1950, 1978 (at Nicoya segment), and 1990 (at southeastern offshore of the Nicoya peninsula).
The ‘‘Nicoya seismic gap’’ is bounded by the rupture areas of two large earthquakes that occurred at both sides of this peninsula; MW 7.0 in 1990 to the southeast and the 1992 MW 7.6 off Nicaragua to the northwest. A strong coupling of the two plates is manifested in this seismic gap by low background seismicity, the sharp edge to the aftershock zone of the 1990 and 1992 earthquakes [2], and the northeastward movement of the Nicoya peninsula recorded with GPS observations [3]. Protti et al. [4] estimated that a MW = 7.5 earthquake is likely to occur in the Nicoya seismic gap. Since the Nicoya peninsula lies above this gap, GPS observations in this peninsula provide invaluable data to evaluate the seismic coupling, the updip and downdip extent of the seismogenic zone and the size of the potential earthquake. Thus, we started a new GPS survey program in the Nicoya peninsula in 2001 by the Observatorio Vulcanolo´gico y Sismolo´gico de Costa Rica, Universidad Nacional (OVSICORIUNA) and the Japan Marine Science and Technology
Center (JAMSTEC) with the support of the Japan International Cooperation Agency (JICA). Seven new benchmarks for GPS campaign observations were installed completing a 10-site transect across Nicoya peninsula [5]. In 2002, the project was progressed by OVSICORI-UNA, Earthquake Research Institute, University of Tokyo (ERI), and JAMSTEC as a cooperative operation of JICA. The second campaign observation was held in autumn 2002, and a third campaign was carried out in spring 2003. We obtained displacement vectors from these data and estimated the inter-plate coupling of the Nicoya seismic gap.
2. Data and velocity field We established an array of GPS campaign sites as shown in Fig. 2. Three sets of GPS campaign observations were carried out in the periods as shown in Table 1.
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
205
Table 1 The periods of GPS campaign observations
We used Trimble 4000SSE/5700 GPS receivers and collected data for several days in each campaign. We analysed the data of campaign sites using GIPSYOASIS II software [6], together with 12 surrounding IGS sites to resolve the phase ambiguities. We then subtracted the Caribbean plate motion model of DeMets [1] from velocities in ITRF2000 of Sella et al. [7]. Table 2 shows estimated all velocity data of 10 campaign sites in Caribbean plate fixed reference frame and also in ITRF2000. Fig. 2 shows the horizontal velocity field in Caribbean plate fixed reference frame. Velocities, as well as Cocos plate motion, have oblique motion relative to the Caribbean plate compared with the direction normal to the Middle American Trench. The velocity field from Lundgren et al. [3] is also plotted in this figure. Their result is not significantly different from our result in tendency. Fig. 3A,B,C shows north – south, east – west, and vertical components of obtained time series in Caribbean plate fixed reference frame at all sites, respectively. Fig. 4 is a plot of the velocity profile with respect to
1st 2nd 3rd
Start
End
2001 – 10 – 01 2001 – 10 – 23 2002 – 09 – 23 2003 – 02 – 12 2003 – 03 – 11 2003 – 04 – 23
2001 – 10 – 07 2001 – 11 – 02 2002 – 10 – 09 2003 – 02 – 16 2003 – 03 – 18 2003 – 05 – 13
the distance from the trench. Here, we used the horizontal components of velocities projected to the line parallel to the GPS array, which is displayed in Fig. 2 by a thick gray line. This GPS array line is set as normal to the Middle American Trench (N47jE).
3. Back slip inversion To estimate inter-plate coupling, we conducted a simple inversion analysis. In this inversion, all GPS sites were assumed to compose a single line array.
GRAN BALL
MATA
SJOS
SJOS
LAJA PALO PUJE UVIT SJUA HUAC SANI JICA SAMA JOBO CARR O E PAQU INDI 7 N4
8.6
5 cm /yr (N2 O Av 3E sli era ) p d ge ire ea cti rth on qu (N ake 33 O E)
GUAR
ZUMA
CABU
AGUJ
3cm/yr
km 0 10 20
Fig. 2. A GPS observation line across the Nicoya Peninsula and estimated velocities from campaign observations. Black circles, vectors, and four-character acronyms represent our sites, velocities, and site IDs, respectively. Gray squares, vectors, and acronyms in italics are sites, velocities, and site IDs of Lundgren et al. [3]. Error ellipsoids represent 95% of confidence limits. The motion of the Caribbean plate was fixed by using plate motion model of Sella et al. [7]. A broken line indicates Middle American Trench and a square of dotted line shows supposed plate boundary fault in inversion analysis. The motion of Cocos plate to Caribbean plate and average earthquake slip direction [1] are also indicated.
206
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
Table 2 GPS site locations and their estimated velocities and standard deviations Site ID
Latitude
Longitude
Vn
rn
Ve
re
Vh
rh
INDI
9.86jN
85.50jW
CARR
9.89jN
85.47jW
JOBO
9.96jN
85.45jW
SANI
9.98jN
85.40jW
HUAC
10.02jN
85.35jW
UVIT
10.07jN
85.31jW
PUJE
10.11jN
85.27jW
PALO
10.24jN
85.22jW
LAJA
10.27jN
85.05jW
SJOS
10.37jN
84.95jW
20.89 24.58 23.04 26.75 25.98 29.71 16.15 19.90 24.13 27.89 12.96 16.74 18.02 21.82 18.11 21.93 16.64 20.52 11.98 15.91
0.33 0.31 0.52 0.48 0.51 0.47 0.51 0.47 1.00 0.93 0.44 0.41 0.45 0.42 0.59 0.55 0.53 0.49 0.54 0.50
6.54 20.20 1.88 15.53 1.50 12.11 4.91 18.52 0.93 14.52 3.37 16.93 2.79 16.34 4.89 8.60 1.35 14.83 1.54 14.98
0.53 0.49 0.82 0.76 0.81 0.75 0.82 0.76 1.66 1.54 0.72 0.67 0.67 0.62 0.84 0.78 0.85 0.79 0.84 0.78
1.33 0.92 7.79 8.20 9.03 8.63 0.82 0.41 25.95 25.54 8.14 7.74 6.15 5.75 7.96 8.36 0.12 0.28 0.89 1.29
1.72 1.60 2.66 2.47 2.55 2.37 2.59 2.40 5.00 4.64 2.25 2.09 2.38 2.21 2.98 2.77 2.69 2.50 2.73 2.53
Vn, Ve, and Vh represent north – south, east – west, and vertical component of the velocities (mm/year), respectively. rn, re, and rh are standard deviations corresponding to each velocity components. Upper columns of each cell are the velocities and standard deviations in Caribbean plate fixed reference frame, and lower columns show ones in ITRF2000.
Horizontal velocity vectors were projected onto the direction that is parallel to the thick gray line in Fig. 2. We used a plate boundary fault that increases its dip angle, as shown in Fig. 5a. The strike of this fault is N43jW and the dip changes as shown in Table 3 [4,8]. It is widely accepted that the inter-plate coupling is strong at an intermediate depth while plate interface is steadily slipping in shallower and deeper parts. This situation is modeled as a superposition of two virtual terms. One term is a steady subduction of oceanic plate and is represented by a uniform forward slip distribution along the entire plate boundary. The other term is a backward slip distribution on the plate boundary, and this term is called ‘‘slip deficit’’ or ‘‘back slip’’ [9]. We can estimate this back slip distribution from the observed surface crustal deformation because the uniform steady slip yields only negligible deformation [10]. We assumed that the distribution of back slip is purely two-dimensional; direction of back slip is normal to the trench and allows only its amplitude to vary along the same direction. Yabuki and Matsu’ura’s [11] inversion method was adopted to esti-
mate the back slip distribution. In this inversion method, the weight of a prior constraint of smoothness of back slip distribution is optimized by minimizing the Akaike’s Bayesian Information Criterion (ABIC). The results of inversion analysis are presented in Fig. 5b. Fig. 5c,d shows predicted surface displacement rate from results of inversion analysis. Because of the lack of data from third campaign, the velocity data of HUAC has large error and could be omitted as an outlier (Fig. 5d). Thus, we performed the inversion analyses with and without HUAC’s velocity data. Since these results were not significantly different, we adopted the result using all data for the present study. Lundgren et al. [3] also estimated inter-plate coupling in the region that contains the Nicoya seismic gap. Their result shows that the dip-slip component of back slip has maximum rate beneath southern coast of the Nicoya peninsula while our result shows that the peak of back slip rate locates offshore of the peninsula. The difference of spatial density of observation sites may cause this discrepancy. Results of the inversion analysis indicate that the downdip limit of the inter-plate coupling zone is at
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
about 30 km. This is consistent with the estimation from seismic data obtained with a dense earthquake observation network [12]. Newman et al. [12] report that the updip limit of seismogenic zone is about 10– 13 km in depth with a few earthquakes as shallow as 8 km. From our results, it is apparent that the updip limit of coupling zone is not deeper than 7 km. Even if it may be difficult to argue the updip limit from our inversion result because of the constraint condition that the back slip should be zero at the trench, we can insist that there is no discrepancy between the esti-
207
mation of updip limit of Newman et al. and ours. Thus, we confirmed that the extent of seismogenic zone estimated by geodesy shows good agreement with that from seismology.
4. Discussion The velocities estimated from GPS observation and Cocos plate motion relative to Caribbean plate (Co– Ca motion) shown in Fig. 2 do not correspond
Fig. 3. (A) North – south, (B) east – west, and (C) vertical components of GPS time series for sites. Each point represents daily average position with error bars of 1 r. Solutions for different sites have been offset, tick marks are at 1-cm intervals. Station IDs correspond to sites in Table 2.
208
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
Fig. 3 (continued ).
to the direction normal to the Middle American Trench, and there are also differences between the azimuth of the Co – Ca motion and GPS velocities. DeMets [1] suggested that the direction of the Co – Ca motion is rotated about 10j counterclockwise from the direction of subduction that is estimated from slip vectors of earthquakes at this plate boundary, and this fact indicates the existence of trench parallel northwestward forearc sliver motion. Thus, surface velocity field reflects not only elastic deformation due to the back slip but also rigid motion as a part of the forearc sliver. Lundgren et al. [3] estimated displace-
ment rate as a result of this forearc sliver in the Nicoya peninsula as about 7 mm/year from GPS observation. Here, we assumed that the average azimuth of the surface velocities corresponds to the subduction direction, N33jE [1], if there is no forearc sliver motion. Then, subtracting trench parallel motion of the sliver from our estimated velocities should provide an average azimuth parallel to the average earthquake slip direction (Fig. 2). The northwestward motion of this sliver can be estimated so that it satisfies the above condition of azimuth, and estimated value is 8.5 mm/
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
209
Fig. 3 (continued ).
year. The difference between this value (8.5 mm/year) and the value estimated Lundgren et al. [3] is less significant because both of them are not estimated from abundant data. Lundgren et al. [3] obtained the value from the GPS observation at only one site and we obtained the value from only 1.5 years of GPS observation. We will be able to estimate this trench parallel rate more precisely in several years. Results in the previous section suggest that the Nicoya seismic gap is preparing for a future earthquake. Here, we estimated the size of this potential
earthquake using the obtained back slip distribution. The increment of the earthquake moment (Mo [Nm/year]) is calculated as Mo ¼ lmA AuðxÞAdx2 , where l, A, and u represent rigidity, fault surface, and slip rate vector, respectively. With respect to slip rate u, we can assume that the slip direction corresponds to subduction direction estimated from earthquake slip directions [1]. Since the slip direction should be N33jE and the trench normal direction is N47jE, the slip vector u is estimated from trench normal component un as juj = un/cos
210
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
Fig. 4. Horizontal (solid circle) and vertical (open circle) components of the velocities at GPS sites versus the distance from the Middle American Trench. Error bars represent F 1r.
14j. The trench normal component of the slip vector is shown in Fig. 5b. Now, assuming that the width of the inter-plate coupling zone is 100 km, referring to the size of the Nicoya seismic gap, and l = 30 GPa, same as used in our inversion analysis, we estimated Mo to be 8.68 1018 Nm/year. If earthquake moment has been accumulating at this rate since the last earthquake in 1950 (MW = 7.7 [4]), the accumulated earthquake moment will be 4.60 1020 Nm. Using the relation between moment magnitude and earthquake moment given by M W=(log10Mo 9.1)/1.5 [13], moment magnitude of the potential earthquake is estimated as MW = 7.7. This value is equal to the moment magnitude of the last earthquake in 1950, which suggests that the next earthquake is impending. Recent studies have shown that a large portion of accumulated strain energy is released by slow events and/or as post-seismic slip (e.g., [14]). Heki et al. [14] suggested that post-seismic slip due to the 1994 Sanriku – Haruka – Oki earthquake released at least 57% of its total moment. If we subtract the same rate of moment release in the present case the resulted seismic moment corresponds to a MW = 7.5 earthquake. Note that this value gives the lower estimate of the potential earthquake. Finally, we would like to note that we have replaced three campaign sites (INDI, HUAC, and PUJE) with continuous site using Trimble 5700 receiver in 2002. These sites can monitor aseismic slip
at inter-plate coupling zone, and will contribute to better estimation of aseismic moment release rate.
5. Summary We constructed a one-dimensional GPS array in the Nicoya peninsula, Costa Rica, to monitor the strain accumulation at the Nicoya seismic gap. Since the peninsula lies above the seismogenic zone, GPS observations at this array provide superior geodetic constraint on the seismogenic zone. We have obtained 1.5 years of data and estimated the surface velocity field from this data. These velocities include the effect of estimated forearc sliver motion at a rate of 8.5mm/ year to the northwest. The results of the back slip inversion using this velocity field indicate that the inter-plate coupling zone corresponds to the seismogenic zone and strain is being accumulated. The size of the potential future earthquake is estimated as large as MW = 7.5.
Acknowledgements We would like to thank Dr. Roland Bu¨rgmann and Dr. Paul Lundgren for their critical reviews of the manuscript. This research was funded by the Japan International Cooperation Agency (JICA) and by the Nicoya Seismic Gap Monitoring Project at OVSI-
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
211
Fig. 5. (a) Geometry of the supposed plate boundary fault referred to Protti et al. [8] and Protti et al. [4]. (b) Estimated back slip rate distribution. Lines of + 1r and 1r indicate the back slip distributions when F 1 standard deviation is added to the estimated values. (c, d) Horizontal and vertical components of predicted surface velocities from estimated back slip on the plate boundary and observed values. Lines of + 1r and 1r indicate the velocity distributions when F 1 standard deviation is added to the estimated values, respectively.
212
T. Iinuma et al. / Earth and Planetary Science Letters 223 (2004) 203–212
Table 3 Dip angle of assumed plate boundary versus the distance from Middle American Trench L f 40 h 8 D
60
80
100
120
140
f
j 10 j 15 j 25 j 35 j 40 j 44 8.8 12.3 17.4 25.9 37.4 50.2
L, h, and D represent distance from Middle American Trench [km], dip angle [degree], and depth [km], respectively [4,8].
CORI. Figures were made using GMT [15]. For this study, we have used the computer systems at the Earthquake Information Center of Earthquake Research Institute, the University of Tokyo. [RV]
References [1] C. DeMets, A new estimate for present-day Cocos – Caribbean plate motion: implications for slip along the Central American volcanic arc, Geophys. Res. Lett. 28 (21) (2001) 4043 – 4046. [2] M. Protti, K. McNally, J. Pacheco, V. Gonza´lez, C. Montero, J. Segura, J. Brenes, V. Barboza, E. Malavassi, F. Gu¨ndel, G. Simila, D. Rojas, A. Velasco, A. Mata, W. Schillinger, The March 25, 1990 (Mw = 7.0 ML = 6.8) earthquake at the entrance of the Nicoya Gulf, Costa Rica: its prior activity, foreshock, afterchocks and triggered seismicity, J. Geophys. Res. 100 (B10) (1995) 20345 – 20358. [3] P. Lundgren, M. Protti, A. Donnellan, M. Heflin, E. Hernandez, D. Jefferson, Seismic cycle and plate margin deformation in Costa Rica: GPS observations from 1994 to 1997, J. Geophys. Res. 104 (B12) (1999) 28915 – 28926. [4] M. Protti, F. Guendel, E. Malavassi, Evaluacion Del Potencial Sismico de la Peninsula de Nicoya (in Spanish), Editorial Fundacion UNA, Heredia, Costa Rica, 2001.
[5] K. Obana, Y. Kaneda, M. Protti, R. Van der Laat, E. Hernandez, GPS observation in Nicoya seismogenic zone, Costa Rica, The Joint Meeting of Earth and Planetary Science, May 27 – 31, 2002 Tokyo, Japan. [6] J.F. Zumberge, M.B. Heflin, D.C. Jefferson, M.M. Watkins, F.H. Webb, Precise point positioning for the efficient and robust analysis of GPS data from large networks, J. Geophys. Res. 102 (B3) (1997) 5005 – 5017. [7] G.F. Sella, T.H. Dixon, A. Mao, REVEL: a model for recent plate velocities from space geodesy, J. Geophys. Res. 107 (B4) (2002) ETG 11-1 – ETG 11-32. [8] M. Protti, F. Gu¨ndel, K. McNally, The geometry of the Wadati – Benioff zone under southern Central America and its tectonic significance: results from a high-resolution local seismographic network, Phys. Earth Planet. Inter. 84 (1994) 271 – 287. [9] J.C. Savage, A dislocation model of strain accumulation and release at a subduction zone, J. Geophys. Res. 88 (B6) (1983) 4984 – 4996. [10] T. Sato, M. Matsu’ura, A kinematic model for deformation of the lithosphere at subduction zones, Geophys. J. Int. 93 (B6) (1988) 6410 – 6418. [11] T. Yabuki, M. Matsu’ura, Geodetic data inversion using a Bayesian information criterion for spatial distribution of fault slip, Geophys. J. Int. 109 (1992) 363 – 375. [12] A.V. Newman, S.Y. Schwartz, V. Gonzalez, H.R. DeShon, J.M. Protti, L.M. Dorman, Along-strike variability in the seismogenic zone below Nicoya Peninsula, Costa Rica, Geophys. Res. Lett. 29 (20) (2002) 38-1 – 38-4. [13] H. Kanamori, The energy release in great earthquakes, J. Geophys. Res. 82 (20) (1977) 2981 – 2987. [14] K. Heki, S. Miyazaki, H. Tsuji, Silent fault slip following an interplate thrust earthquake at the Japan trench, Nature 386 (1997) 595 – 598. [15] P. Wessel, W.H.F. Smith, New, improved version of generic mapping tools released, EOS Trans.-Am. Geophys. Union 79 (1998) 579.
