Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Earthquake focal mechanism solutions for 135 small-magnitude events are inverted for best fitting partial strain rate tensors that characterize contemporary strain in five areas that span the western margin of the Panama microplate in central Costa Rica. The results indicate the predominance of subhorizontal maximum stretching subparallel to the Middle America Trench (MAT) and provide constraints on the role of Cocos ridge collision at the MAT. The trajectory of maximum stretching changes ∼25°–45° over several tens of kilometers from the Central Costa Rica Deformed Belt (CCRDB) where it is nearly E–W to the area inboard of the Cocos ridge where it is NW–SE. This change suggests that background seismogenic deformation reflects the transition from the trailing edge of a fore-arc sliver to an area of the upper plate affected by ridge collision. This diffuse deformation may be localized, in part, on conjugate strike-slip faults of the CCRDB.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
The primary aim of this paper is to use background seismicity to characterize the geometry of three-dimensional (3-D) seismogenic strain across the western margin of the Panama microplate. This is done by inverting focal mechanism solutions for partial strain rate tensors using a micropolar continuum model for crustal deformation [Twiss et al., 1993]. We examine five spatial clusters of small magnitude (2 ≤ M ≤ 4), upper crustal earthquakes that span the western part of the CCRDB, one of which is spatially associated with recently mapped regional strike-slip faults of the Atirro – Río Sucio system [Montero, 2001, 2003]. Small magnitude earthquakes are used because they are widely distributed within the crust and as a consequence provide constraints on the 3-D geometry of non-recoverable strain. Such events have been shown to faithfully reflect detailed kinematics of contemporary deformation at the Cascadia convergent margin [Lewis et al., 2003], the San Andreas fault [Twiss and Unruh, 2007], and in the eastern California shear zone/Walker lane transform margin [Lewis, 2007; Unruh et al., 1996, 2002]. The results presented here indicate that maximum principal stretching, d1, is dominantly subhorizontal with a curving trajectory that is subparallel to the arc, and concave-toward-the-trench. Minimum principal stretching, d3, is also subhorizontal except in one area where background seismicity appears to be related to normal faulting. The overall strain pattern appears to reflect deformation at the trailing edge of a fore-arc sliver [e.g., Corti et al., 2005; Norabuena et al., 2004] that is locally affected by Cocos ridge collision. If this is correct it suggests that the western margin of the Panama microplate is being crosscut by structures accommodating a combination of sliver motion and collision. The findings presented here indicate that (1) NW- and SE-striking faults and nodal planes of the CCRDB likely accommodate this distributed deformation [Marshall et al., 2000] and (2) background seismogenic strain elucidates the geodynamic boundary between sliver motion and collision.
Studies of the Pacific fore arc of Costa Rica reveal trench-parallel gradients in hanging wall kinematics that have been attributed to variations in the roughness and thickness of the downgoing Cocos plate [e.g., Fisher et al., 1998, 2004; Marshall et al., 2000; Sak et al., 2004]. Late Quaternary inner fore-arc uplift is accommodated by (1) steep, margin-normal faults in the area between Puntarenas and Quepos [Fisher et al., 1998] and (2) trench-vergent thrusting on deeply rooted faults SE of Quepos [Fisher et al., 2004; Morell et al., 2008] (Figure 1). This trench-parallel change in kinematics is correlated spatially with the transition in the footwall from rough seafloor NW of Quepos Plateau (QP in Figure 1), to the highstanding hot spot-thickened crust of the Cocos ridge. Seismological studies also indicate trench-parallel gradients in contemporary interplate coupling associated with this southeastward increase in roughness of the downgoing plate. Protti et al.  show that the 1990 Mw = 7.0 earthquake at the entrance to the Gulf of Nicoya reflected rupturing of a subducted seamount and they argue that seamounts locally reduce interplate coupling. In total, geological and seismological observations indicate that mechanical coupling across the plate boundary is sufficient to control the pattern of late Quaternary strain in the Pacific fore arc.
3. Seismogenic Strain in the CCRDB
The CCRDB as defined by Marshall et al.  intersects the fore arc in the area between Puntarenas and Quepos and traverses northeastward along the southern margin of the active Cordillera Volcánica Central, across the inactive Cordillera de Aguacate volcanic arc, into the back arc of the inactive Cordillera de Talamanca arc, ultimately linking with the NPDB (Figure 1). In order to investigate contemporary volumetric strain across the western CCRDB, focal mechanisms for small-magnitude earthquakes (2 ≤ M ≤ 4, color-coded in Figure 2) are inverted for best fitting partial strain rate tensors. A total of 135 events are examined (see auxiliary material1) and all but one of the events occurred from 1990 through 1996. During this time several subduction zone earthquakes triggered widespread seismic swarms in central Costa Rica, mostly during 1990–1993 [e.g., Fan et al., 1993; Goes et al., 1993; Protti et al., 1995]. The modeled seismic clusters are color-coded by area as follows (Figure 2): (1) inactive Cordillera de Aguacate arc, red; (2) NW Cordillera de Talamanca, purple; (3) Cordillera de Talamanca highlands, blue; (4) southern margin of the active Cordillera Volcánica Central, yellow; and (5) northern Cordillera de Talamanca, green.
A micropolar continuum model of crustal strain is adopted [Twiss et al., 1993; Twiss and Unruh, 1998]. In this approach the orientations of seismic P and T axes provide constraints on the orientations of the minimum and maximum principal stretching-rate axes, respectively (Figure 2). Background earthquakes reflect seismic flow [Kostrov, 1974], and therefore provide the opportunity to examine in 3-D contemporary non-recoverable strain in this relatively diffuse plate boundary setting [e.g., Lewis, 2007; Lewis et al., 2003]. The inversion involves solving numerically for four parameters as follows:
d1 Euler angle for the maximum principal strain rate axis; lengthening positive
d2 Euler angle for the intermediate principal strain rate axis
d3 Euler angle for the minimum principal strain rate axis; d1 ≥ d2 ≥ d3
D deformation parameter; (d2 − d3)/(d1 − d3)
The deformation parameter, D, describes the relative magnitudes of the principal strain rates. Because independent constraints on the detailed geometry and kinematics of active faults are not available for central Costa Rica it is not possible to fully exploit a fifth parameter, W [Twiss et al., 1993], which describes relative vorticity [cf. Lewis et al., 2007; Twiss and Unruh, 2007]. As such, the approach is analogous to stress inversion methods which solve for principal stress orientations [e.g., Angelier, 1990; Lopez, 1999].
The analyses presented here provide new constraints on the 3-D pattern of non-recoverable seismogenic strain across the western CCRDB. The inverted focal mechanisms represent a relatively small time window and no attempt is made to examine possible aliasing associated with sample size. Previous seismotectonic investigations, however, indicate that background seismic events and aftershock sequences provide reliable constraints on regional deformation [Lewis et al., 2003; Unruh et al., 1996, 1997].
Maximum principal stretching, d1, is dominantly subhorizontal for all five areas (Table 1 and Figure 2) with the steepest plunge angles occurring in the westernmost two subsets (33° and 32° for red and yellow, respectively). The trend of d1 changes from dominantly E–W in these areas, and in the NW Cordillera de Talamanca (purple), to NW–SE in the Cordillera de Talamanca highlands (blue). The change in trend occurs within the Cordillera de Talamanca over a distance of ∼30 km, resulting in a concave-toward-the-trench trajectory for the trace of d1. The dominant strain geometries favor strike-slip faulting with varying components of crustal thinning or crustal thickening (see V in Table 1).
Table 1. Statistical Summary for 1000 Bootstrap Data Sets at 95% Confidencea
See auxiliary material for a description of the bootstrap method.
Number of earthquake focal mechanism solutions used in the inversion.
Mean angular deviation between slip direction on model nodal planes and data nodal planes.
Negative plunge values indicate that the axis is inclined above the horizontal; in the upper hemisphere.
Standard deviation measured in degrees from the mean bootstrap value.
Deformation parameter D = (d2 − d3/d1 − d3). A value of 0.5 indicates plane strain.
Normalized vertical fraction of the deformation [Unruh et al., 2002] such that negative values indicate net crustal thinning, zero values indicate no net change in crustal thickness and positive values indicate net crustal thickening.
Although the standard deviation suggests non-plane strain, the range of values about the mean at 95% confidence encompasses the 0.5 value (+0.197/−0.343).
The events in the northern Cordillera de Talamanca (green) are characterized by a N-trending d1 and a subvertical d3. The vertical component of deformation is negative (Table 1) indicating that collectively these events accommodate crustal thinning. On the basis of the best fitting model, faulting on north dipping or SSW dipping faults would be favored (Figure 2); however, the scatter of T axes and the relatively large standard deviation for d1 indicate that the strike of these faults is not well constrained. These events occur in the area of a recently identified pull-apart basin near the town of Pejibaye and the inverted strain geometry likely reflects this structure.
4. Relation to Mapped Fault Zones
The relation between the strain inversions and mapped fault zones is assessed on the basis of the locations of the modeled earthquakes. These events have horizontal location errors of ±3–5 km so attributing them to mapped faults is not definitive. Nonetheless, it is likely that many of the inverted events have occurred on mapped faults or splays of these faults. The diversity of orientations of the seismic P and T axes reveals substantial heterogeneity in the geometry of deformation (Figure 2), which in turn suggests that the actual fault geometries are complex. Taking all of these factors into consideration, the partial strain tensors are not interpreted to reflect deformation on individual faults, but rather on systems of structures that may be related to mapped faults.
The inverted strain geometries for the three westernmost seismicity subsets fall within the portion of the CCRDB for which the mesoscale faults indicate the predominance of dextral and sinistral strike-slip on NW and NE striking faults, respectively [Marshall et al., 2000]. The E–W trending, subhorizontal stretching that characterizes seismogenic strain here is consistent with these faults. Mapped regional-scale faults here include the Alajuela, Agua Caliente, Atirro, Azul, Belo Horizonte, Escazú, Navarro, Orosi and Puriscal (Figure 2); however, the modeled earthquakes occur in elongate patterns that are mostly oblique to these structures.
Faults that might accommodate some component of the modeled strain include, for the southern sector of the Cordillera Volcánica Central (yellow), the Agua Caliente, Escazú and Belo Horizonte faults [Montero, 2001, 2003; Montero et al., 2005]. The Alajuela fault accommodates S vergent thrusting driven by collapse of the volcanic edifice [Borgia et al., 1990], and thus the kinematics are inconsistent with the inverted strain (Figure 2b). For the Cordillera de Aguacate area (red) the main upper crustal structure is the NW-striking Puriscal fault system (Figure 2) and seismicity here has been associated with this structure [Montero, 2001; Montero and Morales, 1990]. Historical larger magnitude events on this system were characterized by strike slip and normal mechanisms with WNW trending T axes [Montero, 2001]. The modeled earthquakes occur in an elongate swarm that intersects the fault at a high angle suggesting that a subset of the events may have occurred on this structure.
The principal sinistral fault in this area is the ENE-striking Navarro fault zone [Marshall et al., 2000; Montero, 2001, 2003; W. Montero et al., Neotectonic faulting and fore arc sliver motion along the Atirro-Río Sucio fault system, Costa Rica, Central America, submitted to Geological Society of America Bulletin, 2008]. The map trace of this fault is similar in orientation to the planes of zero extension rate from our inversions (Figure 2b), which equate to planes on which shear rate would be maximized (i.e., they are akin to the nodal planes for single earthquake focal mechanisms). The events in the NW Cordillera de Talamanca (purple) define an elongate E–W pattern of epicenters (Figure 2) that is approximately parallel to, but significantly south of, the trace of the Navarro fault [Montero, 2001, 2003]. The best fitting partial strain rate tensor for these events displays a statistically significant component of crustal thinning (Table 1). This result suggests that slip on steeply ESE dipping faults or moderately SSW dipping faults would be favored. A small subset of these events may thus reflect slip on the NW striking Orosi fault (Figure 2c).
The two easternmost earthquake subsets yield results that are quite distinct from one another, and from the three western subsets. The Cordillera de Talamanca highlands events (blue) yield a subhorizontal d1 and a partial strain rate tensor that is consistent with strike-slip faulting. The trend of d1, however, is NW–SE, rotated ∼25–45° clockwise (looking down) relative the trend of d1 for the three subsets to the NW (purple, yellow and red as described above; Figures 2b (inset) and 2c). A small, but statistically significant component of crustal thinning is apparent for the Cordillera de Talamanca highlands subset (Table 1) implying that slip on faults oriented parallel to the best fitting planes of zero extension rate would accommodate strike-slip motion with a normal component. These planes are steeply W and S dipping (Figure 2b).
The final cluster of seismic events analyzed lies in close proximity to the dextral Atirro – Río Sucio fault zone in the N Cordillera de Talamanca (green) (Montero et al., submitted manuscript, 2008). These events yield a shallow north trending d1 and subvertical d3 indicating that normal faulting on shallow SW-dipping or moderate north dipping faults would be favored. The Atirro – Río Sucio fault zone (Figure 2) here displays a prominent right step to the Azul fault forming a pull-apart basin [Montero, 2003]. More recent mapping suggests that a smaller pull-apart basin near the town of Pejibaye (Figure 2d) occurs along the Tucurrique fault immediately to the SW (Montero et al., submitted manuscript, 2008), in close proximity to the modeled earthquakes.
The fact that background earthquakes accommodate arc subparallel stretching across the CCRDB is consistent with regional geodetic and geologic data. GPS velocities indicate that the Pacific fore arc of Nicaragua and Guatemala are translating toward the WNW at 14 ± 2 mm yr−1 and 14 ± 2.5 mm yr−1, respectively relative to the Caribbean plate [DeMets, 2001]. Recent mapping indicates that this sliver transport is accommodated in part by the El Salvador fault, which records a late Pleistocene-Holocene slip rate of ∼11 mm yr−1 in central El Salvador [Corti et al., 2005]. Fore-arc sliver motion appears to continue into Costa Rica [Lundgren et al., 1999] where geodetic data indicate that the Nicoya block is moving to the NW at 8 ± 3 mm yr−1 [Norabuena et al., 2004]. The arc subparallel stretching documented across the CCRDB appears to reflect distributed strain at the trailing edge of this fore-arc sliver.
The results for the N Cordillera de Talamanca (green) are interpreted to reflect local fault geometries associated with pull-apart basin growth. East of the town of Pejibaye, the Tucurrique fault displays a releasing stepover (Figure 2d) [Montero, 2003]. This area lies SW of the larger releasing stepover in the Atirro – Río Sucio fault system [Montero, 2003; Montero et al., submitted manuscript, 2008]. Collectively these structures are interpreted to be part of a regional-scale releasing stepover that played a role in the localization of the Quaternary Irazú and Turrialba volcanic centers [Montero, 2003; Montero et al., submitted manuscript, 2008]. The pull-apart basin at Pejibaye is floored by well-developed fluvial terraces, upstream of a narrow bedrock canyon of the Rio Pejibaye (Figure 2d). In addition, tributaries to the Rio Pejibaye appear to be deflected, and one displays a basin-margin course that appears to be normal-fault controlled. This basin lies NE of an area of inferred uplift associated with the contractional stepover created where the Tucurrique fault diverges from the Atirro fault [Montero, 2003]. Within the margin of location errors, the inverted earthquakes occur along the Tucurrique fault at Pejebaye, suggesting that they are related to the growth of this structure. At one standard deviation the trend of d1 is poorly constrained (±30.5°, Table 1) so the orientations of the structures that might accommodate seismogenic strain here (Figure 2b) are likewise not well constrained by the inversions.
The ∼25–45° change in the trajectory of maximum stretching across the Cordillera de Talamanca suggests that factors other than fore-arc sliver motion control regional strain patterns here. In particular, this change may be a response to collision of the Cocos Ridge [e.g., Kolarsky et al., 1995; Tapponnier et al., 1982]. The change occurs over a map distance of ∼30 km and likely reflects deformation on a combination of poorly developed faults with diverse orientations and more mature faults akin to those described above [Montero, 2001, 2003].
If the curving trajectory of d1 indeed reflects Cocos ridge collision, the Tucurrique – Atirro fault system and/or parallel counterparts might be expected to continue southeastward, curving across the Cordillera de Talamanca to the area inboard of the Cocos Ridge. Such structures would be analogous to the Altyn Tagh fault which accommodates eastward lateral escape inboard of the India-Asia collision [Tapponnier et al., 1982]. Lineations interpreted as inactive faults have been noted in Cordillera de Talamanca [e.g., Denyer et al., 2003]; however, no active faults have been mapped.
The lack of mapped Quaternary faults crossing the Cordillera de Talamanca as suggested above may reflect the migration of the Cocos ridge indenter. Plate reconstructions indicate that the Panama fracture zone, which marks the eastern margin of the Cocos ridge, has migrated progressively toward the SE during the past ∼1 Ma [MacMillan et al., 2004; Morell et al., 2008]. Morell et al.  show that the Fila Costeña fold and thrust belt has migrated SE during the last ∼3 Ma in response to this process. In addition, this area lies immediately inboard of the indenter and is characterized by relatively high uplift rates [Driese et al., 2007]. Indenter migration and rapid uplift would act in concert with high tropical weathering rates to mute the surface expression of faults that might be kinematically linked to the collision zone. If any such connections exist they would likely occur between the Atirro fault and the N–S striking strike-slip faults near the eastern end of the Fila Costeña (variably ruled area, Figure 3) [Morell et al., 2008]. It is noteworthy that this region experienced numerous M > 4 earthquakes in the months following the April 3, 1983 Mw = 7.4 interplate earthquake at the Gulf of Osa, including the Mw = 6.3 event of July 3, 1983 in the Cordillera de Talamanca (Figure 1). This interplate event has also been dynamically linked to the April 22, 1991 Ms = 7.5 Limon earthquake [Tajima and Kikuchi, 1995].
In contrast, the arc sub-parallel faults that likely mark the northern margin of sliver transport are well expressed, suggesting that these features are geologically relatively mature [Corti et al., 2005; Montero, 2003; Montero et al., submitted manuscript, 2008]. The Tucurrique and Atirro fault systems link with the Río Sucio fault system NW of the Cordillera Volcánica Central forming a regionally significant crustal boundary that accommodates dextral shear and crosses the CCRDB [Marshall et al., 2000; Montero, 2003; Montero et al., submitted manuscript, 2008]. This motion is consistent with GPS velocities for stations that span the CCRDB (see inset in Figure 3) [Lundgren et al., 1999; Norabuena et al., 2004].
The strain geometries documented here have implications for understanding the tectonics of the CCRDB. The overall pattern of strain for the three western subsets of events is consistent with results for mesoscale fault populations in the CCRDB as well as seismic events larger than those evaluated here [Marshall et al., 2000]. Subhorizontal stretching is likely accommodated by steeply dipping NW and NE striking fault planes, and the diffuse distribution of events suggests that this strain is generally not localized on single, large faults, in agreement with the findings of Marshall et al. . Where deformation is localized, such as the Mw = 6.4 Damas event, the strain is likewise in agreement with our results. For the Damas event Pacheco et al.  find a WSW–ENE trending shallow T axis. The change in the trajectory of d1 in the Cordillera de Talamanca suggests that ridge collision is contributing to the contemporary motion of a fore-arc sliver. The recently mapped Atirro – Río Sucio fault system, which extends westward beyond the CCRDB, appears to mark the northern margin of this fore-arc sliver [Montero, 2001; Montero et al., submitted manuscript, 2008]. This implies that the CCRDB contributes to both the motion of the Panama block with respect to the Caribbean plate, and to motion of a fore-arc sliver relative to both the Caribbean plate and the Panama block (Figure 3).
The collision-assisted fore-arc sliver motion suggested by our modeling is in accord with geodetic observations. Recent GPS campaign data indicate that the magnitude of divergent flow across all of Costa Rica is greater than previously suggested, highlighting the probable correspondence between trench subparallel stretching documented here, and the diverging GPS velocities (P. LaFemina, personal communications, 2006, 2008). Whether ridge collision or oblique convergence fundamentally drives this fore-arc sliver motion remains an important question that will require additional earthquake and geodetic modeling, as well as fault mapping. The results presented here suggest, however, that it is possible to use background seismicity to delineate boundaries between different geodynamic regimes.
This work was supported by four Indiana University of Pennsylvania Senate Faculty Research grants to J. C. L. The figures were created with the help of GMT software from W. H. F. Smith and D. T. Sandwell and Stereonet from R. Allmendinger. We are indebted to R. Twiss for his inverse modeling program FLTSLP and plotting script MPSOL and to L. Guenther for automating the statistical procedures for bootstrapping in FLTSLP. The help of I. MacMillan in creating GMT scripts for the figures is much appreciated. We also thank K. McDannell for help with field mapping. Last, we appreciate the thorough reviews by Editor P. van Keken and Associate Editor E. Silver and critical assessments by H. Tobin, J. Marshall, and two anonymous readers, all of which substantially improved the paper.