JOURNAL
OF GEOPHYSICAL
RESEARCH,
VOL. 98, NO. B4, PAGES 6607-6622, APRIL 10, 1993
TomographicImage of the Mid-Atlantic Plate Boundary
in Southwestern
Iceland
INGI TH. BJARNASON1 AND WILLIAM MENKE
Lamont-Doherty Earth Observatory of Columbia University, Palisades, New York
Department of GeologicalSciences,Columbia University, New York
(•)LAFURG. FL6VENZ
Orkustofiaun, Reykjam7•,Iceland
DAVID
CARESS
Lamont-Doherty Earth Observatory of Columbia University, Palisades, New York
The 170 km SouthIcelandSeismicTomography(SIST) profileextendsfrom the westand across
the Mid-Atlantic Ridge spreading center in the Western Volcanic Zone and continues obliquely
through the transformzone (the South Iceland SeismicZone) to the westernedgeof the Eastern
Volcanic Zone. A total of 11 shot points and 210 receiver points were used, allowing precisetravel
times to be determined for 1050 crustal P wave rays and 180 wide-angle reflections. The large
amplitudesof the wide-anglereflectionsand an apparent refractor velocity of 7.7 km/s are interpreted to be from a relatively sharp Moho at a depth of 20-24 kin. This interpretation differs from
the earlier models(basedon data gatheredin the 1960sand 1970s),of a 10-15 km thick crust underlain by a upper mantle with very slow velocity of 7.0-7.4 km/s. Nevertheless,these older data
do not contradict our new interpretation. Implication of the new interpretation is that the lower
crust and the crust-mantle boundary are colder than previously assumed. A two-dimensional totoographic inversion of the compressionaltravel times reveals the following structures in the crust:
(1) a sharpincreasein thicknessof the upper crust ("layer 2A") from northwestto southeastand
(2) broad updomingof high velocity in the lower crust in the Western Volcanic Zone, (3) depth
to the lowercrust ("layer 3") increasesgraduallyfrom 3 km at the northwesternend of the profile
to 7 km at the southeasternend of the profile, (4) a low-velocityperturbation extendsthroughout
the upper crust and midcrust into the lower crust in the area of the transform in south Iceland
(South Iceland SeismicZone), and (5) an upper crustal high-velocityanomalyis associatedwith
extinct
central
volcanos
northwest
of the Western
Volcanic
Zone.
The
travel
time
data
do not
support the existenceof a large (> 0.5 km thick) crustal magma chamber in this part of the
Western Volcanic Zone but do not exclude the possibility of a smaller one.
INTRODUCTION
Geologyof Transect
Iceland was built by volcanic rocks generated at a midocean ridge, beginning about 16 Ma. The volcanic pile
observedat the surface is conventionallydivided into four
the boundary betweenthe volcanicpile originatingfrom the
previous position of the spreadingcenter and the volcanic
pile originating from the present position of the spreading
center in the WVZ. The EVZ is a propagating rift that
started its southern transgressionfrom central Iceland at
3 Ma [$cemundsson,
1979]. The formationof the EVZ has
stratigraphicage groups: (1) postglacial,last 9000-13,000 been interpreted as a ridge jump in progressfrom the WVZ
1974; Seeroundsyears;(2) upperPleistocene,
backto 0.7 Ma; (3) Plio-Pleisto- to the EVZ [Pdlmasonand $cemundsson,
andEir•7csson,
1982;(Jskarsson
et al.,
cene,0.7-3.1Ma; (4) Tertiary,olderthan 3.1 Ma [Seerounds- son,1979;Einarsson
son, 1979].The sourceof the volcanismis a the rift system 1985]. The EVZ is a younggeologicalfeaturein southIcein Iceland. In southwestand south Iceland the rift system
consistsof the overlappingspreadingcentersin the Western
and EasternVolcanicZones(WVZ, EVZ; Plate i and Figure 1). The continuationof the spreadingcenterin central
and northeast
Iceland
is called the Northern
Volcanic
land, and most of its volcanic products are confined within
the zone. Therefore most of the volcanic pile between the
WVZ and EVZ originatesfrom the WVZ or off-axis volcanism betweenthese two overlappingspreadingcenters.
The SouthIcelandSeismicZone(SISZ) is a fracturezone
Zone
(NVZ). The spreadingcenter in southwestIceland made
a ridge jump at 7 Ma from the Snsefellsnespeninsula to
at a nascent stage of development that accommodatesleftlateral transform motion between the the EVZ and rifting
its currentpositionin the WVZ [Pallmason
and $ermunds- on the ReykjanesRidge (Plate i and Figure 1). However,
son, 1974; $eemundsson,
1979]. The BorgarfjSrduruncon- no east-west surface fractures are observed within the zone.
formiry (BUC, at 820 km on the profile,Figure 1) marks The left-lateral motion of the SISZ is accommodatedby
"bookshelf"type faulting (i.e., many parallel, north-south
striking
strike-slipfaults) [Einarssonand Eir•7csson,
1982;
INowat Caxnegie
Institution
of Washington,
Washington,
D.C.
Hackmanet al., 1990; Bjarnasonet al., 1993]. The SIST
Copyright1993by the AmericanGeophysical
Union.
Paper number9ZIB02412.
profile traversesthe SISZ obliquely between 720-735 km
(Figure 1).
Seismic Structure of Iceland, Previous Work
0148-0227/93/9ZIB-02412505.00
The earliest publications on results from seismic refrac6607
6608
Bjarnason et al.: Tomographic Image of Mid-Atlantic Plate Boundary
Latitude
(D
o
oE
o
o
•,-.
o
o
•
ß
epnl!lo-I
Bjarnasonet al.: TomographicImage of Mid-Atlantic Plate Boundary
c:
6609
6610
Bjarnason
et al.: Tomographic
Imageof Mid-AtlanticPlateBoundary
tion experiments
in Icelanddate backto the 1960sand early
the structureof the Iceland-FaeroeRise (IFR) (southeast
1970s[Bath,1960;Tryggvason
and Bath, 1961;Pallmason, of Iceland)is well determined.The IFR is thoughtto have
1971;Pallmason
andScemundsson,
1974;Zverevet al., 1975]. been formed by interaction of the Iceland hotspot with the
Pallmason
[1971]synthesized
the resultsof over80 low-resolu- Mid-Atlanticspreadingcenterat 40-55 Ma [Morgan,1978;
tion (1-4 km station spacing)refractionprofilesand compiled an averagevelocity model for Iceland. The model con-
sistsof four crustal layersand an anomalouslyslowupper
mantle with P wavevelocityof 7.2 km/s. The velocitiesof
the crustal layerscorrespondcloselyto velocitiesof oceanic
crust (layers1, 2A, 2B, and 3) but are 2-3 times thicker
in Iceland. Depth to the lower crust varies considerably
between 3 and 10 km, and the total crustal thickness is
rather poorly constrainedbut likely is between 10 and 15
km [Pallmason,
1971;Fldvenz,1980;F15venzand Gunnarsson,1991]. Fldvenz[1980]showedthat Icelandicrefraction
data could be better interpreted in terms of a continuous
velocity-depth profile rather than a layered structure with
sharp discontinuities.This interpretation is consistentwith
recentmodelsfor oceaniccrust[e.g.,Minshullet al., [1991].
Interpretations of anomalouslyslow upper mantle under
Vink, 1984]. The crustal structure of the IFR is similar
to that of Iceland but is thicker, 28-35 km thick, and has
subcrustalvelocityof 7.8 km/s [Zverevet al., 1975; Bott
and Gunnarsson,
1980].
A low-resolution,
three-dimensional
tomographic
imageof
Iceland,basedon teleseismic
P wavetravel times [Tryggvasonet al., 1983]demonstrates
significantlateralheterogeneity in the crust and upper mantle of Iceland. Compressional
velocitiesin the upper 75 km (the "pixel" height in their
inversion)increasewith lateral distancefrom the spreading
centerby about 4%. (Their type of inversionis not able to
determinethe averagevelocity,only lateral gradients.)
EXPERIMENTALDESIGN, DATA ACQUISITION
AND ANALYSIS
The SIST profile imagesa vertical slice through Earth
Iceland(velocitiesbetween7.0 and 7.4 km/s) werebasedon
the followingreasoning:(1) The highestvelocitiesobserved that strikesnorthwest-southeast
acrosssouthernIceland(Plin these older seismicrefraction experiments are between ate 1). It crossesthe major geologicalfeaturesassociated
7.0-7.4km/s [Bath, 1960;Pallmason,1971; Gebrandeet al., with MAR rifting, suchas the BUC, WVZ, and SISZ. The
[1980];(2) teleseismic
travel timesmeasuredin Icelandare source-receiver
distribution is fine enoughto permit highabout 1.6 s slower than travel times measured on the Euroresolution tomography in the upper crust, where targets
peancontinentand Greenland[Tryggvason,
1964;Longand suchas magma chambers,fault zones,and shallowvolcanic
Mitchell, 1970] and couldbe explainedwith upper mantle intrusionsmay occur. The profile is long enoughthat seisvelocitiesof 7.4 km/s extendingto depthsof 160-240km; mic rays from the longestoffsetspenetrate into the mantle,
(3) hightemperatures
beneathIceland(dueto the hotspot), permitting conventional, low resolution refraction measure1000-1100øCat 10-20km depthproducea largepercentage ments of the lower crust, Moho, and uppermostmantle.
of melt that leadsto low mantle velocities(in the 7.0-7.4
km/s range) [Pallmason
and Scemundsson,
1974; Gebrande Data Acquisition
et al., 1980], (4) a high-conductivitylayer is found under
a large part of Iceland in approximatelythe depth range
10-20 km, indicating partially molten upper mantle in that
depth range [Bebloand BjSrnsson,1980; Eysteinssonand
Hermance,1985].
The multipleoffsetrefractionexperimentwasdesignedin
the followingway: Longrangeshots,necessaryto determine
the crustal structure, were placedat distancesup to 170 km.
Higher shot-receiverdensity was put around the most interesting imaging targets, the WVZ, and the SISZ to improve
The RRISP WorkingGroup[1980]ran an 800-km-long
resolution. The highestshot and receiverdensity was over
refractionprofile that traversedthe flanksof the Reykjanes and around the WVZ and in the South Iceland Lowlands
Ridge and continued on land over central Iceland, in order
(between695and795km onthe profile),with 10-20km shot
to better determine the seismicstructure of the upper manspacingand 0.5-1.0km stationspacing.The resultingprofile
tle under Iceland and the Mid-Atlantic Ridge (MAR). The
was 170 km long, with the WVZ and SISZ approximately
RRISP study determined that 10-m.y.-old crust on the subin its central part, with total of 11 shot points and 210 reoceanicReykjanesRidge is underlainby slow (7.7 km/s)
ceiverpointsdistributedalongthe profile (Table 1). Signifmantle at 10 km depth but found that velocity increasesto
icant geologicaland geographicalfeatures along the profile
a more "normal"upper mantle velocityof 8.2 km/s at 16
are summarizedin Table 2. Although three-componentdata
km depth [Goldflamet al., 1980]. Fundamentallydifferent
upper mantle velocitieswere found beneath Iceland, ranging
were recorded, only P wave arrivals are addressedhere.
from 7.0 km/s at the baseof the crust at approximately15 Data
km depth to 7.4 km/s at the maximumpenetrationdepth
of 30 km in this experiment[Gebrandeet al., 1980]. The
Prominent coherentarrivals are visible to about 120 km,
RRISP experimentdid not supportthe findingsof previous and weak arrivalsare visibleto about 170 km on the profile
lower resolution Soviet studies of 40-60 km crustal thick(Figure 2). Large lateral velocity variationsare observed
nessand upper mantle velocitiesunder Icelandin the range alongthe profile. For example, there is up to I s travel time
7.8-8.2km/s [Zverevet al., 1975]. Accordingly,
the RRISP difference
betweenstationswith the samerange(20-40km)
experimentaffirmedinterpretationsof anomalouslyslowup- from eachof the end shots,AK and JL (Figures2 and 3).
per mantle under Iceland and the relatively "normal" crustal This travel time differenceis an indication of a large difand upper mantle oceanicvelocity structure of the Reyk- ferencein the upper-crustalstructure at the end points of
janes Ridge.
the profile. Consistenttravel time delays are observedat
The seismicstructureof the KolbeinseyRidge,the part of stationsin the SISZ, for both distant and closeshots(e.g.,
the MAR immediatelynorth of Iceland,hasnot beenstudied in 105-120 km range from the AK shot, and in 60-70 km
in any detail [F15venzand Gunnarsson,
1991]. However, rangefrom the JL shot, Figures2 and 3). Travel time ad-
Bjarnason
et al.: Tomographic
Imageof Mid-AtlanticPlateBoundary
6611
TABLE 1. Acquisition Parameters
Parameter
Description
170 km refraction profile, approximately linear crossingridge and transform
11 shot points in water, using 10 to 220 kg dynamite shots
210 receiver points
three-componentMark Product L22D geophones
digital recording with RefTek 72A and Sintrex PRS-4 instruments
digitization at 200 samples/sand 12 and 16 bits/sample
timing to 0.01 s accuracy by Omega radio time
shot and receiver points located by both GPS and aerial photos,
with approximately 30-m accuracy
1050 three-componentseismogramswith good signal to noiseratios
experimentperformedwith small (four person)field team over 2.5 months
TABLE 2. SignificantFeatures
Name
AK (Akrar)
BF (BorgarfjSrdur)
HF (HvalfjSrdur)
SV (Stifiisdalsvatn)
Feature
Profile,km
Latitude*
Longitude*
profile
840.0
64.6000
- 22.4259
shot
shot
shot
shot
841.0
818.3
794.9
774.5
64.6052
64.4812
64.3524
64.2393
-22.4425
-22.0673
-21.6841
-21.3530
TV (?hingvallavatn)
shot
762.6
64.1730
-21.1610
UV (Ulfij(•tsvatn)
HV (Hestvatn)
TA (Thj(•rsa)
YR (Ytri Rang•)
ER (Eystri Ranga
JL (JSkulldn)
shot
shot
shot
shot
shot
shot
753.9
734.4
723.6
708.4
695.2
669.5
64.1243
64.0148
63.9538
63.8677
63.7926
63.6455
-21.0213
-20.7099
-20.5385
-20.2985
-20.0913
-19.6911
profile
Borgarfj5rdur
HvalfjSrdur
fjord
fjord
670.0
813.8-819.0
794.5-798.0
- 19.6989
- 21.9925
- 21.6776
64.2587
-21.4095
Longitude*
64.4851
-22.0788
64.3695
-21.7346
Stardalur
extinct volcano
WVZ
spreadingcenter
753.8-767.0
64.1246
-21.0189
64.1975
-21.2319
volcanic center
transform
740.0-745.6
721.3- 734.0
64.0778
63.9405
-20.7991
- 20.5013
64.1235
-20.8884
64.0125
-20.7035
Grlmsnes
SIS Z
778.0
63.6483
64.4562
64.3502
Latitude*
* All latitudesandlongitudes
are for the pointof projectionof featuresontothe profile.
vances(highvelocityperturbations)
areassociated
with two
extinct centralvolcanosnorthwestof the WVZ (Skardsheidi
and Stardalurcalderain distanceranges31-39km and 62-71
km fromthe AK shot,Figures2 and 3). A numberof later
phasesare observedin the data. The most prominentlater
arrival hasthe characteristics
of a wide-anglereflection.It is
observedin fourrecordsectionsof the shotswith the largest
offsets.The onsetof this phaseis at a reducedtravel time
of 1.5 s andat an offsetof approximately
80 km (Figure2).
Travel times were pickedby hand and are believedto be
accurate to about 0.015 s for first arrivals and 0.050 s for
later arrivals.
Data Analysis
The similarity of Icelandic crust to oceaniccrust is easily
seenin the exemplarytravel time curves(Figure 3). Layer
2A (3.5 + 0.5 km/s) is at the surfaceof the HV shot, followedby layer 2B (5.2 + 0.4 kin/s) and layer 3 (6.5 + 0.3
kin/s). Layer 2B is just belowthe surfaceat the AK shot.
Velocitiesof 7.2 -4-0.2 km/s havebeenattributed to a slow
TomographicInversion
In this study we use the tomographicinversionformula-
tion of Caresset al. [1992]. The velocitystructureis parameterizedas a staggeredtwo-dimensionalgrid divided into
triangles. The velocity within each triangle is determined
by linear interpolationof the vertex nodal values,giving a
continuousvelocity model with constantvelocity gradients
in each triangle. Rays can be traced three dimensionally
throughthe two-dimensional
model,allowingthe actual geometry of the seismicexperimentto be accuratelyincorporated into the inverseproblem. Geometricalray paths and
travel time residualsare calculatedby ray tracing through
a starting velocity model; the residualsare then inverted for
the perturbationto the starting model which is requiredto
satisfy
thetraveltimedata.Theinversion
isregularized
by
the additionof a smoothingoperatorwhichpenalizesroughnessas measuredby the secondderivativesof the velocity
perturbation.
upper mantle in Iceland. However, in this study a higher
The initial data analysisconsistedof one-dimensionalforward modeling of individual record sectionsassociatedwith
maximumapparentvelocity of 7.7 km/s is observedfrom
the shot points. The resulting11 one-dimensional
velocity
the JL shot.
modelswere linearly interpolated into the two-dimensional
starting velocity model used for the tomographicinversion
Further data analysiswas done in two stages. First, the
two-dimensionalcrustal seismicvelocity structure along the
SlST profilewasimagedby tomographicinversionof crustal
travel times. Second,the nature of the observeddeep reflector was explored by two-dimensionalforward modeling of
the reflectedarrivals usingthe tomographyresults.
(Plate 2). We used1020travel time picksin the inversion;
the associated
ray pathsthroughthe startingvelocitymodel
are shown in Plate 2. We inverted the travel time data us-
ing a number of different grid spacingsand with differing
amounts of smoothing applied. We found that somewhat
Bjarnason
et al.:Tomographic
Image
ofMid-Atlantic
PlateBoundary
6612
•.,
z
ILl
[oes]cj'9/X
- J.
[oes]cj'9/X
- .L
Bjarnason
etal.:Tomographic
Image
ofMid-Atlantic
PlateBoundary
2.0-
2.0 -
AK
6613
2.0 - AK and JL
•6.88
6.55
0.75-
4.66
''•.
<50.75-
''.-"'-'r':•-•.
0.75-
6.89
4.90' -••
-0.5
0
I
i
100
200
-0.5
0
I
I
lOO
200
-0.5
1.5 -- HV west
I
200
RANGE [km]
RANGE [km]
RANGE [km]
I
lOO
1.5 - HV east
6.43
ß
'••'•.'17
5.05
0.75-
3.63
3.50'...x• •
0
•
•
62.5
125
RANGE [km]
0
0
•
•
62.5
125
RANGE [km]
Fig.3. P wavereduced
traveltimecurves
fortheshots
AK andJL (which
areat thewestandeastends
ofthe
profile,
respectively),
andshotHV (whichis close
to thecenter)(Plate1). NotethattheP wavehasapparent
velocities
(inkm/s)thatareclose
to thecharacteristic
velocities
oftheoceanic
layers
2A,2B,2C,and3. The
7.74km/sapparent
velocity
observed
at largeranges
forJLshotisslower
thantypical
measurements
oftheupper
mantlebutis muchhigher
thanpreviously
accepted
values
(7.0-7.4km/s). Notethelargedifference
in travel
time(upto I s) between
shots
AK andJL at shortoffsets,
which
isrelated
to largedifferences
in uppercrustal
thickness
betweenthe eastandwestendsof the profile.Note the broadtraveltime delayscenteredat ranges65
km and 110km from shotsJL and AK, whichis due to a low-velocityzonein the SISZ area.
better resolutionwas achievedwith a coarsergrid and less
locitymodelis 0.033s, a 97%variancereductioncompared
smoothing
than with a finergrid andmoresmoothing.The
highestresolution
is obtainedin the areaof the WVZ down
to approximately5 km depth. In this area, stationspacing
is 0.5 km and shotspacingis 10 km (Plate 2). Away from
the WVZ the stationspacingvariesbetween0.5 and 1.5 km,
andthe shotspacingis approximately
20 km. Our preferred
modelwasobtainedusinga 1-kmverticaland 10-kmlateral
grid spacing;
this velocitymeshextendsto a depthof 15
km, hasa lateralextentof 190km, andincludes328veloc-
to the 0.19-s rms misfit relative to the starting model. The
rms misfit of the final model is close to the estimated 0.015-
s average
pickingerror,indicating
that furtheriterationsof
the inversion are unnecessary.
Wide-AngleReflectionModelling
In our data, clearwide-anglereflectionsare observedin
the record sectionsof the four shots with the largest off-
sets.Assuming
theseare fromthe samereflector,we deter-
ity nodesand585triangles.Plate3 presents
the resultsof minedits depthby forwardmodelingof travel timesusing
ray-tracing
program[Cerveny
andPsencik,
1981].
the tomographic
inversion,of the traveltime data, which theSeis81
The upper15 km of the velocitystructureweredetermined
are discussedin detail in the next section of the paper.
inversion.
The lowerpart
The resolvingpowerof the tomographicinversionis as- by the resultsof the tomographic
sessed
with an impulseresponse
method. Synthetictravel of the modelwasadjustedby trial-and-errorforwardmodelthe velocitystructimesare computedfor the actualsource-receiver
configu- ing. Our datasetonlypoorlyconstrains
ration, but with a velocitystructurethat is homogeneous, ture between15km depthandthe depthto the top of the refewraysturn at thosedepths.In thisdepth
exceptfor a perturbationat a singlenode. Thesedata are flector,because
invertedto yield an estimatedvelocitymodel,whichowing rangewehaveuseda velocityof 7.1 km/s at 15 km depth
linearlyto 7.2-7.25km/s immediatelyabovethe
to imperfectresolution
haveperturbations
at morethan one increasing
node. In our case,we find the resolutionto be very good reflector. The travel times of the wide-anglereflectionscan
(10 km horizontally,
I km vertically)downto 8 km depth be modeled with a reflector at 20-22 km depth under the
SouthIcelandLowlands,deepeningto 24 km northwestof
(Figure4).
is modeled
as7.5 km/s
The fit of the final velocitymodelto the traveltime picks the WVZ (Figure6). The velocity
indicatetherearenosystematic
errorsin the modelandthat just belowthe reflectorandhasa positivegradient,sothat
the overallfit is excellent(Figure5). The root-mean-square raysturningjust beneaththe reflectorhavean apparentve-
(rms)misfitof the traveltimedatarelativeto the finalve-
locityof about7.6-7.7km/s and hencefit the traveltimes
6614
Bjarnason
et al.- Tomographic
Imageof Mid-AtlanticPlate Boundary
Starting
0
El 4
!
6
,,=
8
Velocity
Model
6.5
•10
•
12
14
84O
820
800
780
760
740
720
700
680
66O
Northwest to Southeast Distance (km)
7
6
5
4
3
Velocity (km/sec)
Raypaths
0
•
ß•'•,4'.,',.. ',i :.•,':•,"•'•'•.•'.•'•' '
-' '• "•'• "' •'" r,',½,.
,•,•.,,-,,:•2',•,',•,-• •,,•,,•,• ,'
..............
4
•
6
.c
8
•
10
½n 12
_
14
840
820
800
780
760
740
720
700
680
660
700
680
660
Northwest to Southeast Distance (km)
8
7
6
5
4
3
Velocity (km/sec)
Velocity
Perturbation
Model
•zo
• 12•o
840
820
800
780
760
740
720
Northwest to Southeast Distance (km)
1.5
1.0
0.5
0.0
-0.5
Velocity Perturbation (km/sec)
Plate 2. (Top) The two-dimensionalstarting model for the tomographicinversionbasedon interpolatedonedimensionalmodelsalongthe profile. (Middle) The 1050P waveray pathsthroughthe startingmodel. (Bottom)
Perturbations to the starting model as determined by the tomographicinversionof P wave travel time residuals.
Note that the perturbationsare difficultto interpretvisually,becausethe starting modelis laterally heterogeneous.
of the observedmantle refractor in shot JL in Figure 6. We
interpret the reflector to be the Moho and discussthe reasoningfor this interpretation below.
approximately 2.0 km, and east of the WVZ the thickness
varies between 2.0 and 3.0 km. No significant sedimentary
layers occur along the profile.
Earlierrefractionsurveys[Pdlmason,1971;Fl6venz,1980]
RESULTS
AND DISCUSSION
Crust
On the basisof our data and previousknowledgeof the
velocity structure of the Icelandic crust we find it convenient
to subdividethe crust in Iceland into three different units,
upper crust, midcrust, and lower crust.
showthat the upper crust in Iceland varies in thicknessfrom
zero to 2.5 km.
The
lowest
values
are found
in areas of
greatestglacialerosionand high alteration (secondarymineralization) connectedwith extinct central volcanos. The
generally low values in the area west of the WVZ are due
to erosion in glacial times. The maximum thicknessof the
upper crust is controlled by alteration processes.Almost all
1. Upper crust (definedhere as everythingabovethe 5.0 the material of the upper crust is made up of subaerial lava
km/s contour,i.e., approximatelylayer 2A) hassignificant flows that subsidedue to the load of subsequenteruptive
lateral variation and varies in thickness from 0.7 to 3.0 km
material. As the basalt is brought to greater depth it be(Plate 3). The upper crust has a very constantthickness comesincreasingly altered due to secondarymineralization
between 0.7 and 1.2 km west of the WVZ.
At the WVZ
and the compressionalwave velocity increases. When the
there is a sharp discontinuityin the thicknessof the upper temperature reachesapproximately 250øC, heavy minerals
crust. Under the WVZ the thicknessof the upper crust is like epidotestart to form, and the velocityexceeds5.0 km/s
Bjarnason
et al.: Tomographic
Imageof Mid-AtlanticPlate Boundary
BUC
0
W-VZ
6615
SISZ
8
c•1o
,
14
840
6.5
820
800
780
760
,
i
740
720
680
700
660
Northwest to Southeast Distance (kin)
AK
BF
HF
SV
TV
UV
HV
YR
TA
ER
o
•
•
JL
,
..
•o
840
820
800
780
760
740
720
700
680
660
Northwest to Southeast Distance (kin)
8
7
6
5
4
3
Velocity(km/sec)
AK
840
BF
HF
880
SV
600
TV
780
UV
HV
760
TA
740
YR
720
ER
700
JL
680
660
Northwest to Southeast Distance (km)
8
7
6
5
4
3
Velocity (km/sec)
Plate 3. (Top) The major contoursof the compressional
wavevelocityasdeterminedby the tomographicinversion.
The contoursoutlinethe shapeof the uppercrust (above5.0 km/s), the midcrust(between5.0-6.5km/s) and the
lowercrust (below6.5 km/s). The 7.0 km/s contouroutlineshigh-velocityzoneswithin the lower crust. Major
geological
featureson the profileassociated
with the MAR rifting are the BUC, WVZ, and SISZ. (Middle) Compressionalwave velocity in the upper 15 km of the crust as determinedby the tomographicinversion.Locations
of the 11 shotsalongthe profileare indicated(arrows). Seetext for discussion.(Bottom) Upper 5 km of the
tomographic model.
to off-ridgevolcanism.Severaldrill holespenetratealmost
gradientwithin the volcaniczoneand in its surroundings
is through the midcrust to just over 3 km and showthat it is
alwayshigherthan 100øC/km [F16venzand Seemundsson,made of basalticlava flowswith increasingcontentof higher-
which we defineas the top of the midcrust. The temperature
1993]leadingto a maximumthickness
of -,• 2.5 km for the
upper crust. Thicker upper crust can only be formedin
temperature alteration minerals like epidote.
3. The depth to the top of the lower crust (i.e. layer
2-3 boundary, defined by a sharp decreasein the velocity
peraturegradientlowerthan 100øC(e.g.,as a consequence gradientat -,• 6.5 km/s) is on average4.5 km, and it is 14of ridgejumpsor propagationof rifts througholdercrust). 20 km thick (Plate 3.) The depth is relativelyconstantwest
The thick uppercrustin the SouthIcelandLowlandis prob- of the WVZ, between 3.0 and 4.0 km. Under the WVZ the
depth is 4.3 km, and it increasesto 6.0-7.0 km east of the.
ably due to the ridgejumping from the WVZ to the EVZ.
2. The midcrust(definedhereas everythingbetweencon- WVZ. Mutter and Mutter [1993]examininga largesuiteof
tours 5.0 and 6.5 km/s, i.e., layers2B and 2C) has signifi- publishedoceaniccrustal profiles,find that the depth of the
cant lateral variation in thicknessfrom 2.0 to 4.5 km (Plate layer 2-3 boundaryis remarkablyconstant4.5 km belowsea
3). The midcrusthas relativelyconstantthicknessof 2.0- level, when depthsare correctedto paleodepthat the time
2.5 km under and to the west of the WVZ (except around of seafloorcreation. The average depth of this boundary
Iceland when lava is accumulated on older crust with a tem-
810 km on profile). There is rapid thickeningof the mid-
below south Iceland
crust east of the WVZ, where the thicknessis between 3.0
The nature of layer 3 in Iceland is still unknown since the
deepestboreholein Iceland terminatesa few hundredmeters
and 4.5 km. The thickeningof the midcrust may be due
is also about
4.5 km below sea level.
6616
Bjarnason
et al.: Tomographic
Imageof Mid-Atlantic
PlateBoundary
,
o
•1
q'
•)
io
o
•1
ß
ß
.
o
t%l
q'
ß
•-
•
iI0
o
t%l
•-
o
.
o
•1
q-
•
•
o
C•l
q-
o
t%l
qr
,o
•
o
(•)•) mde•
t%l
ql,
thl
q-
,•
to
o
thl
qr
Bjarnasonet al.: TomographicImageof Mid-Atlantic Plate Boundary
1.0-
1.0 -
AK
1.0 -
BF
6617
HF
0.5-
0
I
I
60
120
0
-25
RANGE [kin]
1.0--
•
•
37.5
100
1.5--
0.5-
0
UV
0
•o
o'o
-50
RANGE [km]
1.5--
TA
• 0.750
I
o
-20
0.75-
•
•
-5
RANGE [kin]
40
0
I
-80
0.75-
0
0
-12o
-50
RANGE [km]
RANGE [km]
1.6-
0.75-
0.8-
0
0
-25
JL
- 120
RANGE [km]
8'0
1.5 - YR
0.75-
ER
o
RANGE [km]
HV
RANGE [km]
1.5--
100
RANGE [km]
• 0.75-
1.5--
- 50
RANGE [km]
SV
-70
0
-60
0
RANGE [km]
Fig. 5. P wavetravel time data (dots) and fits (solidcurve)from the tomographicmodel. Noticethat the misfit
rarely exceeds0.03 s.
6618
Bjarnasonet al.: TomographicImageof Mid-Atlantic Plate Boundary
AK
BF
850
800
750
700
660
850
800
Distance(km)
]..8--
750
700
660
Distance(km)
AK
BF
.;'
ß ..•.. ,
•.;•,.%
.-'-- ,,½:..
.
.
o
;"I
I
62
o
0.4
124
i
5
RANGE [km]
52.5
100
RANGE [km]
HF
,-'15
JL
...................
,-
c320
25
850
800
750
700
660
c3ii
••J•J•••J••[•j•
800
850
Distance
(km)
1.3-
750
700
ilt•ll•
60
Distance(kin)
HF
1.7-
,5 -0.65-
0.1
6
I
,
57
108
RANGE [km]
-0.4
18
94.5
171
RANGE [km]
Fig. 6. Traveltime (dots)of wide-anglereflectionsand fit (solidcurve)from forwardmodelingof reflectionswith
Seis81program[Cerven•/andPsencik,1981]. The tomographicmodelwasusedin the upper15 km. The Moho is
best modelledwith slightdeviationsform a linearinterphase(dashedline in the ray diagram),whichdipsgently
to the west.
Bjarnasonet al.: TomographicImageof Mid-Atlantic Plate Boundary
above the boundary.
4.
The
crust
in south
Iceland
is 20-24
km
thick
velocity, km/s
with
0
a sharp reflector at the bottom which we interpret as the
Moho. This is significantlythicker crust than has been postulated in previousinvestigations.A relatively sharp Moho
is probably a common feature in Iceland, at least outside
the volcanic
This
new
6619
4
8
4
zones.
model
does not
contradict
the
older
data
of
Pallmason
[1971]sinceweinterprethislayer4 (compressional
velocity7.2 km/s) as being the lowestpart of the crust,
underlain by a sharp Moho.
Layer 3 is disproportionatelythicker than standard oceanic
crust. The ratio of the thicknessof layer 2 to layer 3 is about
3:5 (- 0.6) for typical oceaniccrust from a slowspreading
centersuchas the MAR [Minshullet al., 1991]but 4.5:17.5
(- 0.25) for southIceland. The disproportionatethickness
of layer 3 has been reported for thick oceaniccrust elsewhere
in the world [Mutter and Mutter, 1993].
Our lower crustal
velocities
are not well constrained
but
reachabout7.2-7.25km/s just abovethe postcriticalmantle
reflector. A jump from 7.2 to 7.25 km/s abovethe reflector to 7.5 km/s just belowit, increasingto 7.6-7.7 km/s a
few kilometers lower, does a reasonablejob in fitting the
mantle
reflection
and refraction
travel
24
time data but is not
especially well constrained. We interpret the reflector to
be the Moho under Iceland, and the prominent reflections
indicate that it is fairly sharp. The key observationsfor
this interpretation are the sharpnessof the reflector and the
7.7 km/s apparentupper mantle velocityjust belowit. It
3O
is, however, necessaryto know in what physical state a 7.7
820 770 760 730 690
km•s fast upper mantle is in, beforeaffirmingthat the re-
range, km
flections are from the crust-mantle boundary. We explore
some possibilitiesbelow.
The Moho, as determined by forward modeling of travel
Fig. 7. Five one-dimensional profiles of compressionalwave ve-
times, dips slightlyto the west (Figure 6). Wide-anglere-
each other. Top scales are for first and last profiles.
flections observedfrom it are probably from the east and
west of the WVZ, with only a few reflection points immediately beneath the WVZ. Consequently,we cannot definitively address whether the Moho is continuous acrossthe
WVZ. Several one-dimensionalprofiles of the final model
are shown in Figure 7.
A wide-anglereflectionobservedon the RRISP profile is
locity throughthe final model. Profilesare displaced2 km/s from
representativeof the NVZ itself. Tomographic inversion of
teleseismicdata have been interpreted by Tryggvasonet al.
[1983]to indicatethe mantlevelocityincreases
by 4% away
from the NVZ to the older areas of eastern Iceland, which
correspondsto an increasefrom 7.4 to 7.7 km/s. Some
RRISP record sectionsthat have rays that bottom at the
interpretedby Gebrandeet al. [1980]to be from a localized periphery of the NVZ have apparent velocities of 7.67 and
high-velocity
lensat 30 km depth[seeGebrandeet al., 1980, 7.7 km/s [Gebrandeet al., 1980,Figure 2, RRISP Working
are in close
Figure1, shotD]. We seeno strongargumentfor this reflec- Group,1980, Figure 4]. These measurements
tor being associatedwith a thin layer and instead interpret
it to be a reflection from -0 30 km deep Moho. Another
agreement with our measurementsof apparent velocity of
RRISP profile [Gebrandeet al., 1980, Figure 2] alsocon-
7.74 km/s for shot JL (Figure 3).
Someauthors[Tryggvason,
1964;LongandMitchell,1970]
tains evidencefor a sharp Moho, sincethe apparent velocity
have proposed that very low upper mantle velocities exist
increases
sharplyfrom approximately7.0 km/s to 7.67 km/s
below Iceland (7.4 km/s down to at least 160 km depth)
at a rangeof 140 km. No Moho reflectionis observedon this
RRISP record section, probably due to the low-frequency
content of the data and spatial aliasing. Hence a relatively
sharp Moho is probably a common feature in Iceland at least
on the basis of teleseismictravel time anomalies. However,
we find that more recent anomaly data and velocity models
outside
the volcanic
zones.
Mantle
5. Upper mantlevelocitiesreach7.6-7.7km/s a few kilo-
[Dziewonski
and Anderson,1983]supportsomewhathigher
velocities,similarto our 7.6-7.7km/s sub-Mohovalue. DziewonskiandAnderson[1983]findthat the teleseismic
P wave
travel time delay of Iceland is between 1.6 and 1.9 s. This
anomaly can be interpreted as a low-velocity zone in the
upper 240 km of the mantle of Dziewonski and Anderson's
[1981]PreliminaryReferenceEarth Model (PREM) from
8.0-8.1km/s to about7.7 km/s but do not requirea velocity
meters beneath the Moho in Iceland. Implications are that
previoustemperature estimates have been overestimatedat
the top of the upper mantle in Iceland. The upper man-
inversion within the uppermost mantle beneath Iceland.
In a dry state the mantle rock, peridotite, has a com-
tle velocitiesof 7.4 km/s reportedby RRISP may only be
pressionalvelocityof 7.4 km/s at 5 kbar pressure(-0 20
6620
Bjarnason et al.: Tomographic Image of Mid-Atlantic Plate Boundary
km depth) and at the solidustemperature1180øC[Murase
and Kushiro, 1979; Sato et al., 1989a; Takahaskhi,1986]
(and compiledby Sato et al., 1989b].)We thereforebelieve
that the 7.7 km/s refractorunder south Iceland is below
the solidustemperature of peridotite, if we assumethat the
upper mantle rocks in Iceland are peridotite. Accordingly,
steady state accumulationsof partial melt do not seem to
occur at the top of the upper mantle, and hencemelt generation must be at greater depths. Unfortunately the lab
measured geotherm in the top 0-2 km part of the crust.
However, a simple, one-dimensionalthermal structure cal-
culationfor a ridge[e.g. SparksandParmentier,1991]shows
that geothermalgradients decreasewith depth. We therefore believethat previoustemperatures estimateshave been
overestimated.
Crustal Anomalies
6. A broad high-velocityzonedomesin the lower crust be-
data of Muraseand Kushiro[1979]and Sato et al. [1989a] neaththe WVZ medianvalley(754-767km) and extends30
do not extend to low enoughtemperatures to determine di-
km farther to the west along the profile. No magma cham-
rectly the temperatureof 7.7 km/s peridotite. An approxi- ber is seenwithin the WVZ at the transect. The velocity is
mate estimate of this temperature can be made by assum- elevatedby about 0.5 km/s (i.e., from 6.5 to 7.0 km/s) at a
ing that the linear velocity-temperaturegradient of olivine
depth of 5 km. The deepestrays, which bottom at a depth
(5.2x10-4km/søC
[Anderson
et al., 1992])appliesto peri- of about 12 km in the WVZ, have apparent velocitiesof 7.13
dotite. This extrapolation gives a temperature of 600øC.
If, however, we extrapolate the very few velocity values of
peridotite that are below the solidusdetermined by Murase
km/s. The shapeof the domeat the bottomof the inversion
model(15 km) is essentiallyunconstrained,
as are the high
velocities(> 7.3 km/s). The domeof high velocitiesunder
and Kushiro[1979]and Sato et al. [1989a]to lowertem-
the ridge axis may representa region of the lower crust that
has been enrichedin olivine by plutonism. This seismically
fast material may be advectedto greater depthsin the crust
peratures, we obtain a temperature estimate of 900øC for
7.7 km/s peridotite. The solidusof gabbroat 5 kbar confiningpressureis 1090øC[Thompson,1975]. Hencegabbro
is below its solidusin the inferred range of Moho temperatures. The only remainingquestionis if crustal rocksat 20
km depth can havethe observedvelocitiesof 7.0-7.2km/s
at the relatively high temperature under Iceland.
Lower oceanic crustal velocities are reported to lie in the
range6.5-7.2 km/s (e.g., Minshull et al., [1991]),but are
usually found in colder crust than in Iceland. The fastest
gabbro sampled from the lower crustal portion of an ophi-
olite site in Papua New Guinea has velocityof 7.69 km/s
at 5 kbar and room temperature[Kroenkeet al., 1976]. A
as it agesand spreads[Pallmason,
1980].
The high midcrustalvelocitiesindicate that a large magma
chamber is not present beneath the WVZ. Careful inspec-
tion of the travel time of the AK, BF, and JL profiles(Figure 5), whichhaverays that passsubverticallythroughthe
midcrust at the WVZ, reveals no pronounced delays that
mightbe associated
with a magmachamber.(In fact, those
portionsof the travel time curvesare generallyfast.) We
feel that the maximum plausible amplitude of a "hidden"
travel time delay is about 0.1 s, which gives a maximum
magmachamberthicknessof 0.6 km (for magmavelocities
of 3 km/s). Vera et al. [1990]modeledthe East PacificRise
a velocitydrop to 7.1-6.7 km/s. Olivine gabbrosampled (EPR) magma chamberas being 2 km wide and 0.7 km
from an ophiolitesite in Oman has a velocity0.12 km/s thick. Our data cannot rule out a small magma chamber
faster than the fastestgabbrofrom New Guinea, when mea- of this type. On the otherhand, Veraet al.'s [1990]crustal
sured at the sametemperatureand pressure[Christensen velocitymodel(their Figure22) showsa broaddepression
of
and Smewing,1981]. It is thereforeplausiblethat olivine velocitiesthroughout all of layer 3 out to a lateral distance
linear extrapolation of this velocity to 600-900øC leads to
chemistsexpect that the lower crust of Iceland should have
a larger proportion of olivine than "normal" oceaniclower
of 5 km. Our inversionrules out the presenceof this kind of
broad low-velocity zone beneath the WVZ.
7. Upper crustal velocitiesare high beneath extinct central volcanoswest of the WVZ. The upper crustal velocities
crust,dueto the influenceof the hotspot(C. Langmuir,personalcommunication,1992). Elevatedtemperaturesunder
are elevatedby • 0.5 km/s in the vicinity of the Stardalur
extinctcentralvolcano(775-785km, Figure I and Plate 3).
Iceland cause melting to occur at higher temperature and
The surfacegeologyof this region is dominatedby massive
basaltic plutons, which are seismically faster than the surrounding layeredflood basalts. The depth to the lower crust
rich gabbro could account for seismicvelocitiesup to 7.2
km/s in the relativelyhot lowercrustof Iceland. Somegeo-
greaterpressure(depth) in the mantle than under normal
ridges, producing olivine-rich melt. Part of this melt will
crystallize into olivine gabbro in the lower crust. Olivine
enrichmenthas the effect of raising seismicvelocities.
Our new interpretation of crustal thicknessin south Iceland, between20 and 24 km, agreeswell with expectedthicknessof crust over hotspotsas determined by geochemicalar-
is shallowestin this area, •
2.5 km. Fl6venz[1980]had
previously noted that of all places measured in Iceland the
shallowestdepth to the lowercrust (2.0 km) occursunder
the Stardalur Caldera. A smaller, fast travel time anomaly
was observedalong the Skardsheidiextinct central volcano
guments[McKenzie,1984;Klein andLangmuir,1987;Sleep, (800-815km, Figurei and Plate 3).
1990].
8. The axisof the spreadingcenterin the WVZ (754-767
The new interpretation for the lower crust presentedhere km) coincides
with a very sharpchangein the thicknessof
is not consistentwith previous estimatesof temperature be-
the upper crust from 0.8 km west of the WVZ to 2.0 km
tween1000and 1100øCin the depthrange10-20km [Pallma- in the WVZ, but a uniformmidcrust(Plate 3). The thin
son and S•rmundsson,
1974]. In this depth and tempera- upper crust west of the WVZ is not confined to our profile
ture range a large proportion of crustal material would be
molten and seismically slow. Hence previous authors con-
cludedthat velocitiesof 7.0-7.2km/s shouldbe attributed
to mantle rocks. The older estimates of temperature at 1020 km depth were obtained by linear extrapolation of the
but extendsnorthward by at least 20 km where Zverev et al.
[1980]havemeasureda 40-km-longprofile(their profileI).
Raysin their profilethat bottomat the layer2A/2B boundary (i.e., apparentvelocity• 5.0 km/s) havetraveltimes
only about 0.1 s slower than our values west of the WVZ,
Bjarnason et al.: Tomographic Image of Mid-Atlantic Plate Boundary
indicatingvery similar upper crustal thicknesses.(However,
rays bottoming at the top of Layer 3 are delayedby approximately 0.45 s in their data set comparedto ours, indicating
that the midcrust is about 1.0 km thicker along their profile
than ours.)
9. A 2.5-km-deepupper crustal velocity low occursunder
the eastern edge of the WVZ and the hyaloclastiteridge,
Bdrfell extendingto the easternedge of the medium valley
6621
Acknowledgments. This research was supported by the U.S.
National Science Foundation, the Iceland National Science Foundation, the National Energy Authority of Iceland, the Incorporated Research Institutions for Seismology, IcelandAir, and
the Lamont-Doherty Geological Observatory of Columbia University. We thank the many landowners in Iceland that graciously allowed access to their land. We thank the Science Institute of the University of Iceland, Roger Bilham of the University of Colorado, Walter Mooney of the U.S. Geological Survey,
Klaus Jacob of the National Center for Earthquake Engineering
of the WVZ (748-754km) (Plate 3). The abrupttermination Researchand Gudmundur Sigvaldasonat the Nordic Volcanologof this featureon the east side (748 km) is probablyrelated ical Institute for loans of equipment. We thank Benedikt Steto an intruding upper crustal velocity high under the off-axis
ingrfmsson, Bryndis Brandsd6ttir, Gunnlaugur Gudmundsson,
Einar Haraldsson, Einar Kjartansson, Halld6r J6nsson, Hilmar
Sigvaldason,Hrappur Magnhsson, J6sep H61mjArn, Larry Shengold, Pall Helgason, and Ragna Karlsd6ttir for technical assistance. We axe grateful for the good work all these people per-
Grimsnesvolcaniccenter(Figure 1).
10. A narrowupperand midcrustalvelocityhigh at (738747 km) is associatedwith the off-axisGrimsnesvolcanic
centerwhichispostglacialin age[Jdhannesson
andSeerounds-formedbut wouldlike to thankshotmaster
J6sepH61mj•rnin
son, 1989] (740-746km, Plate 3). We note that the lower particular. We thank Gunnar Thorbergssonfor use of his geodetic
crust beneath the volcanically active WVZ, as well as the
midcrust under the much smaller Grlmsnes volcanic center,
are characterizedby faster than averageseismicvelocities.
Fast velocitiesmay be causedby high geothermalalteration
programs. This paper benefited from discussionwith Won-Young
Kim, John Mutter, PAll Einarsson, Ge Hu and Marc Spiegelman,
and from commentsfrom anonymousreviewers. Lamont-Doherty
Contribution
5040.
(whichleadsto a reductionin porosity)and by the unusually numerous lower crustal intrusions below the volcanic
centers.
11. A broad crustal low-velocityzoneapproximatelycoincides with the transform
of the South Iceland
Seismic Zone
(721-735 km). The low-velocityzone is approximately15
km wide and extends from the upper crust into the lower
crust to a depth of 10 km between 727 and 740 km. Veloc-
ity reductionis approximately0.3 km/s or 5%.
12. The narrow, slowupper crustal velocity zone near the
REFERENCES
Anderson, O. L., D. Isaak, and H. Oda, High-temperature elastic
constant data on minerals relevant to geophysics,Rev. Geophys, 30, 57-90, 1992.
Bath, M., Crustal structure of Iceland, J. Geophys. Res. 65,
1793-1807, 1960.
Bjarnason, I. Th., P. Cowie, M. H. Anders, L. Seeber, and C.
H. Scholz, The 1912 Iceland earthquake rupture: Growth and
development of a nascent transform, Bull. Seismol. Soc. Am.,
in press, 1993.
Beblo, M., and A. BjSrnsson, A model of electrical resistivity beneath NE-Iceland, correlation with temperature, J. Geophys.,
southeasternend of profile (695-708km) has no clearcorre47, 184-190, 1980.
spondenceto surfacegeology.The surfacein this regionis Bott, M. H. and K. Gunnarsson, Crustal structure of the Icelandcoveredwith glacialoutwash[Jdhannesson
and Scemunds- Faeroe Ridge, J. Geophys., 47, 221-227, 1980.
Caress, D. W., M. S. Burnett, and J. A. Orcutt, Tomographic
son, 1989].
image of the axial low-velocity zone at 12ø50•N on the East
Pacific Rise, J. Geophys. Res., 97, 9243-9263, 1992.
Cerveny, V., and I. Psencik, 2-D seismic ray package, Seis81,
The depth to the lower crust ("layer 3") is asymmetric
Charles Univ., Prague, 1981.
acrossthe WVZ. On averagethe depth to the lower crust is Christensen, N. I., and J. D. Smewing, Geologyand seismicstructure of the northern sectionof the Oman ophiollte, J. Geophys.
3.0 km northwest of the WVZ, 4.3 km in the WVZ, and 6.0
Res., 86, 2545-2555, 1981.
km southeastof the WVZ. The asymmetrymay in part be
Dziewonski, A.M., and D. L. Anderson, Preliminary Reference
explained by greater amount of crustal erosionto the northEarth Model, Phys. Earth Planet. Inter., 28, 297-356, 1981.
west of the WVZ and due to secondarycrustal accretionin Dziewonski, A.M. and D. L. Anderson, Travel times and station
corrections for P waves at teleseismicdistances, J. Geophys.
the South Iceland Lowlandsassociatedwith the ride jumpRes., 88, 3295-3314, 1983.
ing from the WVZ to the EVZ. A broad high-velocityzone
Einarsson, P., and J. Eirfksson, Earthquake fractures in the disdomesin the lower crust beneath the WVZ median valley
tricts Land and Rang&rvellir in the South Iceland Seismic Zone,
and extends30 km farther to the northwestalong the proJSkull, 32, 113-120, 1982.
file. The dome may representa region of the lower crust Einarsson, P., and K. SmmundssonEarthquake epicenters 1982CONCLUSIONS
that has been enrichedin olivine by plutonism. No magma
1985and volcanic
systems
in Iceland(map),in "! hlutarins
chamber
edli", Festschrift .for Thorbj$rn Sigurgeirsson,edited by Th.
Sigffisson,Menningarsj6dur, Reykjavik, 1987.
is seen within
the WVZ
at the transect.
This ex-
periment sampled only a narrow crosssection of the WVZ,
and therefore we cannot generalizeabout the existenceof
magma chambers in other parts of the WVZ. Indeed, the
Hengill central volcano,which has had postglacialeruptions,
is 10 km southwestof the profile.
The crust in southIceland is 20-24 km thick with a sharp
reflector at the bottom which we interpret as the Moho.
This is significantlythicker crust than has been postulated
by most previousinvestigators.A relatively sharp Moho is
probably a common feature in Iceland, at least outside the
volcaniczones.Upper mantle velocitiesreach7.6-7.7 km/s
a few kilometers
beneath
the Moho in south Iceland.
Im-
plicationsare that the lower crust and the top of the upper
mantle are colder than previously postulated.
ETOPOS: Digital reliefof the surfaceof the earth- Bathymetry/Topography data, Rep. 88-M66-2, Natl. Oceanic and Atmos.
Admin, Boulder, Colo., 1988.
Eysteinsson,H., and J.F. Hermance, Magnetotelluric measurements across the eastern neovolcanic zone in south Iceland, J.
Geophys. Res., 90, 10093-10103, 1985.
Fl6venz,(•. G., Seismicstructureof the Icelandiccrustabove
layer three and the relation between body wave velocity and
the alteration of the basaltic crust, J. Geophys., ,/7, 211-220,
1980.
F16venz,
(•. G., andK. Gunnarsson,
Seismic
crustalstructurein
Iceland and surrounding area, Tectonophysics189, 1-17, 1991.
Fl6venz,6. G., andK. Smmundsson,
Heatflowandgeothermal
processesin Iceland, Tectonophysics,in press, 1993.
Gebrande, H., H. Miller, and P. Einarsson, Seismic structure of
Iceland along RRISP-profile I, J. Geophys.,67, 239-249, 1980.
6622
Bjarnasonet al.: TomographicImage of Mid-Atlantic Plate Boundary
Goldflam, P., W. Weigel, and B. D. Loncarevic, Seismic structure
alon•gRRISP-ProfileI on the southeast
flank of the Reykjanes
Ridge, J. Geopkys.•,7, 250-260, 1980.
Hackman, M. C., G. C. P. King, and R. Bilham, The mechanicsof
the South Iceland Seismic Zone, J. Geophys. Res., 95, 17,33917,351, 1990.
J6hannessonH., and K. S•emundsson,Geological map of Iceland,
in Bedrock õeology, 1st ed., Icelandic Museum of Natural History and Iceland Geodetic Survey, Reykjavik, 1989.
Klein, E., and C. H. Langmuir, Global correlationsof oceanridge
basalt chemistry with axial depth and crustal thickness, J.
Geophys. Res., 92, 8089-8115, 1987.
Kroenke, L. W., M. H. Manghnani, C. S. Ral, P. Fryer, and R.
Ramananan-toandro, Elastic properties of selected ophiolitic
rocks from Papua New Guinea: Nature and composition of the
oceanic lower crust and upper mantle, in Geophysics of the
Pacific Ocean Basin and Its Margin, Geophys. Monogr. Set.,
vol. 19, edited by G. H. Sutton, M. H. Manghnani, R. Moberly,
pp. 407-421, AGU, Washington, D.C., 1976.
Long, R. E., and M. G. Mitchell, TeleseismicP-wave delay time
in Iceland, Geophys. J. R. Astron. $oc., 20, 41-48, 1970.
McKenzie, D., The generation and compactionof partially molten
rock, J. Petrol., 25, 713-765, 1984.
Minshull, T. A., R. S. White, J. C. Mutter, P. Buhl, R. S. Detrick,
C. A. Williams, and E. Morris, Crustal structure at the Blake
Spur fracture zone from expanding spreadprofiles, J. Geophys.
Res., 96, 9955-9984, 1991.
Morgan, W. J., Rodriguez, Darwin, Amsterdam, .., A secondtype
of hotspot island, J. Geophys. Res. 83, 5355-5360, 1978.
Murase, T., and I. Kushiro, Compressional wave velocity in partially molten perldotite at high pressures, Year Book Carnegie
Inst. Washington, 79, 559-562, 1979.
Mutter, C. Z., and J. C. Mutter, Variations in thickness of layer 3
dominate oceanic crustal structure, Earth Planet. $ci. Left.,
in press, 1993.
•skarsson,N., S. Steinth6rsson,
and G. E. Sigvaldason,
Iceland
geochemicalanomaly: Origin, volcanotectonics,chemical fractionation and isotope evolution of the crust, J. Geophys. Res.
90, 10,011-10,025,
1985.
in the low-velocity zone: Comparison with heat flow and electrical conductivity studies, J. Geophys. Res., 9,{, 5689-5704,
1989a.
Sato, H., I. S. Sacks, T. Murase, G. Muncill, and H. Fukuyama,
Qp-Melting temperature relation in peridotlte at high pressure and temperature: Attenuation mechanism and implications for the mechanical properties of the upper mantle, J.
Geophys. Res., 9,{, 10,647-10,661, 1989b.
Sleep, N., Hotspots and mantle plumes: Some phenomenology, J.
Geophys. Res., 95, 6715-6736, 1990.
Sparks, D. W., and E. M. Parmentier, Melt extraction from the
mantle beneath spreading centers, Earth Planet. $ci. Left.,
105, 368-377, 1991.
Takahashi, E., Melting of a dry peridotite KLB-1 up to 14 GPa:
Implications on the origin of peridotitlc upper mantle, J. Geophys. Res., 9i, 9367-9382, 1986.
Thompson, R. N., Primary basalts and magma genesis,Contrib.
Mineral., Petrol., 52, 213-232, 1975.
Tryggvason, E., Arrival times of P wavesand upper mantle structure, Bull. Seismol. $oc. Am., 5•i, 727-736, 1964.
Tryggvason,E., and M. Bath, Upper crustal structure of Iceland,
J. Geophys. Res., 66, 1913-1925, 1961.
Tryggvason,K., E. S. Husebye,and R. Stef/•nsson,Seismicimage
of the hypothesized Icelandic hot spot, Tectonophysics, 100,
97-118,
1983.
Vera, E., J. C. Mutter, P. Buhl, J. A. Orcutt, A. J. Harding, M.
E. Kappus, R. S. Detrick and T. M. Brocher, The structure of
O-to 0.2-m.y.- old oceanic crust at 9 ø N on the East Pacific Rise
from expanded spread profiles, J. Geophys. Res., 95, 15,52915,556, 1990.
Vink, G. E., A hotspot model for Iceland and the V 5ring Plateau,
J. Geophys. Res., 89, 9949-9959, 1984.
Zverev, S. M., I. P. Kosminskaya, G. A. Krasilstchikova, and G.
G. Mikhota, The crustal structure of iceland and the IcelandFaeroe-Shetland region, Byu!!. Mosk. Ova. Ispyt. Prir., Otd.
Geol. 50, 72 pp., 1975.
Zverev, S. M., I. V. Litvinenko, G. PAlmason, G. A. Yaroshevskaya,
and N. N. Osokin, A seismic crustal study of the axial rift zone
in southwest Iceland, J. Geophys., ,/7, 202-210, 1980.
PAlmason, G., Crustal structure of Iceland from explosion seis-
inology,GreinarVfsindafdlaõ
Is!., ,{0, 187pp, 1971.
PAlmason, G., A continuum model of crustal generation in Iceland: Kinematic aspects, J. Geophys., ,47, 7-18, 1980.
P Mmason, G., and K. Saemundsson,Iceland in relation the MidAtlantic Ridge, A nnu. Rev. Earth Planet. Sci., 2, 25-50,
1974.
RRISP Working Group, Reykjanes Ridge Iceland Seismic Exper-
iment (RRISP 77), d. Geophys.,,{7, 228-238, 1980.
Seemundsson,K., Evolution of the axial rifting zone in northern
Iceland and the TjSrnes fracture zone, Geol. $oc. Am. Bull.,
85, 459-504,
I. Th. Bjarnason,CarnegieInstitutionof Washington,Department
of Terrestrial Magnetism, 5241 Broad Branch Road, N.W.,
Washington,D.C. 20015.
D. Caressand W. Menice,Lamont-Doherty
Earth Observatory,
Palisades,NY 10964.
•. G. F16venz,
Orkustofnun,
Grensdsvegi
9, 108 Reykjavik,
Iceland.
1974.
Seemundsson,K., Outline of the geology of Iceland, J6kull, 29,
7-28, 1979.
Sato, H., I. S. Sacks, and T. Murase, The use of laboratory velocity data for estimating temperature and partial melt fraction
(ReceivedJune 4, 1992;
revisedSeptember28, 1992;
acceptedOctober6, 1992.)