Veourstofa islandsReport

Ragnar Stefånsson

Earthquake-prediction research in anaturaliaboratory - PRENLAB

Vi-G97041.JA06ReykjavikDecember 1997

Contents

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Key words 2

Introduction 2

Methods and results 6

!IJ Real-time evaluation of earthquake-re!ated-processes anddevelopment of database . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 6

'[I] Development of methods using microearthquakes for monitoringcrustal instability '. . . . . . . . . 12

[Ij Monitoring stress changes before earthqllakes using seismicshear-wave splitting 13

[IJ Borehole monitoring of fluid-rock interaction

lI] Active deformation determined from GPS and SAR

............ 18

........ 22

~ Formation and development of seismogenic faults and faultpopulations .

[Il Theoretical analysis of faulting and earthqllake proeesses

Acknowledgements

References . . . . . .

. 25

.27

. 32

.32

Coordinator and contractors 38

l

ENV4-CT96-0252EARTHQUAKE-PREDICTION RESEARCH IN A NATURAL LABORATORY

PRENLABby

Ragnar StefånssonIcelandic Meteorological Office, Department of Geophysics, Reykjavik, Iceland

SlImmary

PREN LAB is a twa years multinational project of earthquake preeliction research, starting March1,1996. In this project multielisciplinary European technology anel science are applieel in a com­mon a.ction aimed at progress in earthquake prec!iction research and at reducing seismie risk. Thequestiolls are where, when and how large ea rthquake ground mations will occur. Answers are

sought by stuelying the physical processes anel conelitions leaeling to large earthquakes. Icelanel,sometiilles calleel a Naturai Geoph"sical Laboratory, is the test area for the project. The episoeliccharacter of crustal movements anel high seismicity, the \\"ell-elefineel earthquake zones anel shal­low sources, as well as the extensi"e knowleelge of the geology anel geophysics of Icelanel, make itan excellent test area for earthquake prediction research. Frequent short-term variations in strainrate can be utilized as a time variable input to the experiment. Among other significant faci!itiesof this laboratory is the existence of a high qua.lity earthquakc data acquisition and evaiuationsystem. The partners of the projec\ come from various branches of geosciences of 11 institu­tions in 6 European countries: France, Germany, lceland, Italy, Sweelen anel United Kingdom.PRENLAB-2, a continuation of PREi'iLAB is planned to start a two years period March 1, 1998.

Key words

Mitigating risk, Hazarel assessment. Multidisciplinary approach, Geodynamic approach, Mod­elling earthquake processes, Microearthquake technology, Earthquake precursors, Premonitorydlanges, Short-tel'm warnings, Database.

Introdllction

The ciassical hazarel assessment as it has been applied in lceland anel more countries is basedmainly on historical documentation and limited information from instrumental earthquake cata­logues of this century anel a general knowledge of where the earthquake generating plate bound­aries are. Although such hazard assessment has been extremely useful in many aspects, it hasthe obvious limitations that it is only based on a few hundreds of years of history, and in factassumes that we should only expect hazards that are comparable to those \\'hich have happenedwithin this short history. Also it does not take into consideration the exact position of andthe interaction of faults that are expecteel to mave in earthquakes or earthquake sequences. ILintergrates effects over large areas in time and space while it is well knO\\'J1 that by far the largestdestruction is relatecl to the proximity of the faults which are activatecl in the earthquakes eachtime and to the areas where the faults rupture the surface. Hazard assessment basecl mainly oncatalogues of earthquakes and their magnitudes has come to the state that it cannot have moreprogress:

• Until a better moclelling of the [ault processes has been achieved.

• Until we can recognize better the involvecl faults and monitor their movements and inter­action.

2

HazardaSlessment

EarthquakepredlcUon

Space-t1me '\

distributJon ",,AcUve f8ultlng

,,,,,,!,,,,

SEISMOLOGy

Figure 1: In the project multidisciplinary European technology and science are applied in a com­mon action aiming at progress in earthquake prediction research and for reducing seismic risk.Scientists from 11 institutions in 6 European countries participate in the project.

• Until we can monitor better the stresses in the adjacent area and understand better therheological properties and the role of fluids in the crust.

For a progress in earthquake hazard assessment and in general for progress in earthquakepredietion research we must aim at creating dynamic models which can explain multiplicity ofobservations, which means that many disciplines of geosciences must be involved.

This is the basis of the PRE LAB project. It is a multidisciplinary approach in earthquakeprediction research (Figure 1).

The dynamic models to be created must comply with a multiplicity of observations in timescales ranging from seconds to millions of years, ranging from historical seismicity to microearth­quake information, ranging from geological field observations to observations of deformation withspace technology methods and borehole observations. The PRENLAB project collects informa­tion, developes methods and models to base further observations on to create a basis for a moregeneral multidisciplinary modelling (Figure 2).

In the workprogramme of PRENLAB the overall objectives are summarized as follows:

• To develop methods for automatic extraction of all information available in the frequentmicroearthquake recordings, including fault mapping, rock stress tensor inversion, andmonitoring of crustal instability.

• To make use of this information, geological information, historical as well as older seis­mological information for physical interpretation of and modelling the tectonic proeessesleading to earthquakes.

• To improve the understanding of the space and time re!ationship between earthquakes andother observable features associated with crustal deformation.

3

14°W

il lceland Hotspot

Proleet 6tatlon9

• Borehole strainmeters

OGraVlmeters

16°W

&,.(

18°W200W22°W

66°N

64°N

Figure 2: The available 3 eomponent digital earthquake monitoring in leeland. The Mid-AtlantieRidge goes throughleeland from the Reykjanes Ridge along the South leeland seismie zone (SISZ),the rift zones to the Tjornes fmetw'e zone (TFZ) in the north. The most destruetive earthquakesoeeur in the transform zones, SISZ and TFZ. The outlines of the leeland mantle plume at depthare shown in purple. The leeland Hotspot project stations are operated during 1996-1998. Otherstations are permanent. Same other seismologieal and hyd1'OIogieal monitoring stations are oper­ated toa.

• To apply this knowledge for improved real-time evaluations and alert systems and forimproved hazard assessments.

Among repOl'ted results of PREN LAB which are of a great significance for earthquake pre­diction research, the following can be mentioned:

• It has been dernonstrated in severaI studies involving work of seismologists, geologists andgeophysicists that it is possible on basis of mieroearthquakes to map subsurface faults witha great accuracy. A good agreement is between such studies within the project basedon mieroearthquakes and the result.s of studying the faults on the surface when these areexposed.

• It has been demonstrated that stresses inferred from microearthquakes eoincide generallyvery well wit.h what can be expected, based on paleostress st.udies of the geologists in t.hisproject, and have a relevance to the first results gained from the borehole experirnent. aneof the problems of interpl'etation of double-couple focal mechanism sotutions of earthquakes

4

is to distinguish between the fault plane and the auxiliary plane. \Iethods for doing thishave been developed with three different approaches, firstly by comp"risoll with geologiealfault data, seeondly by basing on groups of solutians on lhe same fault and thirdly byapplying stability eriteria on the possible fault plane solutians for each earthquake.

• Among signifieant new resulls that ean be reported is that ehanges of shear-wave splittingat ape of the SIL stations in the South !celand seismie zone (SISZj indieate stress ehangeswitl) time that most probably can be attributed the intrusion of la"a into the crust in thepreparatory stage of the Vatnajiikull eruption that started on September 30, 1996. Theseismie station is 160 km away from the fissure intrusion. Data from volumetrie boreholestrainmeters in SISZ confirm sueh findings, as "'ell as mieroearthquake patterns during thesame period of time. This indieates that il mal' be possible to prediet inereased probabilityfor triggering of earthquakes based on monitoring of stress changes from outside the fault.zone. It is also to be pointed out that ane of the main pillars for using !celand as a test areafor earthquake predietion was that it would be possible to monitor stress or strain ehangeseaused by measurable pulsations of the !celand plume. These results have a consequeneein general for understanding how stresses are transmitted in the crust anywhere.

• The geologieal neid studies and interpretations ha"e revealed signifieant variations in thedireetion of stress fields in the same area. Such changes have also been indieated in evaluat­ing microearthquakes and former modelling of historical activity also shows the significanee

of such changes in time and space. The borehole loggings also indicate the existenee ofstress directions that do not coineide with the prevailing stresses. and also show ehangeswilh time.

• A result of a great significanee is thal il has been shown thal it is possible to use SAR,satellite radar interferometry for measuring stable plate motion during a period of a eoupleof years in the fa,"ourable conditions thal prevail in !celand. This is of enormous significaneefor construeting a dynamieal model of stress build-up in an earthquake area.

• It has been demonslrated on lira oecasions by observations anc! moc!elling how fluid in­trusion may trigger earthquakes. In one case this \Vas a magnitucle .5.8 earthquake l in theother it was an intensive earthquake sequenee. This may be a key to explain or understandforeshoeks whieh are frequenlly reported befare large earthquakes in !celand.

• Modelling work on a large scale based on multidiseiplinary data is still on a preparatorystage. Mueh modelling work has been carriec! out with the purpose of interpreting obser­vations in various fields. An example of lhis is the modell ing of obsen'ecl tension gashes onlhe surface folIOIring an earlhquake, based on knowledge of the upper crustal structure. Bythis modell ing it Iras eonc!uded that it is possible to draw conc!usions about the underlyingstress field from apen fissures striking a fell' degrees away from the fault strike. ModellingIVork has been carried out also with the aim of aclapting and de"eloping existing methodsfor the speeial conditions in iceland, with ils rifting, time dependency of stress field, etc.

• The significanee of the multidisciplinary approach of PRENLAB has been clearly demon­strated in the first steps that have been laken "'ithin the projecl to try to ansIVer the ques­tions where, when and how large earthquakes can be expeeted in the three main earthquakerisk areas of iceland (Figure 2). These are the SISZ, the HLlsavlk-f1atey transform faultwithin the Tjiimes fraeture zone (TFZ) in the north (Figure 3), and the seismie zone nearto the eity of Reykjavik. In the available earthquake history large e"ents do not repeat inthe same IVay in these areas. Predietion, lang or short-term, must t.hus be based on earthrealistic models of the present c1"y leetonical eonditions in and amund the seismie zones[74,76].

5

66.5"

66°

+ +1910M=7

Epleenter ?

-17"

---66.5"

Figure 3: The Tjornes fraeture Ione is a eomplieated right-lateral transform zone whieh eonnectsthe '-ilting in the northern voleanie zone of Iceland to the ESE- WNW rifting in the Mid-Atlantierijt zone to north of Iceland. The general plate motion aeeording to NUVEL-IA is indieated. Thered lines indieate the lines where it is proposed that the most signifieant transform-rijting motionpresently takes plaee. The most pmnouneed feature is the Husavik-Flatey transform fault, whiehwith its proposed westward eontinuation is the only weU developed transform fault of Iceland, upto 10 million years old. Historieal seismicity has been interpreted and is assumed to lie mostlyon this fault. The Grimsey zone is mueh more volcanie and geologieally a more recent feature.These twa zones are pmposed to take up the main part of the plate motion. It is not known whatis the eontrioution of eaeh, and it is assumed this is variable with time, depending on ehangingconditions in and amund the fraeture zone.

In the following the methods and result of the PRENLAB project will be described in moredetail.

Methods and results

[I] Real-time evaluation of earthquake--related-processes and development of data­baseThe work described here is basically a responsibility of the lcelandic Meteorological Office, De-

G

partment of Geophysics (IMOll.DG) in cooperation \\"ith all the other groups involved in PIlEN­LAB. In developing methods for acquisition and evaluations there has been an especially elosecooperation with Uppsala University, DeparLment of Geophysics (UUPP.DGEO).

Extension of the monitoring networksA very significant achievemenL in the data collection is the increase of number of operating seis­mic statiohs in Iceland. Since the start of the PIlENLAB project in \Iarch 1996, the numberof permarfent seismic stations, SIL stations, in operation has increased from 18 to 33. The newstations are funded by Icelandic communities, hydrothermal and hydroelectrical power compa­nies, civil defence funds. a private tunnel-digging company, the Icelandic Research Council, andindirectly by research groups canying out tomographic studies, which can make use of the pow­erfu! SIL acquisition system. The lat'gest supporter of this build-up project of the SIL systemis IMOI1.DG, which besides contributions to the initial costs guarantees the operation cost ofthe system. These stations are as other permanent stations of the SIL netll"ork available for thePRENLAB project and of great significance for it.

From summer 1996 to summer 1998, 29 extra digital broad-band stations are operated con­tinuously at remote places not covered by the SIL system, main!y fo"" collecting teleseismic data.This is a part of the Iceland Hotspot project, lead by Gillian Foulger. rniversity of Durham.Among other participants are Princeton University, \\"ith Jason MOl'gan and Guust Nolet, andBruce Julian of the U.S. Geological Survey, besides 1~IOR.DG. The wa"eform information fromthese stations will be inelucled into the SIL evaluation proeesses, especially as concerns the localseismic activity. This is a very significant addition to the data that we aceOl'ding to the originalplan can approach for the PRENLAB project. As these temporai stations are operated in areaswhere we have only fe\\" SIL stations they can provide us with a complementary overview aboutthe stress conditions in the country as a whole. Figure 2 shows the locations of the seismiestations operated in Iceland during the later part of the PRENLAB project.

Acquisitioll , evaiuatioll and storing of dataA refined and easily accessible database for seismic data, mainly based on microearthCJuakesacquired by the SIL system is under construction. Since 1991, 90.000 earthquakes have beenrecorded by the SIL system. The data were 'llItomatically evaluated and manuaily corrected.

Facilities have been developed to store ail the data on-line on hard disks. Seismogram datais stored using packed binary format, where only the number of bits that is reCJuired to storesample to sample variation, is stored.

Other data are stored in relational database tables. Station parameters such as coordinates,instrument characteristics and time corrections a.re stored in separate tables. This informationis incorporated into headers when data are extracted from the database.

In order to insure against loss of data, procedures and facilities have been developed to backup all data onto magnetic tapes. All new data and moclifications are \vritten to tape each dayand ail data are written to tape approximately eve ry two weeks. Periodicaily, a set of tapes ismoved for storage to a elifferent site. As magnetic tapes only last a few years, anel the long-termstabilty of optical storage media is not well established, this is possibly the most effeetive way topermanently preserve the data, and it has the advantage that the data is ahmys readily accessible[.52].

An interface to the elatabase through the World Wide Web is under de\"elopment. Currentlyanyone with access to the Internet can search through a list of over 65.000 earthquakes that havebeen manually checked since Janllary l, 1995.

Work has been carried out for a new, reevaIuated and refined catalogue of instrumentallymeasured earthquakes in Ic.eland sinee 1926. The cataloglle from 1926-1963 has been reevaluateeland put on digital form. The refinement of the more recent cataloglles is in progress [47].

7

Spatial changes in seismicity have been studied in an area aiong the Reykjanes Peninsula,the South Iceland Lowland. into the eastern micanie zone, and off eoast of north leeland [46,56,66].

\Vork has been carriecl out for refinec1 estimation of magnitucles and locations of historieaiearthquakes as well as felt ewnts, not instrumentally eleteetecl since 1926 [~.)].

~Iuch work has been carrieel out in inlerpreting elata from yolumetric boreholc strainmeters,which has\leacl to very significant reslilts. Premonitory and coseismic changes, volumetric strain,anel foresliocks of the magnitude ·5.8 earthquake at \·atnafji:ill, near the eastem enel of the SouthIcelanel seismic zone, ha,·e been stuelieel. These can be interpreteel ancl ha'·e been moclelleel asl1lagmatic f1uiels intrusion coinciding with the foreshocks anel the main shock fllJ.

A long~term overvic,," (since 1979) of 7 volul1letrie strainl1leters in the SISZ (Figure 2) hasbeen workeel out. Methoels have been elevelopeel for correcting the strainmeter rccorel of weatherinfluences. As a result of this \Vork can be l1lcntionec1 a change in strainrate 5-6 months befarethe start of the 1996 eruption in Vatnaji:ikull, which may be causecl by m"gma intrusion there,i.e. mo're than 150 km from the strainmeter stations (7 , 8, Til

The seismicity of Katla ,·oicano which is beneath the Myrelalsji:ikull glacier has been stuelieel.Erllptions in hatla pose a considerable danger because of enormolls' W(lter- and mudAows whichaccompany the cruptions. Il is one of the objectives of the SIL network anel the attached alertsystem to help to warn for the the future eruptions [-13].

The seismicity of the yoicanic eruption in Vatnaji:ikull was stuelieel as concerns hypocentersanel mechanism of the eanhqui1kes which were linkeel to the eruption. \luch effort was put insaving elata on this remarkable eruption from the seismic net,mrks, both eanhquake eli1ta as wellas elata on voicanic tremor. Vatnaji:iklill is elireetly above the leelanel mamle pillme ancl changesof the pillme activity greatly arreet the seismicity i1long all of the plate bounelary in Icelanel, andis thus or a great significanee for the PRENLAB objectives ['.), 79).

Although the SIL system is a seismie elata aequisition system, that is primarily designed forautomatie aeCjllisition and evaluation of elata from loeal microearthquakes. it can also be useelfor collecting teleseismic anel regional data for deep structure studies. It broaelens the scientificuse of the network anel has maele it easier to obtain funds for extending the network to a largepart of the plate boundar,· in lceland. The SIL station software has no"· been modifieel allowingfor seieetion of waveform data adelitionally at 20 ancl 4 samples per second. This makes iteconomically possible to sa'·e lang time intervals of seismological daUL from the SIL stations. Wehave developed an automatie proceelure to seleet anel slore teleseismie elata in the SIL system,based on USGSjN8IC information on teleseismie events in the whole world. ":hieh are measurablein lceland. From USGS "EIC we reeeive E-mail messages with a single-line information onearthqllakes they have determined, the so-ealled "E" type messages. A selection program reaelsthe messages anel seleets e'·ents that fulfill certain criteria of magnituele and epieentral elistanee.The program uses the L\SPEI91 moelel to compute the first arrival time at each station_ Theteleseismie body wave data are fetched with a sampling rate of 20 samples (in same eases 100samples) per seeonel and the slll-face wave elata with sampling rate of 4 samples per seconel fromthe 1-3 days lang ringbuffer of the SIL site stations. Sinee mid-year 1996 waveform data from230 teleseismic events haw been stored by this automatie procedure ['l-l].

A real~time filter has been introeluceel into the on-line process of the SIL system, to betuned for detecting harmonic tremor and signals which now are nal ielemifieel automatieally.The continuous seismic signal at the SIL sile stations is banelpass filtered at 0.5-1 Hz, 1-2 Hzand 2-4 Hz anel the l minute mean amplitude is seanned and sent to the SIL center. Visualpresentation of this data gives auseful indication of the multiplicity of activity in real~time.

This data has been ealibrated anel proeedllres elevelopeel to estimate magnitlleles of local eventslarger than magnituele 2, independent of the waveform proeessing [53].

The extension of the SIL system in to the highlanels of Iceland has lead to many problems

8

in the automatic detection and analysis which are gradually being solved. The SIL system wasdeveloped for use in the seismic zoncs. Automatic monitoring in the highlands revealed in manyways new problems because the crustal structure is not as well known and the earthquake sourcesare often more complicated. Much work has been carried out to lower the detection thresholdfor earthquakes in the vo1canic Central Iceland. The new real-time filter mentioned above issignificant for this purpose as well as further tuning of all detection parameters [48].

,To search for time and space patterns in the multip!icity of information in the SIL dataResults have been obtained in using seismological data for mapping active earthquake faultsin the Tjornes fracture zone at the north coast of Iceland (Figure 4). The method used is a

670

66.5 0

r

660

-190 -180 -170

•

..km

30

-160

Figure 4: Mapped faults within the Tjornes fracture Jane off the north coast of lee/and. Blacklines indicate faults mapped with conventional ref/ection seismic methods or by direct observa­tions on-land. Red lines are 60 active fault segments mapped using accurate relative locations ofmicroearthquakes recorded by the SIL network. Seismic stations are denoted by triangles. Darkpatehes are sites of recent volcanism. The depth contour interval is 100 m.

9

mliltim·ent method basecl on cross-correlating similar signals at the the same seismic stat ion[65]. Pro"ic!ec! the time accuracy of the seismic c!ata of the SIL system. close to l millisecond,active faulls can be mappec! with accuracy of the orc!er of 10 meters. Faull plane solutions basec!on spectral amplituc!es of P- anc! S-waves are then usec! as a part of the mapping to reveal thesense of the fault motion. SeveraI speeial fault mapping efforts have been carriec! out relatec! toongoing earthquake sequences in other parts of the country, so graduall,' information on faultarrangement in c!ifferent parts along the plate bounc!ar)' is being eolleclec!.

Resulis have been obtained on basis of ilwestigation of recent high seismie activit)' near theHengill triple j unetion in SIN-kelanel. spatial anc! ternporal variations of aClivity have been stuc!­iec! anc! migration within the area. Fault plane solutians of more than 20.000 earthquakes havebeen stuc!iec! in the area anc! faults have been rnappec! b,' multievent loealion procec!ure (rigure5) [63,6-11.

Introc!uclion of new algorithms imo lhe alerl syslem anc! other evalualions of the SIL systemThe basic option of the SIL seismie system techniques is to use microearlhqwlkes to bring to thesurface information from the source areas of earthquakes. Basec! on c!etailec! microearthquakeanaiysis it is possible to monitor active fallits and mo\·emenls across these, as well as stressesanc! stress changes in their surroundings. The smaller the earthquakes are which can be usec! thecloser \Ve are to continuous moni oring of such features. anc! the more c!elailed information weobtain of the spatial conc!itions. Therefore it is sa signincant to be able to obtain aUlomatically'LS detailec! and secure information as possible [50].

'Nork is going on for introc!ucing ACIS into the automatic procec!ures of the SIL system. ACISis an acronym for "Reducing manual checking by Automatic Correlation of Incoming Signais".As has been shown in the work on multievent analysis for c!etailec! mapping of faults mostseismic events correlate well with each other \Vithin some areas. Work is going on for testing anc!introc!ucing ACIS in the automatic operation of the SIL network in Iceland. A geographicallyindexed database is being created where different classes of earthquakes are stored. As newearthquakes are recorded bl' t.he netIVork, the system automatically looks for similar waveformsin the c!atabase, and if founc!, takes lhe onset anc! the first motion direction picks from there. Jfno existing entry in the database eorrelates with the newevent, the event is checkeel interactivelyb.v the netIVork operators. This approach will improve the accuracy of the alltomatic analysisand reelllce the need of IVork for interactive checking of the elata without loss of lIseful signais.

Preliminary testing bas demonstrated thal the approach described here is possible. It isexpected that lhe first version of the algorit.hm lViII be reaely for aut.omatic routine operationIVithin the SIL system in October 1998, anel thlls as a basis for an enhanced alert c!etectoralgorithms.

'Nork has been carrieel out for stuelying anel renning the alert thresholds for the SIL relateelalert system in Iceland. An alert detector monitoring large amplitudes and backgrolInd noise(tremor) in botb unnlterecl anel filtered baneIs of tbe seismic wavefonn data is operated on allthe SIL stations. It has been tuned for different types of sensors as the SIL system operatesaccOl'eling to neeel with l second, 5 seconds and broadband sensors [9, 10, ~9J.

The continllous seismic signal al the SIL site stations is banc!pass filtered in three channelsand the l minute mean amplituc!e is ca1culated and sent to the SIL center, where the inter­pretation of characteristics of the tremor is canieel out anel linked to the alert system. Thesefrequencies show to be useful in eliscriminating naise of different origin. The 10IVer frequenciesare typical for harmonic vo1canic tremor, while the highest frequency seems to be expressingnoise created by very intensive aclivity of very small earthquakes, although these are not dis­criminated as such. Such an activity is more typical in the approaching of an eruption and mal'passibly be of signincance in the intraeluctionary phase of earthquakes. ~Iuch work remains tobe done to analyze the noise anel halV it is relatec! to other activities of the crustal fOl·ces. This

10

N

64.15°

s

64.1°

/

64.05°

63.95°

km

o 5

I

-21°-21.1 °-21.2°-21.3°-21.4°63.9° ~----.=====l---"""-'=====----~

-21.5°

Figure 5: Earthquakes in the Hengill volcanic area, SW-Iceland. The red lines show the locationand orientation of fault planes estimated from the relative location of earthquakes. The blacklines are mapped surface faults, yellow circles are relocated earthquakes. The inset rose diagramshows the orientation of the 28 faults mapped using accurate relative locations of earthquakes.

11

noise monitoring is al ready no\v llsed to monitor volcanic acivity. The experiencc \Vill also be agood basis for designing a new detector in the SIL system whieh will be "imed at detecting anda..utomatically evaluating "slow eal'thquakes1J (mcaning small earthqllake:'3 \\"ith corner fl'equenciesof the order of l Hz) which are aften observed in Iceland.

Retrieving of da.ta a.nd othcr prepa.ratory \Vorl.; fal' modelling destrllcti\"e earthquakes

\Vork has' started to retrieve seismic data and older models which ha\'e direct relevance formodelling where, when and how catastrophic earthqllakes occlIr in three of the main scismie

risk zones of Iceland. These are the srsz, the TFZ and the seismic lOne close to Heykjavik.Models based on older information from seismological, geological and geodetic information willbe a significant, input to the modelling work based on the new mullidisciplinary approach ofPRENLAB [78,80].

A tentative model of earthquake occurrence based mainly on data on historical and recent

seismicity and older geological data and tectonical interpretations has been set up for the TFZ(rigul'li 3) [81].

Although much of the pl"te motion is probably taken up by " main transform fault it wouldbe wrong to assurne that the e"rthquakes on this fault would follo\v a simple kinernatic modelof stable/unstable strain release as a result of an even plate motion in ilS surroundings. Basedon geological evidence it is indicated that during the last tens of thousands to hundreds ofthousands of years this transform fault only takes up a small fraction of the total plate divergency.

However, the earthquake activity during the last 200-300 years suggesls that presentil' most ofthe transform motion is taken up by this fault. Also it is evident from these data that largeearthquakes repeat in a different manner on the same fault segments.

What is saiel here highlights the significance of earth realistic modell ing of the present dal'general seismotectonic conditions in this zone for lang or short term prediction of hazards.

Evidence from historical seismicity or the lack of evidence in the other main risk zones high­light the same for these zones. We do not see the repetition of "nearIl' identical" hazardousearthquakes, highlighting the significance of earth-realistic dynamic models for these zones.

[Ij Development of methods liS ing microearthquakes for monitoring crustal instabil­ity

The SIL microearthquake system produces dclailed results of autamalie analysis of large num­ber of microearthquakes. To be able to work effieiently with this kind of information a specialinteractive program had previously been created. During the PRENL.-\B project, this programhas now been further devclopeel anel allows now the results from single event location, from

multievent loeation, from fault plane solution, anel from rock stress tensor inversion as input.The program can now be useel for steering results from ane analysis to another, for examplecan the relative locations give constraint on the fault plane orientation which can be used inthe input for both fault plane solutians and rock stress tensor inversion. The development of

this interactive software has also requireel modifications in all other software to facilitate theinformation flow between the different algorithms. The work detailed in this chapter is basicailythe responsibility of Uppsala University, Department of Geophysies (UUPP.DG EO) [26,65,72,73].

Methoels for subcrustal mapping of faulLsThe algorithm for absolute and relative location of microearthquakes has been implemented inthe SIL system routine analysis. This software has been applied in a search of cmstal faults, bothin the TFZ anel the SISZ. This work has been in cooperation with geologists within PRENLABanel these faults, found from the microearthquakes, show a remarkable agreement with the fault

information available from sea bottom anel lanel surveys. These stuelies also inelicate the powerof the multievent location teehnique for discriminating the fault pl"ne anel the auxiJiary plane.

12

An example of the results using this algorithm is shown in Figure 'I.

Methods for monitoring the local rock stress tensorNew methods and software have been developed to estimate a regional or local stress tensorbased only on microearthqua.ke focal mechanisms and locations [72]. The inversion scheme isbased on existing techniques which has been improved along twa major lines of work:

,• The, method takes advantage of the SIL fault plane solution algorithm and is tims able

to handle a range of acceptablc fault plane solutians for each Hent which signifieantlyimproves the stress tensor osLim<ltion (Figure 6) .

• The eruci<ll point of ehoosing which nod<ll pl<lne to inciude in the il1\'€l'sion, i.e. ehoosing thefault pl<lne, has been taekled boLh the trac!itional way, by gooc!ness of fit between observec!slip anc! estimated shear-stress c!irection, anc! with t,m new approaehes. The fault planecan be ehosen as the one least stable, using a simple i\lohr-Coulomb criterion, in the testec!stress regime or with information from the fault mapping SIL syslem algorithrn for absoluteanc! relative loeation. Difficulties concerning the generality ofthe stability algorithm havereeenlly been solvec!. Both new methods show very promising resuIts (figure 7).

The software has been c!evelopec! anc! implementec! in the SIL system context. Tests with syn­thetie data, semi-synthetic (geologieal) c!ata anc! real data from IcelanC! ha,'e been performec! [55].

Methoc!s for monitoring of stable/unstable fault movementsMethods aro in development for monitoring what usually is charaeterizeC! as aseismie faultmovement by use of mieroearthquakes. The commonly observed in leraction between miero­earthquakes, aften over large c!istances comparec! to the earthquake sizes. is most probably re­lat.ec! to defarmation expressed by stable aseismic slip. Utilizing the extensi,'e information carriedby the large amount of microearthquakes has the potential lo find a rock-mechanical connectionbetween microearthquakes during episodie activity. In principal this apens indireet possibilities\.0 aehieve knowledge about the aseismie fault slips. Such an a""lysis mal' be performed bydoducing possible aseismic faulL movements from the microearthquakes anc! vary unknown pa­rameters to put the earthquakes into physieal chains of eflccts anc! consequences. This approachis totally physical (rock-mechanical) anc! can be expectod, together with lheorctical moc!els, todevelop models of fault slip proeess. Such models basec! on laboratory stuC!ies havo alreac!y beenproposed anc! have found gre"t support from numorical modelling anc! comparisans with earth­quake observations [25,27,67].

Methods for monitoring crustal wave velocities from microearthquakesWork has been carried out to finc! 3-D erusU,1 velocity struclure in SW-Iceland from localearthquake tomography. The first results show a very interesting velocity anomal)' at the lowerboundary of the brittle crust which ma)' significantly influence the the moc!elIing of the tectonicsof the SISZ (Figure 8) [84].

o Monitoring stress changes before earthquakes using seismie shear-wave splittingThe World Wide Web is being used to aecess seismic data from the SIL seismie network atIMOIl.DG. Seismicit)' maps demonstrate that a nllmber of stations are sited over sllfficient seis­micity for analysis of shear-wave splitting to be viable. This work is mainl)' the responsibiIity ofUniversity of Edinburgh, Department of Geology and Geophysics (UEDIN.DGG).

Observed stress dependent shear-wave splittingShear-wave splitting is widely observed (figure gl, c!isp!aying lInllsllally large time-delays prob-

13

HESTFJALL,ICELAND

56 FauU Plane Solutions

STRESS TENSOR

Slip angle

0246810MuJliples of20

SELECTED NQDAL PLANES

N

Optimal rps

---l---le

0,68%

0195% • Opt. al

• 0l68% ... OPl.O]

• 0"395

%

wl-+--1f---'

Acceptable rps Slip angleSELECTED NQDAL PLANES

N

•

56 FauU Plane SolutionsSTRESS TENSOR

al 68% S S

0\ 95% • Opt.al• aj 68% ... Opt. aj• O 2 4 6 8 10o] 95% Multiplcs of20

wt--t--t-

Figure 6: Example of stress tensor inversion using 56 microearthquakes in the SIL a'·ea. Thefigure shows lowe,' hemisphere equal area prajections, in the left column are the resulting principalstress directions with 68% and 95% confidence limits and the optimal solution marked by a blacksquare, al, and a black triangle, a2. Deviation is the average angular misfit for the optimalsolution, R = (al - a2)/(al - a3), is the relative size of the intermediate principal stress. Theblack rose diagram around the circle is the 95% confidence limit for the direction of maximumhorizontal compression. The right column shows the nodal planes that the inversion algorithmpicked as fault planes. The plus signs are the poles to the individual planes and they have beenoverlayed by a K amb contour diagram. The upper row shows the result when only the optimalfault plane solutians (fps) were included in the inversion and the lower row is the result whena range of acceptable fps were used for each event. We see that although the confidence limitsfor the stress directions do not change much the deviation decreases dramatically when usingacceptable fps. There is also a large difference in the chosen fault planes. Jn the optimal fpscase the algorithm picks NW-SE striking subvertical planes and a large number of subhorizontalplanes. The acceptable fps case on the other hand picks mostly NE-SW striking subvertical planesand fewer subhorizontal planes [72}.

14

HESTFJALL,ICELAND

56 Fault Plane Solutions

STRESS TENSOR

Slip angle

SELEC'fED NQDAL PLANES

Deviation: 3.2 deg N N

wf-+--t----i

0"1 68% S

0", 95% • °PI.O",

• o) 68% .... Opt. 03

lB 0"3 95%

E w

s

0246810Multiplcs of20

E

56 Fault Plane SolntionsSTRESS TENSOR

InstabiJitySELECTED NODAL PLANES

Dcviation: 3.8 deg N N

Ew

• 0"\ 68% S S

0"1 95% • oPL al

• a) 68% .... 0pLa) o 2 4 6 8 10• a) 95% Multiples of20'

wf-+--1-1~c--+--t-1 E

Figure 7: Same 56 mieroearthquakes as in the previous figure but now inverted with differentnodal plane pieking algorithms. Both examp/es in this figure have been inverted with aeceptablefault plane solutions (fps). The upper row is the same as the lower row in the previous figure andused a plane selection eriterion based solely on the fit of the theor'dieally caleulated shear-stressdirection to the direction of observed slip on the p/ane. Whiehever p/ane had the smallest misfitangle was ehosen. In the /ower row a more physiea/ eriterion was app/ied. The p/ane with /oweststability, caleulated using a simple Mohr-Coulomb failure eriteria, was chosen as the fault plane.This eriterion gives a s/ightly higher average deviation, but as can be seen in the figure, muehm.ore well defined prineipa/ stress direetions. The direetion of maximum horizontal eom.pressionis likewise more weU defined and the stress regim.e is now more deeisive/y strike-slip. The ehosennodalplanes are preferably NE-SW striking subvertieal planes for both algorithms although theinstability method pieks them more frequently and dDes include very few subhorizontal p/anes {72}.

15

20mI I

P-Wave Ve10city (kmls)

10m

6.5 6.6 6.7 6.8 6.9

Figure 8: 3-D P-wave veloeity at 10 km depth in the erust beneath SW-Ieeland estimated fromloeal earthquake tomography.

ably eaused by high temperatures and/or high pore-fluid pressures both of whieh may well bepresent in the upper erust in Iceland (Figure 10) [26,28,29,30,31,32,33,34,35,36,37,54,89,90J.

The remarkable observations are that, the first 200 days of the first SIL seismie station exam­ined, which was the station SAU in the SISZ, showed variations in shear-wave splitting similarto those observed before earthquakes. Such behaviour can be interpreted (and numerically mod­elled) as the effects of increasing stress on the stress-aligned intergranular microcracks presentin almost all rocks. These variations at SAU were reported at the internal PRENLAB workshopin Reykjavik, September 10, 1996, but since this was the first data analyzed from SIL, the inter­pretation was not eonfirmed. In fact, SAU is about 160 km WSW of the eruption on September30, 1996, beneath the Vatnaj6kull iee cap. Since the eruption more data have been analyzed andit is planned to better constrain the observations before the eruption by more observations alsoin that time period. Figure 11 shows the different behaviour of shear-wave splitting betweenray path directions 0°-15° and 15°-45° to stress directions. It is suggested that the changes inshear-wave splitting at SAU were the result of increasing pressure as magma was injected intothe lower crust for some five months before the eventual eruption [62,79].

The observed variations at SAU suggest that shear-wave splitting may be a very sensitivemonitor of current stress changes, and does not depend on the complicated interactions in earth­quake preparation zones. It is very encouraging to see the effect of changes so early in theproject.

16

SIG GRl

flj~

9/9&4197

9196-2/97

1/97-4/97

9196-2197

9196-4197

HAU

9196-4197

9196-4197

Figure 9: Map demonstrating ongoing work in studying shear-wave splitting polarizations belowseismie stations in leeland, where seismie data are available with angle of incidenee within theshear-wave window, during September 1996 - April 1997.

• They indicate that shear-wave splitting has the potential for monitoring the detailed stressbehaviour in Iceland.

• They indicate that Iceland is an active naturaliaboratory for research on earthquake sourcezones and vo!canic manifestations.

• They suggest a number of new projects to improve stress-monitoring of the crust beneathIceland.

Identify optimum areas and developing routine techniquesThe ongoing work of will gradually allow the identification of suitable areas for deployment ofmore closely spaced SIL stations for more effective studies of precursory changes. Development isplanned of routine techniques that can be used for real-time monitoring of splitting parametersin Iceland [35].

17

Example of shear-wave splitting from statlon SAU

p

SLOW(N3WE)

SI

U90 ms

Date: 1996 July 10, 0728:30.4 ± 0.12Depth: 9.8 ± 1.6 km Epicentral Distance: 5.0 ± 0.9 kmAzimuth: N172°E Magnitude: 0.6

Figure 10: Three-eomponent seis11l0grams of an earthquake (at 07:28:30.-1, July 10, 1996) atSA U 1Vith time 11I01·ks at every second. Upper traces are iV-S, E- W, and uertieal components.The lower traces are horizonta! eomponents rotated to the faster (iV225' E) and slowe,' (N315° E)shear-wave polO1'izatiol1s showing time-de/ays between spid shear-waoes.

@] Borehole monitoring of fluid-rock interaetionGeophysical loggingsA pilot study is angoing to obtain a time series of lags in the SISZ. An 1100 m deep borehole(11-03, "Nefsholt") inside the zone (63.92°N, 20.41°W, 7 km south of the seismic station SAU) isused and provides the unique opportunity to perform measurements much nearer to earthquakesources than usual. Hypocenter depths at that location range between 6 and 9 km. iVloreover,data can be obtained for a depth interval of more than 1000 m, uninfluenced by the sedimentarycover and less disturbed by surface noise.

In the preparational phase of an earthquake, stress accumulation is expected to be connectedwith the creation of borehole breakouts, changes in the number anel size of cracks, a possiblevariation of the stress elirection, etc. Therefore, the following set of geoparameters is monitored:

Temporal changes visible in these lags \Vill be correlated with elata obtained by other meth­ods used in the whole project, seismicity, anisotropy observeel in S-wa\"es, crustal defarmation,gravity, etc.

There are twa main objectives: The first is to measure stress ineluceel changes in physicalparameters of rocks by repeated logging. The measured parameters are the acoustic velocityand the coneluctivity of the rock surraunding the borehole as well as the geometry and degree offracturing of the borehole wall. The seconel aim is to get supplementary information about the

18

Time-delay Variations at Station SAUN N

I31 May 97

I- __ e

+ l I

~ I +To O~":30~":60~":9'::-O~120':'--':c150':'--:-:'8ll':'--:-:21'=O~2:"40:::"-'2:::'7O::0~300:::":~33:::":O~3:::'60O::-3='90'

time (djYS after 01 May 1996)

30 Sep(slert of enlplion)

I01 May 96

E 20~§.

~ l l

~ t !-~ 10 -- ,

~, +g ,

Figure 11: Variations of shear-wave splitting at SA U for 390 days from May 1, 1996. Polarequal-area maps out to 45° of left shear-wave polarizations, with rose diagram indicating averagedirection, and right circles scaled to normalized time-de/ays (ms/km). Variation with time ofnormalized time-delays for ray paths in bonds with incidence 00 to 15° to the crack face (sensitiveto crack density), and for ray paths in bonds with incidence 15° to 45° to the crack face (sensitiveto aspeet-ratio). Lines are least-squares fits befare and after the eruption. Dashed lines in thelower plot repeat lines in the upper ploto Ereor bars are approximate.

19

stress fieIei anel its changes insiele the SlSZ by the eletection of borehole breakouts.T'his work is mainly under the responsibHit.v of GeoForschungsZentrunl PotscIam} Division

.5-Geomechanies anel Management of Drilling Projects, Section 5.3-Roek Meehanies and StressFielel of the Earth's Crust (GFZ.DR.DBL). Besieles the other partners of the PRENLAB thereis a speeial cooperation with Valgarelur Stefånsson at the Icelanelic Energy Authority in this work.

Equipmen,t anel methoelsThe following tools are used: the borehole compensated sonie tool (BCS) to measure the P-wave"elocity of the rock, the elual-ineluction-Iaterolog (OIL) to get an estimate of the resistivity inthree different penetration elepths, the gamma-ray (GR) and spectral gamma-ray (SGR) tools tohell' evaluating the lithological sequenee, and the borehole tele"iewer which provieles a completeimage of the borehole wall. The BCS also allows registration of the whole wavetrain which ishelpful for fracture ielentification and elistinguishing the elifrerent phases of al'l'iving waves. Withthe BCS, the OIL anel the GR severai ru liS are performed one immeeliately after the other. Af­ter a earefu! depth mateh these runs are a"eraged. This proeeelure is repeated after elifferenttime intervals anel is inteneled to be continueel. So far, two repetitions have been performed,the first after a time interval of three months, the seeonel approxiinately one year later. Theborehole televiewer allows eleteetion of borehole breakouts, whieh are stress induced elongationsof the borehole indicating the orientation of the lesser principal horizontal stress which is per­penelieular to the greater horizontal principal stress. Aelditionally, fractures can be pickeel indepth a.nd azimuth and c10seel fractures can be distinguished from open ones. Thus repeatedtelcviewer-logging allows to observe changes in borehole geometry anel in apen ing of fraetures [61].

ResultsConcerning the repeateel logging it can be saiel that the repeatability showed to be extremelygood. Significant changes in the measured curves are not due to instrumental errors and can beassumed to be real. Some changes in the P-wave velocity have been observed between last yearand this year. The same is observeable in the DIL-Iogs.

The breakouts founeI in the borehole at Nefsholt (Figure 12) seem to be in agreement withthe secondary stressfield in this area found by the geologists as eIescribed in section 6 (NNE-SSWto NE-SW extension). The analysis of fractures shows open, mainly steeply dipping fracturesOl' bedeling interfaces with an average strike of 4.)0. This coincieles with the strike of fraeturesmeasureel at the surfaee. Examples of televiewer elata for a fractureel borehole-section anel aninterval eontaining breakouts is shown in Figures 12 anel 13.

These primary results are examined further by carefully combining the available elatasets andevaluating the theol'etical background by moelelling.

Radon reIated to seismicity in the South IceIand seismie zoneThis work is carrieel out at the Division of Geophysics, University of Icelanel (U/CE.DG).

Build an improveel LSC apparatus to measure the raelon content of water and gas samplesThis \\'ork is in progress and funds have been secured to appl)' it for regular monitoring in theSISZ [82,83].

Revive the radon sampling program in South IcelandA radon sampling program started in the SISZ in 1978, anel samples \\'ere collecteel on aregularbasis for 17 years. This monitoring will be reviveel with improveel techniques, by using theimproved LSC apparatus which is being developeel. As a part of reviving the program an overvie\\'inl'estigation has been earrieel out on observeel interrelationshil' between seismic aetivity anelradon signais. The main results of this can be summarizeel as follows:

20

o(}epth9775

•978.0

978.5

979.0

979.5

980.0

9805

981.0

9gj5

982.0

90

AmJ)lituile

81 161

180

241

270 o

1.8

90

RllidilJS

9.8

180

17.8

270

Figure 12: Televiewer'logs run in bordwle Lrr03 in a depth interval containing breakollts.

• Of all anomalies that eould be expected from the magnitude-distanee selection eriterion,24% were actually cletected.

• 35% of all measured anomalies are related in time to seismicity.

• 80% of earthquake related anomalies are positive.

• If a positive anomaly is detectecl, there is 38% probability of an earthquake following it.

• Anomalies were detected before 30 events out of 98 possible events. There is thus 31%prob­ability of a measured anomaly before an earthquake, that fulfills the rnagnitude-distaneecriterion.

• For most of the events the associated raclon anomaly was cletected at one station onIy. Fiveevents were preeecled by anomalies cletected at two stations, one event at three stations,and one event at five stations.

• The sampling sites are not equally sensitive. The sensitivity is related to the loeal geo!ogicalformations. The statistics can be improved consiclerably by eliminating stations with lowsensitivity.

21

BHTV LL03 - JULI 96

oClfpth

4lH.O

,4G4.5

'!-OH

405.5

<406.0

406.5

"01.0

401.5

'00.0

i08.5

411!l,O

4095

'10.0

AmplitlJlfa61 161

~IJS

9.64 17.84

Figure 13: Televiewer IOg8 run in borehole LL-03 in a jractured section.

• Radon anomalies were detected that seem to be related to eruptions of the neighbouringHekla volcano. Eight anomalies were found, five of which occurred prior to the eruptions[51].

[I] Active defarmation determined from GPS and SARThe objective is to measure crustal deformation, understand how it relates to seismicity and dis­tribution offaults, and use it for better understanding of earthquakes. In particular, deformationmonitoring will be llsed to improve the understanding of elastic strain accllmulation, tectonicsetting of seismic zones, coseismic slip during earthquakes, and aseismic slip on faults duringinterseismic periods. The work is in first hand carried out by Centre National de la R.echercheScientifique, UPR. 0234-Dynamiqlle Terrestre et Planetaire (CNRS.DTP) and the Nordic Yol­canological Institute (NYI) in special cooperation with Uppsala University (UUPP.DGEO), Uni­versity of Iceland (UICE.DG) and Icelandic Meteorological Office (IMOR..DG).

Analysis of SAR images from the ERS-l and ERS-2 sateIlitesTechniqueIn applying the SAR inteferometry the entire deformation field is recorded with an unsurpassedspatial sampling density (~lOOO pixeIs/km2 ) at intermittent times. For the interferometry workwe have used data acquired by the European ERS-l and ERS-2 statellites. They pass over a

22

gi"en study area at an altitude of 785 km: transmitting along ray paths at an avel'age angle of23° from the vertiea!. These satellites proviele SAR images, each of whieh is a map of the grounelreflectivity sorteel by range. The phase of e<tch <1 by 20 Ol pixel mcasures both thc range anelthe phasc shift duc to reftcction of the wa"c from thc grounel surfacc. The l<tttcr quantity canbe eliminatcd betwccn twa images of the same <trea if the elielectric characteristics of the grounelremain constant anel the orbits satisfy the conelitions necessary for coherence. The remainingpath differ€nce, known only to within an intcgcr nlllllber of wavelengths. canta-ins informationfrom three sources: relative orbital positions, topography as seen in stereo by the satellite from81ightly eliffcrent orbital passes, anel any change in position of the grounel. from a suitable p<tirof images, we l'econstruct the phase of each pixel using a phase-preser"ing cOl'relator. \Ve adjlIstthe satellite orbital parameters to minimize the num ber of fringes at the foul' corners of theimage, assuming that the far field elisplacement is negligible. The stereoscopic path difference iseliminated using a eligital eievation moele!. Thc rcsulting interferogram is a contour map of thechange in range, i.e. the component of the elisplacement which points towarel the satellitc. Eachfringe torresponels to ane cycle (28 mm or half the 56 mm wavelength) of the ERS-l SAR. Theaccuracy of the llleaSUl'ement is bet ter than sc"eral cm in range.

~Iain results

• We have demonstrateel that satellite radar interferometry can be useel to measure platemotion and acclImulation of strain in seismic zones: under favorable conditions.

• We have aelvanceel the unelerstanel of earthquake triggering in voleanic areas.

Satellite raelar interferometry has been applieel to map the satellite-vie,,: component of a crustal"elocity fielel, as well as volcano defarmation, at the Reykjanes Peninsula in SW-Iceland. Thearea is the elirect anlaneI structural eontinuation of the submarine Miel-Atlantic Ridge. Obliqucspreading between the :"orth American anel Eurasian platcs of 1.9 cm yr occurs there, causingboth shearing and extension across thc plate boundary. Using ERS-l images from thc 1992-1995periDd we have formed interferograms, spanning up to 3.12 years. Interferogram spanning 2.29yea.rs and a moclel referring to that time period is displayed in Figure l-l.

Coherence is presen-ed: and time-progressive fringes caused by (rustal cieformatioll are ap­parent. The most ob"ious defarmation is time-progressive defiation of the Reykj<tnes central"olcano, averaging to l·) mm/yr, probably caused by compaction of a geothermal reservoil' inrcsponsc to its utilization by a power plant. The deflation we infer is in good agreement withleveIling elata. This gh'es confielence in the interpretation of more subtle eleformation signal inthe interferograms, fringes aligned in the elireetion of the plate boundary causeel by plate bound­ary eleformation. Relying partly on geologic evidence we assumc the shape of the horizontaland vertical crustal velocity field. We estimate best-fit made! parameters by maximizing theglobal eoherence of the residual interferograms, the difference between observed anel model in­terferograms. The elata constrain the locking depth of the plate boundary to be about 5 km.Below that leve! the plate maveOlents are accommodated by continuous eluctile defarmation, notfully balanced by infto\\" of magma from depth, causing about 6.5 mm/yr subsidence of the platebaunelary. Previous regional geodetic elata agrees with this interpretation [85,86,69].

Another interferometry study of the Krafta volcano in North lceland has clearly demon­strated the usefulness of raelar interferometry for volcano monitoring. In that study we eletectedthe reaeljustment of the the Krafta spreaeling segment, to angoing postrifting readjustment of thespreading segment to rirting episode from 1975 to 1984 [71].

GPS geodesyIt measures relative position between stations using signals transmitteel by the satellites of theGlobal Positioning System (G PS). The technique has been used in lcelanel since 1986, but only

23

A

B

Figure 14: SAR study of the Reykjanes Peninsula, SW-Iceland. lnterferogram 2.29 years (A)and model interferogram showing best-fit simulated 2.29-year deformation (8). Time-progT'essivefringes appearing consistently in the interferograms are indicative of erostal deformation. Phasescoded into bytes (8 bits) are represented with a false colaur tab/e. A complete colour cyete, forexample from blue to Mue, represents ane complete fringe; a 28-mm change in range in the caseof ERS. Concentric fringes are located at the Reykjanes central volcana and manifest a time­progressive increase in range to the satellite. An increase in range along the whole plate boundary,indicative of spreading, is visible as a central fringe in the 2.2g-years (A).

as intermittent measurements perforOled every summer. We have instal!ed a semi-continuouslyrecording GPS station in the SISZ to monitor deformation there in real-time.

Since July 1994 an unusual!y persistent swarm of earthquakes (M<4.0) has been in progressat the HengiIl triple junetion, SW-Iceland. Activity is clustered around the center of the Hr6­mundartindur vo1canic system. Geodetie measurements indicate a few cm uplift and expansion ofthe area, consistent with a pressure source at 6.5±3 km depth beneath the center of the vo1canicsystem. The system is within the stress field of the South Iceland transform zone, and majorityof the recorded earthquakes represent strike-slip fauiting on subvertical planes. We show thatthe secondary effects of a pressure source, model!ed as a point source in an elastie halfspace,include horizontal shear that perturbs the regional stress. Near the surface, shear stress is en­hanced in quadrants around the direction of maximum regional horizonta! stress, and diminishedin quadrants around the direction of minimum regional stress. The recorded earthquakes showspatia! correlation with areas of enhanced shear. The maximum amount of shear near the surfacecaused by the expanding pressure source exceeds l millistrain, sufficient to trigger earthquakesif the crust in the area was previously dose to failure [68,70].

24

@] Formation and development of seismogenie faults and fault populationsTids \I'ork includes extensive geological field work and related theoretical work based on tint,concentrating on the SISZ and the TFZ. It also includes geodetic investigation and comparisanof results and methods with seismological il1\'esligations. The partners mostIl' respansible forthis work are the Nordic Volcanologicallnstitute ("VI) and the Centre "ational de la ReehereheSeientifique, URA 1759-Tectonique (CNRS.TT). There is a \-er)' close eollaboration with all theother part'ners in this work, espeeiaIll' with the liniversitO' of Bologna (l'BLG.DF), the lcelandie~'leteOI'ologieal OffIce (IMOn.DG) and the lippsala University (UUPP.DGEO). Also, there iseooperation with Thierry Villemin of Universit'; de Savoie and Olivier Dauteuil, GeosciencesRennes, in france.

Paleostress tensorThe paleostress tensors of 2.5 localities in the SISZ have been calculated, u,ing fault-slip datasets,and cOlilpared to focal mechanisms of earthquakes in the SIL dataset. Preliminary studies of thepa.leostress field in the TFZ were carried out during the summer of 199, [:39].

Stress fields and mechanisms of seismogenic faultsIn the SISZ, the recent and present-dal' tectonic mec!Ianisms are studied by geological andgeophysical means. The determination of the stress regimes is done through calculation of stresstensors, involving the use of inverse methods. Data inversion is applied in a nearIl' identical wayto fault slip data and to sets of double-couple foeal mechanisms of earthquakes.

The methodology applied includes the following foul' geological and geophysical criteria: (l)the consideration of the nodal plane attitude (strike and dip), as compared with that of allgeologiealfaults and planes of mechanical \I'eaknesses known from field observation at the sites,geological mapping and aerial p!Iotograph analyses; (2) the comparisan v:ith the recent faultmec.hanisms observed in the field (including pitch and sense of motion as well as strike and dip offault); (3) the mechanical probability of either of the twa fault plane solutians, based on takingfriction into aeeount (e.g. a shallow dip is more likely than a steep ane for a reverse fault); (4·)the best fit criterion rela.t.ive t.o the st.ress t.ensors calculated. Jf ext.ernal (i.e. comparisans wit.hgeological t.ensors), il is of good value. Il' internal (i.e. the nodal planes resulting in the best.possible fit. wit.hin the data inversion process). it is some\l'hat circular, hence disputable.

The systematic use of the foul' criteria for nodal plane selection within a weight.ed approach al­lowed more aecurate determination or the stress regimes in the SJSZ than earlier st.udies and thusbet.ter understaneling of t.he geological significance of the eart.hquake mechanisms [1,2,4,20,87].

Determinat.ion or t.he paleostress tensor in t.he SISZIn t.he SISZ present-dal' tectonic activity is mainIl' associated with a conjugate system ofNNE-treneling right-lateral strike-slip faults anel ENE-trending left-Iateral anes; NW-trendingfaults are also present. All these faults affect basaltic lavas anel hyaloc1astites, Upper Tertiary­Pleistocene to Holocene in age. In the SISZ we collected about 700 brittle tectonic data at 25sites. The major stress regime, " NW-SE treneling extension, inc1ueles 70% of the total popu­lation of fault.s, whcreas the minor ane, a NE-SYV t.reneling extension, inclueles 30%. Of a totalof 718 fault slip elata, 55% indieat.e primarily normal-slip and 4.5% primarily strike-slip. Theratio normal/strike-slip faults is lower than in other areas in Iceland. The above results indicatethat the dominating stress field in the SISZ favours strike-slip faulting, ",ith an horizontal 0'3

axis treneling approximately vVNW-ESE to NW-SE. We point out, hO\ye\'er, that in addition,there is a cont.rasting miner, stress field, characterized by approximately J\NE-SSW to NE-SWextension. These results were compared to the stress regimes determined from eart.hquake focalmechanisms. This comparison reveals t.hat the paleostress and stress regimes are ielentical and

25

that the ehanges in the stress field of the SISZ are eharaeterized by stress permutations betweenal and a2 and between al and a3 [3,18,19,20,42,871.

Oetermination of the paleostress tensor in the TI'ZDuring the summer of 1997 a very detailed study IVas made of the paleostress field of the TI'Z.The study was earried out at severnJ sites within the on-land part of lhe fraeture lOne onthe Flateyjarskagi Peninsula, as well ns at severai sites to the south of the zone itself, on theI'lateyjarskagi Peninsula. In addition, fault-slip data sets IVere eolleeted along the eastern partof the Oalv!k Iineament.

Preliminary results indieate a rather camplex stress histor,. of the TFZ, partieular1y in thcin-land parts on the north eoast of I'late.rjarskagi. The datasets are eurrently being analyzed inmore detail, and it is hoped that the final results \\'ill be ready in earl,. 1998. The paleostressdata from the Tjiirnes fraeture lOne and the Dalvik lineament lViII be compared with the eurrentseismie dnta from these areas and with results of geodetic IVork and other ongoing geologiealinves~igation ongoing in the area [16,17,411.

Field and theoretical studies of fault populationsOetailed study of faults in the TFZ and the SISZMa.ps have al ready been made of somc of the most important faults in the Holaeene part of theSISZ.

Oetniled field data on faults in the Pleistoeene part of the SISZ have also been eolleeted.An analytieal study has been made of the controlling stress field and the secandary fraeture

pattern of the best-exposed faults in the Holaeene part of the SISZ.As regards the TFZ, ficld studies wcre made in thc summer of 1997 of the infrastrueture of

the on-land parts of that lOne at the north coast of the Flateyjarskagi Peninsula. This workfoeussed on the overall evolution of the fault rock from large-seale blocks, bounded by majorfaults, to fault breceia and erushed rock. The results indieate that many parts of the TFZ eontainzones of eompletely erushed rock where the erustal movements may be mostly by aseismie ereep.

Another aspeet of the work on the infrastrueture of the TFZ eoncerns the effeets of erustalf1uids on faulting in that zone. The results indicate that there are sets of very intense mineral­ization that presumably formed in areas of transtension along the main strike-slip fault [38,40,85).

Oetailed teetonie map of the TFZA detailed set of G PS points within and in the vieinity of the TFZ was installed in 1994 by theNV!. This network was extended in 1995 and again in the summer of 1997 by NV! in eooperationwith Villemin and his group. Altogether there are now 44 GPS points in this area. In 1997,32points were remeasured, as well as the 12 new points. The results from 1997, eompared with1995, are not .ret available but should be so in early 1998.

Mapping of the whole Krafta fissure swarm has also been macle and a cletailed map of thejunction between the Hllsavik fault and the rift lOne is under the way. A SAH. study is plannedof the TFZ and its junetion with the rift lOne [88).

Boundary-element studies of the TFZNew general models for the erustal deform<ttion and the distribution of seismieity within the TFZare being developecl, cambining the results of analytieal <tnd boundary-element studies with thefield results.

Analog models of the the TFZThis work is foeussed on l<trge-se<tle Holoeenc erustal defarmation in the area south of the TFZusing gl<teio-morpho!ogic<tl data. The purpose of this work is partly to try to deteet the Holaeene

26

crustal eleformation in the area that couleI be relateel to the controlling stress nelel of the TFZ [24).

[Il Theoretical analysis of faulting anel earthquake proeessesThis invoh·es moelelling of earthquake anel rifting related space and time behaviour of the stressfield in Iceland. The modelling work is based on results of geological, seismological and geodeticinvestigations. The main responsible partners for this work are Department of Physics, Univer­sity of Bologna (VBLG.DF) and GeoForschungsZentrull1 Potsdam (GrZ.DR.DBL). This workis based on the work of all other partners.

Crust-mantle rheology in Iceland and Mid-Atlantic Ridge from studies of post­seismie rebounclHere the Imrk is mai nil' fOClIsseei on modell ing the stress and the displacement nelds due toearthquakes and to rifting episoeles in Iceland. Emphasis is posed on the interaction betlVeen thetwa proeesses follolVing stress relaxation in the asthenosphere. From these studies we mal' obtaina signiflcantly improved unelcrstanding of t.he space and time relationships betlVeen earthcruakesanel other geophysical phenomena governing the stat.e of stress in the erusl.

Global st.udies of post-scismie and post-rifting rebound in Iceland lVere performed employingspherical, radially st.ratined Earth model [57,.58j. Earthquakes and rifting episodes are modelledin t.erms of suitable distributions of equivalent body forces. The method of solution is based ona speetrai approach to the equations which govern the defarmations of a spherical Earth due toseismie sources loeated within the erust. The method, has the advantage of including a real­istic manlIe layering and il self-consistent deseript.ion of t.he gravitat.ional effects. Post-seismie<lnel post-rifting deformat.ion are proved to be signincant transient components of plate motion.Studies on a local scale <tre performed employing the theory of elastic disloeations in layered me­dia. Comparison is cOi1stantly made with observations obtaineel in the framework of structuralgeology [12,14,15). Near fielel studies on the stress !leid induced by ridge activity are performedemploying original methods of theoretical fraeture mechanics in plane-strain connguration. Thesingular integral equations governing fault and ridge dynamics are solved by means of suitablepolynomial expansions, yielding a linear inverse problem which is solved by standard numericalmethods [13,21J.

Far-neid displacement !leleI following rifting episodesl3y means of a viscoelastic model we have computed the defarmations associated lVith the dy­namics of a spreading ridge in aspherieal, rheologically str<lti!led Earth. A simple Earth modelis preliminarly employed, which includes a lOa-km thiek elastie Iithosphere, a uniform mantlelVith Maxwell rheology, and a fluid inviscid core. The source of defarmation consists of a 200km lang tensile fault buried at a depth of 50 km. Figure 15 portrays the coseismic surface dis­placement u (in centimeters) observed at a given distance from the fault along different azimuthsalpha (namely, 0°, 45° and 90° from tap to bottom). The surface displacement is decomposedalong the sphericalunit vectors ,·,8, and </I (dash-dotted, solid, and dotted curves, respectively).Figure 16 shows the lang term relaxation of the displacement !leid. The time-scales governingthe transition from eoseismic to postseismie displacements depends essentially on the viscositystratification of the mantle. For an upper mantle eharaeterized by a relatively low viscosity(such as the mantle beneath lceland) these time-seales amount to a felV years. Large amounts ofrelaxation mal' affect all of the components of the displacement !leid. In particular, we observeamplincations by a factor of 2 for the 8 and T components of displacements. Another interestingfeature of Figure 16 is the large spatial scale of the region experiencing horizontal motions in thepostseismie regime [6].

Stress changes indueed by magma up rise in a layered medium

27

I

Coseismic Displacement (cm)2S .--- ,

·75 0.:::;0 0

.. 00 ," '" "00 10000

50

25A 8t

.•..::..•...2S

,·50 A

.1$ r a=45°.'00 , 00 '" "00 10000

75

50

2S

·25.,,-- .. _.. -

·75

..00'L~~~~u_~_~~~l 10 100 1000 10000

Distance (km)

Figure 15:

Postseismie Displacement (cm)

2Sr-----------,

----------~---..-~==-..-.-,

·50

·75

"

a=900

"00 L..~~~~c..c__~__-'

l 10 100 1000 10000

50 .---A".---.,.-----,$ .•.. A../ \ e

E.-_.~...."'....~ \_._.; ..:.:~.~._...

,A

ra=45°-

" '" 0000 ooסס,

75,-- ---,

"-----------; ...--

.2S

·50

·75.'00 '-__~_~~_~_~...J

l 10 100 1000 ooסס1

Distance (km)

Figure 16:

28

4

/

2

--/ /

/ "',. .t" "

, '{

-2

---......... >-..--,

. ... ........ \

"\\\\' I { ...-

\ ' I I .. \ I .

.... "'0 \ \ I I ....'\, "" \ ~ i I ,: / //~ z':,.-:-1.5 km ~ \1 ,,/ //

........\~=-o-.{i kro . '. .,/ /\. '. i i ~ \. ..,/ /

z .....-O.l laD I' l ./ /\. "'0 / j I '\ - i

z= ~.1 1':01 ",,- .• ,.' it" ../ /\ ' l i

z= o.~. km I • /; l ,-

Z= 1.5\,m r ~ .I-8 L-~~_~--L~~~_'--"--,--,-\~~_~-.L.1_~__~..J

-4

'"~b(fJ -4

"~§N'':.2 -6

---

Figure 17:

Magma ascent throllgh a mid-oceanic ridge can be modeled as a tensile crack within which theoverpressure is determined by magma blloyancy. Explicit analytic sollltions are given for theelementary dislocation problem in the most simple layered medium, made up of two welded halfspaces characterized by different elastic parameters. Particularly interesting appears to be thecase of a crack cutting across the boundary between the two media. In lhis case a further singll­larity appears in the kernel of the integral equation, which is connected with the presence of theboundary surface. The problem can be solved by splitting the crack into two interacting opencracks. The application of a constant overpressure within the crack is found to produce c1rasti­cally different stress regimes in neighbollring regions located on opposite sides of the interface;this feature may provide a straightforwarel explanation for the episodic reversal (from sinistralto dextral) of strike-slip mechanisms observed in the South-Iceland seismie zone. If the rheo­logical discontinuity betwecn the lithosphere and the asthenosphere is considered, moelel resultspredict a much larger horizontal flow in the asthenosphere than is accomplished by motion oflithospheric plates. Furthermore, the stress field near the transition eleplh is strongly controlledby elifferential shear flow in the asthenosphere, tlllls yielding a simple explanation for the differ­ent stress regimes prevailing in the seismogenic zones of Iceland. Figure 17 shows graphicallya result of great interest in interpreting Iceland seismicity: if a dyke 4 km long cuts across theinterface between the half-space z>O, with rigidity la times greater lhan the half-space z<O,the horizonta! compression in the stiffer medium may be much higher than magma overpressure(5 MPa) in the magma [22,231.

Surface cracks in fault regions as indicators of the state of stressMost earthqllakes in the South Iceland seismic zone occur on NNE-trending dextral and ENE­trending sinistral strike-slip faults. Many of the earthquake fractures rupture the surface inbasaltic (pahoehoe) lava flows of Holocene age. The resulting rupture zones display complex en­echelon patterns of secondary structures including arrays of (mostly) NE-trending fraetures andhillocks (push-ups). The field data indicate that the arrays consist of both mixed-mode cracksand pure mode-] cracks, concentrated in a narrow belt trending in the elirection of the strike­slip faults in the Pleistocene bedrock bllried by the Holocene lava flows. For the dextral faults,

29

,

,Fa111t slip 1'--1

Figure 18:

the angle between the strike of the fault array and the strike of individual secondary fracturesranges over severai tens of degrees, but is com mani)' 10°-30°. Modelling indicates that if thearrays consist of pure tension (made-I) fractures, the angle between the strike of the hidden faultand each tension fracture must be between 22.5° (if the prestress dominates with respect to theseismic stress) and 45° (if the prestress is negligible), assuming that faulting occurs according tothe Coulomb-!'Iavier failure criterion and the prestress is purely deviatoric. If the arrays consistof mixed-mode cracks, the angle between the fault str'ike and individual cracks is lower than22.5°, this value being attained if the seismic stress dominates (Figure 18). Modelling suggeststhat all fractures, being narrowly concentrated near the fault strike, fonn as a consequence ofslip-induced local stresses during major earthquakes, small angle fractures being predominantlymixed-mode cracks, while higher angle fradures mal' be pure mode-I cracks. The role played bythe regional prestress field is found to be significantly dependent on the rigidity contrast betweenthe shallow layer and the basement rock. Useful inferences on the regional stress field can beextracted from such modelling [12,14,15J.

Comparison between fiat and spherical Earth modelsTwa different approaches were compared to the study of post-seismic deformations. In the firstone, we considered a fiat earth model fOlTed by a vertical strike-slip fault embedded in an elasticlithosphere. In the second ane, we have solved the same problem in spherical geometry. Inboth cases, we have computed the coseismic displacements and the delayed post-seismic dis­placements associated with the viscoelastic relaxation of a uniform mantie. In Figure 19 wecompare coseismic (Ieft) and postseismic (right) horizontal displacements computed accordingto the spherical model (dashed lines) with those predicted on the basis of the fiat ane (solidlines). The twa tap panels refer to moderate source-observer distances (i.e., O < d < 120 km),whereas the far-field respanses are portrayed in the bottom panels, with 120 < d < 4000 km. Inthis case study we have employed a vertical strike-slip fault source of width W=50 km, whichbreaks the lithosphere-mantie boundary, located at a depth of 100 km. As expected, there is adose agreement between the two models in the coseismic regime for moderate source-observerdistances (O < d < 120 km, top left panel). In the postseismic regime (top, right) the sphericalmodel predicts a displacement which sensibly differs from the one obtained by means of a fiatmade!. Differences between spherical and fiat models are particularly large in the far field (bot­tom paneis) . An analysis similar to that performed in Figure 19 has also been carried out onthe stress fields induced by a strike-slip earthquake. Significant corrections to both the time-

30

Coseismic

Displacements (cm)

Postseismie60 l»

......lO 1110

. ...

" so30 60

'" ,O

10 '"o'" ,O 60 eo 1110 1'" o '" ,O 60 80 100 1'"

" 140

35 h 1'"......... -. ......

30 100 spherieal

" l. SO

'"1560

10 'O Ilal

, ....... ..... _- .. Xl

o oo 11100 1J)JO 3000 '1100 o lCOO 1J)JO 3000 '000

Distance (km) Distance (km)

Figure 19:

evolution and the spatia] paLtern of the stress tield have been found even at distances from theridge much less than the radius of the Earth. We have observed that stresses due to lithosphericearthquakes in a spherical Earth decay slowly with increasing distance from the fau!t, in contrastwith predictions based on flat madels [.5].

Modelling the earthquake related space-time behaviour of the stress field in thefault system of southern IcelandTo model the space-time development of the stress tield using data on strain and stress 'changesfrom the other experiments and from databases.

In detail, this aims at the modelling of:

• The changes in crusta! strain and stress due to earthquakes and aseismic movement in thefault system of the SISZ.

• The formation and growth of faults and their interaction.

• The mutual influence between vo!canic and earthquake activity, e.g. magmatic upwellingand shearing at fault zones.

Following models are addressed:

• Calculation of the stress tield due to motions on the main faults.

• Comparison with the seismic moment release.

• Stress build-up by these motions and stress release by the major earthquakes.

31

• Forward modelling of the rheologieal parameters of the lithosphere asthenosphere in south­em Iceland using data of postseismic deformations.

~Iethods anel resultsThe moelelling is performeel by applying static elislocation theory to geoeletic elata anel elata ob­taineel through seismic moments from seismic measurements. The methoel and software useelallows to ,calculate displacements, strain anel stresses elue to elouble-couple and extensionalsources in layered elastie and inelastic Earth structures. Besieles the change in displacementeluring seismic or other short lived events, the changes caused by the mo\·ement of plates can beincluded.

The existing programs programs have been t.uned for faster performance and ext.ended toallow calcu!ation of six instead of five layers to account for elet.ailed knolI·leeIge of velocity elepthprofiles.

Preparational stuely has bcen made on the infiuence of severai layers. i.e. their elastic con­stants and t.hickness, on surface eleformation. Doing so, speeial attention \las paid to the effectsof the physical propert.ies of the source \;tyer on amplification or diminishing displacement at thesurface.

Data are gathered on strong (historieal) earthqltakes on Iceland anel its surroundings.Two moelels are being prepareel:

• Ascheme comprising the main rieige parts on Iccland and the North Atlantic Ridge to thenorth anel to the south of IceIanel.

• A moeleI of the SISZ anel the aeljacent part of the eastem volcanic zone.

The models incluele both faults and the load due to the Katla, Hekla. etc. volcanoes, i.e.driving fotTes from rifting and from the mantle plume below Iceland [59.60).

Acknowledgements

PREN LAB is funded by the European Commission as a part of the Em'ironment and ClimateProgramme, ClimatoIogy and Natural I·l azards, Seismic Risk. This is acknowledgeel. Speeialthanks to the scientific officer Maria Yeroyanni for her help and guielance. The support andsignificant contribution of the institutions of the partners of this project is also acknowledged.Special thanks to the Icelandic Meteorological Office. headed by director ~1agnus J6nsson, forits financial and administrative cont.ribution.

References

[1.] Angelier, J., S.Th. R6gnvaldsson, F. Bergerat, A. Guclmundsson, S. Jakobscl6ttir & R. Stefånsson1996. Earthquake focal mechanisms and recent faulting: a seismotectonic analysis in the V6rdufell area,

South Icelancl seismie zone. In: B. ThorkeIsson (editor). Seismology in Europ•. Papers presentecl at theXXV ESC General Assembly, September 9-14, 1996, Reykjavik, Iceland. ISBl\ 9979-60-235-X, 199-204.[2] Angelier, J. & 1'. Bergerat 1997. The South Iceland Seismie Zone. Part Il: Stress fielcls and mecha­nisms of seismogenic faults. Ill: Abstracts from the PREN LAB-1 workshop, OClober 24-25, 1997, Paris.13] Angelier} J., F. Bergerat, O. Dauteuil & T. Villemin 1997. Tension~shear relationships in extensional

fissul'e swarms 1 axial rift of northeastern Iceland: morphological evidences. J. Strucl. Geology 19, 673­685.[4] Angelier, J., S.Th. ROgnvaldsson, F. Bergerat & A. Gudmundsson 1997. Earthquake focal mecha­nisms and regional stress: the seismotectonic behaviour of the South Iceland Seismie Zone revealed bythe V6rdufell earthquake activity. In: Abstracts from the nintlt biennial EUG meeting, Marclt 23-27,

32

1997, St rasbourg.

[5] Antonioli, A., A. Piersanti &. G. Spada 1997. Stress diffusion following large 5trike-slip carthquakes:Et. comparison bctween srherical and flat Earth modeis. Gfophys. J. ]111., in press.

[6] Antonioli, A., A. Piersanti, G. Spada & M. Bonafede 1997. Time-dependent stress ficld associated\vith rift clynamics. In: Abstracts from the ninth bicllnial EVG meeting, March 23-27, 199i l Strasbourg.[7] Agu5tsson, I\. 1996a. ConlilltlOliS slmin measllremenls III SW-Iceland wilh locus O" lhe 1987 lialna­

fjoll farlhquake and the 1991 Hekla er"plion. Fil.lic. thesis. Uppsala University. :34 p.[8] Agustsson, 1\. 1996b. l>yngdarmælingar {t Suourlandi (gravity measurements in S\V-lceland). Ril

Veo"rsloj" hlands. Research report Vl-R96001-JA01, Reykjavik, 37 p.[9] Agiistsson, 1'\. 1996c. Alvakinn (testing and tUlling of the seismie alert system). Greinarge,'o

Veo"rstoj" islands. Report Vl-G96031-JA03, Reykjavik, 13 p. + figures.[10] Agllstsson, 1\. 1997. Overview of seismie aetiviLy in Ieeland January 1995 - :'\ovember 1996. Prelim­inary report on SIL data. G1'eillal'gero l/eoursloju islallds. Report Vi-GQi002-J.-\02, Reykjavik, 7 p. +figures.

[11) Åg~stsson, I,., A. Linde &:. R. Stefånsson 1996. The 1987 Vatnafjoll earthquake South Iceland viewedby strainmeters: information on source processes from aS::lociated deformation. In: B. Thorkeisson (edi­tor), Seisl1lology in El/rope. Papers presentecl at the X:\V ESC General Assembly. September 9-14,1996,Reykjavik, !eeland. ISBN 9979-60-235-X, 175-180[12) Belordinelli M.E. &:. M. Bonafede 1996. Stress fields and tensile fraetures in a transform domain. In:Abstracts from the XXV ESC General Assembl)', September 9-14, 1996, Reykjavik, !eeland. !eelandie

Meteorological Office, ~linistry for the Environment, University of !eeland.(13J Belardinelli M.E. & M. Bonafede 1997. Near-field stress evolution after a strike-slip earthquake in

a layerecl viscoeiastic half spacc. Phys. "nd C'hem. oj the E",·th 21, 231-235.[14) Beiardinelli, M.E., M. Bonafede & A. Guclmundsson, 1997a. Secondary earthquake fraetures gener­ated by a strike-slip fault in the South Icelancl Seismie Zone. J. Geophys. Res.. submittecl.[15J Belarclinelli, M.E., M. Bonafecle & A. Gudmundsson. 1997b. lnferenees on the regional stress fieldfrom the study of secondary earthquake fractures. In: Abstracts from the ninth biennial EVG meeting,Mareh 23-27, 1997, Strasbourg.[16) Bergerat, F., .1. Angelier, O. Dauteuil & T. Villemin 1996a. Morphologie et deformation dans unrift oceallique cmcrge: exemple du champ du fissurcs du hrafla (Nord-EsL Islande). In: Abstraets from

166me Reun. Se. TeITe, April 10-12, 1997, Orleans.[17] Bergerat, F., J. Angelier, O. Dauteuil & T. Villemin 1996b. Present-clay tension-shear deformatianand morphology in the Krafta fissure swarm, axial rift ~one of northeastern lcelancl. In: Abstraets from

the XXV ESC General Assembly, September 9-14, 1996, Reykjavik, leeland. !eelandie MetearologicalOffice, l\'Iinistry for the Environment, University of lccland.

[18] Bergerat, F. & J. Angelier 1997. The South Iceland Seismie Zone. Part I: \"eotectonie evidence fromfield stuclies of recent faulting. In: Abstraets from the PRENLAB-l workshop. October 24-25, 1997,Paris.

[19] Bergerat, F., J. Angelier, A. Guclmunclsson & S.Th. Rognvalclsson 1997. Joint study of recent fault­ing and earthquake focal meehanisms in thc SouLh !eeland Seismie Zone: the Vordufell area as a case

example. In: Abstracts from the ninth bienniai EUG meeting, March 23-27, 1997, Strasbourg.[20] Bergerat, F., A. Gudmunclsson, J. Angelier & STh. Rognvaldsson 1997. Seismotectonies of thecentral part of the South Ieeland Seismie Zone. Tectonics, submittcd.

[21) Bonafede, M., S. Danesi & M. Olivieri 1996. Modelling surface topography of rift zones in terms ofcrack theory. In: Abstracts from the XXI EGS General Assembly, Ma)' 6-10, 1996, The Hague.[22] Bonafede, M. & S. Danesi 1997. Near-fielcl modifications of stress inducec! by dyke injection atshallow clepth. Geophys. J. lill. 130,435-448.[23J Bonafede, M. & E. RivaIta 1997. Displacement anc! stress fields producec! by a tensile crack in alayered medium. In: Abstracts from the PREN LAB-l workshop, October 24-25, 1997, Paris.[24] Bourgeois, O., O. Dauteuil, F. Bergerat, T. Villemin, J. Angelier, S. Verrier, C. Homberg & V. Ferber

33

\997. Et.udc structurale des mecanismes du rifting occaniquc en blande. In: .-\bstracts from ColloqucNational Dorsales, November 24-2,), ][197, Paris.

[2.=)] Bodvarsson, H" S.Th. Hogn\'alclsson &. R. Slunga 1996. \Vaveform correlation as a tool for aULomatic

phase picking. In: Abstracts from the XXV ESC General Assembly, September 9-14, 1996, Reykjavik,

lceland. Icelandic l\'leteorological Office, i\'Iinistry for the Environment, University of lceland.

[26] Boc!varsson, R., S.Th. Rogll\"alelsson, S. Jakobsel6ttir. R. Slunga s.: R. Stef.n;;on 1996. The SIL data

acquisition\anc! monitoring system. Seism.. Res. Lelt. 67.35-46.

[2i] B6c1vm'sSOI11 R., S.Th. Rogn\"aldsson & TI.. Slunga 199T. Adaptive methocls for real-time earthquake

nnalysis in the SIL network. In: .-\bstracts from the ninth biennial EUG meeting. 1\Iarch 23-27, 1997,

Strasbourg.

[28] Crampin, S. 1996. Earlhquake predietion and earthquake forecasting. In: .-\bslracts from Earth­

qnake Research in Turkcy: state-of-the-art, September 30-0ctober 5, 1996, Ankara.

[29] Crampin: S. 1997a. Going .i..PE: monitoring and modell ing rock defarmation with shcar-wave split­

ting. In: Seism. i\/incs, Proceedings of the 4th International Symposium on Rockbursts, Krak6w , in

press.

[30] Crampin, S. 19D7b. Earthquake forecasting: a viable alternative to prediction. Geophys. J. Int.,

submittecl.

[31] Crampin, S. & S.V. Zatsepin 1996a. Forecasting earthquakes with APE. In: B. ThorkeIsson (eelitor),

Seismology in Europe. Papers presenled at the XXV ESC General Assembly. September 9-14, 1996,

Reykjavik, Ieeland. ISBN 99,9-60-235-X, 318-323.

[32J Crampin, S. & S.V. Zatscpin 199Gb. Rocks, dead or alive'?: Theor)' prompt; new understaneling of

fluid-rock interaction and sheal'-w3ve splitting. EOS 11,281 and 28G.

[:33] Crampin, S. & S.V. Zatsepin 1996c. Opportunilies for eal"thquake forecasting. In: Abstracts from

Earthquake Research in Turkey: state-af-the-art, September 30-0ctober 5, 1996. Ankara.

[34] Crampiu, S., H.J. RowIands &. R. Stefånsson 1997. Monitoring stress during magma injeetioll, in

preparation .

[35] Cralllpin, S. & S. V. Zatsepin 1997a. Changes of strain before earthquakes: the possibility of routine

monitoring of both long-term and short-tenTI precursors. J. Phys. Earth, in press.

[36] Crampin, S. & S.V. Zatsepin 199'b. Moclelling the compliance of crustal rock: Il - response to

temporai changes befare earthquakes. G'eophys. J. Inl. 129,495-506.

[3'J Cralllpin, S. & S.V. Zatsepin 199'c. The possibility of forecasting earthquakes. In: Abstracts from

3ssessment of schemes for earthquake predidion, Geological Society, November 7-8, 1996, London.

[38] Dauteuil , 0., J. Angelier, F. Bergerat, V. ferber , S. Verrier & T. Villemin 1997. Deformation pattern

and morphology in the Ilorthern Icela-nd rift. In: Abstracts from the ninth biennia! EVG meeting, lVlarch

23-27,1997, Strasbourg.

[39] Guclmunelsson, A. 1997. Stress fielels controlling strike-slip fau!ting in Icelanel. Phys. and Chem. aj

the Earth 21, 261-265.

[40] Gudmunclssoll, A. 1998. Formation and development of tension fractures. in a lava pile, and their

evolution into faults. In: P. Hancock and A. Gudmundsson (eclitors), Tensilt [rccluring in the Earth'scrust: a book to honouJ' Professor Neville J. p,.ice. University College Press, Londall, in press.

[41] Gudmundsson, A., F. Bergerat & J. Angelicr, 1996. Off-rift and rift-zone paleostresses in Northwest.

lcelanel. Teetonophysics 255, 211-228.

[42] Guelmunelsson, A., 1'. Bergerat & J. Angelier 1997. Strike-slip fau!ts in the central part of the South

Icelancl Seismic Zone. In: Abstracts from the ninth biennial EUG meeting, ~[arch 23-27, 1997, Stras­

bourg.

[43J Gudmunelsson, G. 1996. Seismicity in the central volcanoes beneath Myrc!als- anel Eyjafjallajokull.

I,,: Abstracts from the XXV ESC General Assembly, September 9-14, 1996, Reykjavik, lceland. icelandic

Meteorological Office, Ministry for the Environment, University of Icelanel.

[44] Guelmunelsson, G., S. Jakobscl6ttir & R. Boelvarsson 1996. Automatic se!ection of teleseislllic e1ata in

the SIL system. In: Abstracts from the XXV ESC General Assembly, September 9-14,1996, Reykjavik,

34

lceland. Ieelandic Meteorological Officc, Ministry for thc Environment, Uni\"ersity of Icelancl.[·15] Halldårsson, P. 1996a. EsLinmLions of magnitudes of hisLorieal earthqllakes. In: Abstraets from the:\XY ESC General Assembly, September 9-14,1996, Reykjavik, !celand. lcelandie ~leLeorologieal Office.~Iinistry for the Environment, University of Iceland.

P6J Halldårsson, P. 1996b. SpaLial ehanges in seismiciLy on Lhe Reykjanes Peninsula and South IcelandLowland. In: Abstracts from the XXV ESC General Assembly, September 9-14. 1996, Reykjavik, Ice­lancl. Icelanclic Meteorological Office, Ministry for the Envirollment, University of Icelancl.

[47] Hallcl6rssol1, P., Th. Skaftad6ttir & G. Gudmundsson 1996. r\ new catalogue of earthquakes in Iee­

land 1926-1974. In: AbsLracts from the XXV ESC General Assembly, September 9-14,1996, Reykjavik,Ieeland. Icelandic Meteorological Office, i\Iinistry for the Environment, Uni\'ersity of lceland.[48] .Jakobscl6ttir, S. HH)6a. The SIL neLwork: The need of automatic processing in seismicaliy active

areas. Ill: AbstracLs from the XXV ESC General Assembly, September 9-14. 1996, Reykjavik, leeland.Icelandic l\tleteorological Office, MinisLry for the En\"irollment, University of Iceland.

[49J ,lakobsdåttir, S. 1996b. Alert-deteetor in the SIL net,,·ork. In: AbstracLs from the XXV ESC GeneralAssembly, September 9-14, 1996, Reykjavik, lcelancl. Icelanclic Meteorological Office, Ministry for the

Environment, University of Jceland.[50J JakobsciåLtir, S. & F. Scherbaum 1996. Effects of the acausal response of zero phase FIR filLers onthe onset time c1etermination of P waves for intermediate and big earthquakes. In: Abstracts from the

XXV ESC General Assembly, September 9-14,1996, Reykjavik, Iceland. lcelanciie Meteorologieal Office,~Iinistry for the Environment, University of Iceland.

[5!] Jånsson, S. & P. Einarsson 1996. Radon anomalies and earthquakes in the South Iceland seismiezone 1917-1993. In: B. Thorkeisson (editor), Seis17lology ill Enmpe. Papers presented at the XXV ESCGeneral Assembly, September 9-14, 1996, Reykjavik, Icelanc!' ISBN 9979-60-235-X, Reykjavik, 247-252.[52] I":jartansson, E. 1996. Database for SIL earthquake data. In: Abstraets from lhe XXV ESC GeneralAssembly, September 9-14, }996, Reykjavik, Iceland. Icelanciic Meteorological Office, 'Iinistry for theEnvironment, University of Iceland.

[53] Kjartanssol1, E., S. Jakobsciottil', P. Erlendsson & G. Foulger 1997. Stafræn un'innslagos6r6a (digital

evaluation of voicanic noise). In: Abstracts [rom the symposium on the eruption in Vatnajokull 1996 ,Pebruar)' 22, 1997. Geologica! Society of Icelanel.

[54] Linde, AT., l.S. Sacks, R. Stefansson, J":. Ågustsson, O. l":amigaiehi & 1..:. i..:anjo 1996. Strain mea­surements and vo!canic eruptions. In: Abstracts from the XXV ESC General Assembly, September 9-14,

1996, Reykjavik, Iceland. lcelandie Meteorological Office, ~Iinistry for the Environment, University ofleelanel.

[55J Lund, B. & R. Slunga 1997. Stress tensor inversion of microearthquake fault plane solutians. Awayto monitor erustal stress ehanges. In: Abstracts from the lASPEl 29th General Assembly, August 18-28,1997, Thessaloniki.

[56] Moehizuki, M., H. Shimamllra, R. Stefånsson, H. Shiobara, G. Gudmundsson, & B. Brandsdåttir1996. Mieroseismieity and ernstal struetnre offshore north of Iceland. In: Abstracts from the XXV ESCGeneral Assembly, September 9-14, 1996, Reykjavik, lcelancl. Icelanclic MeLeoroiogical Office, Ministry

for the Environment, University of Iceland.

[57J Piersanti, A., A. Antonioli, ~l. Coceo & C. Nostro 1996. Global post-seismic deformation: stressfield ehanges. In: Abstraets from the AGU 1996 fall meeting, December 15-19, 1996, San Francisco.[58J Piersanti, A., G. Spada & R. Sabaciini 1997. Global postseismie rebound of a viscoelastie earth:Lheory for finite faulLs and application to the 1964 Alaska earthqnake. 1. Geophys. Res. 102,417-492.[59] Roth, F. 1996a. Surfacc deformution due to a source below severaI soft sedimentary layel's. In:

Annales Geophysieae. Abstrael.s from the XXI EGS General Assembly, May 6-10, 1996, The Hague.Supplement l to vo!. 14.[60] Roth, F. 1996b. Deformation paUern of displacement source below severallayers. In: Abstracts fromthe XXV ESC General Assembly, Seplember 9-14, 1996, Reykjavik, Jceland. leelandie MeteorologicalOffice, Ministry for the Environment, University of lceland.

35

(Gl} Rolh, f. 1997. VOl'stellung des Projekts zu \,Viederholungsmessungen ill einer Bobrung im Siidis­landischcn Seismizitiitsgebiet. In: C. Biic:ker (eclitor), Spt'clal vohune 1/1997 of the proceedings of theGermall Geop!lys. Svciety. Paper prcscllted at the 3rd workshop of the Fh-:PE working grollp, HBoreholeGeophysies and Rock Physics", Oclober 17-18, 1990, Hannover, 52~57.

[62J Rowlanels, II.J., S. Crampin &: R. Slefansson 199/. Forecasling eruplions. In: Abstracls from thellinth biennial EUG meeting, !\Iarch 23-27, 1997, 5tr<lsbourg.

[63] Rognvaldsson, 5.Th., G. Gudmundssoll, K. AgllstssOIl. S. Jakobsd6ltir & R. Stefånsson 1996: Recentseismicity near the l-Iengill triple-jullction, S\V-Iceland. In: B. ThorkeIsson (editor), Seismology in Eu­mpe. Papers presenleel al the XXV ESC General Assembly, September 9-14. 1996, Reykjavik, Icelanel.ISBN 99/9-60-235-X, 461-466.[641 Rognvaldsson, S.Th., G. Cudmundsson, h-:. Aglistsson, S. Jakobsd6ttir, R. Slunga & R. Stefånsson

1996. 5eisIllicity in the I-lengill volcanic area, S\V-Icelancl. \ 'olcallology a"d S€ls/Ilology, submitted.[65] Rognvaldsson, 5.'1"h., A. Gudmunclsson 8:. R. Slunga 1997. Seislllotectonic analysis of the Tjornes

Fradure ZOlle, an aclive transform faulL in North Icelancl. J. G'eophys. Res., subllliUed.[G6] Shiinamllra, H., R. Stefansson, ~1. MochizlIki, T. \\"alanabe, H. Shiobara, G. GlIc1munelsson & P.Einarsson H196. Norlhern Reykjanes Ridge microseismicity revealed by dense OBS anays. In: Abstl'acts

from the XXV ESC General Assembly, September 9-14, 1996, Reykjavik, icelanel. !celanelie Meteorolog­ieal Office, Ministry for the Environment, University of Icelancl.[67] Shomali, Z.H. &: R. Slunga 1996. Determination of the mechanism and source parameters of earth­

quakes by using moment tensor inversjon of earthquakes in the South Icelancl seismie zone. In: Abstracts

from lhe XXV ESC General Assembly, September 9-14, 1996, Reykjavik, Ice!anc1. leeIanelic Meteorolog­ieal Ofl1ce, Ministry for the Environment, University of Icelanc!.

[68J Sigmllnelsson, F. 1996. Inversion of geoeletie elata from Hengill lrip!e junelian, 1celanc1: rift subsi­dence anc! volcano inflation. In: Annales Geophysicae. Abstracts from the X\.I EGS General Assembly,

May 6-10, 1996, The Hague. Supplement I to 1'01. 14.[69J Sigmunelsson, F. & H. Vaelon 1996. Oblique spreaeling at the Miel-Atlantie Rielge, SW-Icelanel: 3Dcrustal veloeity field inferred from satellite radar interferometry and geologic evidenee. In: Abstractsfrom lhe AGU 1996 fal1meeting. EOS Transaclions Ameriean Gcophysieal Union, 77.[70] Sigmundsson l F., P. Einarsson, 5.Th. R6gnvalclsson, G .R. Foulger, lei\!. Hodgkinson & G. Thor­

bergsson r997. 1994-1995 seismicity and defol'matioll at the Hengill triplc junction, leelancl: triggeringof eal'thquakes by a small magma injection in a zone of horizontai shear streES. J. Geophys. Res. 102,

15151-15161.[71] Siglllunelsson, F., H. Vaelon &: D. Massonnet 199/. Ileaeljustment of the I\rafia spreaeling segment tocrustal rifting measured by sate!lite raelar interferometry. Geophys. Res. Lellers 24,1843-1846.172] 51unga, R. & B. Lund 1996. Stress tensor inversion basecl on microeal'thquake foeaI meehanisms. In:

Abstracts from the XXV ESC General Assembly, September 9-14, 1996, Reykjavik, Icelancl. Icelanclic

Meteorologieal Office, Ministry fOl" the Environment, Universily of 1eelanel.[73] Slunga, R., Lund, Il. & R. Boclvarssoll 1997. Estimates of ellrrent Icelanclic stress tensors from theinversion of mieroearthquake fault plane soilitions. In: Absiraets from the ninth biennial EUG meeting,

Mareh 23-27, 199/, Strasbonrg.[74] Slefansson, R. 1996. Towarels earthquake preelidion in Icelanel. In: B. Thorkeisson (eelitor), Seismol­

ogy ill Eu rope. Papers presented at the XXV ESC Genera! Assembly, September 9-14, 1996, Reykjavik,Iceland. ISBN 9979-60-235-X, 3-8.[75] StefånssoJl, R., R. Boclvarsson & G. Cuelmundsson 1996. Iceland plume tectonics. Same specuiationsanel fads. In: B. ThorkeIsson (eelitor), Seismology ill Europe. Papers presenteel al the XXV ESC GeneralAssembly, September 9-14, 1996, Reykjavik, leelanel. ISBN 9979-60-235-X, 505-51!.[76] Slefansson, R., G. Palmason, H. Palsson, P. Einarsson, R. Sigbj6rnsson, &: P. Ha!lel6rsson 1996.Tillogur um aoge,'oir til 00 draga lii- hællu aJ viildum jal'oskj6lfta (proposals for actions for mitigating

earthquake risk). Reykjavik, 63 p.[71] Stefånsson, R. & le Ågllstsson 1997. Continuolls measurements of strain anel hydrologica! effects in

36

the SISZ. In: Abslracls from lhe PREN LAB-I workshop. Oelober 24-25, 199,. Paris.

[78] SLefånsson, R.: h~. Ågt'istsson & P. Halld6rssol1 1907. The dua.l mechanism madel revivecl in light of

the obscl'ved fluid intrusion following the Vatna fjoll earthquake sequence of 1967. In: Abstracts from the

PREi\L.-\B-I workshop, Odober 24-25, 1997, Paris.

[79] Stefånssol1, R., S. Crampin, G. Gudl1lUndsson, 1\ ..'-\gt'ist 55011 , P. Halld6rs50n &: 5.Th. Rognvaldsson

190T. Jaroskjålftar og spennuhreytingar Lengdar innskori k\'iku og eldgosi i Yatnaj6kli. (earthquakes

and stress 'changes related to magma intrusioll and erup[ion in Vatnajokull). In: AbstracLs frol11 t.he

symposium. on the eruption in Vatnaj6kulJ 1996, February 22. HJO/. Gcolocical Society of lcelancl.

[SD] StefånsSoIl, R. & P. Hallcl6rsson 1997. On the possibility of a magnitude 7 earthquake in the BlåfJoll

area, neal" Reykjavik. In: Abstl'aets from the PRENLAB-l wOl'kshop, October 2-l-2:j, 19~17, Paris.

[8i) S,erånssoll, R., P. Halldorsson, S.Th. R6gn\·aldsson k G. Gudmundsson 199,. Husavik earlhquakes

and their reialion to seismicily, faulL 1l10VCmellts and rif[ing episodes. In: Abstracts from the PRENLAB­1 workshop, Oetober 24-25, 1997, Paris.

[821 Theocl6rsson, P. 1996a. Improved automatic radon monitoring in gl'ound \\·ater. In: B. ThorkeIsson

(editoJ.:); Seisl1wlogy in Europe. Papers prescnted at the XXV ESC General Assembly, September 9-1<1,

199G, Reykjavik, Iceland. ISBN 9979-GO-235-'>:, Reykj"·ik. 2.53-257.

[83] Theocl6rsson, P. 1996b. I'vJeasuremenL of weak radioacti\"ity. \Vorld Scientific Publishing Co., Singa­

pore, 333 p.[841 Tryggvason, A., S.Th. ROgnvaldsson 1:.: Q. Flovenz 199G. 3-D P- and S-",,,·e velocily slruclure be­

neath South-\Nest Jceland deri ved from localearthquake tomography. In: Abstracts from the XXV ESC

General A55embly, Seplember 9-14, 1996, Reykjavik, !ce1and. !celandic Meleorological Office, Ministry

for the Environment, University of Ieelancl.

[85J Vaclon l H. & F. Sigmunclsson 1996. Plate ll1ovement::s and volcano cleformation from satellite radar

inlcrlerolllelry: 1992-1995 eruslal deformalion al lhe llIid-Allanlic Ridge, S\\"-1celand. In: Abslracls

from the AGU 1996 fall meeting. EOS Transactions American Geophysical Union. 77.

[8GJ Vadon, H. & F. Sigmundsson 1997. CrH~tal derormation from 1992 to 199.s al the Mic!-Allantic

Ridge, SW-1celand, mapped by salellile radar illlerferometry. Science 275,193-197.

[87] Verrier, S. 1997. Ett/de sismo-teclonique d'lIll segment lmnsfol'mant emuge de la ,-ide medio­

atian/iq,,": la Zon" Sismiq"e S"c/-lsianc/aise. 111emoire de D.E.A. (lILSe. lhesis), ·50 p.

rS8] VilleminI 'I., f. Sigmundsson, F. Jouanne &: A. Guclmunclsson 1997. Present-day surface cIeforma­

tion at t.he rift-transformjunciion in North Iccland. In: .--\bstracis from the ninth biennial EUG meeting,

~larch 23-27, 1997, Slrasbourg.

[S9] Zalsepin, S.V. & S. Cralllpin 199G. 111odeIling rock111ass c!eforlllation \\"ith APE: a mcchani5m for

monitoring earthquake preparation zones with seismie shear-waves. In: Abstracts from the XXV ESC

General Assembly, September 9-14, 1996 1 Reykjavik l Icelancl. Icelanclic Meteorological Office, Ministl'Y

for the Environment, University of Iccland.

[90J Zalsepin, S.V. & S. Cralllpin 1997. Modelling lhe cOlllpliance or c"u5tal rock: I - response or

shear-wave splitting to differential stress. Geophys. J. /nl. 129, 477-494.

37

Coordinator

Ragnar StefanssonIcelandic Meteorological OfficeDepartment of GeophysiesBlistadavegur 9IS-150, Reykjaviklcelan~1

Other partners

Reynir BiidvarssonUppsala UniversityDepartment of Geophysicsynfavagen 16S-752 36 UppsalaSweden

Stu art CrampinUniversity of EdinburghDepartment of Geology and GeophysicsI\ings BuiIdings, West Mains RoadEdinburgh, EH9 3JWUnited Kingdom

frank RothGeoforschungsZentrum PotsdamTelegrafenberg A34D-1'1473 PotsdamGermi:l.ny

hurt FeigICNRSDelegation Regionale Midi-PyrennesUPR 23414, Avenue Edouard BelinF-31400 ToulouseFranee

Agust GudmunssonNm'dic Yolcanological InstituteGrensasvegur 50IS-lOS ReykjavikIceland

3S

Cont ract: ENY4-CT96-02.52Starting date: 01.03. 1996Durarion: 24 monthsClimatologyand ;atural HazardsHead of Unit: Anver GhaziEC Sei. Officer: Maria Yeroyanni

~Iaurizio BonafedeUnil'ersity of BolognaDepartment of PhysicsViale Berti Pichat 6/21-40127 BolognaItaly

Påll EinarssonUnil'ersity of IcelandScience InstituteDunhagi 3IS-I07 ReykjavikIceland

Freysteinn SigmundssonI\'ordic Yolcanologieal InstituteGl'ensåsvegur 50IS-lOS ReykjavikIceland

Franc;oise BergeratCNRS - DR2Departement de GeotectoniqueUnil'ersite Piene et ~Iarie Curie4, plaee J llssieu752·52 - Paris CEDEX O·)Franee