On the interrelation of fluid-induced seismicity and crustal deformation at the Columbo submarine volcano (Aegean Sea, Greece)

(1)

On the interrelation of

fluidinduced seismicity and crustal deformation

at the Columbo Submarine Volcano

(Aegean Sea, Greece)

Dissertation zur Erlangung des

Doktorgrades der Naturwissenschaften

im Department Geowissenschaften

der Universität Hamburg

vorgelegt von

Martin Hensch

aus Koblenz

Hamburg

2009

(2)

der Universität Hamburg

auf Grund der Gutachten von

Prof. Dr. Torsten Dahm

und

Prof. Dr. Matthias Hort

Hamburg, den 11. Januar 2010,

____________________________________

Prof. Dr. Jürgen Oßenbrügge

Leiter des Departments Geowissenschaften

(3)

Joint seismologi al anddeformation studiesareamighty toolto study thedynami sof

mag-mati systems at a tive vol anoes. While GPS measurements and InSAR are su essfully

applied at onshore vol anoes, the monitoring of submarine vol anoes is mostlyrestri ted to

islandbasedortemporaryseismologi al measurements. Wetherefordevelopedafreefall,self

levelingO eanBottomTiltmeter(OBT)toobservegrounddeformationontheseaoor,using

atwo omponent tiltsensorwitharesolution ofabout

15 nrad

. Thetiltmeter ismountedon thepreexistingHamburgO eanBottomSeismometer(OBS) arriersystem. Itisadditionally

equippedwithahydrophonetoassessseismi dataandanabsolutepressuresensortoobserve

uplift andsubsiden e.

Between June 2006 and Mar h 2007, four of these OBT systems were deployed along a

prole over the slopes and on top of the Columbo Submarine Vol ano. The network was

ompleted byfour OBSsinthevi inity ofthe seamount andadditional seismometers on the

surroundingislands. Columbo ispartofthe Santorinivol ani omplex,lo atedinthe enter

of theHelleni Vol ani Ar , Aegean Sea (Gree e), approximately

8 km

north-east of Thira island(Santorini). Thevol anohasattra tedattentionsin eislandbasedmonitoringindi ates

ahighseismi ityrate lusteringaroundtheseamountand possible rustaldeformation. Both

might represent uidmigration inthesubsurfa e.

Through this 10 months long lo al experiment, azimuthal gapsbetween theislands were

losed and the magnitude threshold of the permanent network was signi antly de reased.

The installation of zero oset seismi stations on top of the vol ano enabled us to derive

high pre ise depthlo ationsof earthquakes. Purpose of the study was to nd eviden es for

swarm triggers,su h aspossiblyuidmigration, bypre isely relo ating theevents bymeans

ofmultipleeventsmethods. About4000eventshavebeenmanuallypi kedandsixearthquake

swarms dire tly o urring atColumbo havebeen analyzedfor migration velo itiesof seismi

fronts. Four of these swarms were lassied as supposably dike-indu ed, the two remaining

swarms as the expression of in reased hydrothermal a tivity. Moment tensor solutions of

strongerearthquakes

(M

W

> 3)

were al ulated toeviden e our ndingsinterms ofpossible stresseld perturbationsindu ed bythe postulated triggers.

Simultaneouslytotheseismologi al observations,general unrestintermsofnoisein rease

(4)

earthquake luster entroid. For one swarm o urring lose to the tiltmeter prole, strong

near-eldtermswereobservedandsu essfullymodeledasanas endingvolumesour e. Both

ndingsaredis ussedextensivelywithrespe ttoapossiblelinkagebetweentheseismi luster

andtheoriginofthedeformationsignals. Furtherpointsofdis ussionarethegeneralte hni al

fun tionality of the newly developed OBT as well as additional ndings like long period

deformation signalsandtrendssuggestingtheuplift ofthe ompleteregionbetweenColumbo

and Santorini.

We on lude witha hypotheti al modelon deformation signals a ompanying theas ent

of a volumetri sour e. Thishypothesisis basedon our preexisting modelabout thepattern

of dike-indu ed earthquake swarms. We show, that the same migration velo ities found by

seismologi al observations an be independently derived by analyzing the hange of the

de-formation signal ofa propagatingvolumetri sour e. Finally,we eviden ethat our approa h

(5)

Abstra t iii Table Of Contents v List Of Figures ix List Of Tables xv Chapters 1 Introdu tion 1 1.1 Outline . . . 1

1.1.1 Stateof the art . . . 1

1.1.2 Goal . . . 3

1.2 Geologyof theEastern Mediterranean . . . 5

1.2.1 Regionalsettings . . . 5

1.2.2 Lo al settings andmonitoring networks . . . 8

1.3 TheColumbo Experiment . . . 10

1.4 Stru tureof thethesis . . . 11

2 Data a quisition 13 2.1 Passive measurements . . . 14

2.1.1 TheColumbo OBS/OBTExperiment . . . 14

2.1.2 TheEGELADOS Experiment . . . 17

2.2 A tive measurements . . . 18

2.2.1 Ree tionseismi proling. . . 19

(6)

3 The new Hamburg O ean Bottom Tiltmeter (OBT) 23

3.1 Te hni al des ription . . . 23

3.2 First experien es inpra ti al use . . . 25

3.2.1 Calibration inthe laboratory . . . 25

3.2.2 Testmeasurements intheBla k Forest Observatory (BFO) . . . 27

3.2.3 Overviewon signalsobservedat Columbo . . . 28

4 Theory of seismi and deformation sour es 35 4.1 Seismi sour es . . . 35

4.1.1 Earthquake me hanism . . . 35

4.1.2 The Earthquake Moment Tensor . . . 37

4.2 Deformationsour es . . . 39

4.2.1 The FaultSour e Model . . . 40

4.2.2 The MogiSour eModelfor spheri volumes . . . 47

4.2.3 The Lenti ularVolume Sour efor dike intrusions . . . 54

5 Methods and Results 1 Seismologi al analysis 57 5.1 Pre ise relative earthquake relo ation . . . 57

5.1.1 SEISAN: Pi king andfurther parameterization . . . 57

5.1.2 Velo itymodel . . . 61

5.1.3 3 omponent ross orrelation . . . 63

5.1.4 HYPOSATsingle event lo ation . . . 65

5.1.5 HYPODD relative earthquake relo ation . . . 67

5.2 Moment TensorInversion . . . 70

5.2.1 MomentTensorsolutions . . . 73

5.2.2 Stressinversion . . . 78

6 Methods and Results 2 Cluster and deformation analysis 81 6.1 Analysisof earthquake swarms . . . 81

(7)

6.1.2 Resultsfor lustermigration velo ities at Columbo . . . 83

6.2 Additional ndings . . . 87

6.2.1 Further swarm parameters . . . 87

6.2.2 Magnitude distribution. . . 87

6.2.3 A tivation ofadja ent faults . . . 88

6.3 Analysisofdeformation data . . . 89

6.3.1 Re-orientation offreefall tiltmeters . . . 89

6.3.2 Removaloflevelling- and highfrequen y jumps . . . 92

6.3.3 Resultsof tiltmeasurements. . . 94

6.3.4 Absolutepressuredata . . . 101

7 Dis ussion 105 7.1 Classi ationof earthquake swarms . . . 105

7.2 Correlationof deformation signalsand short lastingswarms . . . 107

7.2.1 Modelfor dike-indu edearthquake swarms. . . 107

7.2.2 A ompanying deformation . . . 110

7.2.3 Joint hypotheti al modelfor dike-indu ed earthquakeswarms anda ompanying deformation signals . . . 119

7.3 Globaltrends and aseismi deformation signals . . . 121

7.4 Possible te hni al improvements of the OBT . . . 123

8 Con lusions 125 Appendi es A Sensor parameters and velo ity models 127 A.1 Seismi stations . . . 127

A.2 Instrumentation . . . 129

A.3 Bandpasslters . . . 131

(8)

C Instrument alibration 137 C.1 Seismometer alibration . . . 138 C.1.1 PMD sensor . . . 139 C.1.2 EP300 sensor . . . 140 C.1.3 Comparison . . . 140 C.2 Tiltmeter alibration . . . 143

D Synopsisof earthquake swarms 149 D.1 Swarm on 28thofJuly 2006 (CS-1) . . . 150

D.2 Swarm from23rd Sept. until 1st ofO t. 2006 (CL-1) . . . 154

D.3 Swarm from6th until 8th ofDe ember2006 (CL-2). . . 158

D.4 Swarm on 10thofJanuary 2007 (CS-2). . . 162

D.5 Swarm on 18thofFebruary 2007 (CS-3) . . . 166

D.6 Swarm on 26thofFebruary 2007 (AS-1) . . . 170

D.7 Swarm on 1stofMar h 2007 (CS-4) . . . 174

E Synopsisof fo al me hanisms 179

F The EGELADOS working group 207

(9)

1.1 Overviewmap oftheHelleni Subdu tionZone . . . 6

1.2 Target area: The Santorini-Amorgos Fault . . . 7

1.3 Combined bathymetryand lowersedimentlayers: TheSantorini-Ana Basin. 8 1.4 Re ent seismi ityand deformationat Santorini . . . 9

1.5 3Dviewof theColumbo bathymetry . . . 11

2.1 Experimental setuparoundColumbo Seamount . . . 15

2.2 Tiltmeterprole overColumbo Seamount . . . 15

2.3 S ienti rew ofthere overy ruise (RV Aegaeo) . . . 17

2.4 TheEGELADOS network . . . 18

2.5 Prolelines ofa tive measurements (RV Poseidon) . . . 19

2.6 Joint seismi and magneti prole overColumbo . . . 20

2.7 Magneti heterogeneity aroundSantoriniand Columbo . . . 21

3.1 Sket h andphotoof theOBTsensorsphere . . . 24

3.2 Photosof absolutepressure gauge andOBTdeployment . . . 25

3.3 Tilt-tableusedfor tiltmeter alibration . . . 26

3.4 Examplefor tiltmeter alibration . . . 26

3.5 Regionalearthquakeobserved at theBFO . . . 27

3.6 OBTtra esof BFOmeasurement . . . 28

3.7 Noise omparison: BFO vs. Columbo . . . 29

3.8 Lo al earthquake on OBT54 andOBS50 . . . 30

3.9 Teleseismi event on OBTs and STS2 . . . 31

3.10 Shortperiodtilt-signal simultaneousto an earthquake . . . 31

(10)

3.12 Global trendof OBT57 . . . 33

3.13 Tiltmeterpowerspe trum(OBT 54) . . . 33

4.1 Dening parameters of afault . . . 36

4.2 Equivalent body for esand for e ouples . . . 37

4.3 Example for ahighfrequen y tiltsignal . . . 40

4.4 Re tangular sour e model . . . 41

4.5 Displa ement-and tilt-elds for typi al faults . . . 43

4.6 Minimummoment overdepthand distan eto triggerdeformation sensors . . 44

4.7 Minimummoment overdepthand distan efor astrike-slip event . . . 45

4.8 The Mogipoint sour e model . . . 47

4.9 The MogiModel: Displa ement and tilt . . . 51

4.10 The MogiModel: Amplitude distributionon X-and Y- omponents . . . 51

4.11 The MogiModel: Volume-and depth-dependen ies . . . 52

4.12 Threshold-Volume dependingon depthanddistan e to triggertheOBT . . . 53

4.13 Gridded volume thresholdoverdepthand distan e . . . 53

4.14 Deformationdue to magmati diking . . . 54

5.1 Overview: STA/LTA triggered andpi ked events . . . 58

5.2 Typi al event at Columboseamount (OBS 50 andIOSI) . . . 59

5.3 Earthquake swarm at Columbo . . . 60

5.4 Velo itymodelfor

v

P

and

v

S

. . . 61

5.5 Wadati diagrambased onSEISAN pi ks . . . 62

5.6 Example for timewindows uto ontinuous data . . . 64

5.7 Station orre tionsfor P- and S-phases . . . 66

5.8 Comparison oforiginal HYPO71 andimproved HYPOSAT lo ations . . . 67

5.9 Stru togram of datapro essingand moment tensor inversion . . . 72

5.10 Fo alme hanismsof all events

M

L

≥ 3.5

. . . 73

5.11 Fo alme hanismsof events

3.5 > M

L

≥ 3.0

. . . 74

(11)

5.14 Resultplotof stressinversion . . . 78

6.1 Fromzt-distribution to gridded event density . . . 82

6.2 Centroid depthestimation withGaussian urve t . . . 82

6.3 Exampleof migration velo ities for both swarm types. . . 86

6.4 Magnitude-timedistribution for all earthquake swarms . . . 87

6.5 Timedependen eof average and maximum magnitudes . . . 88

6.6 A tivation ofadja ent faults . . . 89

6.7 Correlation oe ients between OBT54 andlandseismometers . . . 90

6.8 Rosediagrams ofOBS orientations . . . 91

6.9 Examplefor theremoval ofa levelingevent on OBT54. . . 92

6.10 Corre tionexample for ajumpdue to an earthquake on OBT54 . . . 93

6.11 Lo ation of potential deformation sour es . . . 95

6.12 Tilt-walk of the tiltmeterpendulum duringan earthquake swarm . . . 96

6.13 Tiltmetertra esrotated inba k-azimuth dire tion . . . 97

6.14 Tilttra esfor two monthsof sparsea tivity . . . 100

6.15 Trendsof OBTs overthe omplete experiment . . . 102

6.16 Resultsofabsolute pressuremeasurements . . . 103

6.17 Comparison: Tilt measurement on apoint andover distan e . . . 104

7.1 Result-Plotfor swarmCS-1 lose to OBTs 54 and55 . . . 108

7.2 Hypotheti al modelfor short lastingearthquake swarms . . . 110

7.3 Unrestandination-deation signal onOBTs during theCS-1 swarm . . . 111

7.4 Depthdependen y ofthetiltamplitude . . . 114

7.5 Distan edependen y of thetiltamplitude . . . 115

7.6 Tiltamplitude overepi entral distan e, depthand time . . . 117

7.7 Spatialandtemporaltilt maxima . . . 118

7.8 Tiltamplitude versus epi entral distan e . . . 118

7.9 Hypotheti al modelfor deformation signals a ompanying short swarms . . . 120

(12)

A.2 Powerspe tra of EP300/DPGoshorestationOBS50 . . . 132

B.1 Analysisof orrelation urves . . . 133

B.2 Correlation oe ients1 . . . 134

C.1 Tilt-table for instrument alibration . . . 137

C.2 Instument spheresof PMD endEP300 seismometer . . . 138

C.3 Calibration urvesofPMD s/n512 seismometer. . . 139

C.4 Calibration urvesofPMD s/n4 seismometer . . . 140

C.5 Calibration urvesofEP300 s/n10535 seismometer . . . 140

C.6 Calibration urvesofEP300 s/n10536 seismometer . . . 141

C.7 Comparison ofPMD and EP300 sensors . . . 141

C.8 Ba k-shift velo itya ording to immersiondepthof thesensor(Stokes). . . . 142

C.9 Calibration measurements oftiltmeter 1 . . . 144

C.10Calibration measurements oftiltmeter 3 . . . 145

D.1 Swarm CS-1: lat-lon, lat-z, lon-z,map . . . 150

D.2 Swarm CS-1: zt-distribution . . . 151

D.3 Swarm CS-1: Tilt tra es . . . 152

D.4 Swarm CS-1: Tilt orientation . . . 153

D.5 Swarm CL-1: lat-lon, lat-z, lon-z,map . . . 154

D.6 Swarm CL-1: zt-distribution . . . 155

D.7 Swarm CL-1: Tilt tra es . . . 156

D.8 Swarm CL-1: Tilt orientation . . . 157

D.9 Swarm CL-2: lat-lon, lat-z, lon-z,map . . . 158

D.10Swarm CL-2: zt-distribution . . . 159

(13)

D.13SwarmCS-2: lat-lon,lat-z, lon-z,map . . . 162

D.14SwarmCS-2: zt-distribution . . . 163

D.15SwarmCS-2: Tilttra es . . . 164

D.16SwarmCS-2: Tiltorientation . . . 165

D.17SwarmCS-3: lat-lon, lat-z,lon-z,map . . . 166

D.21SwarmAS-1: lat-lon, lat-z, lon-z,map . . . 170

D.22SwarmAS-1: zt-distribution . . . 171

D.23SwarmAS-1: Tilttra es . . . 172

D.24SwarmAS-1: Tiltorientation . . . 173

D.25SwarmCS-4: lat-lon, lat-z,lon-z,map . . . 174

D.29Tilttra esoverevent rate . . . 178

E.1 Fo al me hanism ofevent 2006.07.2812:11:14.5 . . . 180

(14)

E.14Fo alme hanismof event 2006.11.1412:45:29.7 . . . 193

E.23Fo alme hanismsof events

M

L

< 3.5

(plot 1) . . . 202

M

L

< 3.5

(plot 2) . . . 203

M

L

< 3.5

(plot 3) . . . 204

(15)

2.1 Listof automati triggers ofdierent networks . . . 14

5.1 Velo itymodelsfor

v

P

and

v

S

. . . 62

5.2 hypoDD initial ontrolparameters . . . 68

5.3 Listof Moment Tensor solutions . . . 77

6.1 Results: Migration velo ities . . . 85

6.2 Further swarm parameters . . . 87

6.3 Orientation ofo eanbottomsensors . . . 91

6.4 Listof luster entroids andother possible deformation sour es . . . 98

6.5 Results: Deformation . . . 99

6.6 Results: Long period trends . . . 101

6.7 Results: Absolute pressuredata . . . 103

A.1 Listof seismi stations . . . 127

A.2 Listof station orre tions . . . 128

A.3 Instrumentation . . . 129

A.4 Dataloggersand sampling rates. . . 130

A.5 Poles and zerosof seismi sensors . . . 130

A.6 Bandpassltersapplied to seismi data . . . 131

C.1 Tilt-stepsfor seismometer alibration . . . 139

C.2 Tilt-stepsfor tiltmeter alibration. . . 143

C.3 Tilt-stepsfor tiltmeter 1(OBT54) . . . 144

C.4 Tilt-stepsfor tiltmeter 3(OBT55) . . . 145

(16)

C.6 Tilt-steps fortiltmeter 5 (OBT57) . . . 147

D.1 Deformationof the 28thof July2006 swarm (CS-1). . . 152

D.2 Deformationof the23rdof Sept.-1st ofO t. 2006 swarm(CL-1). . . 156

D.3 Deformationof the 6th-8th ofDe ember2006 swarm(CL-2) . . . 160

D.4 Deformationof the 10thof January2007 swarm(CS-2). . . 164

D.5 Deformationof the 18thof February2007 swarm(CS-3) . . . 168

(17)

INTRODUCTION

1.1 Outline

The assessment of vol ani hazards in populated regions is the main goal of all

vol anolog-i al dis iplines, whereas vol ano seismology mostly deals with the lo ation and tra ing of

magmati and other vol ani uid reservoirs and their migration in the subsurfa e, based

on seismologi al observations. Therefor, fundamental studies at a tive vol anoes are as

in-dispensable as the monitoring itself, be ause they have the potential to improve permanent

surveys and the general knowledge on vol ani pro esses. In ase of the Santorini-Columbo

Vol ani Complex, lo ated in theHelleni Vol ani Ar (Aegean Sea, Gree e), a dense

pop-ulation, in reased bytourists during summer on Santorini and thesurrounding islands,and,

of mu h greater impa t in ase of a vol ani hazard in this region, the possible ut o of

theNorthern Aegean and thus the Bla kSea (Turkey,Bulgaria, Ukraine et .) from

interna-tional sea routes make a duly preparation on a possible eruption ne essary. The Columbo

proje t was a ompletely fundamental resear h proje t, using dataof an amphibian seismi

network, in luding seismometers on the adja ent islands, O ean-Bottom-Seismometers and

newlydeveloped O ean-Bottom-Tiltmeters to assesssubmarine deformation data.

1.1.1 State of the art

Magma propagation at depth, su h as dike as ent, is often a ompanied by vol ani

earth-quakeswarms (e.g.Rubinand Gillard, 1998b; Battagliaetal.,1999). The omplexity ofthe

hypo enter patterns and migration paths has been intensively dis ussed for several types of

earthquake swarms and today itis known, that despite of magma movement also other

ori-gins have to be onsidered asswarm triggers: Te toni mainsho k-aftersho ksequen es (e.g.

Haukssonetal.,2001) and in reasedhydrothermala tivity due to vol ani unrest (e.g.Lupi

etal.,2007).

While the separation of purely te toni earthquake sequen es is relatively simple due to

(18)

the lassi ation of dierent types of vol ani earthquake swarms is rather ompli ated. It

is exhaustively dis ussed in Hens h (2005) for earthquake swarms oshore North I eland,

by regarding verti al migration velo ities and the depth over time pattern (zt-distribution)

of these swarms: Supposedly dike-indu ed earthquake swarms show lear seismi ity fronts

whi h resemble the propagation of lo alstress a umulation. A fast migrating front during

the initial phase of the swarm is dis ussed to be aused by ra k opening and/or degassing.

A se ond front is supposed to ree t the a tual position of thehead region of the as ending

dike. Thisba kfront isas endingmu hslowerthan theinitial frontand markstheboundary

to theregionbelow,whereseismi ityisla king. Resultofbothfrontsisthetypi altriangular

shapeofthezt-distributionofthis typeofearthquakeswarm. Inoppositeto magmaindu ed

earthquake swarms, hydrothermally triggered swarms or swarms due to gas propagation are

more s atteredandofalessstru turedshape. Theseswarmsaresupposedtobetriggeredby

the wide spread unrest of hydrothermal uids and thus do not show hara teristi fronts of

seismi ity.

Similar studiesledto omparable results,e.g. for the Izu(Japan) swarmin2000 (Ukawa

andTsukahara,1996)oranearthquakeswarmheraldingthe2001Mt. Etnaeruption (Patané

etal.,2002),whi harebothdis ussedtobelinkedtodikeas ent. Inboth ases,fastmigration

paths during the initial phase of the swarm and the triangular shape of the zt-distribution

due to a seismi ba kfront were observed. Additionally, these swarms ended within a few

days. Fluid- or gas-indu ed earthquake swarms have been dis ussed by e.g. Fis her (2003)

or Fis herandHorálek(2003)for a lusterinNW-Bohemia (Cze hRepubli ). Theseswarms

are omparable with hydrothermally indu ed lusters, the stru ture of their zt-distribution

is more ompli ated, often without lear migration paths and the temporal length of these

swarms ismu h longer,partlya week andmore.

Theobservationofvaryingtemporallengthsforbothtypesofvol ani earthquakeswarms

ts well with our results of re ent studies. A maximum length of

2 d

for supposedly dike indu edswarmsin ontrasttodurationsuptoaweekforhydrothermallytriggeredswarmsisa

generalndingforswarmsoshoreNorthI eland(Hens hetal.,2008). Both,magmaintrusion

and in reased hydrothermal a tivity, must be onsidered as possible trigger me hanisms for

earthquake swarms measuredat Columbo.

A se ond important expression of magma propagation in shallow depths is lo al rustal

deformation. It anbe auseddire tlybyavolumesour e,su hasamagmareservoiroran

as- endingdike,orbynear-eldtermsofshallowvol ano-te toni earthquakes. Earlyapproa hes

on modeling these sour es have been done e.g. by Mogi (1958) for a spheri volume sour e

or Okada (1992) for point and re tangular fault sour es. With today's te hni al

(19)

su h aslenti ular volumes whi h aremore omparable with dikes aremodeled and observed

(e.g.Pollard et al., 1983;Hautmann etal., 2009). While both te hniques, InSAR and GPS,

are mighty tools in surveying onshore vol anoes, they fail for oshore observations.

Assess-mentsofthestateofa tivityofsubmarinevol anoesisstillmainlyrestri tedtoseismologi al

measurements, mostlyeven limitedto temporary amphibian networks.

Onlyveryfewstudiesta kledtheeldofsubmarinedeformationmeasurementsuptonow.

Alreadyinthe late1980s,Fox(1990)installedabsolutepressuresensorsatthesummitofthe

axialvol ano( entralJuandeFu aRidge,Pa i O ean)andsu essfullymeasureddeation

afterana tive episodein1998intherangeofafewmeters(Fox, 1999). Tolstoyetal. (1998)

developed short-and long-baseline tiltmeterswithresolutions of

50 nrad

thatwere deployed asfreefallinstruments. Those instruments partlyworked su essfullyinanother deployment

at theJuan de Fu a Ridge. The most re ent approa h by Fabian and Villinger (2007) and

Fabian and Villinger (2008), University of Bremen (Germany), was to study deformations

aused by hydrothermal a tivity at the Logat hev Vent Field at the southern Mid-Atlanti

Ridge: Asingleprototypewassu essfullyre overedafteralongtermdeploymentandproved

thepossibilityto measure tiltintherangeofsome

µrad

ontheseaoor.

RestraintsofalreadydevelopedO ean-Bottom-Tiltmeters(OBTs)weretheirpoor

resolu-tionfor absolutepressure sensors,their limitedoperating timeofonly some weeksor in ase

of theBremen OBTthe fa tthat ithas to be deployed and re overed bya Remote

Operat-ingVehi le (ROV). But the studies ited above have all shown that submarine deformation

measurementsare inprin iple possible andthat itis worthto advan einvestigations on this

eld. Thus,wedevelopedafreefallOBTsystemforlongtermdeployments(upto10months),

withtheoreti al resolutions of

15 nrad

for tilt and

1 mm

or

0.1 mbar

for absolute pressure to measure uplift and subsiden e. Additionally, it is equipped witha hydrophone to re ord

seismi signals. These sensors were mounted on the pre-existing Hamburg

O ean-Bottom-Seismometer (OBS)frame.

1.1.2 Goal

Aimofthisstudyistoperformdeformationmeasurementsatana tivesubmarinevol anoand

to study orrelations between possible deformation signals and theo urren e of earthquake

swarms. Columbo was hosen be ause of its re ent and ontinuous seismi and possibly

deformationa tivity,be auseofitsfavorablepositionbetweenmanyislandstomountonshore

seismometers anddue to the ex ellent ooperation and data-ex hange with our olleagues of

theEGELADOS proje t(Friederi h and Meier, 2008), another amphibian network installed

parallely inthe wholeAegeanregion (see Ch. 2.1.2). Thisstudy solely deals withbasi and

(20)

The main te hni al innovation of this work was the development of the prototype of

an oshore deformation sensor and its su essful test in a pilot experiment at a submarine

vol ano. A basi question was, whether this instrument is inprin iple working and apable

for submarine deformation measurements.

The most important s ienti innovations of my work despite of te hni al developments

an besummarized asfollows:

•

The improvement of our hypotheti al models for vol ani earthquake swarms, whi h were derived fromre ent studies

•

The rst assessment of moment tensor solutions for earthquakesin luding hydrophone dataof our OBS/OBTsystem

•

The rst submarine deformation measurement using the preexisting arrier system of theHamburg O eanBottom Seismometer

•

The investigation of 10 months long tilt timeseries for trends and short period defor-mationsignals ina joint analysiswithearthquakeswarm data

S ope and omplexity of the following investigations using the obtained data required

exhaustive theoreti and methodi works, whi h are sorted as three re urrent themesin the

same orderas given below (swarm behaviour, sour e me hanisms and deformation) for ea h

hapter, to in rease the readability of the thesis and to allow skipping of spe i se tions.

FromChapter4 on, hapters arethus subdividedea h into themaingoals of thisthesis:

•

Classi ation of earthquake swarms by the parameterization of luster pattern and migration velo ities, with the ba kground to on lude for trigger me hanisms of

earthquake swarms o urring in the Columbo region and to generally arm similar

ndings ofre ent studiesinother regions.

•

Estimationoffo alme hanismsin ludinghydrophonedatatodraw on lusions ontheregionalstressregimeandpossible lo alperturbationsbyinvertingfor thestress

tensor. The knowledge of the stress eld is of great importan e for the dis ussion of

geometry and orientation ofpossiblyas endingdikes.

•

Study of a ompanying submarine deformation signalsand ndingpossible in-terrelationstosimultaneouslyo urringseismi a tivityisthenalpurposeofthethesis.

The obje tive is to substantiate previous ndings, su h as migration velo ities of

(21)

1.2 Geology of the Eastern Mediterranean

The European Mediterranean Sea has been built up by the opening of Pangea during the

Triassi rifting about 220-230 Ma ago and thelater northwards motionof the Afri an plate

ausedbythesea-oorspreadinggeometryoftheNorthandSouthAtlanti o eanduringthe

Creta eous and Tertiary (Gealey, 1988). Strong redu tion of the northward motion of the

Afri anplateafterthe ollisionwithEurasia30-35Maagoinitiated Mediterraneanextension

and a southward retreat of the o eani slab that is subdu ting below the European plate.

Thisretreatissuggestedtohave ausedtheformationofseveralextensionalbasinswithinthe

Mediterranean Sea, one ofwhi h is nowknown astheAegean Sea, whi hhas been further

opened by re ent slab rollba k of the Helleni Subdu tion Zone (Pi hon and Angelier,

1979;Kahle etal.,1998; Jolivetand Fa enna, 2000).

1.2.1 Regional settings

Fig. 1.1 gives an overview on the regional te toni settings of the Aegean Sea, positions of

sedimentary ar and vol ani ar and the ve vol ani enters. The Helleni Subdu tion

Zone (HSZ) is the seismi ally most a tive region in Europe (Bohnho et al., 2006). Due

to slab rollba k, it is a typi ally extendedsubdu tion zone: The onvergent plate boundary

between the Aegean mi roplate and the Afri an plate is pla ed in the Lybian Sea, around

100-150 km south of the Helleni Ar . The sedimentary ar is lo ated (from west to east,

see Fig. 1.1) between the Peleponessus peninsula, Kythera, Crete and Rhodos (Papaza hos

and Panagiotopoulos, 1993; Meier et al., 2004). The overall rate of onvergen e of the HSZ

is about 3.5-4 m/a, splitinto a major ontribution of the Aegean plate with about 3 m/a

SW-ward propagation and 0.5-1 m/a N-ward migration of the Afri an plate, the dip angle

ofthe subdu tinglithosphereisonaverage

30 − 40

◦

(Papaza hosandPanagiotopoulos,1993;

Knapmeyer, 1999; Jolivet and Fa enna, 2000; Bohnho et al., 2001). Following the

sub-du ting lithosphere around 150-200 km to thenorth, it rea hes a depth of 100-150 km (e.g.

Meier et al., 2004) and the absen e of strong earthquakes below 150 km is assumed to be

linkedto the hightemperature ofthematerial inthis region(Papaza hosand

Panagiotopou-los,1993). Hotmaterial as ends from these deep zones and possibly intrudes into the rust

alongthefra ture zones. Extensionalfaultsopenthe pathways forthesemagmati intrusions

(Papaza hos and Panagiotopoulos, 1993; Dimitriadis et al., 2009). This assumption is

sup-ported bythe o urren eof vol anism alongtheHelleni Vol ani Ar (HVA)dire tlyabove

thisregion,separatedintoveseismovol ani lusters(fromwesttoeast): Sousaki, Methana,

Milos,Santoriniand Nysiros.

(22)

Com-22˚

22˚

23˚

24˚

25˚

26˚

27˚

28˚

29˚

34˚

35˚

36˚

37˚

38˚

50 nautical miles

Strabo trench

Ionian trench

GREECE

TURKEY

Crete

Sousaki

Milos

Santorini

Nisyros

Methana

Pliny trench

VOLCANIC ARC

SEDIMENTARY ARC

0.5−1cm/a AFRICAN PLATE

3−3.5cm/a

AEGEAN

PLATE

VOLCANIC ARC

longitude [°]

latitude [°]

extension

faults

Figure1.1: OverviewmapoftheHelleni Subdu tionZone: Subdu tionofAfri anlithospheresouth ofCreteleadstoasedimentaryar andfurthernorthwardstoatypi alvol ani ar withtheknownvol ani entersofSousaki,Methana,Milos,SantoriniandNysiros. TargetareaofourstudiesistheSantoriniVol ani Complexanditsadja entregiontothenortheast,theSantorini-Amorgosfaultzone. (Lo ationsofextension faults takenfromPapaza hosandPanagiotopoulos(1993)).

(Heikenand M Coy,1984), andits adja ent extensionalfaultsystem,theSantorini

Amor-gos Fault (SAF). Both, vol ani and te toni a tivityare observed inthis region: At least

nine re ent eruptions of Santorini have been proven for the last 600 years, the last one in

1950 (Dimitriadis et al., 2009)and furthermore strong and shallow earthquakes o uralong

the extensional faults, su h as the 1956 events at the SAF: Two M 7.5 and M 6.9 normal

faulting events (lo ation see Fig. 1.2) within 13 minutes on 9th of July 1956, a ompanied

byatsunami. Faultplanes ofthe1956earthquakes learly t withtheassumedstress model

(Papaza hos andPanagiotopoulos,1993).

Thetargetareaofourstudyislo atedbetween

36.3 ◦

N − 36.9

◦

N

and

25.25 ◦

E − 26.15

◦

E

, see Fig.1.2. The dominant te toni feature isthe Santorini-Amorgos Fault with a ouple of

minor faults inits vi inity and around theColumbo Seamount. The SAFis slightly bended

(23)

25˚15'

25˚30'

25˚45'

26˚00'

36˚30'

36˚45'

25˚15'

25˚30'

25˚45'

26˚00'

36˚30'

36˚45'

3 nautical miles

0

200

400

600

800 1000

longitude [°]

latitude [°]

depth [m]

Amorgos

Anafi

Ios

Santorini

Kameni line

(minor faults)

Anidros

Santorini −

Santorini − Columbo line

M 7.5

M 6.9

Amorgos − Fault

Figure1.2: Bathymetrymapoftargetarea-TheSantorini-AmorgosFault:Thedashedlinemarks thedominant te toni feature,theSantorini-Amorgos-Fault. Dottedlinesmark adja ent minorfaults. The 1956 earthquakesaregivenwithred dots,dire tion ofextensionismarkedwitharrows (a slightbendingof thestressaxisisvisible). Dataonfaults andtheearthquakesaretakenfromDimitriadisetal.(2009). Main vol ani featuresaroundSantoriniaregivenwithsolidlines: TheSantorini-ColumbolineandtheKameniline (takenfromtheVol ani HazardZonationMapoftheInstitutefortheStudyandMonitoringoftheSantorini Vol ano (ISMOSAV),http://ismosav.santorini.net). Additionalsmallseamounts around Anidrosisland are visibleinthebathymetry. TheColumbo alderaismarkedbyawhitedot.

foundaroundAnidrosislandonanelongated axisSantorini-Columbo-Anidros, paralleltothe

SAF, whi h indi ates the o urren e of vol ani a tivity along a fault parallel belt and not

onlyatone spotatSantorini. Averyimportantfeature ofthisvol ani beltistheColumbo

Submarine Vol ano(white dot inFig.1.2), whi h isthefo usofthis study.

A ross se tion derived of ree tion seismi proling is given in Fig. 1.3 (Note: View

from NE!). The SAF is dire tly lo ated SE of the es arpment between Santorini, Anidros

andAmorgos. Lateral tiltingof thesediment layerssuggestsudden stressreleases inform of

(24)

Ios

Anafi

_Anidros

_Columbo

N

E

Z

Santorini − Anafi Basin

Amorgos Basin

Figure1.3: Combinedbathymetryandlowersedimentlayersmeasuredbyree tionseismi proling oftheSantorini-AnaandtheAmorgosBasins. Thedepths aleistentimesexaggerated,themaximumdepth rangeoftheusedhighfrequentseismi wavesisabout2km.Themainte toni feature,normalfaultingatthe Santorini-Amorgos Es arpment,ismarkedbyared line. Dashedlines marksedimentarylayers,supposably tilted by several strongnormalfaulting earthquakesdueto theextensional stressregime. Theview isfrom NE,seamountsaroundAna,theColumbovol anointheba kgroundand anothersmallerbasin ausedby adja entminorfaultsarevisible.

1.2.2 Lo al settings and monitoring networks

The Columbo Submarine Vol ano is lo ated about 8 km northeast of Santorini at

36.55 ◦

N

and

25.45 ◦

E

. It has a well dened aldera of about 1500 m diameter, a maximum depth inside the aldera of about 500 m and a minimum depth of the rater rim (the Columbo

Reef) asshallowas17 m(Sigurdsson etal., 2006;Dimitriadis etal.,2009). The most re ent

eruption of Columbo is do umented to have happened on 26th of September 1650, lasting

until De ember of the same year and ausing a tsunami on Thera island (Santorini). Sin e

geologi al and other observations were thoroughly do umented in Gree e already at that

time, it is known that earthquake a tivity in reased about two weeks before the eruption.

Furthermore,the o urren eof asmall isletduringtheinitial phaseoftheeruption hasbeen

reported (Dominey-Howesetal.,2000, and referen estherein).

Currently, Columbo shows seismi and geothermal a tivity: Sigurdsson et al. (2006)

ob-served widespreadhydrothermalvents,emissionplumesof10mandmoreanduid

tempera-turesofupto

220 ◦

C

fromvent himneysofupto4mheight. Hydrothermala tivityismainly on entrated on the northern part of the inner Columbo aldera. In ontrast to Columbo,

only low-temperature venting is observed inside the Santorini aldera with temperatures of

about

15 − 17

◦

C

duringthesame marinesurvey(Sigurdsson etal.,2006).

Lo al seismi ity is also suggested to derive mainly from a vol ani origin: Fig. 1.4 is

an overview on seismi a tivity and observed deformation. While some events ares attered

alongtheknownfaultsandhavealreadybeendes ribedtobelinkedtoextensionofthebasins

(25)

25˚15'

25˚30'

25˚45'

36˚30'

36˚45'

2 nautical miles

25˚15'

25˚30'

25˚45'

36˚30'

36˚45'

Ios

_Amorgos

Santorini

Anafi

25˚18'

25˚24'

25˚30'

36˚21'

36˚24'

36˚27'

36˚30'

-120

-100

-80

-60

-40

-20

0

0 25˚18'

25˚24'

25˚30'

36˚21'

36˚24'

36˚27'

36˚30'

-100

-50

0

0 25˚18'

25˚24'

25˚30'

36˚21'

36˚24'

36˚27'

36˚30'

25˚18'

25˚24'

25˚30'

36˚21'

36˚24'

36˚27'

36˚30'

longitude [°]

latitude [°]

+/−0

+15

+1

−4

−11

−10

−5

−11

−57

−130

−35

−57

Cape Columbo

Columbo seamount

8km

THIRA

THIRASSIA

ASPRONESI

−KAMENI

NEA−

PALEA−

vertical deformation

in mm

SANTORINI

deformation

seismicity

15 km

APE (Naxos)

latitude [°]

NOA permanent

seismometer

EGELADOS

AUTH / ISMOSAV

GPS station

(ISMOSAV)

contour lines of

vertical deformation

longitude [°]

latitude [°]

longitude [°]

Columbo

seamount

8 km

Figure 1.4: Re ent seismi ity and deformation at Santorini. Joint ataloguedata ofthe National observatoryAthens(NOA,www.gein.noa.gr)andtheInstitutefortheStudyandMonitoringoftheSantorini vol ano(ISMOSAV,ismosav.santorini.net)fortheyear2006(lefthandsideplot).Reddotsmarksingleevents, yellow diamondsEGELADOStemporary landstations,green inverse triangles stations runby the Aristotle University Thessaloniki and ISMOSAV, blue triangles mark permanentNOA stations, ofwhi hthe se ond losestonNaxosis slightlyoutsidethe map. Whiletheseismi ity rateatSantoriniis relatively low,events o ur along the Santorini-Amorgos es arpment and a large luster is observed abut8 km NE,around the ColumboSubmarineVol ano. Parallel(plotontherighthandside),GPSmeasurements(ISMOSAV)suggest tiltingofSantoriniwithhighestupliftrates afewkmeastof CapeColumbo. Browninversetriangles mark positions of GPS stations. Deformation data here shown is of the period from 1994 to 2001 (taken from ismosav.santorini.net).

et al. (2009) to be related to vol ani , possibly hydrothermal, a tivity. The o urren e of

earthquakes before the 1650 eruption (Dominey-Howes et al., 2000) also suggests magma

as ent asapossibletrigger ofsu h events. However, an investigationontemporal lustering,

su hastypi al for vol ani ally indu edearthquake swarms,hasnot been arriedout yet.

A permanent seismi network is run by theNational Observatory Athens (NOA, for

pa-rameterdataseewww.gein.noa.gr), withone stationonSantoriniwhi h ismostlyshutdown

or noisy and a sparse net of stations on the surrounding islands. The se ond losest

sta-tion is Naxosat a distan e of about 50 km, whi h is denitely insu ient to analyze weak

vol anote toni earthquakes. The magnitude threshold of the NOA network to su essfully

lo ate events around Santorini is about

M

L

= 3

. Re ently, an additional network is being established on Santorini and the surrounding islands to better lo ate events oshore

San-torini and to de rease the magnitude threshold to about

M

L

= 1.5

(for parameter data see ismosav.santorini.net).

(26)

Vol ano (ISMOSAV,ismosav.santorini.net) isrunning anonshore GPSnetwork on Santorini

that is permanently installed to measure possible deformation signals. Here shown data

(Fig. 1.4)suggests slightuplift oftwo spotsonThira(main islandofSantorini): Onespot in

thenortheast, losetoCapeColumbo,andpossiblyanotheroneinthesoutheast. Theoverall

uplift nearCapeColumbo was

15 mm

fortheperiodfrom1994-2001,whileforthesametime subsiden eofThirassiaislandofabout

130 mm

wasobserved. Butalsoforthesedeformation signals, no further investigations on short-term behaviour or relation to the seismi a tivity

have been been arriedout until now.

1.3 The Columbo Experiment

Tostudythetemporalbehaviourofvol ani earthquake lusters,amu hdenser networkand

seismi stations in loser epi entral distan e, some preferably near zero-oset, are ne essary.

Similar experiments at oshorevol anoes(e.g. Hens h etal.,2008) have alreadyshown that

animprovement oftheevent lo ationresidualsbyafa torof10-20ispossible,on eazimuthal

gaps around the sour e region are losed. Hypo entral depths an be estimated more

pre- iselyusingzero-osetstations andeventsarerelativelyrelo atedusingwaveform- orrelation

te hniques.

We therefor installed a network of 4 O ean-Bottom-Seismometers (OBS) and 4 of the

new OBTs ontop of the Columbo Reefand its vi initybetween 10th of June 2006 and 27th

of Mar h 2007. Fig. 1.5 shows the bathymetry and the positions of the OBT prole over

Columbo and the losest OBSs. For a detailed des ription of the amphibian network and

stationpositionssee hapter2. OBTswereadditionallyequippedwithhydrophonestore ord

seismi signals. Te hni al details of thenew Hamburg OBT system are given in hapter 3.

The tiltmeter prole (Fig. 1.5) was deployed along the third prin ipal stress axis

σ

3

, sin e possible intrusions su h as a dike wouldopen along an axisperpendi ular to

σ

3

(Rubin and Gillard, 1998a). Thesmallestdeformation wavelengthand thus largesttiltgradientsarethus

expe tedalong

σ

3

. Eventheshapeofthe Columbovol anois learlyelongatedperpendi ular to

σ

3

,indi ating the predominan e ofNE-SW orientation ofdikes.

Paralleltoourexperiment,theEGELADOSexperiment,anotheramphibiannetworkwith

alargenumberoflandseismometersandadditional24OBSs(see hapter2.1.2),wasperformed

in the omplete Aegean Sea region. In ollaboration with our German, Greek and Turkish

partners, all networkswere joint to alarge database for severalsubproje ts andthus we had

a essto mu h moreseismi dataofthesurrounding islandsto omplete our dataset.

Thesetemporal improvements enabled us to pre isely investigate sixearthquake swarms

(27)

25.45

25.50

25.55 longitude [deg]

36.50

36.55 latitude [deg]

0

250

500 depth [m]

0

125

250

375

500 N

depth [m]

Santorini

Caldera

σ3

5 km

uplifted region

Columbo Reef

OBT 55

OBT 54

OBT 56

OBT 57

OBS 50

OBS 51

OBS 52

OBS 53

Figure 1.5: 3D view of the Columbo bathymetry: Bathymetry plotof Columbo derivedfrom ship e hosounderdataofRVPoseidon, view withanazimuthof

135 ◦

and aplungeof

40 ◦

. Highestelevationis theColumboreef,lo ated betweenthe alderaandanuplifted regionbetweenColumboand Santorini. An OBSwithanabsolutepressuresensorwaslo atedontopoftheupliftedregion,anotherOBSontheopposite siteofthe aldera. A proleof4 OBTs was deployedoverthe ColumboReefandperpendi ular toanaxis Santorini-Columbo,parallelto thelowestprin iplestressaxis

σ

3

. Twoadditional OBSsslightlyoutside this mapse tion losedazimuthalgapsbetweenthesurroundingislands.

1.4 Stru ture of the thesis

The omplete Columbo experiment, a tive measurements parallel to the OBS/OBT

deploy-mentandthepartnerproje tEGELADOSareintrodu edinChapter2,followedbyadetailed

te hni al des ription of our newly developed O ean-Bottom-Tiltmeter inChapter 3. Within

thisOBT hapter, questionsonthete hni al fun tionalitywillbe learedbasedontest

mea-surementsinthe laboratory,ina mineand nallythepra ti al test at Columbo.

Anexhaustiveintrodu tiononseismi andespe iallydeformationsour esisgivenin

Chap-ter 4: Despite the seismi moment tensor inversion, predominantly the theory of near-eld

terms of re tangular fault sour es and volume sour es is shown. Deformation sour es are

introdu ed very detailed, in ludingmodelings ontheir shape and range.

Chapter 5 summarizes all methods and results of the seismologi al analysis (earthquake

(28)

overviews ondata-pro essing steps.

All observations are nallydis ussed together withmodelling approa hes andother

nd-ings of all main topi s in Chapter 7. Results of a tive measurements during the Columbo

deployment ruise in June 2006, a tive seismi and magneti proling, and additional

(29)

DATA ACQUISITION

TheColumboSeamountO ean-Bottom-Seismometer(OBS)andO ean-Bottom-Tiltmeter(OBT)

Experiment tookpla eat and aroundtheColumbo submarine vol anobetween 10th ofJune

2006 and27thofMar h 2007. The dataset is omposed ofownOBS/OBTdataand passive

seismi data of the amphibian EGELADOS network (see se tion 2.1.2), whi h overed the

wholeAegeanSea, major Aegean islands andGreek and Turkish mainland inthevi inity of

theAegean Seaduring thesametime. We alsohad a essto dataofthepermanent network

of the National Observatory of Athens (NOA). Parallel to the deployment of 4 OBSs and

4 newly developed OBTs (for te hni al details see hapter 3), 1.500 km of high frequen y

ree tionseismi , magneti andgravimetri prolesweremeasured.

WhiletheinstallationofonshorestationswasorganizedbytheEGELADOSproje t

part-ners of Bo hum, Thessaloniki and Istanbul, the deployment of oshore stations and a tive

measurementswas arried out duringthree ship ruises:

•

18th- 30thofMay2006: Heraklion(Crete) - Piraeus

deployment ofEGELADOS OBSsinthesouthern AegeanSea

RVPoseidon, ruiseP337, hiefs ientistProf. Dr. Wolfgang Friederi h

•

1st- 12thofJune 2006: Piraeus- Piraeus

deployment ofHamburg OBSs/OBTs at Columboand a tive measurements

RVPoseidon, ruiseP338, hiefs ientistPD Dr. ChristianHübs her

•

27thofMar h - 11thofApril2007: Piraeus- Piraeus

re overy ofEGELADOS and COLUMBOOBSs/OBTs ina joint ruise

RV Aegaeo (Helleni Center of Marine Resear h), hief s ientist Prof. Dr. Torsten

Dahm

Itookpartinallthree ruises,onthe2ndand3rd ruiseIwasresponsiblefortheHamburg

OBS/OBTdeploymentandre overy. This haptergivesanoverviewonthedierent legsand

(30)

2.1 Passive measurements

Dierenttypesofoshoreseismi anddeformationsensorshavebeendeployedaroundColumbo

to better analyzeseismi ityand to get lose enough to potential deformationsour es.

2.1.1 The Columbo OBS/OBT Experiment

Re entstudiesontheseismi ityoftheColumboseamountandthete toni faultsinitsvi inity

werebased on datare orded on islands (e.g. Bohnho etal.,2006; Dimitriadis et al.,2009).

Althoughtheareaallowedtheinstallationofdensenetworksonthesurroundingislands,these

experiments had to deal with large un ertainties in lo ation, espe ially depth, of theevents

due to partly large azimuthal gapsand missingzero-oset stations. With thedeployment of

seismi sensors in the dire t vi inity of the seamount and even on its top, azimuthal gaps

between the surrounding islands were losed and the sensitivity of the network, i.e. the

magnitude thresholdof dete table eventssigni antly de reased (seeTab.2.1):

network remarks 3 losest stations automati triggers

M

L

threshold

NOA permanent SANT, APE,AMOE 4 3.5

EGELADOS temporary NEAK, ANID,IOSI 800 2.0

COLUMBO temporary all OBSs/OBTs 14.000 0.5

Table 2.1: Listof automati triggers ofdierentnetworks. ThesparsepermanentNOA networktriggered onlyeventsabovemagnitude3.5. Thedenserthe temporarynetworksare, themorethe numberoftriggers in reased,while themagnitudethresholdsank. FortheColumbonetwork,thesewereover14.000triggersof whi haround90%were"real"eventsandonly10%mistriggersornoise,thethreshold ouldbede reasedto

M

L

= 0.5

.

O ean bottom stations were installedas shown inFig. 2.1: OBSs between theislands of

Ios, Anidros and Ana, between Columbo and Santorini as well as on the ba k side of the

Columbo alderatowardstheSantoriniAmorgosFault. Furthermore,allOBTswereequipped

with hydrophones inorder to measureseismi signals. Dierential Pressure Gauges (DPGs,

S ripps Institution of O eanography, for te hni al details see Cox et al., 1984) to measure

longperiodrelative pressuresignals from60 s upto 5Hz were mountedonOBSs50 and 51,

i.e. those OBSs losest to the seamount.

TheTiltmeterstationsweredeployed dire tlyontopoftheseamount(seeFig.2.2),sin e

rustal deformation due to magmati as ent is best re orded lo ally. The prole is designed

asymmetri due to unknown depth and size of a possible migrating uid bat h: A deep

sour e auses slight deformation of a larger region, while a shallow sour e auses stronger

deformation of a mu h smaller region. Depth and volume dependen ies of the deformation

signal are modeled inChapter 4.2.2. Furthermore, the OBT prole was deployed along the

(31)

25˚15'

25˚30'

25˚45'

26˚00'

36˚15'

36˚30'

36˚45'

37˚00'

25˚15'

25˚30'

25˚45'

26˚00'

36˚15'

36˚30'

36˚45'

37˚00'

3 nautical miles

0

200

400

600

800 1000

25˚27'

25˚30'

36˚30'

36˚33'

25˚27'

25˚30'

36˚30'

36˚33'

0.5 nautical miles

0

100

200

300

400

500 depth [m]

OBS51

SANTORINI

B)

A)

NAXOS

SCHI

IOSI

NEAK

IOS

ENO26

OBS52

OBS53

OBT55

OBT54

OBT56

OBT57

OBS50

depth [m]

SANT

NAXO

AMOS

AMORGOS

ANIDROS

AMOE

ANAFI

ANAF

ANID

longitude [°]

latitude [°]

Figure 2.1: Experimental setup aroundColumbo Seamount. A)Seismi networkat Columbo and the surrounding islands: triangles mark OBSs with OAS-hydrophone, diamonds OBSs with DPG; inverse triangles mark OBTs, white ir les NOA and EGELADOS landstations. Colors of OBSs/OBTs stand for green=permanentlyrunning,yellow=in ompletedata,red=nodata.

B)DensenetworkatColumboSeamount,symbolsand olors sameasgraphi A.

subsiden e). Adikeissupposedtohaveitssmallestextensionandthusthetiltsignalwiththe

shortestwavelength indire tion of

σ

3

(e.g.RubinandGillard, 1998a). Thus, anasymmetri prole along

σ

3

allows thelargest overage ofpossible depthsand sizes ofan intrusion, even withonly fourtiltmeters.

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4 TWT

5250

5500

5750

6000

6250

FLDR / 1

6500

6750

7000

7250

7500

7750

HH06-01_chan25

25˚24'

25˚36'

36˚24'

36˚36'

25˚24'

25˚36'

36˚24'

36˚36'

25˚24'

25˚36'

36˚24'

36˚36'

25˚24'

25˚36'

36˚24'

36˚36'

time [s]

sigma_1

A)

B)

0.2

0

0.4

0.6

0.8

1.0

1.2

1.4 Columbo

caldera

distance [km]

0

10

20

30 shotnumber (every 5s at 5kn, i.e. 12.5m)

profile

Figure2.2: TiltmeterproleoverColumboSeamount.A)Asymmetri proleoverColumboSeamount along

σ

3

. Tiltmetersaremarkedwithredtriangles,graysymbolsmarkOBSs.

B)Ree tionseismi rossse tionoverthetiltmeterprole(redtriangles),10-timesexaggerated. Despitethe topographyoftheseamount,both,vol ani depositsandpossibleintrusionsarevisibleandwillbedis ussed in hapter2.2.1.

Additionally, all OBTs andOBS50 (between Columbo and Santorini island, seeFig. 2.1)

were equipped with absolute pressure sensors to measure uplift and subsiden e. This was

deemedimportant in ase ofhaving pla ed thesensorontop ofthepossiblyinatingregion,

(32)

positions,polesand zeros et . is giveninAppendix A.

Thedeployment ampaignstarted on 1stof June 2006 fromthe harbour ofPiraeus with

theGermanresear hvesselRVPoseidon. Therstweekoftheexpeditionwasusedfora tive

measurements (see hapter 2.2), whileOBSs and OBTs were assembled on board. The nal

deployment tookpla eon the 10thof June.

There overy ruise nallystartedon midnight of26th/27th of Mar h2007 from Piraeus

withtheGreekresear hvesselRVAegaeooftheHelleni CenterofMarineResear h(HCMR),

in ollaboration with the EGELADOS OBS re overy of the RUB. The target area around

Columbo was rea hed during the evening of the 27th. During an overnight re overy, all

Hamburg OBSs/OBTs ouldbe re overedwithin lessthan 12 hours.

Ex eptfor OBT56,allOBSs/OBTs ouldbesyn hronized withGPSto interpolate their

lo k-drift overthe omplete dataset. All syn hronized re orders showed driftssmallerthan

0.8 s and an average drift of around

0.05 s

month

. OBT 56 shut down only 35 days after its

deploymentduetoawaterleakageinthebattery ylinderwhi hae tedboth,tiltandseismi

re ording as well as the absolute pressure sensor and its logger. All stations were showing

unusualstrong orrosiondamages,su hasrustys rewsand riti ally orroded releaser

hous-ings. The so alled"stainless"-steel of the absolutepressure sensorswaspartly dissolved so

that waterran into the sensors anddestroyed them.

Anotherinteresting aspe twas,thattherewasstrong lime-a retionon allOBSframesif

theyweredeployedabovethe riti aldepthwherelimebe omes ompletelydissolvedinwater

(thelyso linefor arbonatedissolution anbeexpe tedatabout3.500-4.000 mdepth

a ord-ing to Berger, 1973). This a retion was observed for all Hamburg OBSs, OBTs 54 and 57

and mostEGELADOSOBSsdespiteofthosedeployed inthedeep-seatren hsouthofCrete.

But for both OBTs 55 and 56 whi h were standing losest to the aldera rim of Columbo,

no a retion, but strongest orrosion damages were observed after deployment. This might

indi ate ana idi hemi al regimedue to fumarolea tivity insidethe aldera.

Sin e a tive levelling of the OBTs was exe uted the rst time on 11th of June and one

re-levelling(every 48hours) wasawaited aswell asthepassivelevelling oftheOBSsrequires

several hours to rea h its perfe t verti al adjustment, we used data between the 1st of July

2006, 00:00 UTC, and 27th of Mar h 2007, 12:00 UTC, for pro essing. Smooth nonlinear

trends on the tiltmeters for the rst days of data olle tion suggested an initial phase of

sinking and setting of theinstrument frame of about 3weeks afterdeployment. Thisperiod

wasnotused for datainterpretation.

(33)

Figure2.3: S ienti rew ofthere overy ruise(RVAegaeo). Resear hvesselAEGAEO(Helleni CenterofMarine Resear h,Athens)inthe harbourofPiraeus(left-handside photo)andthes ienti rew of there overy ruise: Upperrow, from left to right, Torsten Dahm, MartinHens hand SvenWinter (all University of Hamburg, UHH),WolframGeissler (AWI), KasperFis her (Ruhr-University Bo hum, RUB), Tun ay Taymaz (Istanbul Te hni al University), Reinhard S hrutzky (SEND GmbH), mid row, Wolfgang Friederi h(RUB),ErikLabahn(KUMKiel),Me hitaS hmidt-Aurs h(AWI),frontrow,Celia Rios(UHH), Domeni osVamvakaris(AristotleUniversityofThessaloniki),AndreasS hmidt(RUB).

2.1.2 The EGELADOS Experiment

TheEGELADOSexperiment("ExploringtheGEodymani sofsubdu tingLithosphereusing

an Amphibian Deployment Of Seismographs", (Friederi h and Meier,2008)) was a passive

seismi experiment in the Helleni Subdu tion Zone, ondu ted within the framework of

the Collaborative Resear h Center CRC526 "Rheology of the Earth" at the Ruhr

Univer-sity Bo hum(RUB),funded bythe German Resear hFoundation(DFG).

TheCOLUMBOproje tis arriedoutin lose ooperationwithEGELADOS,both

exper-iments were performed parallel and data is ex hanged between all parti ipating institutions

(seeAppendixF). Between autumn2005andspring 2007,51broadbandseismometersofthe

RUBandtheGermanpoolofamphibianseismographs(DEPAS)weremountedinadditionto

theexistingNOAnetwork on the Aegeanislands and theadja ent Greek andTurkish

main-land. BythemiddleofMay2006,thedensiednetworkwas ompletedwith24DEPASOBSs

between theAegeanislandsandsouthofCretealongtheHeleni tren h. Theoshore

deploy-ment was also arried out by RV Poseidon, shortly before the Columbo expedition and the

re overy wasdone in ooperationwith theColumbo proje twithRV Aegaeo(27thof Mar h

- 11thofApril2007).

A stationmapof EGELADOS isgiven inFig.2.4, a listofstations is given inApp. A.1.

For ourrelo ation andmoment tensor inversionroutines, weonly usedEGELADOSstations

beinglessthan80 kmawayfromColumbofor tworeasons: Limited diskspa e (the omplete

(34)

Figure 2.4: The EGELADOS network onsisted of 104 seismi stations, of whi h 24 were installed oshore inthe entireAegeanSeabetweenlatitudes

34.5 ◦

_N

and

38 ◦

_N

and longitudes

21.5 ◦

_E

and

29 ◦

_E

. In additionto oshore stations,the Aegeanislands, aswellasGreek andTurkish mainlandwere overed with seismometers.

(a 1D model of the Santorini-Columbo vol ani omplex was used, the regional model for

the Aegean Sea is faster and thus led to dieren es in pi ked and al ulated traveltime up

to

3 s

for stations

> 200 km

away). SEISAN and MTInvers are not able to handle a more dimensional velo itymodel.

2.2 A tive measurements

Whilepassiveseismi measurementsasdes ribedabovedeliveranimageofpresenta tivityof

thetargetarea(lo ation,frequen yandstrengthofa tivity), a tivemeasurements wereused

toimagethesubsurfa e(sedimentthi kness,layersofformereruptionset .). Theinvestigation

of thea tive data is the task of two diploma students at theInstitute of Geophysi s at the

(35)

25˚00'

25˚30'

26˚00'

36˚30'

37˚00'

25˚00'

25˚30'

26˚00'

36˚30'

37˚00'

5 nautical miles

25˚00'

25˚30'

26˚00'

36˚30'

37˚00'

0

200

400

600

800 1000

1200

longitude [°]

latitude [°]

depth [m]

Santorini

Ios

Amorgos

Anafi

Naxos

Folegandros

Piraeus

Figure2.5: Prolelinesofa tivemeasurements(RVPoseidon). Tra klines(bla k)ofa tiveproling (ree tionseismi ,magneti andgravimetry,plottedoverthebathymetryofthetargetregion.Denseproles (500mdistan e)wereshotoverthe Columbovol ano,a sparsernetof prolesoverthe Santorini-Amorgos-Zone. AdditionalprolesaroundandthroughtheSantorini aldera ompletethedataset.

2.2.1 Ree tion seismi proling

To investigate shallow expressions of te toni pro esses and/or magmati intrusions, that

mightresultina tivefaulting,theColumbovol anoitselfandtheadja ent

Santorini-Amorgos-Zoneweremappedindetailbymeansofmulti hannelree tionseismi . The ompleteprole

length of the experiment is about 1.500 km (Fig. 2.5). For this a tive measurement, a

GI-Airgunwitha main frequen yof 100 Hz wasusedastheseismi sour e. Data werere orded

bytwo seismi streamers (sensor ables)of 150 mand600 mlength.

Fig.2.6shows ajoint plotof amagneti and aseismi prole,measured dire tlyoverthe

Columbo aldera. Aree tion-freespotunderneath the alderaisinterpretedtobe ausedby

gasand/orhydrothermaluidsinshallowdepths. Thedistributionofvol ano lasti deposits,

espe ially sta ked ones beneath the present aldera rim, suggest at least two eruptions of

Columbo inthe past.

S ope of the ree tion seismi experiment is to identify a tive te toni s as well as to

(36)

Santorini-field intensity

double traveltime [s]

distance

Figure 2.6: Joint seismi and magneti prole over Columbo. Seismi (time-migrated se tion) and magneti (gradiometer)dataa ross the Columbosubmarine vol ano. Vol ano lasti (VC) depositsare hara terized by weak and haoti ree tions. The so alled Poseidon Ridge aligns along the Kameniline. Strongestsignalsofthemagneti eldarefounddire tlyabovetheCaldera(Hübs heretal.,2006).

2.2.2 Gravimetri and magneti proling

TheHamburgSea Gravimeterwasrunningduringthe ompletedeployment ruise,in luding

referen e measurementsonshore inPiraeusat the beginning and at theend ofthe ruise, all

in all around 2.500 km. Magnetometer proles(gradiometer) have been arriedout parallel

to the seismi proling,whi h ledto about 1.500km ofmagneti proles.

The dete tion of magneti heterogeneities an help to model size and depth of possible

former intrusions, whi h makes it an interesting parallel tool to study present and re ent

a tivity at a vol ano, espe ially when depth and spatial extension of possible deformation

sour es areunsurebeforetheexperiment,asitwasthe asefor Columbo.

Fig.2.7 showsmagneti anomalies aroundtheSantorini-Columbo vol ani omplex.

(37)

25˚27'

25˚30'

36˚30'

36˚33'

200

400

400 25˚27'

25˚30'

36˚30'

36˚33'

−200

0

200 25˚15'00" 25˚22'30" 25˚30'00" 25˚37'30" 25˚45'00"

36˚15'00"

36˚22'30"

36˚30'00"

36˚37'30"

36˚45'00"

25˚15'00" 25˚22'30" 25˚30'00" 25˚37'30" 25˚45'00"

36˚15'00"

36˚22'30"

36˚30'00"

36˚37'30"

36˚45'00"

−200

0

200 restfield [nT]

restfield [nT]

B

A

Profile

Tiltmeter

Columbo−Seamount

Santorini

latitude [°]

longitude [°]

Figure 2.7: Magneti heterogeneity around Santorini and Columbo. Magneti resteldon e the internationalgeomagneti referen eeld(IGRF)hasbeensubtra tedfromthedata. Alargedipoleunderneath Santoriniandseveralsmallanomalies on entratedatColumboandthesurroundingseamountsareobserved. Theseanomaliesaresuggestedtoree tformerintrusions.

suggestedto derive from former magmati intrusions, whi h intensify thereferen e elddue

to their ferromagneti sus eptibility on e they ool below the Curie-Temperature. Results

of the magneti measurements will be partly in luded in our nal dis ussion. Additionally,

(38)

(39)

THE NEW HAMBURG OCEAN BOTTOM

TILTMETER (OBT)

While ommon te hniques, su h asInSAR or GPS measurements, have already been proven

to be valuable inmonitoring onshore deformation signals due to magmati re harging, both

methods are not appli able to assess the state of a tivity for submarine vol anoes or the

submarine slopesof vol ani islands. A rst steptowards measuring deformation relatedtilt

oftheseaooristhedevelopmentoftheHamburgO eanBottomTiltmeter(OBT),whosefour

prototypeshavebeendeployedinthepilot-experiment atColumbo. This haptersummarizes

te hni al details oftheOBT andrst experien es inpra ti al use.

3.1 Te hni al des ription

Core of the new Hamburg OBT is a highly sensitive 2 omponent tiltmeter, manufa tured

by Lippmann Geophysikalis he Messgeräte (Germany). Ea h omponent of thetiltsensor is

a pendulum whi h is kept at the zero-position in an ele tri eld. On e the instrument is

tilted,this ausesaslighthorizontala eleration(asasmall omponentofthegravity)onthe

pendulum. Tomaintainthependulumatitszero-position, theinstrumentadjuststheele tri

elda ordingly. The hange involtageusedtokeepthependulum atitszero-positionisthe

measuredvariableto al ulate for tilt.

The instrument in luding a gimbal system is mounted in a 17" glass sphere whi h are

alsoused for the Hamburg O ean-Bottom-Seismometer (Dahm et al.,2002) andthus an be

deployed usingthe standard Hamburg OBS-frame. Thetiltmeter hasatheoreti alresolution

of 2 nrad. Re ording the data with the SEND Geolon MLS (SEND GmbH Hamburg,

Ger-many)withasamplingrateof50Hz(18 bitsresolution),results inanalresolution ofabout

15nrad. Therangeofthesensoris

±4 mrad

(

4000 µrad

),buttherangeisextendedtoseveral tensof degrees through me hani ally relevellingthe sensorplatform(external gimbaling, see

(40)

Be ause this highly sensitive sensor is mounted on a freefall o ean bottom instrument

frame, theinstrument design alledfor ertainrequirements:

•

Atiltoftheterrainofupto

45 ◦

shouldbea ommodatedbytheexternalgimbalsystem

•

The systemshouldlevelthetiltmeter sensorplatform to better than

±50 µrad

•

During operation, thesensor ismounted on thebottom ofthe glasssphere at 3 points bygravity,i.e. it standsfreeon thebottom ofthesphere

Thoserequirementswereobtainedbydevelopingatwophaseslevelingpro ess: The

exter-nalgimbaling systemof made ardani aluminiumrings isxedto thelowerhalfoftheglass

sphere. A highlysensitive leveling stage of thesensorplatform(manufa tured byQuante) is

xed below the enter of theexternal gimbaling's aluminiumrings (seeFig. 3.1). The sensor

platform is onne ted to the external gimbal system through three nylon strings that are

length ontrolledbyasmallele tri motor. Thisallowstolowerandraisethesensorplatform

with an a ura y of better than

±5

◦

. On ethe platform rests on thebottom of the sphere,

theinternallevelingdevi eofQuante levelsthesensordowntoana ura yofabout

10 µrad

.

EXTERNAL GIMBALLING

ARREST

1 GND

2 GIMBAL

2 POWER

3 CHAN

ELECTRONICS

CONTROL

A)

B)

SENSOR

TILT

INTERNAL

GIMBALLING

Figure3.1: Sket handphotooftheOBTsensor sphere. A)Sket hofthesensorsphereoftheOBT: The internal gimbaling system is hostingthe tiltsensor and is hanging inthe external ardani gimbaling. Nylon stringsenable tomountand unmountthesensor. B) Thealuminiumrings ofthe externalgimbaling and howtheyaremountedtothesphereareshownintheupperphoto. Ontop,thesteeringele troni sare visible,throughtheglasssphere,thelevelingstagehostingthetiltmeter anbe rudelyseen. Thelowerphoto

(41)

On e the system is deployed, it an be programmed to rest for a given number of days

beforethe rstleveling ofthe systemis arried out. Afterwards, thesystemisreleveledon a

regularbasis(inour aseevery48hours, sin eitwasun lear howfastthesignals ouldrea h

thelimitsofthe instrument's dynami rangeduringthispilotexperiment). Onedisadvantage

ofthesystemisthattiltsteps ausedbylevelingofthesensorplatform(internalandexternal

gimbaling)arenotmeasuredbyanindependentsensorandhavetobeestimatedandremoved

fromdataduring postpro essing.

A)

B)

Figure 3.2: Photos of absolute pressure gauge and OBT deployment A) The absolute pressure gauge(verti al ylinder)ismountedatthesideoftheOBS/OBT-frame. Itusesapowersupplyindependent fromthe tiltmeterand MLSdatalogger, itsdataa re orded onaseparatelogger (horizontal ylinder). B) DeploymentoftheOBTsystematColumboseamount(Santoriniislandintheba kground). Theinstrument sinkstothesea bottomusing anan hor-weight (ironrails), it anbere overed byreleasing theweight and as endingduetoitsownbuoyan y.On eitrea hestheseasurfa e,radiobea onandashlighthelptondit eveninroughseaorduringthenight. ThebatteriesandtheMLSdataloggerarestoredinthelargepressure tubesonthebesideoftheframe.

In addition to the two tilttra es, we also re orded the temperature inside the sphere in

order to dete t a possible temperature dependen eof the tiltsignals. The instrument frame

alsohosts a OAS-hydrophone to olle t seismi data(lower orner frequen yis

f

c

= 0.3 Hz

) and ahighly sensitive absolute pressuresensor (seeFig. 3.2) manufa tured by Paros ienti

to observe possible uplift and subsiden e with a resolution of

0.1 mbar = 10 P a

whi h orrespondsto averti aldispla ement of

1 mm

.

3.2 First experien es in pra ti al use

Following subse tions give a short overviewwhat kinds of signals an be measured withthe

OBT- fromits rst stepsin thelaboratoryto its pilot-deployment intheAegeanSea.

3.2.1 Calibration in the laboratory

The tiltsensor was alibrated in the laboratory using a tilt-table as shown in Fig 3.3 by

On the interrelation of fluid-induced seismicity and crustal deformation at the Columbo submarine volcano (Aegean Sea, Greece)