caracteritzaciÓ estructural i sismotectÒnica de...

Consell Superior d’Investigacions Científiques (CSIC)

Institut de Ciències de la Terra “Jaume Almera” Departament d’Estructura i Dinàmica de la Terra

Programa de doctorat Ciències de la Terra

Bienni 2006-2007

CARACTERITZACIÓ ESTRUCTURAL I SISMOTECTÒNICA DE LA LITOSFERA EN EL

DOMINI PIRENAICO-CANTÀBRIC A PARTIR DE MÈTODES DE SÍSMICA ACTIVA I PASSIVA

Tesipresentada per

Mario Ruiz Fernández

Al Departament de Geodinàmica i Geofísica de la Universitat de Barcelona

Directors:

Dr. Josep Gallart Muset Dr. Jordi Díaz Cusí

Tutor: Dr. Francesc Sàbat Montserrat

Estudi de les rèpliques del terratrèmol del 21 de Febrer del 2002

95

Capítol 5

5.- ESTUDI DE LES RÈPLIQUES DEL TERRATRÈMOL DE MAGNITUD 4.1 Lg OCORREGUT EL 21 DE FEBRER DEL 2002 A LA SERRA D’ARALAR

5.1.- Síntesi del treball presentat en forma d’article a Geophysical Journal International

El dia 21 de febrer del 2002 va tenir lloc a les rodalies de la població d’Irurtzun, a

menys de 15 km al nord-oest de la ciutat de Pamplona, un terratrèmol de magnitud 4.1 mbLg

que va ser sentit a tota la regió. Per tal de fer un estudi detallat de l’origen de la crisi sísmica,

entre el dia 22 de Febrer i el 29 de Març del 2002, es va desplegar una xarxa temporal a la

zona epicentral, ocupant un total de 16 emplaçaments.

La geometria de la xarxa va anar variant a mesura que es disposaven de localitzacions

precises de les primeres rèpliques, a fi de millorar la cobertura del dispositiu. Es va treballar

amb 13 estacions digitals del tipus Reftek, MarsLite i Mars88 que es varen distribuir en un

àrea circular d’aproximadament 25 km de radi entorn de l’epicentre de l’event principal. Un

total de 6 instruments varen funcionar en mode registre continu, mentre els 7 restants

funcionaren en mode detecció. Totes les estacions estaven equipades amb sensors

Lennartz de 3 components d’1 s de període propi.

Es varen extreure de la nostra base de dades els terratrèmols catalogats pel serveis

sismològics a fi de relocalitzar-los amb totes les dades disponibles. L’IGN va detectar 13

rèpliques en el període comprès entre el 21 de Febrer i el 18 de Març, mentre que CEA

només en va detectar 11 i RENASS 5. L’aplicació dels algoritmes de detecció sobre els

registres continus obtinguts per les estacions temporals va permetre obtenir 319 events, 106

dels quals tenien suficient cobertura d’estacions per ser correctament localitzats. A fi de

completar la memòria de la tesis, a l’apartat 5.3 s’adjunten les determinacions hipocentrals

dels 106 events analitzats i que no havien estat incloses a l’article (veure la taula 5.1 de

l’apèndix).


96

Les 13 rèpliques catalogades pels serveis sismològics IGN, RENASS o CEA presenten

una distribució epicentral molt dispersa deguda a la baixa cobertura de les estacions

permanents en aquesta regió dels Pirineus Occidentals. Les relocalitzacions obtingudes per

aquests events, barrejant les lectures de les xarxes temporals i permanents, milloren

clarament la resolució de la zona epicentral agrupant tots els events inicialment dispersos

entorn de l’event principal (veure figura 2 de l’article).

Els epicentres obtinguts per les 93 noves repliques identificades i localitzades en

aquest estudi es concentren també en la mateixa àrea, definint una zona epicentral de 4

km2, amb la major part dels events distribuïts entre 1 i 4 km de fondària. De les localitzacions

obtingudes en aquest estudi és possible deduir una relació directa de la sismicitat

enregistrada amb l’extrem sud-est de l’encavalcament d’Aralar, relació que no s’havia pogut

deduir a partir de les determinacions obtingudes amb les xarxes permanents (veure figura 3

de l’article).

Es varen calcular els mecanismes focals de l’event principal i de les 12 rèpliques amb

major cobertura azimutal. El mecanisme focal del sisme principal mostra una solució de falla

de cisalla dextra i amb un dels plans nodals orientats en la direcció NNO-SSE. Els

mecanismes focals obtinguts per a la majoria de les rèpliques mostren una solució de falla

normal amb certa component de cisalla i un dels plans nodals orientats en el direcció ENE-

OSO (veure figura 3 de l’article). Aquest tipus de mecanismes suggereixen que després de

la ruptura principal, com a falla de cisalla dextrògira i localitzada a la falla d’Egarrieta,

d’orientació NNO-SSE i que creua de nord a sud tota l’estructura d’Aralar, hi va haver un

reajust de les capes encavalcants E-O del sud de l’estructura, amb moviments de falla

normal, produint el relaxament d’aquestes estructures desprès de la compressió produïda

pel terratrèmol principal.

Les dades enregistrades a les dues estacions en registre continu més properes a la

zona epicentral (EGA i OLL) han estat objecte d’un anàlisis acurat a fi de trobar events que

escaparen de la detecció efectuada amb la xarxa completa. Durant els 36 dies d’experiment,

l’estació d’OLL va detectar 511 rèpliques, mentre que l’algoritme aplicat a la xarxa completa

en va detectar 319. L’estació d’EGA només va estar operativa durant els 15 dies finals, però

en aquest període va detectar 279 sísmes (veure figura 6 de l’article). Aquestes dades es

van emprar per estudiar l’evolució temporal i la distribució del nombre d’events en funció de

la magnitud de la crisis sísmica.

La tassa de rèpliques produïdes pel terratrèmol principal s’ha modelitzat mitjançant la


97

llei d’Omori, la qual descriu el decaïment exponencial del nombre d’events en funció del

temps. S’ha obtingut un valor del paràmetre P igual a 0.9 ± 0.2, indicant un decaïment de

l’activitat entre moderat i lent (veure figura 6 de l’artícle).

A fi d’obtenir el valor b de la relació acumulativa de Gutenberg-Richter, es va ajustar

una fórmula de la magnitud local a partir de la duració de l’event i la distància epicentral

emprant de referència els 13 events catalogats per l’IGN. Es va obtenir un valor b=0.80 ±

0.02 que està en el rang baix de valors i és indicatiu d’una zona epicentral reduïda i de plans

de falla molt ben definits (veure figura 8 de l’article).

Aplicant les tècniques de correlació creuada, es varen trobar 10 families amb un mínim

de 3 events cada una (veure figura 9 de l’article). Aquestes famílies agrupen la meitat dels

sismes analitzats, però no s’aprecia cap patró en l’evolució temporal dels grups de

terratrèmols.

La relocalització dels events associats en famílies es va realitzar seguint dos

esquemes diferents. Per una banda es va emprar el mètode clàssic, basat en la correcció de

les lectures dels events secundaris mitjançant els temps relatius, que s’obtenen a partir de

les correlacions creuades entre l’event de referència de cada família i els restants. Els

residus de localització de l’event de referència s’introdueixen al procés d’inversió hipocentral

com correccions d’estació (Deichmann i García-Fernández, 1992). Per altre banda, es van

invertir els temps de propagació diferencials, obtinguts per correlació creuada, mitjançant

l’algoritme de dobles diferències HypoDD (Waldhauser, 2001).

Les relocalitzacions obtingudes milloren clarament la resolució de la zona epicentral,

definint una franja E-O d’uns 2 km de llarg per 1 d’ample que ressegueix el límit sud de les

capes encavalcants de l’estructura d’Aralar. El rang de fondàries hipocentrals també apareix

més ben acotat, amb la major part dels events localitzats entre 1.5 i 2 km de fondària. Les

seccions en fondària en la direcció E-O mostren una distribució d’events discontínua,

marcant els diferents sectors de la falla que s’han activat durant la crisi (veure figura 11 de

l’artícle).

Finalment, s’han localitzat tots els sismes analitzats amb el programa HypoDD

barrejant els events que pertanyen a les famílies, i per als quals es disposa de temps de

propagació relatius, i els que no s’adscriuen a cap familia. Els resultats són molts similars als

obtinguts amb la metodologia anterior. Les localitzacions hipocentrals apareixen millor

acotades, tant en superfície com en fondària, que a les localitzacions inicials, i s’observa

l’existència d’alguns events més profunds que podrien estar marcant un cert aprofundiment


98

cap al sud (veure figura 12 de l’artícle).

Aquest estudi ha posat de manifest la rellevància sismotectònica de l’estructura

d’Aralar, fet que no havia estat possible identificar a partir de les dades de les xarxes

permanents presents en aquesta regió. Als catàlegs dels serveis sismològics l’estructura

més activa dels Pirineus Occidentals és el sector central de la falla de Pamplona. Aquesta

falla és una estructura a escala cortical i probablement controla l’estat d’esforços de la regió.

El sistema d’Aralar, amb una orientació principal en la direcció E-O, canvia a NO-SE en el

seu extrem oriental, al aproximar-se a la falla de Pamplona, per tant es pot interpretar com

una estructura secundaria conjugada a la falla principal. L’activitat sísmica analitzada en

aquest estudi pot ser explicada en termes de moviments relatius afectant a tot el sistema de

falles.

5.2.- Article. Seismotectonic constraints at the western edge of the Pyrenees: aftershock series monitoring of the 2002 February 21, 4.1 Lg earthquake.

Els resultats d’aquest estudi es troben recollits a l’article que s’adjunta a continuació.

La seva referència és:

Seismotectonic constraints at the western edge of the Pyrenees: aftershock series monitoring of the 2002 February 21, 4.1 Lg earthquake. M. Ruiz, J. Díaz, J. Gallart, J.A.

Pulgar, J.M. González-Cortina i C. López (2006). Geophysical Journal International. V. 166

(1), 238-252. doi: 10.1111/j.1365-246X.2006.02965.x.

Geophys. J. Int. (2006) 166, 238–252 doi: 10.1111/j.1365-246X.2006.02965.xG

JISei

smol

ogy

Seismotectonic constraints at the western edge of the Pyrenees:aftershock series monitoring of the 2002 February 21, 4.1 Lgearthquake

M. Ruiz,1 J. Dıaz,1 J. Gallart,1 J. A. Pulgar,2 J. M. Gonzalez-Cortina2 and C. Lopez2

1Dpt. Geofısica i Tectonica, Institut de Ciencies de la Terra ‘Jaume Almera’ IJA-CSIC, Lluıs Sole i Sabarıs s/n, 08028 Barcelona, Spain.E-mail: [email protected]. Geologıa, Universidad de Oviedo, Arias de Velasco s/n, 33005 Oviedo, Spain

Accepted 2006 February 10. Received 2006 February 9; in original form 2005 February 15

SUMMARY

Seismic data recorded from a temporary network deployed at the western edge of the Pyreneesis used to study the aftershocks series following a magnitude 4.1 earthquake that took placeon 2002 February 21, to the NW of Pamplona city. Aftershock determinations showed eventsdistributed between 1 and 4 km depth in a small active area of about 4 km2, E–W orienteddelineating the southern sector of the Aralar thrust unit. This seismogenic feature is supportedby focal solutions showing a consistent E–W nodal plane with normal faulting following themain strike-slip rupture. The Aralar structure with its shallow activity may be interpreted as aconjugate system of the NE–SW deep-seated Pamplona active fault nearby. Cross-correlationtechniques and relative location of event clusters further constrained the epicentral domain to2 km long and 1 km wide. Statistical relations and parameters established indicate a ratherlow b-value of 0.8 for the Gutenberg–Richter distribution, denoting a region of concentratedseismicity, and a P-parameter of 0.9 for the Omori’s law corresponding to a low decay of theaftershock activity in this area. More than 100 aftershocks were accurately located in this high-resolution experiment, whereas only 13 of them could be catalogued by the permanent agenciesin the same period, due to a much sparser distribution. The results enhance the importance ofusing dense temporary networks to infer relevant seismotectonic and hazard constraints.

Key words: aftershocks parameters, clustering analysis, Pamplona fault, seismicity,temporary array, western Pyrenees.

1 INTRODUCT ION

On 2002 February 21 a magnitude 4.1 Lg (4.8 mb) earthquake tookplace in North Spain at the western edge of the Pyrenees, in the so-called Basque Cantabrian basin that marks the transition betweenthe Pyrenean Chain and the Cantabrian Mountains. The SpanishNational Geographical Institute (IGN) network located the epicen-tre near the Irurtzun village, less than 15 km NW from Pamplonacity (Fig. 1). The event was widely felt in the region, with EMSintensity values from V at the vicinity of the epicentral area to IIIin Pamplona and Alsasua cities. This work deals with the analysisof the aftershock series, as recorded by a dense temporary networkcovering a circular zone of about 25 km of radius centred at theepicentre of the main event.

The region is characterized by a moderate and sparse seismic-ity, mainly concentrated westwards of Pamplona where events, dis-tributed in depth between 5 and 20 km, are related to the centralsegment of the Pamplona Fault (Grandjean et al. 1994; Souriau &Pauchet 1998; Ruiz et al. 2006). This accident is a NE–SW-strikingdeep structure that runs from the Palaeozoic Basque Massifs to the

Ebro Basin. It is only expressed at surface in its central segment, inthe Pamplona basin, where a line of Triassic salt diapirs marks thefault trace (Martınez-Torres 1989; Turner 1996; Faci et al. 1997).Historical catalogues report three earthquakes felt with intensity VIin the Pamplona area during the end of the 19th and the beginning ofthe 20th centuries (Olivera & Gallart 1987; Ruiz et al. 2006). Instru-mental catalogues show a present-day moderate to low-magnitudeseismicity WNW of Pamplona city, in the study area, with magni-tudes near 5 reported in past decades (see Ruiz et al. 2006). Between1982 May and June a seismic crisis produced up to 28 earthquakeswith magnitudes between 2.7 and 4.9 (Olivera & Gallart 1987).On 1996 February 25 and 1998 October 27, 4.5 and 4.8 magnitudeearthquakes were also reported in the International SeismologicalCentre (ISC) and Observatoire Midi-Pyrenees (OMP) catalogues,respectively.

The origin of the Basque Cantabrian basin is related to the exten-sional period that leads to the opening of the Gulf of Biscay and theindividualization of the Iberian Peninsula as a subplate. The exten-sionwas particularly intense duringAptian–Albian interval owing totranstensive movements related to the counter-clockwise rotation of

238 C© 2006 The Authors

Journal compilation C© 2006 RAS

Seismotectonic constraints at western edge of Pyrenees 239

Figure 1. (a) Geological sketch of the study area. (b) Blow up of the area covered by the network (Geology modified from IGME 1978, 1987). The epicentreand isoseistes of the February 21 main event are shown. EMS intensity values of V were reported in the Irurtzun village for the IGN service.

Iberiawith respect to stableEurope (Montadert et al.1979;Rat 1988;Garcıa-Modejar 1996). The eastern limit of the Basque–Cantabrianbasin is usually placed at the Pamplona Fault (Fig. 1). This fault, alsonamed Estella Fault, is a NE–SW crustal structure, which acted as atransfer zone during Cretaceous extension and played an importantrole in controlling the Mesozoic sedimentation.

During the Alpine (Pyrenean) orogeny in Tertiary times, theroughly N–S convergence of Iberia and Eurasia resulted in theclosure of the Mesozoic basins and the formation of the Pyre-nean Chain (Choukroune, 1992), leading to the inversion of theBasque–Cantabrian basin in the Palaeogene. The Pamplona fault,acted as a major traverse structure during the Alpine compression as

C© 2006 The Authors, GJI, 166, 238–252


240 M. Ruiz et al.

Table 1. Location and description of the instruments deployed during this experiment. The network geometry and the location of some continuous recordinginstruments was changed to improve the hypocentral determinations.

Station Recording Latitude Longitude Height Instrument Recording Sampling Stationname period (N) (W) (m) mode rate correction (s)

GOM 02/02/22–02/03/13 42◦ 42.395′ 1◦ 48.272′ 592 Reftek Continuous 100 sps 0.11BER 02/02/22–02/03/29 43◦ 0.667′ 1◦ 50.382′ 814 Reftek Continuous 100 sps 0.15TRI 02/02/22–02/02/24 42◦ 48.314′ 1◦ 57.981′ 1000 MarsLite Continuous 62.5 sps 0.18OLL 02/02/22–02/03/29 42◦ 52.108′ 1◦ 51.438′ 555 MarsLite Continuous 62.5 sps 0.10USI 02/02/23–02/03/29 42◦ 54.577′ 1◦ 40.632′ 632 MarsLite Continuous 62.5 sps 0.11IHA 02/02/23–02/03/12 42◦ 55.632′ 1◦ 55.142′ 500 MarsLite Continuous 62.5 sps 0.09IZC 02/02/23–02/03/29 42◦ 48.572′ 1◦ 46.843′ 445 Mars88 Trigger 62.5 sps 0.08VIL 02/02/24–02/03/12 42◦ 43.692′ 1◦ 57.388′ 622 MarsLite Continuous 62.5 sps 0.11

02/03/12–02/03/29 Mars88 Trigger 62.5 spsURD 02/02/24–02/03/29 42◦ 55.288′ 2◦ 09.115′ 593 Mars88 Trigger 62.5 sps 0.11INT 02/02/24–02/03/12 43◦ 0.828′ 2◦ 0.010′ 420 Mars88 Trigger 62.5 sps 0.08

02/03/12–02/03/28 MarsLite Continuous 62.5 spsBAQ 02/02/25–02/03/29 42◦ 46.857′ 2◦ 07.068′ 664 Mars88 Trigger 62.5 sps 0.12ETX 02/02/25–02/03/12 42◦ 58.29′ 1◦ 47.284′ 420 Mars88 Trigger 62.5 sps 0.08

02/03/12–02/03/29 MarsLite Continuous 62.5 spsEGA 02/03/13–02/03/28 42◦ 55.544′ 1◦ 51.808′ 536 Reftek Continuous 100 sps 0.10OLA 02/03/13–02/03/29 42◦ 57.445′ 1◦ 36.698′ 591 Mars88 Trigger 62.5 sps 0.11ARR 02/03/13–02/03/29 43◦ 00.417′ 1◦ 38.746′ 602 Mars88 Trigger 62.5 sps 0.11TOR 02/03/14–02/03/29 42◦ 52.415′ 1◦ 59.520′ 596 Mars88 Trigger 62.5 sps 0.11

evidenced by the strong lateral changes of the crustal structureon both sides of the fault (Pedreira et al. 2003). This fault hasbeen classically interpreted as a major strike-slip fault with ei-ther sinistral (Engeser & Schwentke 1986; Rat 1988; Faci et al.1997), dextral (Muller & Rodgers 1977; Turner 1996) or combined(Martınez-Torres 1989) sense of movement. More recently, the ab-sence of significant palaeomagnetic rotation around the fault hassupported a new interpretation of the Pamplona Fault as a hanging-wall drop fault resulting from variations of the geometry and thick-ness of Mesozoic sequences on both sides of the fault (Larrasoanaet al. 2003).

Two domains have been classically defined within the Basque–Cantabrian basin (Feuillee & Rat 1971), the Basque Arch at thenorthern part, with a northward vergence and a significant structuralcomplexity and the Cantabro-Navarrian Domain to the South, witha general South vergence of structures. The epicentral zone hereconsidered is located at the Aralar range, one of the subdomains ofthe Basque Arch, situated in the eastern prolongation of the BilbaoAnticline and characterized by several north-directed thrust sheets.

2 DATA ACQUI S I T ION

AND PROCESS ING

From 2002 February 22 to March 29, a dense temporal networkcomposed by a total of 16 stations was deployed in the epicentralzone (Fig. 1 andTable 1). Up to 13 stationswere operated at the sametime with epicentral distances lower than 25 km. Reftek, MarsLiteand Mars88 digital dataloggers were used in this experiment. A to-tal of six instruments worked in continuous recording mode, whilethe other seven operated in triggered mode. All the stations wereequipped with Lennartz three-component 1 Hz seismometers. Asfirst accurate hypocentral locations of the aftershocks series wereobtained, the network geometry was modified to improve the cov-erage of the active zone.

Events from the published catalogues of IGN, RENASS(Pyrenean French network) and CEA (French’s Atomic EnergyCommission) networks were extracted from the data set obtainedduring this temporary experiment. From February 21 to March 18

Figure 2. Epicentral determinations of the aftershocks reported by the IGNservice (triangles) and relocations using data from permanent and temporarynetworks (circles). (Geology modified from IGME 1978, 1987).

the IGN service reported 13 earthquakes following the main event,with magnitudes between 1.8 and 3.5, sparsely located in the Pam-plona Region (Fig. 2). The RENASS network reported only five ofthese earthquakes, with magnitudes between 2.6 and 3.7, and the

C© 2006 The Authors, GJI, 166, 238–252



Table 2. P-velocity model used forhypocentre inversion.

Depth Z (km) VP (km s−1)

0.0 5.53.0 6.012.5 6.222.0 6.738.0 8

CEA catalogue contains 11 events with magnitudes between 2 and4.1. All these events were relocated using the additional informationprovided by our network.

A STA/LTA detection algorithm was applied to the data of eachof the stations operated in continuous mode, and the events detectedby the triggered stations were also included in the detection lists.An event was retained only if it was present on 3 or more stations inan 8 s time interval. A total of 319 uncatalogued aftershock eventswere detected using this method, and up to 106 events from this setof quakes had enough station coverage to be correctly located.

3 HYPOCENTRAL DETERMINAT IONS

AND FOCAL MECHANI SMS

Phase picking and seismogram processing was done using SAC(Seismic Analysis Code) package (Goldstein & Minner 1996;Goldstein et al. 2003). First hypocentral determinations were ob-tained from the Hypo71 code (Lee & Lahr 1975), using a homoge-neous five-layered velocity-depth model derived from wide-angleseismic profiles across the Basque–Cantabrian basin (Pedreira et al.2003) (Table 2). A time correction is considered for each station totake into account their altitude (Table 1).

The epicentres of aftershocks reported by IGN, RENASS andCEA seismological services show a sparse distribution probablyrelated to the low station coverage of this region.Relocations of theseaftershockswere done using together the phase picking of temporaryand permanent seismic networks. The relocations clearly improvedthe resolution on the epicentral zone, grouping all the sparse eventscloser to the main event, around Egarrieta village (Fig. 2).

The epicentres of the 93 new events identified and located in thisexperiment concentrate also at the same area, defining an epicentralzone of less than 4 km2 (Fig. 3a),most of the events being distributedbetween 1 and 4 km depth (Fig. 3c). The mean qualityQ obtained forthese locations is always A or B, due to the dense network coverage.The results obtained have a mean rms value of 0.08 s and it is alwayslower than 0.2 s. The mean horizontal and vertical errors are 0.3 and0.7 km respectively. The gap has a mean value of 113◦ and is alwayslower than 200◦, showing two peaks corresponding to the two stagesof the network (Fig. 4).

The accurate locations reveal a direct structural relation of thisseismicity with the southeastern edge of the Aralar thrust sheets(Figs 3a and b), which could not have been inferred from perma-nent network determinations (Fig. 2). This unit, approximately 30km long and up to 10 km wide, is situated at the eastern prolongationof the Bilbao Anticline. It is composed of Jurassic and Early Creta-ceous carbonated terrains disposed along a general E–W directionand forming a north-verging thrust imbricate (Fig. 3b). Towards itseastern edge, the Egarrieta Fault, with a NNW–SSE direction and adextral component, crosses all the structure and the thrust directionchanges to the SE, probably affected by the NE–SW Pamplona fault(Martınez-Torres 1989; Faci et al. 1997).

Fault plane solutions of the main event and of the aftershocks withbest azimuthal coverage (12 events with gaps less than 100◦ and atleast six P-polarities were retained) have been computed using theFPFIT algorithm (Reasenberg & Oppenheimer 1985) (Fig. 5 andTable 3). The focal mechanism of the main event has been derivedusing 13 P-wave polarity readings from permanent networks, re-sulting in a dextral strike-slip solution with one of the nodal planesoriented N23W (Figs 3a and 5). The strike and rake uncertainties areequal to 5 degrees and zero for the dip error. The most common so-lution obtained for the aftershocks shows a normal fault with somestrike-slip component, having one nodal plane oriented ENE–WSW(Figs 3a, d and 5). The average errors on themaximum likelihood so-lutions remained less than 20◦ for strike, dip and rake, respectively.The accuracy of these solutions was checked by analysing differentparameters provided by the FPFIT package to assess the quality ofthe results. An estimation of the misfit of the solution is provided bythe F-value (F = 0 representing a perfect fit to the data and F = 1a complete misfit), which is in our case lesser than 0.15. Solutionsare robust for station distribution ratio (STDR) greater than 0.5,and in our case it is comprised between 0.50 and 0.71. Hence, thesolutions obtained appear to be stable and robust, even if in someparticular cases small variations in the fault plane orientations cannot be excluded.

The aftershock mechanisms suggest that after the main ruptureas a shear fault, there has been a readjustment of the E–W thrusts interms of normal fault movements. The local stress field is markedby maximum P directions oriented NW–SE (Table 3).

Possible directional dependences in the hypocentral depths dis-tribution were tested by cross-sections over the Aralar structure indifferent directions (Fig. 3c). Hypocentres are mainly distributedbetween 1 and 4 km depth, and no clear dipping structures are ob-served. A geological cross-section with a N–S orientation was per-formed. The projection of the hypocentres over this cross-sectionshows as the events are related to the southern E–W thrusts slicesof the Aralar structure (Fig. 3b).

The location of themain event over theNNW–SSEEgarrieta fault(Fig. 2) and the focal mechanism obtained, strongly suggest that adextral movement of this N–S fault, generating a compressionalstress field in the eastern sector of the Aralar edge, is the origin ofthis seismic crisis. Both the accurate location of the aftershocks,with an E–W distribution, and the focal solutions obtained for themost significant events, mainly normal fault solutions, suggest thatthe aftershocks are due to a relaxation process of the E–W thrustingslices after the main event.

4 STAT I ST ICAL PARAMETERS

OF THE AFTERSHOCKS

The data recorded by the two stations closer to the epicentral area(EGA and OLL, see Fig. 2) have been specifically analysed. Allthe detections resulting from STA/LTA algorithms were extractedfor these two stations, revealing a set of additional events that werebeyond the detection capacity of the whole network. During the36 days of the experiment, theOLL station detected 511 aftershocks,whereas the algorithm for the whole network resulted in 319 events.The EGA station was operated only for the last 15 days, but up to279 aftershocks were detected there in that period.

These additional data have been used to better constrain the evo-lution of the seismic series, including the characterization of thetime decrement of the seismic activity and its frequency–magnitudedistributions.

C© 2006 The Authors, GJI, 166, 238–252


242 M. Ruiz et al.

Figure 3. (a) Epicentral locations and fault plane solutions obtained. (Geology modified from IGME 1978, 1987) (b) Geological cross-section, approximatelyin a N–S direction. (c) Cross-sections over the Aralar structure, showing events distributed between 1 and 4 km depth. (d) Dip angle of T-axes versus dip angleof P-axes for the 13 fault plane solutions obtained.

4.1 Temporal evolution of the seismicity

Fig. 6(a) shows the temporal evolution of the seismicity as de-tected by the network. During the first operating period, from Febru-ary 22 to March 12 (Julian days 53–71), the expected aftershock

exponential decrease is not clearly imaged from the whole net-work (Fig. 6a), because most of the low-magnitude events werenot recorded in the nearby stations operated in triggered mode. Thisdecrease is better imaged at the OLL station, near the epicentralzone and equipped with a continuously recording device (Fig. 6b).

C© 2006 The Authors, GJI, 166, 238–252



Figure 4. Error analysis of the hypocentral determinations obtained using Hypo71 program. The horizontal and vertical errors, Gap and RMS are shown.

Figure 5. Fault plane solutions (lower hemisfere equal-area projection) of the 13 studied events. For each mechanism the date (year, month, day), the origintime (hour and minute), the focal depth in km (z) and the magnitude (M) are reported. P and T denote the P- and T-axes positions. Open circles and black dotsindicate dilatations and compressions, respectively.

In the second part of the experiment, from March 13 to March 29(Julian days 72–88), the detection threshold of the network was im-proved after the stations redistribution. The most significant obser-vation is an abrupt increment of the seismicity related to the March

18 magnitude 3.5 event. The aftershock sequence triggered by thissecondary event, with the characteristic exponential frequency dis-tribution, is best imaged at EGA station that recorded up to 225events (Fig. 6c), 139 of themalso recorded atOLL station.

C© 2006 The Authors, GJI, 166, 238–252


244 M. Ruiz et al.

Table 3. Focal solutions obtained. FM: Number of polarities used in the inversion. The strike (Strk) and dip (Dip) angles for each plane, as well as the azimuth(Azm) and plunge (Plng) angles of the P- and T-axes, are reported. All the angles expressed in degrees.

Date Origin Latitude Longitude Depth Mag FM Plane 1 Plane 2 P-axis T-axis

Time (◦N) (◦E) (km) Strk Dip Strk Dip Azm Plng Azm Plng

02/02/21 10:21:49 42.9300 −1.8600 1.82 4.1 13 70 70 337 81 25 7 292 2102/02/25 12:41:47 42.9243 −1.8352 2.45 2.1 8 150 75 252 52 104 38 206 1502/02/26 02:34:37 42.9253 −1.8455 1.71 1.6 6 55 70 309 53 278 42 178 1102/02/27 06:13:46 42.9258 −1.8327 2.03 3.0 15 40 75 254 18 296 59 138 2902/02/27 06:17:15 42.9270 −1.8360 1.98 2.7 10 135 60 279 36 86 68 211 1302/02/27 08:44:27 42.9253 −1.8327 1.45 2.7 13 75 65 331 63 294 38 203 102/02/27 13:38:25 42.9235 −1.8308 1.93 2.3 11 130 30 40 90 103 38 337 3802/02/27 19:13:38 42.9232 −1.8362 1.39 2.3 10 75 85 178 21 147 37 6 4602/03/03 00:43:23 42.9263 −1.8252 1.87 2.2 9 110 25 268 67 161 67 5 2102/03/15 08:20:48 42.9270 −1.8463 0.83 2.6 15 10 70 163 22 92 24 296 6402/03/16 03:44:41 42.9253 −1.8475 1.44 2.1 7 110 50 260 44 85 74 186 302/03/18 15:18:53 42.918 −1.8295 2.41 3.5 14 60 70 299 36 292 55 172 1902/03/19 09:28:41 42.925 −1.8273 2.03 2.1 9 145 70 251 53 102 42 202 12

The occurrence rate of aftershocks sequences is empirically de-scribed by the modified Omori formula:

n(t) = K (C + t)−P (Utsu 1961),

where n(t) is the frequency of aftershocks per unit time (days) afterthe main shock. P is a rate constant for aftershocks decay, indepen-dent on the magnitude of the main event but reflecting mechanicalconditions of the crust on a regional scale (Ogata 1999; Utsu 2002).Usual values for P lie between 0.7 and 1.8 (Kisslinger & Jones1991; Guo & Ogata 1997). Constant C is related to the incomplete-ness of recordings immediately following themain shock (Hauksson& Jones 1989; Utsu 2002) and often shows strong dependence onmagnitude of the main event. Constant K is also dependent on themagnitude of the main event and on the total number of quakes inthe sequence (Kisslinger & Jones 1991).

This expression was fitted by a least-square method to obtain theconstant K and the P parameter for the aftershocks of the magni-tude 3.5 earthquake of 2002 March 18, as recorded at EGA station(Fig. 6c). We assumed in this case C = 0 as the station is locatedin the vicinity of the epicentral zone and was operating prior to theevent. We obtained a constant K equal to 32 ± 1 with a P-valueof 0.9 ± 0.2 (Fig. 6d), that is in the lower range of the expectedvalues, indicating a moderate to low decay of the aftershocks in thesoutheastern edge of the Aralar region.

4.2 Local magnitudes and Gutenberg–Richter

distributions

In order to fix the magnitudes of the aftershock events, we firstestablished a local magnitude (MD) formula, defined as an empiricalrelationship involving epicentral distance and total signal duration.We have used a relation of the form (Lee et al. 1972; Ruiz et al.2006):

MD = a + b Log T + cD,

where T is the total signal duration in seconds, D is the epicentraldistance in kilometres and a, b and c are constants determined sta-tistically. The signal duration and the epicentral distances obtainedat each station for the 13 events catalogued by the IGN service, withtheir known local magnitudes, were used to adjust the coefficients ofthe formula in a least-squares analysis. This resulted in the following

equation, with a correlation coefficient of 94 per cent:

MD = −1.983 + 2.573 Log T + 0.0312 D

Fig. 7 shows the local magnitude distributions of the earthquakesdetected by the whole network, and by OLL and EGA stations re-spectively. For the events detected in those stations but not accuratelylocated, we used for parameter D an average of the distance fromthe station position to the midpoint of the epicentral area, consid-ering that it extends less than 4 km2. EGA station, nearest to theepicentral zone, detected events with magnitudes between−1.2 and3.5, while at OLL station, distant around 6.5 km, magnitude de-tection threshold was −0.4. For the whole network, the detectionthreshold was 0 for the first recording period, and improved to −0.4for the second one. After the March 18 earthquake, a great amountof low-magnitude earthquakes were recorded in all the stationsnearby.

The catalogues of the two closer stations EGA and OLL were alsoused to obtain, after a linear regression adjustment, the b-value ofthe cumulative Gutenberg–Richter distribution:

Log N = a − bM,

where N is the total number of events with magnitude greater than

M .A statistical analysis of the b slope variability was performed test-

ing different intervals ofmagnitude increments, and a stable solutionwas always obtained. A mean b-value of 0.78 ± 0.02 was inferredat EGA station using the 225 events following the March 18 event,considering magnitude increments of either 0.25, 0.2, 0.15 or 0.1.At OLL station, a mean b-value of 0.81 ± 0.02 was obtained usingthe complete catalogue, 511 events, and for the same magnitudediscreteness (Fig. 8). The error bar of both calculations includes thestatistical error from least-squares fitting, and the b-value variationsdue to magnitude increment.

The b coefficient is normally close to 1, but it may range between0.5 and 1.5 depending on the region (Guo & Ogata 1997; Nanjoet al. 1998; Utsu 1999). The fractal dimension,D, of the active faultsystem involved in the seismic process is related to the b-value bya simple relationship: D = 2b (Turcotte 1992; Guo & Ogata 1997).Therefore, low b-values are characteristic of regions with concen-trated seismicity and well-defined fault planes (Njike-Kassala et al.1992). The dimension D = 1.6 obtained in our study shows a

C© 2006 The Authors, GJI, 166, 238–252



Figure 6. (a) Frequency distribution of events obtained by the detectionalgorithms applied to the whole network. (b) Same at OLL station, 6.5 kmsouthwards from the epicentral zone, (c) Same at EGA station, only 1.9 kmwestwards from the epicentral area. (d) Adjustment of the P-parameter ofthe modified Omori relation at EGA station. The frequency n(t) is computedas the number of events every 12 hr, and t is the time after the main event.

geometry intermediate between a line and a plane, correspondingto a well-defined hypocentral zone located in a small area of about4 km2 around the southern thrusts of Aralar system, with depths notexceeding 4 km.

Figure 7. Time distribution of local magnitudes of the earthquakes detectedwith the whole network, and at OLL and EGA stations respectively.

The b value resulting from this experiment is lower than the oneobtainedbyNjike-Kassala et al. (1992) in another area of thewesternPyrenees near the Pamplona Fault: b = 0.96 ± 0.1, the activity ofwhich is sparser in comparison with the one related to the Aralarstructure.

5 CLUSTER ING ANALYS I S

Aftershocks sequences are often characterized by numerous eventswith striking waveform similarities at individual stations, usuallyreferred to as clusters or earthquake families. The source–timefunction, propagation path, station site and recording instrumentsare factors contributing to the seismogram similarity in a given

C© 2006 The Authors, GJI, 166, 238–252


246 M. Ruiz et al.

Figure 8. Cumulative Gutenberg–Richter distribution and b-value fittingusing 0.25 magnitude increments at EGA and OLL stations. EGA catalogueholds the events related to the 3.5ML secondary event while OLL stationbegan recording in February, 22 and holds part of the aftershocks of themain event.

station (Tsujiura 1983; Poupinet et al. 1984; Maurer & Deichmann1995). Therefore, clustering analysis can only be performed if theepicentral zone is narrow compared to the dimensions of near-source heterogeneities and to the observed wavelengths (Maurer &Deichmann 1995). The detection of each of these clusters can givesome clues to the understanding of the whole rupture and relaxationprocesses.

The high waveform similarity of events of a given cluster mayalso be used to highly improve the picking of the phase onsets, al-lowing an accurate relative hypocentre determination that can pro-vide enhanced resolution on the location and extent of the seis-motectonic structure (Fremont & Malone 1987; Deichmann &Garcıa-Fernandez 1992; Got et al. 1994; Augliera et al. 1995;Shearer 1997; Rubin et al. 1999; Saccorotti et al. 2002).

5.1 Cluster detection

The algorithm described by Maurer & Deichmann (1995), based oncorrelation analysis ofP- and S-wave data, was applied to detect andassociate events with similar waveforms. In this approach, the eventcluster separation is achieved by close inspection of row patterns inthe correlationmatrix, avoiding a typical problemof theEquivalenceclass method (Aster & Scott 1993), which could result in eventsbelonging to different clusters.

The waveform similarity is checked station per station for P andS waves. S-wave data provides a more reliable measure of similaritythan P-wave data, as their waveforms are generally more complexthan thefirst arrivingPwavelets (Maurer&Deichmann1995).How-ever, P-wave correlation must also be done to verify the consistencyof the onset polarities.

Events belonging to the same cluster should exhibit similarities,station per station over the whole network, but stations located closeto the epicentral area with good signal-to-noise ratio will give highercorrelation coefficients. Therefore, seismograms from OLL stationwere used in our cluster search, although consistency was alwaysguaranteed by stationsBERandUSI,which also recorded all the 106retained earthquakes. A 3–15 Hz bandpass filter was applied priorto the correlation analysis to improve the semblance of the time-series. The normalized cross-correlation among all the independentpair of signals was finally calculated for P and S waves. P-wavecross-correlation matrix was obtained correlating a time windowof 1 s, starting 0.25 s before the P phase picked, and the S-wavecross-correlation matrix resulted from the correlation of 2 s, fromthe N–S channel, starting 0.15 s before the S picking. A cross-correlation threshold of 85 per cent was imposed to the time-series,hence only events with a correlation coefficient in S and P waveshigher than 0.85 were kept to perform the cluster separation.

Using these parameters, up to 10multiplets or clusters were foundin our data set, containing at least three similar earthquakes. In total,they group 59 events, 55 per cent of the 106 events revised. Most ofthe earthquakes in each family occur through all the recording periodwithout any relevant temporal distribution, except families G and Hthe events of which concentrate in 3 days, both between March 18and 20. Fig. 9 shows the vertical component seismograms of the 10families recorded at OLL. Most of the families show an impulsive Parrival, except families F, G and H, which have an emergent P onset.Clusters C, J and I show a strong secondary arrival approximately0.5 s after theP onset. The Swave arrivals show, for all the families, amaximum of amplitude for frequencies between 7–8.5 Hz. ClustersB, C, F and H also show strong amplitude peaks for frequencies of11, 13 and 16 Hz.

5.2 Relative event relocation using waveform similarity

Events with high waveform similarities allow for a very precisedetermination of the arrival time differences for a same phase. Thesetime differences can be calculated using either the cross-spectralanalysismethod (Poupinet et al. 1984; Fremont&Malone 1987;Gotet al. 1994; Rubin et al. 1999), or the time-domain, cross-correlationapproach (Deichmann & Garcıa-Fernandez 1992; Augliera et al.1995; Saccorotti et al. 2002), which is used in this study. In bothapproaches, cross-correlation functions are calculated to obtain thetime-shift between pair of signals, improving the precision of therelative arrival time estimations.

For each family, the event recorded by the greatest number ofstations, usually the highest magnitude earthquake, was selected as

C© 2006 The Authors, GJI, 166, 238–252



Figure 9. Vertical component seismograms, 3–15 Hz bandpass filtered, recorded at OLL station, of the ten families obtained by correlation analysis. Boldtraces are the master events used to cluster event location.

C© 2006 The Authors, GJI, 166, 238–252


248 M. Ruiz et al.

Figure 10. Pairs of master and slave events of the G cluster, 3–15 Hz bandpass filtered, recorded at EGA and OLL stations and their discrete cross-correlationfunctions, indicated by grey and black dots. Each pair of seismograms is aligned according to the maximum of their cross-correlation function and the originaland calculated P onsets are shown for each slave event. The continuous bold curves show the parabolic function fitted through the convex part of the maximumpeak of the discrete cross-correlation function. The maximum of the interpolated cross-correlation function is indicated by a discontinuous vertical line andallows the retrieval of the time-shift between onsets of master and slave events.

the master event. Its phase picking were then used as a reference tocorrect the onsets of the slaves events of its family.

As signal to noise ratio at far-off stations and low-magnitudeevents can sometimes be moderate, the time-shift was estimated byfitting a parabolic function, with a least-squares method, through thediscrete points that form the convex part of the maximum peak ofthe cross-correlation curve (Deichmann & Garcıa-Fernandez 1992;Augliera et al. 1995). A time window of 1 s, starting 0.5 s before theP phase picked during the location procedure, was used to obtain thediscrete cross-correlation functions needed to correct the P onsetsof the slave events (Fig. 10). The S onsets corrections were obtainedusing time windows of 2 s starting 1 s before the S phase picked.The interpolation of the cross-correlation curve provides a time-shiftmeasure between the traces smaller than the sampling interval.

The results obtained in this way may be influenced by the con-fidence of the phase picking of the master event and its signalsemblance with the slave one. Repeating the correlations with allpossible pairs of events within a given set, and applying a least-squares fitting procedure, it is possible to determine arrival-timedifferences between master and slaves events, which minimize dis-crepancies, allowing amore precise determination on the onset times(seeDeichmann&Garcıa-Fernandez 1992 andSaccorotti et al.2002for the mathematical formulation of the procedure). Consistencyamong the synchronization of all the stations was always insured byGPS or DCF systems, and simplifications proposed by Saccorottiet al. (2002) were applied.

The relocation of the event clusters was performed following twoschemes. First, the relative location procedurewas applied followinga classical master event approach. The earthquakes defined as mas-ter events were located using the standard Hypo71 program. Thenthe arrival times of the slave events were corrected with the time-shifts, calculated by cross-correlation, and the travel time residuals

resulting from the master event localization, which are handled asa new station correction (Deichmann & Garcıa-Fernandez 1992;Saccorotti et al. 2002). Finally the corrected arrivals of the slaveevents are inverted, obtaining their relative positions to the mas-ter events (Fig. 11). Relative locations of the cross-correlated eventfamilies were also performed with the double-difference hypocen-tre locations approach (Waldhauser & Ellsworth 2000), using theHypoDD program (Waldhauser 2001).

Fig. 11(a) shows the standard locations of the event familieswithout any correction, to be compared with improved locationswhich display the slave events clustered around their master event(Figs 11b and c). The relocated events, particularly those with thedouble difference approach constrain a much narrower epicentralarea (Fig. 11c). A well-defined E–W region of about 2 km long andless than 1 km wide is imaged, which follows the southern thrust-ing slice delineating the southern sector of the north-verging Aralarthrusting system. The depth range appears also better constrained,with most of the events located between 1.5 and 2 km depth.

The hypocentral distributions imaged in N–S and E–W cross-sections (Figs 11b and c) probably delineate the sectors of the faultthat have been active and which could not be resolved with the stan-dard locations (Fig. 11a). The E–W cross-sections (A-A′ in Figs 11band c) show a thin discontinuous E–W pattern in which two groupsof events can be distinguished. The westernmost one is formed byfamilies A, I and J, distributed mainly in a thin line at 1.5 km depth,but with a few events reaching the surface. The other families, in theeastern fault sector, are distributed at 1.5 to 2 km depth, althoughsome events may extend to 4 km depth. The N–S cross-sections(B–B′ in Figs 11b and c) suggest a southward-dipping distribution,specially in the southern half of the transect where the events layfrom near the surface down to 4 km depth, which may delineate thefault plane responsible of the seismic crisis.

C© 2006 The Authors, GJI, 166, 238–252



Figure 11. (a) Standard absolute locations of the events belonging to the ten families obtained. Master events are indicated with a symbol greater than slaveevents and marked with the label of each family. Two, N–S and E–W, cross-sections show the distribution in depth of these events. (b) Relative locations of theten earthquake families obtained from cross-correlation and using the master event approach technique. All the earthquakes appearing in this figure have meanquality Q equal to A or B. (c) Relative locations of the cross-correlated events obtained using Double Difference approach.

C© 2006 The Authors, GJI, 166, 238–252


250 M. Ruiz et al.

Figure 12. Relocation of the complete catalogue using the Double Difference approach. Squares show the cross-correlated events belonging to earthquakefamilies, and circles the non-cross-correlated events.

Finally, the double difference approach was also applied to relo-cate the events not belonging to any of the defined families. Fig. 12shows all the DD relocations. The epicentral area appears moreclustered than with standard locations (Fig. 3a) and the E–W cross-section shows again the two sectors imaged with the clusters, eventsbeing mainly distributed between 1.5 and 2 km depth.

6 D I SCUSS ION AND CONCLUS IONS

The deployment of a temporary seismic network after the 2002February 21 earthquake has provided new seismotectonic insightsat theBasque–Cantabrianbasin, on thewestern edgeof thePyrenees.Accurate relocation of the 13 aftershocks already recorded by per-manent networks and hypocentral determinations of 93 previouslyunreported events provided relevant constraints on the properties ofthe seismically active area, the southern sector of the Aralar thrustunit.

The geographical location of the hypocentral area and the focalmechanisms obtained for the most significant events, showing nor-mal fault solutions with one of the nodal planes roughly orientedE–W, both favour as the seismogenic structure the relaxation of theE–W thrusts that bounds the southern side of the Aralar thrustingsystem, activated by a dextral movement of the Egarrieta Fault, in-stead of other structures located nearby, like the Pamplona fault.The local stress field derived from focal solutions of the aftershocksshows maximum P directions oriented NW–SE. Although a ho-mogeneous stress field for the whole Pyrenean range can not bedefined from the present-day seismicity pattern of moderate magni-tude (Nicolas et al. 1990; Delouis et al. 1993; Herraiz et al. 2000;Souriau et al. 2001), focal mechanisms of earthquakes occurredin the western part of the Pyrenees tend to suggest a NNW–SSEto NW–SE direction of maximum stress. The dips of the P-axes

range from subhorizontal to subvertical, thus cases of normal or re-verse faulting and a variable component of strike-slip are present inthis region (Gagnepain et al. 1980; Gagnepain-Beyneix et al. 1982;Delouis et al. 1993; Souriau et al. 2001). Three focal solutions, com-ing from events with magnitudes between 4.2 and 4.9, are availablefor the 1982 seismic series, showing normal faulting with smallstrike-slip component and one of the nodal planes approximatelyE–W oriented (Olivera & Gallart 1987).

The seismotectonic relevance of theAralar system revealed in thisstudy could not be inferred from the sparse distribution of hypocen-tres shown in the catalogues established by permanent networks dueto the low coverage of stations operating in this area. In these cata-logues, the most active structure in this western Pyrenean area ap-pears to be the central segment of theNNE–SSWoriented Pamplonafault. This fault is a crustal scale accident separating different tec-tonic domains and probably controlling the regional state of strain.The Aralar system is an E–W structure changing to NW–SE in itseastern edgewhen approaching the Pamplona fault, and correspondsto a secondary conjugate structure of this main fault. Therefore theshallow seismic activity detected and studied here may be explainedin terms of relative movements affecting the whole fault system.

Statistical relations and aftershocks parameters were also estab-lished from the new data. A P-value of 0.9 resulted from the ad-justment of the modified Omori law. This value, lying in the lowerrange of the expected values, reflects a moderate decay of the after-shock activity in this region. A local magnitude adjustment was alsoperformed. The b-value of the cumulative Gutenberg–Richter distri-bution obtained, close to 0.8 and therefore relatively low, is charac-teristic of regions with concentrated seismicity and well-constrainedfault planes. This solution agrees with the epicentral zone deducedfrom our data.

Further constraints on the hypocentral region have been obtainedafter a detailed analysis of the seismogram waveforms. First, a

C© 2006 The Authors, GJI, 166, 238–252



correlation analysis has been applied to detect and associate eventswith similar waveforms. 10 clusters or families of earthquakes,containing a minimum of three events each, are obtained. Themost significant differences between these families are the impul-sive/emergent character of the first arrival, the relevance of an in-termediate P phase arrival and the frequency content of the S-wavearrival. These differences have to be associated with differences inthe rupture mechanisms and in the geographic hypocentral location.No remarkable temporal patterns were observed in the occurrenceof the different families. Secondly, cross-correlation techniques, to-gether with least-squares fitting of the relative arrival times havebeen applied to these event clusters, resulting in highly improvedhypocentral locations. The final results, obtained with a double dif-ference approach, show an E–W epicentral domain of only 2 km2,where different rupture zones can be identified. The seismicity ismainly constrained to the 1.5–2.0 km depth range, although somedeeper events may delineate a southward dip. The improved loca-tions depict a narrower epicentral zone tightly clustered, and allowto associate as the seismogenic structure the E–W thrusts delineat-ing the southern sector of the northward-verging Aralar thrustingslice.

The results obtained in this study enhance the importance of tem-porary monitoring by portable network deployments after a rel-evant earthquake, even of moderate magnitude. High-resolutionaftershock experiments are needed to reveal the seismogenic fea-tures of areas poorly covered by permanent networks and provideinvaluable constraints on the regional seismotectonics.

ACKNOWLEDGMENTS

This work was carried out with the support of the MARCONI(REN2001-1734) project from the Spanish MCYT. MR and CLbenefited from a PhD grant from the Spanish MCYT. We wouldlike to thank to the responsible of the IGN, RENASS, OMP andCEA databases for sharing the seismological data used in this study.We acknowledge Dr. A. Villasenor for his numerous suggestionsand comments. The authors also wish to thank Dr M. Herraiz andan anonymous reviewer for their constructive comments on thismanuscript.

REFERENCES

Aster, R.C. & Scott, J., 1993. Comprehensive characterization of waveformsimilarity inmicroearthquakedata sets,Bull. seism. Soc.Am.,83(4), 1307–1314.

Augliera, P., Cattaneo, M. & Eva, C., 1995. Seismic multiplets analysis andits implication in seismotectonics, Tectonophysics, 248, 219–234.

Choukroune, P., 1992. Tectonic evolution of the Pyrenees, Annu. Rev. EarthPlanet. Sci., 20, 143–158.

Delouis, B., Haessler, H., Cisternas, A. & Rivera, L., 1993. Stress tensordetermination in France and neighbouring regions, Tectonophysics, 221,413–438.

Deichmann, N. & Garcıa-Fernandez, M., 1992. Rupture Geometry fromhigh-precision relative hypocentre locations ofmicroearthquakes clusters,Geophys. J. Int., 110, 501–517.

Engeser, T. & Schwentke, W., 1986. Towards a new concept of the tectoge-nesis of the Pyrenees, Tectonophysics, 129, 233–242.

Faci, E. et al., 1997.Mapa geologico deNavarra, Escala 1:200.000.Gobiernode Navarra. Spain, p. 142.

Feuillee, P. & Rat, P., 1971. Structures et paleogeographies Pyreneo-Cantabriques, in Histoire Structurale du Golfe de Gascogne, Vol. 1,

pp. 1–48, eds Debyser, J., Le Pichon, X. & Montadert, L., Publ. Inst.Francais du Petrole, Ed. Technip.

Fremont, M.J. & Malone, S.D., 1987. High precision relative locations ofearthquakes at Mount St. Helens, Washington, J. geophys. Res., 92(B10),10 223–10 236.

Gagnepain, J., Mediano, T., Cisternas, A., Ruegg, J.C., Vadell, M., Hatzfeld,D. & Mezcua, J., 1980. Sismicite de la region d’Arette (Pyrenees-Atlantiques) et mecanismes au foyer, Ann. Gefis., 36, 499–508.

Gagnepain-Beyneix, J., Haessler, H. & Modiano, T., 1982. The Pyre-nean earthquake of February 29, 1980: an example of complex faulting,Tectonophysics, 85, 273–290.

Garcıa-Modejar, J., 1996. Plate reconstruction of theBay ofBiscay,Geology,24, 635–638.

Goldstein, P., Dodge, D., Firpo, M. & Minner, L., 2003. SAC2000: Sig-nal processing and analysis tools for seismologists and engineers. Invitedcontribution to ‘The IASPEI International Handbook of Earthquake andEngineering Seismology’, Vol. 81B, eds Lee, W.H.K., Kanamori, H., Jen-nings, P.C. & Kisslinger, C., Academic Press, London.

Goldstein, P. & Minner, L., 1996. SAC2000: Seismic signal processing andanalysis tools for the 21st Century, Seis. Res. Lett., 67, 39.

Got, J.L., Frechet, J. & Klein F.W., 1994. Deep fault plane geometry inferredfrom multiplet relative relocation beneath the south flank of Kilauea, J.geophys. Res., 99(B8), 15 375- 15 286.

Guo, Z. & Ogata, Y., 1997. Statistical relations between the parametersof aftershocks in time, space and magnitude. J. geophys. Res., 102(B2),2857–2873.

Grandjean, G., Daignieres, M., Gallart, J. & Hirn, A., 1994. Repartition dela sismicite dans la partie occidentale des Pyrenees. C.R. Acad. Sci. Paris,t. 319, serie II, P. 527–533.

Hauksson, E. & Jones, L.M., 1989. The 1987 Whittier Narrows earthquakesequence in Los Angeles, Southern California: seismological and tectonicanalysis, J. geophys. Res., 94(B7), 9569–9589.

Herraiz, M. et al., 2000. The recent (upper Miocene to Quaternary) andpresent tectonic stress distributions in the Iberian Peninsula, Tectonics,19, 762–786.

Instituto Geologico y Minero de Espana (IGME), 1987. Serie Magna, HojaAlsasua-114, (24–07) Escala 1:50.000. Mapa Geologico de Espana. Ser-vicio de publicaciones del Ministerio de Industria.

Instituto Geologico y Minero de Espana (IGME), 1978. Serie Magna, HojaGulina-115, (25–07) Escala 1:50.000. Mapa Geologico de Espana. Ser-vicio de publicaciones del Ministerio de Industria.

Kisslinger, C. & Jones, L.M., 1991. Properties of aftershock sequences inSouthern California, J. geophys. Res., 96(B7), 11 947–11 958.

Larrasoana, J.C., Pares, J.M., Millan, H., del Valle, J. & Pueyo, E.L., 2003.Paleomagnetic, structural, and stratigraphic constraints on traverse faultkinematics during basin inversion: the Pamplona fault (Pyrenees, NorthSpain), Tectonics, 22(6), 1071, doi:10.1029/2002TC001446.

Lee, W.H.K., Bennet, R.E. & Meagher, K.L., 1972. A Method of estimatingmagnitude of local earthquakes from signal duration,USGeol. Surv.Openfile report 1–28.

Lee, W.H.K. & Lahr, J.C., 1975. HYPO71 (revised): A computer programfor determining hypocenters, magnitudes and first motion pattern of localearthquakes, US Geol. Surv. Open File report 75–311.

Martınez-Torres, L.M., 1989. El Manto de los Marmoles (Pirineo Occiden-tal): geologıa estructural y evolucion geodinamica. PhD thesis, Univ. PaısVasco, Spain, p. 294.

Maurer, H. & Deichmann, N., 1995. Microearthquake cluster detection onwaveform similarities, with an application to the western Swiss Alps,Geophys. J. Int., 123, 588–600.

Montadert, L., Charpal, O., Roberts, D.G., Guennoc, P. & Sibuet, J.C., 1979.Northeast Atlantic passive margins: rifting and subsidence processes,Am-mer. Geophys. Un. Revue, 3, 154–186.

Muller, J. & Rodgers, P., 1977. L’evolution des Pyrenees (domaine central etoccidental). Le segment Hercynien, la chaine de fond alpine, Geol. Alp.,53, 141–191.

Nanjo, K., Nagama, H. & Sotomura, M., 1998. Rates of aftershock decayand the fractal structure of active fault systems, Tectonophysics, 287, 173–186.

C© 2006 The Authors, GJI, 166, 238–252


252 M. Ruiz et al.

Nicolas, M., Santoire, J.M. & Delpech, P.Y., 1990. Intraplate seismicity: newseismotectonic data in Western Europe, Tectonophysics, 179, 27–53.

Njike-Kassala, J.D., Souriau,A.,Gagnepain-Beyneix, J.,Martel, L.&Vadell,M., 1992. Frequency-magnitude relationship and Poisson’s ratio in thePyrenees, in relation to earthquake distribution,Tectonophysics, 215, 363–369.

Ogata, Y., 1999. Seismicity analisis throught point-processs modeling: areview, Pure appl. Geophys., 155, 471–507.

Olivera, C. & Gallart, J., 1987. Sismicidad de la region de Navarra (PirineosOccidentales), Rev. de Geofısica, 43, 221–234.

Pedreira, D., Pulgar, J.A., Gallart J. & Dıaz, J., 2003. Seismic evidence ofAlpine crustal thickening and wedging from the Western Pyrenees to theCantabrian Mountains (North Iberia), J. geophys. Res., 108(B4), 2204,doi:10.1029/2001JB001667.

Poupinet, G., Ellsworth, W.L. & Frechet, J., 1984. Monitoring velocity varia-tions in the crust using earthquakes doublets: An application to the Calav-eras Fault, California, J. geophys. Res., 89(B7), 5719–5731.

Rat, P., 1988. The Basque-Cantabrian basin between the Iberian and Eu-ropean plates; some facts but still many problems, Rev. Soc. Geol. Esp.,1(3–4), 327–348.

Reasenberg, P.A. & Oppenheimer, D., 1985. FPFIT, FPPLOT and FPPAGE:Fortran computer programs for calculating and displaying fault-plane so-lutions, US Geol. Surv. Open File report 85–739, p. 25.

Rubin, A.M., Gillard, D. & Got, J-L., 1999. Streaks of microearthquakesalong creeping faults, Nature, 400, 635–641.

Ruiz, M., Gallart, J., Dıaz, J., Olivera, C., Pedreira D., Lopez, C., Gonzalez-Cortina, J.M. & Pulgar, J.A., 2006. Seismic Activity at the Western Pyre-nean Edge, Tectonophysics, 412, 217–235.

Saccorotti, G., Carmona, E., Ibanez, J.M.&Del Pezzo, E., 2002. Spatial char-acterization of Agron, southern Spain, 1988–1989 seismic series, Phys.Earth planet. Inter., 129, 13–29.

Shearer, P.M., 1997. Improving local earthquake locations using the L1 normand waveform cross correlation: application to the Whittier Narrows, Cal-ifornia, aftershock sequence, J. geophis. Res., 102(B4), 8269–8283.

Souriau, A. & Pauchet, H., 1998. A new synthesis of Pyrenean seismicityand its tectonic implications, Tectonophysics, 290, 221–244.

Souriau, A., Sylvander, M., Rigo, A., Fels, J.F., Douchain, J.M. & Pon-solles, C., 2001. Sismotectonique des Pyrenees: principales constraintessismologiques. Bull. Soc. Geol. France., 172, 25–39.

Tsujiura, M., 1983. Characteristic frequencies for earthquake families andtheir tectonic implications: evidence from earthquake swarms in theKantoDistrict, Japan, Pure appl. Geophys., 4, 573–600.

Turcotte, D.L., 1992. Fractals and Chaos in geology and geophysics,Cambridge Univ. Press, New York, p. 221.

Turner, J.P., 1996. Switches in subduction direction and the lateral termina-tion of mountain belts: Pyrenees-Cantabrian transition, Spain, J. Geolog-ical Society, 153, 563–571.

Utsu, T., 1961. Statistical study on the ocurrence of aftershocks, Geophys.Mag., 30, 521–605.

Utsu, T., 1999. Representation and analysis of the earthquake size distribu-tion: a historical review and some new approaches, Pure appl. Geophys.,155, 509–535.

Utsu, T., 2002. Statistical features of seismicity. Invited contribution to ‘TheIASPEI International handbook of earthquake and engineering seismol-ogy’, Vol. 81A. Edited by Lee, W.H.K., Kanamori, H., Jennings, P.C. &Kisslinger, C. Academic Press, London.

Waldhauser, F., 2001. HypoDD—a program to compute double-differencehypocenter locations, US Geological Survey. Open File Report 01–113.

Waldhauser, F. & Ellsworth, W.L., 2000. A double-difference earthquakelocation algorithm: method an application to the northern Hayward fault,California, Bull. seism. Soc. Am., 90(6), 1353–1368.

C© 2006 The Authors, GJI, 166, 238–252



114


115

5.3.- Apèndix: Determinacions hipocentrals de la sèrie sísmica d’Aralar

Taula 5. 1.- Determinacions hipocentrals obtingudes amb el programa Hypo71 pels 106 events analitzats durant l’experiment del 2002 a la serra d’Aralar.

Data Origen HoraOrigen

Latitud (N)

Longitud (E)

Z(km) Mag Gap

(º) Rms(s)

ERH(km)

ERZ(km) Q SQD

23/02/02 18:58:39 42.9215 -1.8662 5.51 1.63 105 0.06 0.4 0.7 B A|B 24/02/02 08:59:51 42.9198 -1.8590 5.85 1.52 101 0.13 0.7 1.2 B A|B 24/02/02 19:03:14 42.9173 -1.8590 2.63 0.86 104 0.15 0.6 0.8 B B|B 25/02/02 02:54:39 42.9182 -1.8440 2.56 1.46 97 0.07 0.3 0.5 B A|B 25/02/02 03:44:27 42.9262 -1.8377 2.64 1.67 99 0.07 0.3 0.4 B A|B 25/02/02 03:56:56 42.9288 -1.8447 1.46 1.13 96 0.08 0.3 1.4 B A|B 25/02/02 04:01:18 42.9273 -1.8390 0.89 1.10 99 0.06 0.3 1.6 B A|B 25/02/02 12:41:46 42.9243 -1.8352 2.27 2.14 99 0.09 0.3 0.6 B A|B 25/02/02 16:34:30 42.9235 -1.8465 2.58 1.38 92 0.03 0.2 0.5 B A|B 26/02/02 02:34:36 42.9255 -1.8453 1.71 1.48 83 0.12 0.3 0.7 B A|B 26/02/02 12:41:42 42.9260 -1.8420 2.60 1.62 90 0.09 0.3 1 B A|B 27/02/02 00:51:57 42.9260 -1.8470 2.28 1.40 176 0.08 0.5 1.3 B A|C 27/02/02 01:39:30 42.9310 -1.8488 0.39 1.37 99 0.09 0.3 2.8 B B|B 27/02/02 06:13:46 42.9260 -1.8322 1.98 3.33 78 0.09 0.2 0.3 B A|B 27/02/02 06:17:15 42.9272 -1.8342 1.91 3.30 87 0.13 0.3 0.4 B A|B 27/02/02 07:57:48 42.9255 -1.8365 2.75 1.78 87 0.12 0.4 0.7 B A|B 27/02/02 08:44:26 42.9253 -1.8318 1.40 3.01 85 0.13 0.3 0.5 B A|B 27/02/02 12:16:42 42.9292 -1.8472 1.81 2.50 177 0.04 0.3 0.9 B A|C 27/02/02 13:38:24 42.9240 -1.8310 1.36 2.40 83 0.1 0.3 0.5 B A|B 27/02/02 19:13:38 42.9237 -1.8355 1.30 2.41 85 0.17 0.4 0.9 B B|B 28/02/02 13:42:02 42.9218 -1.8437 4.35 1.78 171 0.02 0.3 0.6 B A|C 01/03/02 23:55:31 42.9288 -1.8452 1.29 1.63 96 0.08 0.3 0.8 B A|B 02/03/02 08:42:06 42.9208 -1.8528 1.68 1.43 94 0.09 0.4 1.5 B A|B 02/03/02 12:27:37 42.9240 -1.8410 1.07 2.32 88 0.15 0.5 1.5 B A|B 02/03/02 22:49:31 42.9290 -1.8408 2.25 1.52 99 0.19 0.6 1.1 B B|B 02/03/02 23:10:25 42.9192 -1.8558 5.87 1.24 186 0.06 0.6 1 C A|D 03/03/02 00:05:55 42.9272 -1.8340 2.28 2.51 87 0.15 0.4 0.6 B B|B 03/03/02 00:25:48 42.9282 -1.8458 3.82 1.29 175 0.05 0.7 1.3 B A|C 03/03/02 00:43:22 42.9263 -1.8263 1.87 2.39 83 0.14 0.4 0.6 B A|B 03/03/02 07:41:36 42.9288 -1.8623 0.29 1.17 104 0.14 0.5 4.2 B B|B 03/03/02 08:08:00 42.9225 -1.8495 2.24 1.49 91 0.05 0.2 0.9 B A|B 04/03/02 01:52:41 42.9247 -1.8317 1.34 1.74 101 0.09 0.3 0.8 B A|B 04/03/02 03:41:08 42.9302 -1.8433 2.08 2.53 95 0.17 0.4 0.7 B B|B 04/03/02 10:16:02 42.9192 -1.8482 2.52 1.51 93 0.07 0.5 1.4 B A|B 05/03/02 03:13:25 42.9277 -1.8420 2.28 2.20 92 0.11 0.3 0.5 B A|B 07/03/02 17:41:15 42.9223 -1.8452 4.31 1.58 94 0.07 0.6 1.4 B A|B 08/03/02 22:28:19 42.9240 -1.8517 1.21 2.27 92 0.16 0.4 0.9 B B|B 09/03/02 05:22:32 42.9240 -1.8542 0.60 1.58 93 0.16 0.5 3.3 B B|B 09/03/02 08:39:05 42.9288 -1.8488 2.50 1.27 97 0.09 0.3 1.2 B A|B 09/03/02 21:06:20 42.9210 -1.8447 3.81 1.48 172 0.1 0.6 1.1 B A|C 10/03/02 07:45:52 42.9297 -1.8457 1.07 1.16 175 0.06 0.4 1.8 B A|C 10/03/02 09:52:26 42.9260 -1.8385 2.38 1.41 154 0.06 0.3 0.8 B A|C 11/03/02 00:43:44 42.9163 -1.8547 2.88 0.90 101 0.1 0.4 0.4 B A|B 13/03/02 23:50:26 42.9273 -1.8362 1.91 1.32 156 0.05 0.2 0.3 B A|C 14/03/02 04:19:29 42.9273 -1.8495 1.33 2.10 75 0.05 0.2 0.3 A A|A 15/03/02 01:55:02 42.9292 -1.8392 1.34 1.11 93 0.07 0.3 0.6 B A|B 15/03/02 08:20:48 42.9332 -1.8440 0.99 2.68 58 0.18 0.4 0.7 B B|A 15/03/02 22:18:44 42.9273 -1.8373 1.81 1.34 155 0.08 0.3 0.6 B A|C 16/03/02 03:44:41 42.9243 -1.8467 1.44 2.06 88 0.07 0.2 0.4 A A|A 17/03/02 10:38:23 42.9213 -1.8322 1.94 1.32 166 0.06 0.3 0.5 B A|C 17/03/02 18:29:09 42.9233 -1.8330 2.30 1.19 164 0.05 0.2 0.4 B A|C 17/03/02 22:01:12 42.9260 -1.8422 1.56 1.19 150 0.04 0.2 0.3 B A|C 18/03/02 13:34:52 42.9263 -1.8435 1.33 1.64 110 0.04 0.1 0.3 B A|B 18/03/02 15:18:53 42.9175 -1.8277 2.10 3.43 67 0.1 0.2 0.3 A A|A 18/03/02 15:32:52 42.9248 -1.8350 1.52 1.37 117 0.06 0.2 0.5 B A|B 18/03/02 16:39:46 42.9212 -1.8345 2.21 1.51 126 0.04 0.2 0.3 B A|B 18/03/02 17:36:38 42.9250 -1.8303 1.74 1.39 102 0.06 0.2 0.5 B A|B 18/03/02 19:12:40 42.9238 -1.8283 2.10 1.53 104 0.08 0.3 0.6 B A|B 18/03/02 19:21:52 42.9298 -1.8360 2.84 1.34 100 0.03 0.2 0.2 B A|B 18/03/02 19:26:10 42.9315 -1.8413 3.08 2.50 146 0.05 0.4 0.4 B A|C 18/03/02 22:16:53 42.9292 -1.8293 2.27 1.39 93 0.05 0.2 0.4 B A|B 18/03/02 22:51:20 42.9208 -1.8297 1.52 1.95 80 0.07 0.2 0.6 A A|A


116

19/03/02 00:53:58 42.9273 -1.8268 2.14 1.73 66 0.09 0.3 0.6 A A|A 19/03/02 01:54:05 42.9222 -1.8383 2.93 1.11 99 0.04 0.3 0.3 B A|B 19/03/02 02:38:30 42.9310 -1.8398 3.51 0.92 107 0.04 0.2 0.3 B A|B 19/03/02 02:57:32 42.9233 -1.8503 1.65 0.83 142 0.03 0.3 0.3 B A|C 19/03/02 03:07:23 42.9197 -1.8362 2.62 1.07 163 0.04 0.3 0.4 B A|C 19/03/02 03:31:46 42.9188 -1.8378 2.05 1.34 101 0.07 0.2 0.5 B A|B 19/03/02 04:18:33 42.9195 -1.8397 2.26 1.04 158 0.03 0.3 0.4 B A|C 19/03/02 09:07:21 42.9203 -1.8375 2.45 1.86 90 0.09 0.3 0.3 A A|A 19/03/02 09:28:40 42.9245 -1.8278 2.03 2.08 68 0.08 0.2 0.4 A A|A 19/03/02 12:49:32 42.9273 -1.8437 1.18 1.09 147 0.06 0.3 0.5 B A|C 19/03/02 13:25:31 42.9258 -1.8413 1.21 1.21 151 0.07 0.3 0.5 B A|C 19/03/02 14:13:31 42.9248 -1.8265 1.43 1.94 70 0.07 0.2 0.6 A A|A 19/03/02 19:46:28 42.9290 -1.8493 3.76 2.50 139 0.02 0.2 0.1 B A|C 19/03/02 20:27:05 42.9238 -1.8275 1.62 1.28 170 0.06 0.4 0.7 B A|C 19/03/02 22:57:47 42.9203 -1.8365 1.94 1.72 101 0.07 0.3 0.5 B A|B 20/03/02 00:07:38 42.9220 -1.8355 2.28 1.66 101 0.08 0.3 0.5 B A|B 20/03/02 00:19:37 42.9217 -1.8368 2.31 1.52 100 0.08 0.3 0.5 B A|B 20/03/02 04:28:00 42.9198 -1.8390 2.33 1.66 100 0.04 0.2 0.3 B A|B 20/03/02 05:16:41 42.9242 -1.8428 1.66 1.50 94 0.07 0.2 0.4 B A|B 20/03/02 11:59:46 42.9227 -1.8350 2.13 1.15 101 0.06 0.2 0.4 B A|B 20/03/02 20:44:00 42.9215 -1.8400 2.33 1.66 98 0.05 0.2 0.3 B A|B 20/03/02 21:04:10 42.9252 -1.8295 2.35 1.69 71 0.08 0.2 0.3 A A|A 21/03/02 04:05:35 42.9273 -1.8365 1.89 1.05 96 0.07 0.2 0.5 B A|B 21/03/02 22:25:59 42.9273 -1.8492 1.83 0.86 140 0.03 0.3 0.3 B A|C 21/03/02 23:14:25 42.9213 -1.8350 1.66 0.77 163 0.04 0.2 0.4 B A|C 22/03/02 11:17:40 42.9275 -1.8495 1.51 1.46 88 0.07 0.2 0.4 A A|A 22/03/02 13:57:45 42.9198 -1.8372 2.19 1.67 91 0.07 0.2 0.4 B A|B 22/03/02 18:43:26 42.9225 -1.8302 0.94 1.41 104 0.08 0.3 0.8 B A|B 22/03/02 20:48:25 42.9233 -1.8315 1.73 1.07 166 0.07 0.3 0.6 B A|C 22/03/02 22:06:26 42.9203 -1.8317 1.64 2.30 70 0.07 0.2 0.3 A A|A 22/03/02 23:17:09 42.9193 -1.8423 2.08 1.99 100 0.07 0.2 0.4 B A|B 24/03/02 05:06:14 42.9262 -1.8340 1.25 1.15 98 0.06 0.2 0.6 B A|B 24/03/02 11:36:37 42.9253 -1.8355 1.29 1.11 159 0.06 0.3 0.5 B A|C 24/03/02 13:05:16 42.9278 -1.8260 0.78 1.71 132 0.06 0.2 0.8 B A|B 25/03/02 00:10:07 42.9177 -1.8420 1.34 0.98 157 0.05 0.2 0.4 B A|C 25/03/02 12:15:01 42.9250 -1.8303 1.41 1.35 166 0.06 0.3 0.6 B A|C 25/03/02 20:21:04 42.9187 -1.8452 1.48 1.86 107 0.08 0.3 0.5 B A|B 25/03/02 21:50:19 42.9212 -1.8422 0.50 1.05 154 0.05 0.2 0.6 B A|C 26/03/02 04:14:05 42.9273 -1.8258 1.40 1.27 169 0.07 0.3 0.8 B A|C 27/03/02 00:12:19 42.9187 -1.8448 1.27 0.86 152 0.05 0.2 0.4 B A|C 27/03/02 00:41:24 42.9215 -1.8275 1.34 1.19 172 0.07 0.4 0.8 B A|C 27/03/02 20:40:35 42.9187 -1.8340 2.18 1.73 89 0.07 0.2 0.4 A A|A 28/03/02 00:18:21 42.9273 -1.8373 1.54 1.65 95 0.07 0.3 0.5 B A|B

caracteritzaciÓ estructural i sismotectÒnica de...

Documents