J. F. Pacheco

The March 25, 1990 (Mw = 7.0, ML = 6.8), earthquake at the entrance of the Nicoya Gulf, Costa Rica: Its prior activity, foreshocks, aftershocks, and triggered seismicity

On March 25, 1990 a large earthquake (Mw = 7.0, ML = 6.8) occurred at the entrance of the Nicoya Gulf, Costa Rica, at 1322:55.6 UTC, producing considerable damage in central Costa Rica and generating much interest about whether or not the Nicoya seismic gap (Nishenko, 1989) had broken. The local country-wide seismographic network recorded 6 years of activity prior to this large earthquake, 16 hours of foreshocks, the mainshock, and its aftershocks. This network is operated jointly by the Costa Rica Volcanological and Seismological Observatory at the National University (OVSICORI-UNA), and the Charles F. Richter Seismological Laboratory at the University of California, Santa Cruz (CFRSL-UCSC). We obtained high resolution locations from this network and located the mainshock at 9°38.5′N, 84°55.6′W (depth is 20.0 km) and the largest foreshock (Mw = 6.0, March 25, 1990, at 1316:05.8 UTC) at 9°36.4′N, 84°57.1′W (depth is 22.4 km). We find that the aftershock zone abuts the southeast boundary of the Nicoya seismic gap, suggesting that the seismic gap did not rupture. Since the installation of the local network in April 1984 to March 24, 1990, nearly 1900 earthquakes with magnitudes from 1.7 to 4.8 (318 with magnitude 3.0 or larger) have been located at the entrance of the Nicoya Gulf, one of the most active regions in Costa Rica. The March 25 earthquake occurred at the northwest edge of this region, where a sequence of foreshocks began 16 hours prior to the mainshock. The spatial-temporal distribution of aftershocks and directivity analysis of the mainshock rupture process using teleseismic records both indicate a southeast propagating rupture. The mainshock ruptured an asperity of approximately 600 km2 of area, with this area expanding to 4000 km2 after 7 days. We present evidence that suggests that the ruptured asperity is produced by the subduction of a seamount. Inversion of teleseismic broadband and long-period P and SH waves yields a thrust faulting mechanism with the shallow plane striking 292°, dipping 26°, and with a rake of 88°, in agreement with the subduction of the Cocos plate under the Caribbean plate. Local first motions for the largest foreshock and the mainshock agree with this solution. We also present evidence suggesting that the March 25, 1990, earthquake triggered and reactivated several seismic swarms in central Costa Rica and temporally decreased the activity in the epicentral area of the July 3, 1983 (Ms = 6.2), Perez Zeledon earthquake.

Inslab Earthquakes of Central Mexico: Peak Ground-Motion Parameters and Response Spectra

We developed equations to predict pseudoacceleration response spectra (5% damping), peak ground acceleration, and peak ground velocity at free-field rock sites for intermediate-depth, normal-faulting inslab earthquakes of Central Mexico. The data set comprises 16 earthquakes (5.2 ≤ M w ≤ 7.4; 35 ≤ H ≤ 138 km) recorded at local and regional distances ( R ≤ 400 km). It represents a homogeneous catalog with respect to tectonic regime, fault mechanism, and soil class. Our results show larger amplitudes in the epicentral area from inslab events than from interplate events, a consequence of higher stress drops during the former type of earthquakes. Peak ground accelerations from moderate to large ( M w >6.0) inslab events significantly exceed those from interplate events with similar magnitude, reaching almost three times for the largest events. The ground motion due to inslab events, however, decays faster than for the thrust events. Our results are in reasonable agreement with other studies based on Japanese and the worldwide data.

The October 9, 1995 Colima‐Jalisco, Mexico Earthquake (Mw 8): An aftershock study and a comparison of this earthquake with those of 1932

Data from portable seismographs and a permanent local network (called RESCO) are used to locate the aftershocks of the October 9, 1995 Colima-Jalisco earthquake (Mw 8.0). The maximum dimension of the aftershock area, which is rectangular in shape, is 170 km × 70 km. Our study shows that the mainshock nucleated ∼24 km south of Manzanillo, near the foreshock of October 6, 1995 (Mw 5.8), and propagated ∼130 km to the NW and ∼40 km to SE. The aftershock area lies offshore and is oriented parallel to the coast. The observed subsidence of the coast is a consequence of this offshore rupture area. The aftershocks reach unusually close to the trench (within 20 km). This may be due to lack of sediments with high pore pressure at shallow depth. There are some similarities between this earthquake and the two great earthquakes of 1932 (3 June, Ms 8.1; 18 June, Ms 7.8) which occurred in this region. In both cases the aftershocks were located offshore and the coastline subsided. The sum of seismic moments and the rupture lengths of the 1932 events (1.8×1021 N-m and 280 km, respectively), however, were greater than the 1995 earthquake. Also a comparison of seismograms of 1932 and 1995 earthquakes show great differences. It seems that the 1995 event is not a repeat of either June 3 or June 18, 1932 earthquakes.

Seismicity and state of stress in Guerrero segment of the Mexican subduction zone

[1]xa0We take advantage of a relatively dense network of seismic stations in the Guerrero segment of the Mexican subduction zone to study seismicity and state of stress in the region. We combine our results with recent observations on the geometry of the subducted Cocos plate imaged from receiver function (RF) analysis, an ultraslow velocity layer mapped in the upper crust of the subducted slab, and episodic slow slip events (SSEs) and nonvolcanic tremors (NVTs) reported in the region to obtain a comprehensive view of the subduction process. Seismicity and focal mechanisms confirm subduction of the Cocos plate below Mexico at a shallow angle, reaching a depth of 25 km at a distance of 65 km from the trench. The plate begins to unbend at this distance and becomes horizontal at a distance of ∼120 km at a depth of 40 km. Some of the highlights of the inslab seismicity are as follows: (1) A cluster of earthquakes in the depth range of 25–45 km, immediately downdip from the strongly coupled part of the plate interface, revealing both downdip compressional and extensional events. This seismicity extends from ∼80 to 105 km from the trench and may be attributed to the unbending of the slab. (2) The slab devoid of seismicity in the distance range of ∼105–160 km. (3) Sparse inslab seismicity revealing downdip extension in the distance range of 160–240 km. The NVTs are also confined to this distance range. The episodic SSEs occur on the horizontal segment, in the distance range of ∼105–240 km, where an ultraslow velocity layer in the upper crust of the slab has been mapped from waveform modeling of converted SP phases. Thus, in the distance range of ∼105–160 km, SSEs occur but NVTs and inslab earthquakes are absent. This suggests that metamorphic dehydration reactions in the subducting oceanic crust and upper mantle begin at a distance of ∼160 km, giving rise to both the inslab earthquakes and NVTs. No inslab earthquake occurs beyond 240 km. The receiver function images and P wave tomography suggest that the slab begins a steep plunge at a distance of ∼310 km, reaching a depth of 500 km around 340 km from the trench. The negative buoyancy of such a slab should give rise to large extensional stress in the slab. Yet inslab seismicity is remarkably low, which may be explained by a slab that is not continuous up to a depth of 500 km, but is broken at a shallower depth. The resulting slab window may permit subslab material to flow through the gap. This may provide an explanation for the recent rift-related basalts found near Mexico City. The fore-arc, upper plate seismicity, which during the period of study (1995–2007) consisted of a moderate earthquake (Mw5.8) near the coast (H = 12 km), its numerous aftershocks, and two shallow events farther inland, demonstrates a trench-normal extension in the upper plate near this convergent margin, a state of stress that may be explained by tectonic erosion and/or seaward retreat of the trench. Seismicity, location of the mantle wedge, and rupture areas of Mexican earthquakes suggest that the downdip limit of rupture during large/great earthquakes in Guerrero may be 105 ± 15 km. Shallow-dipping, interplate thrust earthquakes are not the only type of events that affect the seismic hazard in the region. The magnitudes of inslab downdip compressional and extensional earthquakes that occur within ∼20 km inland from the coast, in the depth range of 25–45 km, may reach 6.5 and 7.5, respectively. In addition, we now identify normal-faulting earthquakes in the upper plate. These sources need to be taken into account in the hazard estimation.

Inslab Earthquakes of Central Mexico: Q, Source Spectra, and Stress Drop

We analyze 17 intermediate-depth, normal-faulting, inslab earthquakes of Mexico (4.1 M w 7.4; 35 km H 118 km), recorded on hard sites at local and regional distances (R 600 km), to study spectral attenuation of seismic waves, quality factor Q, source spectra, and Brune stress drop. Assuming 1/R geometrical spreading, the quality factor is given by Q(f ) 251f 0.58 . Although there is consid- erable uncertainty in Q due to the trade-off between geometrical spreading and Q, this uncertainty does not influence strongly the estimation of source spectra and stress drops. We find that source spectra of nine events (4.1 Mw 6.4) follow the x 2 model, while those of the other eight (5.8 Mw 7.4) significantly deviate from it. Interpreting the high-frequency level of the source spectra with the x 2 model yields a nearly constant stress drop, Dr, with a median value of 304 bars. This is more than 4 times greater than the corresponding value for interplate earthquakes in central Mexico. The observed source acceleration spectra, S(f ), is, however, better fit by an empirical source spectrum characterized by two corner frequencies, f a and f b, such that S(f ) f 2 M0/({1 (f /f a) 2 } • {1 (f /f b) 2 }) 1/2 , where f a 4.962 10 10 /M0 0.454 , f b 4.804 10 5 /M0 0.213 , and M0 is in dyne centimeters. This empirical source spectrum may be useful in predicting ground-motion parameters using sto- chastic methods.

Q of the Indian Shield

We analyze spectral attenuation of Lg waves to determine Q of the Indian shield. The dataset consists of four earthquakes recorded in the distance range of 240-2400 km. The new estimate, Q(f ) 800f 0.42 (0.1 Hz f 20 Hz), is based on a larger dataset than a previous estimate, Q(f ) 508f 0.48 (1 Hz f 20 Hz). The new Q(f ) of the Indian shield is comparable to the Q in regions of eastern North America where its value is relatively high, for example, the Adirondack Mountains. It is also in agreement with previously reported Lg coda Q at f 1 Hz for the Indian subcontinent. The revised Q(f ) has important implications for the estimation of ground motions in the Indian shield region using the stochastic method since the recorded data, presently, come mostly from large distances.

The Energy Partitioning and the Diffusive Character of the Seismic Coda

Following recent theoretical developments of the radiative transfer the- ory of elastic waves, we propose to use the ratio R of energies of curl and divergence part of the ground displacement to distinguish between the different possible domi- nant scattering regimes in the lithosphere. A consequence of the diffusion regime is the partitioning of the energy between different vibrational modes, which is inde- pendent of time. It results in the stabilization of R. This behavior is not expected if low-order diffraction (such as single scattering) is dominant. We apply our technique to seismograms recorded by a small-aperture seismic array operated during June- August 1997 in Guerrero (Mexico). We estimate the energy ratio R in the coda of 13 earthquakes recorded by the array. We find it to be very stable in the coda window, while the energy level itself changes by several orders of magnitude. The value of R is 7 1, independent of the earthquake location and the magnitude. The observed stabilization of R is a strong indication of the diffusive regime of the seismic coda.

The 1995 Colima-Jalisco, Mexico, Earthquake (Mw 8): A study of the rupture process

In this study we map rupture characteristics of the great, shallow, thrust earthquake of October 9, 1995 which caused extensive damage to the coastal towns of Colima and Jalisco. To isolate the earthquake rupture details, we deconvolve surface waves with two empirical Greens functions, the aftershock of October 12, 1995 (Mw 5.9) and the foreshock of October 6, 1995 (Mw 5.8), from the corresponding mainshock records. Specifically, we use a spectral water-level deconvolution to obtain 80 Apparent Source Time Functions (ASTF) at 62 stations (Rayleigh and Love waves). Durations of the ASTF, as a function of azimuth indicate that the rupture propagated toward N70°W. The duration of the Source Time Function (STF) is around 62 s with a large pulse at 45 s. To map the main characteristics of the rupture, we use an inverse Radon transform of the ASTFs, assuming a ribbon fault-model aligned in the direction of the rupture propagation. Our analysis indicates that the rupture initiated about 20 km offshore of Manzanillo and propagated almost unilaterally for 150 km towards N70°W, with an average rupture velocity of approximately 2.8 km/s. The earthquake was a composite of three significant subevents, the largest occurred 45 s after the initiation of the rupture and was located about 100 km away. This result is in good agreement with the inversion of deformation data, measured with GPS [Melbourne et al., 1997].

Estimation of Ground Motion for Bhuj (26 January 2001; Mw 7.6) and for Future Earthquakes in India

Only five moderate and large earthquakes (Mw 5.7) in India—three in the Indian shield region and two in the Himalayan arc region—have given rise to multiple strong ground-motion recordings. Near-source data are available for only two of these events. The Bhuj earthquake (Mw 7.6), which occurred in the shield region, gave rise to useful recordings at distances exceeding 550 km. Because of the scarcity of the data, we use the stochastic method to estimate ground motions. We assume that (1) S waves dominate at R 100 km and Lg waves at R 100 km, (2) Q 508f 0.48 is valid for the Indian shield as well as the Himalayan arc region, (3) the effective duration is given by fc 1 0.05R, where fc is the corner frequency, and R is the hypocentral distance in kilometer, and (4) the acceleration spectra are sharply cut off beyond 35 Hz. We use two finite-source stochastic models. One is an approximate model that reduces to the x 2 -source model at distances greater that about twice the source dimension. This model has the advantage that the ground motion is controlled by the familiar stress parameter, Dr. In the other finite-source model, which is more reliable for near-source ground-motion estimation, the high- frequency radiation is controlled by the strength factor, sfact, a quantity that is phys- ically related to the maximum slip rate on the fault. We estimate Dr needed to fit the observed Amax and Vmax data of each earthquake (which are mostly in the far field). The corresponding sfact is obtained by requiring that the predicted curves from the two models match each other in the far field up to a distance of about 500 km. The results show: (1) The Dr that explains Amax data for shield events may be a function of depth, increasing from 50 bars at 10 km to 400 bars at 36 km. The corresponding sfact values range from 1.0-2.0. The Dr values for the two Himalayan arc events are 75 and 150 bars (sfact 1.0 and 1.4). (2) The Dr required to explain Vmax data is, roughly, half the corresponding value for Amax, while the same sfact explains both sets of data. (3) The available far-field Amax and Vmax data for the Bhuj mainshock are well explained by Dr 200 and 100 bars, respec- tively, or, equivalently, by sfact 1.4. The predicted Amax and Vmax in the epi- central region of this earthquake are 0.80 to 0.95 g and 40 to 55 cm/sec, respectively.

S wave velocity structure below central Mexico using high‐resolution surface wave tomography

Shear wave velocity of the crust below central Mexico is estimated using surface wave dispersion measurements from regional earthquakes recorded on a dense, 500 km long linear seismic network. Vertical components of regional records from 90 well-located earthquakes were used to compute Rayleigh-wave group-velocity dispersion curves. A tomographic inversion, with high resolution in a zone close to the array, obtained for periods between 5 and 50 s reveals significant differences relative to a reference model, especially at larger periods (>30 s). A 2-D S wave velocity model is obtained from the inversion of local dispersion curves that were reconstructed from the tomographic solutions. The results show large differences, especially in the lower crust, among back-arc, volcanic arc, and fore-arc regions; they also show a well-resolved low-velocity zone just below the active part of the Trans Mexican Volcanic Belt (TMVB) suggesting the presence of a mantle wedge. Low densities in the back arc, inferred from the low shear wave velocities, can provide isostatic support for the TMVB.