GEOPHYSICALRESEARCHLETTERS,VOL. 25, NO. 6, PAGES805-808,MARCH 15, 1998
Thickness of a submarine lava flow determined from near-bottom
magnetic field mapping by autonomousunderwater vehicle
Maurice A.Tivey, • H.PaulJohnson, 2Albert Bradley, • andDana Yoerger • Abstract. Magneticfield surveysobtainednearthe seafloorcan mapthe boundaries of recentvolcaniceruptionsandcanprovide thicknessestimatesof these lava flow units independentof bathymetrydifferencingmethods.Magneticthicknessestimation requiresknowledgeof the intensityof magnetizationof the new lava and surroundingterrain,but this can be satisfactorily obtainedby representative samplingof the variousvolcanicunits. While bathymetrydifferencing requirespre-existing datato assess the thicknessof new lava eruptions,magneticsurveyscanbe obtainedafter an eruptionhasoccurred.In this study,near-bottom magneticsurveyswere obtainedusingan autonomous underwater vehicle (AUV), which operateswithout a tetheror humanintervention. AUV technologyoffersrapid deploymentand an efficientsurveyingapproachfor remotelymappingrecentlava erup-
The rock magnetizationof the new flow and older surrounding terrain must be measuredin additionto underwaymagneticfield mapping,in order to determinethe thicknessof a new eruption andto correctfor topographiceffectsuponthe observedmagnetic field. Assumingthat remanentmagnetizationintensityis relatively uniform throughoutthe flow and older terrain, these rock
measurementsprovide constraintsfor forward and inversemodeling of the observedmagneticfield. The directionof magnetization must alsobe determined,althoughfor youngcrustthis would be in the local geomagneticfield direction. Magnetic surveys alsooffer somepotentialadvantagesover differentialbathymetric mapping. Surfaceship swath bathymetryhas a relatively large footprintof 100 m and a limiting depth resolutionof 5 to 15 m [Fox et al., 1992], althoughnear-bottombathymetricmapping tion sites on the seafloor. could improve on this resolutionby an order of magnitude. Depending on the geometry and density of the magnetic survey tracks and the magnetizationcontrastsof the lavas, near-bottom Introduction magneticmappingcould haveboth a smalleffectivefootprintand the ability to map flows thinnerthan5 m. The upperoceancrustis comprisedof a sequenceof extrusive In this paper, we presentthe resultsof a survey over a young lavas that form a layer several hundreds of meters thick and seafloor lava flow and calculate the thickness of the flow based which is thoughtto be the primary sourceof magneticanomalies on an estimateof its magnetization. We chosethis approachto in young crust [e.g., Tivey, 1996]. Understandingthe construc- demonstrate that an estimate of lava flow thickness could be obtional fabric of this layer is importantin elucidatingthe processes tained independentlyof differential bathymetricmapping, alof crustalaccretion,magmadelivery and supply,and the evoluthoughclearly the combinationof thesedata setswould also be tion in physicalpropertiesof oceancrust. Determiningthe extent highly informative. This surveyis also uniquein that it is the and thickness of individual lava flow units is difficult, and has first scientific results collected by an autonomousunderwater only been successfulfor very large lava flows [Macdonaldet al., vehicle, specifically as part of an ongoing research program. 1989] or for known volcanic eruption events [Embley et al., Autonomous underwater vehicles are intelligent roving robots 1991; 1995; Gregg et al., 1996; Chadwicket al., in press]. Rethat operatewithout a tether or human supervision,and are relapeat swath mapping of seafloorbathymetry[Fox et al., 1992; tively fast and efficient survey platforms. The Autonomous Chadwick et al., 1995] providesimportantinformation on the Benthic Explorer "ABE", under developmentat Woods Hole extent and volume of the lava products,critical for quantifying OceanographicInstitution, was used to carry out a near-bottom and characterizingeruptionparameters[e.g., Gregg et al., 1996]. magneticsurveyover a lava flow eruptedin 1993 on the CoAxial Pre-eruptionsurveysare not always available, however, which segmentof the Juande Fuca (JDF) Ridge [Yoergeret al., 1996; precludesusing differential bathymetricmappingto determine Tivey et al., 1997]. ABE was used in "lawn-mower"survey lava flow thickness. mode, where straight tracklines were navigated across the Near-bottommagneticfield surveyscan provide alternative seafloorcollectingdata with a sensorsthat included:a 3-axis constraintson the thicknessand spatial extent of young lava flows, and furthermore, this information can be obtained after the fluxgate magnetometer,depth and altitude sensors,CTD, and digital video cameras. lava hasbeenemplaced. Young, newly eruptedbasaltis highly magnetizedand can producesignificantnear-bottommagnetic field anomalies; in somecasesup to 50% of Earth'sfield intensity Geologic Setting [Tiveyand Johnson,1995]. The boundariesof younglava flows Seismicityassociated with the 1993 seafloorlava eruptionon should provide sufficient magnetizationcontrastwith the surthe CoAxial segment of the JDF was monitoredin real-timeby roundingolder lava to producewell-definedmagneticanomalies. the SOSUS hydrophoneslocatedoff the west coastof the continental United States[Fox et al., 1995]. T-phaseepicenters,ini•Woods HoleOceanographic Institution, Woods Hole,Massachusetts. tially located near 46ø15'N, 129ø53'W, migrated 40 km north 2Sch. ofOceanography, University ofWashington, Seattle, Washington.over of period of 2 days to a region near 46ø31.5'N, 129ø35'W [Dziak et al., 1995]. Subsequentcruises confirmed that a seafloor eruptionhad occurredat this latter site, forming a lava Copyright1998by theAmericanGeophysical Union. flow up to 30 m thick, 2500 m long and 400 m wide [Chadwick et al., 1995; Embley et al., 1995]. Remotely-operatedvehicle Papernumber98GL00442. 0094-8534/98/98GL-00442505.00 (ROV) and submersibledives mappeda fresh basalticlava flow 805
806
TIVEY ET AL.: MAGNETIC THICKNESS OF A SUBMARINE LAVA FLOW 132øW
128øW
124øW
120•W 129' 35'W
52•N
14-6
14-5 14-4
50øN
14-3 11-6
48'N 14-2 11-5 11-4
SITE JDF
14-1
PLATE
17-2 11-3
42'N
11-2 19-1
17-3[10000 nT 19-4
46' 31 'N
19-3
9-5 19-6 19-7
19-8
19-9
•
• 500 m
35.4
35.2
129' 35'W
35
34.8
3•-.6
'
34.4
3•.2
34
LONGITUDE129' WEST (MINUTES)
Figure 1. Bathymetrymap of CoAxial eruptionsite (contourinterval 10 m) with the 1993 lava flow shadedgray. ABE track-
Figure 2. Observedmagneticanomalyfield datameasuredalong the ABE tracklines (Figure 1), projectedeast-westand plotted lines are shown in bold and identified as indicated. Lava flow overlying the 1993 lava flow extent shown in gray. Y-axis is boundaryis basedon ROV, ALVIN dives and sidescanimagery arbitrary. (W.W. Chadwick and R.W. Embley, personal communication, 1997; Tiveyand Johnson,unpublished data). stunt thickness layer whose upper surface is defined by the bathymetry. A thicknessof 30 m was used, basedon estimates from differential Sea Beam mapping [Chadwick et MI., 1995]. with hydrothermalventingandbacterialmatsdistributedoverthe This inversion techniquealso assumestwo-dimensionalitypercrestof the extrudedlava [Embley et MI., 1995]. Hydrothermal pendicularto the profile and a magnetizationvectorin the direcactivitywas also locatedalongnarrowgrabensto the north and tion of the presentfield (inclination67.2ø, declination19.8øE). A southof the flow, orientedalongthe sametrend as the 1993 lava flow (Figure 1). The grabensand associated hydrothermalactivity are interpretedas evidenceof the presenceof a shallowsubsurfacedike that fed the 1993 lava flow [Chadwickand Embley, 14-5 in press]. 14-4 Magnetic field data were collectedby ABE during two con14-3 secutivefield seasonsin the CoAxial area (Figure 1). In 1995, 11-6 ABE surveyswere conductedas a night-time componentto an 14-2 ALVIN deepsubmersibledive programand collectedapprox.35 11-5 11-4 km of tracklinesover the northernhalf of the newly eruptedlava. In 1996, ABE operatedalternatelywith ROV JASON and col14-1 lected an additional
23 km of tracklines over the central and
17-2 11-3
southernendof the 1993 lava flow (Figure'l).
11-2 17-3
Results and Discussion
19-1
50A/m The 1993 lava flow producesan extremelystrongmagnetic 19-4 19-5 field anomaly of approximately15,000 nT in the near-bottom 19-6 field (Figure 2). The observedanomalyarisesfrom a combina19-7 tion of the magnetizationof the new flow and the topographic 19-8 effectof the ridge upon which the new flow erupted. The ABE 19-9 magneticfield data were first correctedfor the magnetizationof 500 m the vehicle [e.g., Tivey, 1996] and then upward continuedfrom 34.8 35.4 3•.2 •5 ' 3•-.6 34.4 ' 3•-.2 34 the unevenobservationpath to a level planeabovethe topograLONGITUDE 129' WEST (MINUTES) phy. Traditional magneticfield analysisinverts the magnetic field data for crustal magnetizationto remove the effects of toFigure 3. Magnetizationinversionprofiles calculatedfrom the pographyand the skewnesseffect of latitude [Parker and Huestis, observedmagneticfield for a constant30 m thick layer plotted 1974]. We invertedfor crustalmagnetization,assuminga con- overlyingthe 1993 lava flow shownin gray. Y-axis is arbitrary I
I
TIVEY ET AL.' MAGNETIC THICKNESS OF A SUBMARINE LAVA FLOW
807
a) :::$.......... ?..? 55 m 3O
ß
.:
.:
..
ß
..:
. :........ ....
46 ø 31 'N
46' 3 ....
.
..
I 3oo rn J I
I
129 ø 35'W
129 ø 35'W
Figure 4. (a) Magneticallydeterminedlava thicknessmap with outlineof observed1993 lava flow shownby bold line and surveytracksshownby light lines. (b) Differentialbathymetrylava thicknessmap from Chadwicket al. [19951.
meancrustalmagnetizationof 60 A/m was obtainedfor the 1993 lava, consistentwith an averagefrom rock samplemeasurements of 67 A/m [Johnsonand Tivey, 1995]. The maximum anomaly gradientin magnetizationis generallylocatedat the edge of the new flow as determined from differential bathymetry, sidescan mapping and on-bottom observations[Embley et al., 1995; Chadwicket al., 1995; W.W. Chadwickpersonalcommunication, 1997]. There is clear variability in the computedmagnetization of the 1993 lava flow, but this likely reflectschangesin lava flow thicknessrather than variationsin magnetizationintensity(Figure 3). Thus, the assumptionof a constantthicknesssourcelayer is unlikely to be appropriatefor the 1993 lava flow, requiring a differentapproach. The lava flow thicknessrequiredto createthe observedanomaly can be estimatedby assuminga mean magnetizationfor both the old and new lava and then calculatingthe resultantanomaly usinga variablethicknesslayer. An iterativeforwardmodeling approachis usedbecausedirectinversionfor sourcelayer thicknessis an unstableprocess. Specifically,anomalydata were upward continuedto a level plane and the topographiceffect of the present-dayseafloorwas removed,assuminga meanmagnetization of 26 A/m, estimatedfrom rock sample values of the surroundingterrain [Johnsonand Tivey, 1995]. The residualmagnetic field was then modeledby calculatingan iterative forward model that varied the thicknessof the sourcelayer, assuminga constantmagnetizationof 34 A/m (i.e. the differencebetweenthe new and old magnetization,60-26 A/m). The initial starting thicknesswas assumedto be zero and then incrementedby 1 m in thicknessfor each iteration. The computedmagneticfield was
The resultant"magnetic"thicknessmap is remarkablyconsistent with both the mappedflow boundariesand the lava thickness
derivedfromdifferentialbathymetry (Figure4). Fromthethicknessmapwe estimate a totallavavolumeof 8.8 x 106ms, which includesa correctionestimateof 1.8 x 106 m3 for the northern and southernextremesof the flow andeasterntonguenot traversedby ABE (Figure4a). Thisvolumeis consistent withprevious estimatesof the 1993 CoAxial lava flow [Chadwicket al.,
1995; Chadwicket al., in press]. The high-resolution magnetic data identifiesthe presenceof the 1993 lava flow where it was too thin for detectionby the differentialbathymetrymethod (Figure4). Differentialdepthanomalyis alsolessreliablein ar-
easof steepslopesbecause smallshiftsin navigation canproduce artificialdepthanomalies [Chadwicket al., 1995]. An example of thisis shownin the differentialdepthanomalyat the southend of the eruption, where the 1993 lava flow abutsthe side of the
smallseamount,and flow thicknessis overestimated (Figures1 and4). Magneticthickness estimates thusprovidean independent measurement thatis unaffectedby thesebathymetryproblems. The magneticestimatesdo, however, have their own set of reso-
lutionlimitationsandproblems.
The errorin themagnetic thickness estimate depends primarily on the constantremanentmagnetization assumption.The extent to which remanentmagnetization varieswithin a singleflow is not well-known, but it is reasonable to assume that the 1993 lava
flow has an overallconstantmagnetization becauseit eruptedas one singleflow and cooledrelativelyquickly. From rock magnetic measurements, the standard deviation estimate for the 1993
lava magnetization is +15 A/m or -20% of the meanmeasured
compared to the residualmagneticfield andthe thickness of the value [Johnsonand Tivey, 1995], which translatesinto a -20% new lava was adjustedappropriately.Finally, the resultantthicknessprofiles were interpolatedto producea map of lava flow thickness(Figure 4). The top of the sourcelayer was assumedto be the bathymetrymeasuredby ABE along-trackor from highresolutionbathymetryobtainedfrom ALVIN and JASON scanning sonarsurveys.
error in the thicknessestimate. There is greaterlikelihood of variationin remanentmagnetizationof the olderterrainbecauseit maybe composed of severalflowsfrom differentagesanderuption events. Rock magneticmeasurements of the older terrain found a standarddeviationof +15 A/m [Johnsonand Tivey,
1995]. Uncertainty in thisvalueaffectsthetopographic modeling
808
TIVEY ET AL.: MAGNETIC THICKNESS OF A SUBMARINE LAVA FLOW
and increasesthe total error in the thicknessestimatesbasedon
lavaemplacement andcrustalaccretion canbe obtained by stud-
the differencein magnetization of the old andnew lava. By
ies focusedat the axis of spreading.
quadratic addition of theerrorestimates we obtaina standard deviationof +21 A/m for the error in the magnetization difference,whichwouldresultin a thickness estimateuncertainty of about -•30%. It is difficult to estimatehow well the magnetic
modeling fitsthetruelavaflowthickness because theonlyconstraintsarethe differentialbathymetry thickness data,whichrep-
resent a spatially-filtered andthickness truncated version of reality. Nevertheless, differential bathymetry doesprovideanupper
Acknowledgments. We thank the crew of Atlantis II and WHOIDSOG for their effortsin makingthe ABE surveya success.We thank Rod Catanach,A1Duester,SteveLiberatore,and Hanu Singhfor their efforts as part of the ABE team and D. Van Pattenfor her help on the cruise. We thank Bill Chadwick and Bob Embley for their comments that helpedimprovethe manuscript.M. Tivey was supportedby NSF grantOCE-9505514;H.P. Johnsonwas supportedby NSF grantOCE9405078. A. Bradley,D. Yoergerand ABE development was supported by NSF grantsOCE-8820227andOCE-9216775.
constrainton lava thickness[Chadwicket al., 1995]. We also estimatedhow well the computedboundaryof the
References
magnetic thickness modelfit theknownspatial extentof thelava flow. A fewprofileswereeditedfor spurious anomalies outside of themainbodyof the 1993lavaflow,butthesewerelimitedto
Chadwick,W.W., R.W. Embley,and C.G. Fox, SeaBeamdepthchanges associatedwith recent lava flows, CoAxial segment,Juan de Fuca Ridge:Evidencefor multipleeruptions between1981-1993,Geophys.
the more convolutedeasternboundary(Figures1 and 4). Thus,
magnetic thickness estimates neededitingfor spurious datasimilar to differentialbathymetry data. An averageof 55 m is obtainedfor the mismatchbetweenthe magneticlava flow edge(2 m contour)andthe actuallava flow boundaryalonga givenprofile, as determinedby observationand sidescandata [W.W.
Chadwick, personal communication, 1997]. Anothersource of
Res. Lett., 22, 167-170, 1995. Chadwick,W.W., and R.W. Embley, Grabenformationassociatedwith recentdike intrusionsandvolcaniceruptionson the mid-oceanridge, J. Geophys.Res.,in press. Chadwick,W.W., R.W. Embley,andT.M. Shank,The 1996 GordaRidge eruption:Geologicmapping,sidescansonar,and SeaBeamcomparisonresults,Deep SeaRes.,in press. Dziak, R.P., C.G. Fox, and A.E. Schreiner, The June-Julyseismoacousticevent at CoAxial Segment,Juande Fuca Ridge, Geophys. Res. Lett., 22, 135-138, 1995.
error in this studyarisesfrom the two-dimensional algorithm Embley,R.W., W.W. Chadwick,M.R. Perfit,andE.T. Baker,Geologyof used in the anomalymodeling,which can resultin incorrect the northernCleft segment,Juan de Fuca Ridge: Recentlava flows, thickness estimates if the anomalies have three-dimensional
seafloorspreading,and the formationof megaplumes, Geology,19,
sources.Theseproblems couldbeovercome by a tightertrackline spacing andtheuseof a three-dimensional modeling ap-
771-775, 1991.
proach.
Finally,thereareissues concerning thespatial resolution of the data. Previouson-bottomsubmersible profilesacrossthe 1993 lava flow observeda narrow (25 m wide) centrallow in the
Embley, R.W., W.W. Chadwick, I.R. Jonasson,D.A. Butterfield, and E.T. Baker, Initial resultsof the rapid responseto the 1993 CoAxial event: relationshipsbetween hydrothermaland volcanic processes, Geophys.Res.Lett., 22, 143-146, 1995. Fox, C.G., W.W. Chadwick,and R.W. Embley,Detectionof changesin ridge-crestmorphologyusingrepeatedmultibeamsonarsurveys,J. Geophys.Res.,97, 11,149-11,162,1992.
anomaly fieldinterpreted asdueto a non-magnetic feederdike Fox, C.G., W.E. Radford, R.P. Dziak, T-K. Lau, H. Matsumoto, and A.E. zone[TiveyandJohnson, 1995]. The ABE trackline spacing Schreiner,Acousticdetectionof a seafloorspreadingepisodeon the (-•150m), surveyaltitude(>10 m) andupwardcontinuation reJuande FucaRidge usingmilitaryhydrophone arrays,Geophys.Res. Lett., 22, 131-134, 1995. quirements reduce thespatial resolution required to properly resolvethis feature. Betterresolutionwouldrequirelessthan5 m
altitude andlinespacing of tensof meters overthewholeflow. Conclusions
High-resolutionnear-bottommagneticfield surveyscanhelp to quantify the geometryof young, recently-eruptedseafloorlavas and can provide thicknessestimatesthat are consistentwith other observationssuchas differentialbathymetryestimates.Magnetic fields offer someadvantages:e.g., in mappinglava flows that are too thin for detectionby the differentialbathymetrymethod'and where steep topography causes the differential bathymetry method to be unreliable. Magnetic surveyscan also be undertaken after an eruption, removingthe need for pre-existingdata over the site. Near-bottommagneticfield mappingcan complement differential bathymetry mapping of newly erupted lava flows, although the relationshipbetween magneticanomaly and thickness of the extrusive unit can be complicatedby crustal magnetization variations due to thermal structure, subsurface feederdikes, hydrothermalalterationand three-dimensional magnetic sources. For magneticanomalydata to provideindependent estimatesof flow thickness,crustal magnetizationover the old and new lavas must be determinedby representativerock sampling. Rapid decayin magnetizationwith age, althoughstill not well-determined(seeJohnsonand Tivey, 1995), alsolimits this techniqueto the zone of active crustalaccretion,where the contrast in magnetizationbetween young and older lava is at a maximum. Nevertheless,importantinsight into the processesof
Gregg,T.K.P., D.J. Fomari, M.R. Perfit, R.M. Haymon, and J.H. Fink, Rapidemplacementof a mid-oceanridgelava flow on the EastPacific Rise at 9ø46'-51'N, Earth Planet. Sci. Lett., 144, El-E7, 1996.
Johnson,H.P., and M.A. Tivey, Magneticpropertiesof zero-ageoceanic crust;a new submarinelava flow on the Juan de Fuca Ridge, Geophys.Res.Lett.,22, 175-178,1995.
Macdonald, K.C.,R.M. Haymon, andA. Shor,A 220 km2 recently eruptedlava field on the East Pacific Rise near lat 8øS, Geology,17, 212-216, 1989.
Parker,R.L., and S.P. Huestis,The inversionof magneticanomaliesin thepresenceof topography, J. Geophys.Res.,79, 1587-1593,1974. Tivey, M.A., The vertical magneticstructureof oceancrustdetermined from near-bottommagneticfield measurements, J. Geophys.Res., 101, 20,275-20,296, 1996.
Tivey, M.A., and H.P. Johnson,Alvin magneticsurveyof zero-agecrust: CoAxial segmenteruption,Juande Fuca Ridge 1993, Geophys.Res. Lett., 22, 171-174, 1995.
Tivey, M.A., A. Bradley,D. Yoerger,R. Catanach,A. Duester,S. Liberatore, and H. Singh, Autonomousunderwatervehicle maps seafloor, EOS, 78, 229-230, 1997.
Yoerger,D., A. Bradley,R. Bachmayer,R. Catanach,A. Duester,S. Liberatore,H. Singh,B. Walden,and M.A. Tivey, Near-bottommagnetic surveysof the CoAxial Ridge segmentusingthe AutonomousBenthic Explorersurveyvehicle,RIDGE Events,7, 5-9, 1996.
M. A. Tivey, WoodsHole Oceanographic Institution,WoodsHole, MA 02543-1542(email:
[email protected]) H. P. Johnson,Schoolof Oceanography, Universityof Washington,
Seattle,WA 98195(email:
[email protected]) A. Bradley,WoodsHole Oceanographic Institution, WoodsHole,MA 02543-1542(email:
[email protected]) D. Yoerger,WoodsHole Oceanographic Institution, WoodsHole,MA 02543-1542(email:
[email protected]) (ReceivedOctober14, 1997;revisedJanuary13, 1998; acceptedJanuary22, 1998.)
