Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 94, NO. B10, PAGES 14,127-14,144, OCTOBER 10, 1989 ResistivityCrossSectionThroughthe Juande Fuca SubductionSystem and Its TectonicImplications PI-IILIPE. WANNAMAKER, 1JOaNR. BOOKER, 2ALANG. JONES, 3ALAN D. Canv•, 4 JEANH. FILLOUX ,5 HA• S. WAFF,6ANDLAWmEK. LAW? A resistivity cross section to depths exceeding 200 km has been derived from magnetotelluric observationsalong a profile near latitude45øN from the Juande Fuca spreadingcenter,acrossthe coastal subductioncomplex,the High Cascadesvolcanicarc, and into the back-arcDeschutesBasin region. In this two-dimensional interpretation, emphasiswas placedon dataapproximating the transversemagneticmode sincethese data are relatively robustto commondeparturesfrom the two-dimensionalassumption. The vertical magneticfield, however,has been very valuablein defining structureof the offshoresediments,of the oceanicasthenosphere and below the arc volcanicsof the Westem and High Cascades. The transverse electric data on land suffer a variety of three-dimensionaleffects,making their interpretationvery difficult. In contrast,the greateruniformity of upper crustalconditionson the seafloorallowed a good fit to both modesof theimpedance plustheverticalmagnetic fieldat leastdownto 104s' Important components of our model resistivity structureand their hypothesizedtectonicimplicationsare as follows. Measurements on the seafloorindicatevery low resistivityin the eastwardthickeningsedimentarywedge of the abyssal CascadiaBasin providedits depthextentis constrained.The majority of thesesedimentsare inferred to be off-scrapedor lose most of their interstitial water before being carriedmore than ten or so kilometersdown the trench. In the oceanicuppermanfieof the Juande Fucaplate, moderatelylow resistivitiesfrom about 35-215 km depth (nominal) presumablyreflect up to several percent partial melt attending regional upwellingin the vicinity of the ridge. Very low resistivitiesbelow 215 km depthin the oceanicmantleare difficult to explain physically,but problemswith very long-periodsourcefield effects or bias error may hamper the modeling here. Under the Oregon Coast Range, a low-resistivity layer dips inland at approximately 20ø. Its conductancedecreasesby an order of magnitudefrom the coastto around60 km inland and its position appearsconsistentwith the Juan de Fuca plate subductiondecollement.Possible causesof the low-resistivity include residual sediments,pore waters, perhapssulfides, and dehydrating oceaniccrust carrieddown the subductionzone. East of about60 km inland, there is a much stronger, subhorizontalconductorwith a depth to top of around 25 km. To explain it, a massive breakdownof greenschistminerals is hypothesizedto liberate most of the water carried in the altered ocean crust well before the volcanic arc is reached. The resistivitymanifestationof the arc magmatismitself is not strong but consistsof a low-resistivityaxis in the middle crustbelow and to the west of the arc at the surface. It presumablyrepresentsfluids releasedfrom crystallizingmagmas.East of the High Cascades,both the MT and geomagneticdata detect a stronggradientin resistivitywith lower values to the south.It is suggested that the Brothersfault zone signifiesan electrical, thus possiblya thermal and magmatic,boundarywith the northernBasin and Range. INIRODUCTION A magnetotelluric(MT) transectof unprecedentedbreadth and quality has been collected across the Juan de Fuca subductionsystem [EMSLAB Group, 1988; Wannamaker et al., this issue] (Figure 1). This profile, called the Lincoln Line becauseit crossesthe northern Oregon coast near the town of Lincoln City, reveals a rich variety of resistivity structuresassociatedwith the subductionprocess. Possible examplesincludedeepstructureof the Juande Fuca spreading regime, the near-offshoreaccretionaryprism, the subduction decollement beneath western Oregon, the High Cascades 1University of UtahResearch Institute, SaltLakeCity. 2University of Washington, Seatfie. 3Geological Survey of Canada, Ottawa, Ontario. 4AT& T BellLabs,MurrayHill,NewJersey. •Scripps Institution of Oceanography, LaJolla,Califomia. 6University of Oregon, Eugene. ?Geological Survey of Canada, Sydney, British Columbia. Copyright 1989 by the AmericanGeophysicalUnion. Paper number 89JB00681. 0148-0227/89/89JB-00681 $05.00 volcanic arc, and the back-arc Deschutes Basin region [Wannamakeret al., this issue]. While a numberof aspects of the subsurfaceresistivity can be deduced straight from inspection of the data, computer simulation of the observations helps to quantify more rigorously the permissibleor requiredstructures.In this paper,we derive a two-dimensional resistivity model, with a presumednorthsouth strike, by trial-and-error fitting of our MT measurements with a finite element forward modeling algorithm. Constraintson model geometryfrom independent geological or geophysical investigations have been incorporatedwherejustified. Our paper is divided into three major sections. First, we explain the approachtoward two-dimensionalmodeling. On land, the transversemagnetic (TM) impedancefunctionsare emphasizedbecausetheory and experienceshow that they are more robust to plausible three-dimensional effects in the region than are the vertical magneticfield or, especially,the transverseelectric(TE) impedance. Upper crustalstructureon the seafloor appears to be rhuch simpler than on land, however,and both modesof impedanceplus the vertical field are fit fairly well. Second, the model cross section is described and its goodnessof fit to the data demonstrated. Model uniqueness is investigated by perturbing certain features of the model and assessingthe increasein misfit. 14,127 14,128 WANNAMAKER ETAL.:RESI• 50 ø CROSS SECTION, JUAN DEFUCASYSTEM 124 ø 128 ø 120 ø 116;00 ßMT ß ß...... T el_ ß 48 ø ß 48ø PACIFIC ß JUAN PLATEMTß MT, V,P ß ß ß ß ß DE • ßß Mß 8eMT'V ZeMT, v •eMTMT 6ø ßM,P ßMT, ep M 46 ø FUCA 6e ßMT, P 44 ø ßM,P M,V41• i• MT•eLU ßMT, P PLATE M'• M Magnetometer "%e•/ o o MT Magnetote!luric // 42 ø P Pressure Se,sor Mendocino Fracture . ./ ß • ß ß 0 440 - BASIN...._. ß 0") ß ß o o 40 ø ORDA ß •A o' U V Vertical Telluric ß Volcano •' ß 42 ø elI ß AND RANGE ß ß PLATE L____... 40 ø Zone 116ø 132 ø 128ø 124ø 120ø Fig. 1. Instrument locationof entireEMSLABarray. On land,theunlabeled dotsindicatelocations of three-component magnetovariometers. TheE-W trending concentration of dotsin northwestern Oregondenotes thelong-period MT recorders alongthe landportionof the LincolnLine. Seafloorarraysitesarelabeledaccording to instrument typeandsites0-8 are givenfor theseafloor extension of theLincolnLine. However, nodatawererecovered fromseafloor sitesSF0andSF6. This is not a complete method of model appraisal,but it allows us to test our model structure against important competinghypotheses.Finally, our resistivitymodel offers new insightinto the tectonicsof the Juande Fucasystemand of the subduction processin general. This is realizedwithin a framework of independent geological and geophysical knowledge, which includes primarily heat flow, seismic velocity, and slab mineralogy. VALIDITY OF Two-DIMENSIONAL MAGNETOTELLURIC the observation that some MT functions are much less affectedby variationsalongstrike than others. As reviewed more fully with the data presentationby Wannamaker et al. [this issue], we measureand interpret the complex-valuedtensorimpedancerelating the total electric (E) and magnetic (H) field vectorsat the surfaceof a threedimensional earth, i.e., Z• ] (1) INTERPRETATION We have only one detailedprofile of MT observations in the EMSLAB projectand so must interpretthe observations within primarilya two-dimensional framework. Formately, a substantialamountof progresshas been achievedin the past several years in understandingthe conditionsunder which two-dimensionalinterpretation is permissible. In essence, the possibility of two-dimensional approaches dependson the commonpreferredorientationof tectonism (roughlyN-S in theUnitedStatesandin ourstudyarea)andon A right-handedcoordinatesystemwith z positivedown is assumed.Over a purely two-dimensional earthwith x as the strike direction, the on-diagonal impedance elements Zxx and Zy• arezerowhile Zxy and Zy• represent the independenttransverseelectric (TE) and transversemagnetic (TM) modes of polarization. The TE mode is defined as having the electricfield (currents)parallel to strike,while the TM mode has the field (currents) across strike. Apparent resistivities (e.g.,p• and p•) andimpedance phases(0• WANNAMAKER ET AL.: RESISTIVITY CROSSSECTION, JUANDE FUCASYSTEM and ½yx)are formedfrontthe complex elements for subsequentinterpretation.Similarly, the relationship 14,129 downwardbias persiststo arbitrarily long periods, tending to make the three-dimensional impedance appear more anomalousthan the two-dimensional one. In attempting to replicatethe three-dimensional effectin p• and ½• usinga two-dimensionalTE routine, one would thus need to place additional (i.e., false) low-resistivity material beneath the is defined for the vertical magnetic field. Over a two- dimensional earth, Mz,, is zeroand Mzy arisesfromtheTE mode of polarization. Our recommendationfor two-dimensional interpretation is to derive a resistivity cross section emphasizing the TM mode data since the two-dimensional assumption for this mode is much more robust. For example, it has been shown that two-dimensional modeling of synthetic TM mode data across simple, elongate three-dimensionalstructuresyields accurate resistivity cross-sections [Wannarnaker et al., 1984]. The profile in fact does not need to be close to the center of the three-dimensional body for this to occur. Furthermore, a two-dimensional TM interpretation of p•,, and ½• is possibleevenwhenuppercrustalinhomogeneity is more complex. In this case, such an interpretationyields a cross section which may be somewhat incorrect in its surficial features but represents accurately the regional resistivitystructureof interestprovided the regional structure itself is not arbitrarily three-dimensional [Jones, 1983; Bostick, 1986; Kariya, 1986; Wannarnaker, 1989, in press]. For two-dimensional modeling, the MT functions should be defined using a fixed coordinatesystem approximatingthe strike direction to overcome unwanted effects of near-surface structure,specificallyon the impedancephase[Wannamaker et al., 1984]. The Lincoln Line profile was designed to meet the foregoing criteria for three-dimensional modeling emphasizing the TM mode. Observed geological and physiographic trends in the region are oriented predominantlyN-S [Baldwin, 1981]. This is implied alsoby most of the gravity data [Riddihough and Seeman, 1982]. Somedeparturefrom an assumedN-S geoelectricstrike in the upper crust is indicated along the east side of the Willamette Basin and the west side of the High Cascades graben. However, most resistivity variations, and especially the deeperonespresumablyassociatedwith subduction,appearto trend mainly N-S beneathour data profile accordingto their effecton Mz• and Mz• [Wannarnaker et al., this issue]. Furthermore,the preliminarymini-EMSLAB soundingstaken about 50 km southof the Lincoln Line [Young et al., 1988] also exhibit the major features of the present study, in trueconductor. However,M,• appears lessanomalous overa finite length conductor than over a fully two-dimensional one. This occursbecausethe depressionof electric field (and thus current) within the three-dimensionalbody reduces H, in its vicinity relative to the two-dimensional case. Finally, modeling of the TE results should not compromisethe TM mode fit. If the foregoing conditions cannot be met, then three-dimensional effects are probably at hand and require appropriatethree-dimensionalmodeling. MODEL RESISTIVITY CROSS-SECTION Our preferredtwo-dimensional resistivitymodel for the Lincoln Line is presentedin this section. It has been obtainedby trial-and-error fittingof the datausingthe finite elementforwardalgorithmdescribedby Wannarnakeret al. [1986, 1987]. Triangularelementshapesallow simulationof polygonal resistivity media, including a precise representation of the oceanbathymetry.Consequently, the oceanicMT responses of ourmodelcouldbe calculatedat the bottom of the seawater where the data were measured. For the detailed MT profile on land (Figure 2), the TM mode impedance functions p•,, and½•xweregivengreatest weight throughmostof the fitting process. The verticalmagnetic field, however,wasvaluablein understanding severalareasof the subsurface structure, especially those under the continentalshelfandslopeanddeepunderthe High Cascades. The TE quantitiesp• and ½• on landhavebeenusefulin revealing important off-line resistivity variations. In contrast, a single two-dimensionalmodel explains both the TE and TM mode impedance functions plus the vertical magneticfield on theseafloorat leastfor periodsup to 104s. Several constraintson model geometry were incorporated. First, the bathymetriccontoursof the Juande Fuca plate relief map of the Pacific GeoscienceCenter [Geological Survey of Canada]were usedto specifythe depthof the seawater,which was given a resistivity of 0.3 ohm m [Filloux, 1987]. Second, thicknesses of conductive sediments over resistive p,,• and ½•,yshouldalso improvethe fit to the vertical marine basalts in the near-offshoreNewport Basin and the abyssal Cascadia Basin were simplified from the geologic sectionof Snavely [1987]. The base of conductivesediments of the Willamette Basin was modeledas a sharpinterfacewith resistive rocks below in keeping with deposition over igneousor metamorphicbasement. The resistivity model in the High Cascadesgrabenand vicinity was patternedafter the volcanic stratigraphyand structurepresentedby Priest et al. [1983, 1987]. At 400 km depth, a very low resistivity (0.3 ohm m) half-space was imposed to represent the global decrease in resistivity near there [Banks, 1972; Roberts, magnetic field. This is becausethree-dimensionaleffects on the TE impedancetend to be of an oppositesensethan on 1986]. Generally, our model was constructedfrom the near-surface particular thesubtlepeakin ½• belowtheCoastRangeand its substantialincreaseto the east. Consequently,a fixed N-S strike direction is defined for two-dimensionalmodeling of the Lincoln Line data. In contrast,three-dimensionaleffects can be quite severe for the TE mode and reliable interpretationsof this data require care. Specifically, any structureintroducedto model M,•. For example,oversimplethree-dimensional conductive (short periods) downward. Deep structurewas included only inhomogeneities,finite strike length typically depressesthe electric field, and thus the impedance,relative to the twodimensionalTE response[Wannarnakeret al., 1984]. This when the data could not be explained by shallow structure. The abruptnature of the resistivity boundariesin the model, which are an implicit constraint of the finite element 14,130 WANNAMAKER ETAt..:RESiSTiViTY CaossSECTION, JUAND• Fuc^ SYSTEM 124ø 122ø 123 ø I - /-- _, x 121ø ' P / /' "................... / • -- • '- ( 5 ' OXUS ' / I / • • • • { Basinr • I , E } • + Broadband • • •, ••S•• • 0 MT • 50 Lava Plai o B ns o 100 km ß Long Period MT Fig.2. Broadband andlong-period magnetotelluric soundings along thelandward portion of theLincoln Lineandalso alongthemini-EMSLAB testtraverse farther south [Young et al., 1988].Urban areas include Portland (P),Salem (S), Eugene (E),Newport (N),Lincoln City(L),Bend(B),Madras (M),andTheDalles (D). Cascades volcanoes arelabeled as triangles: MountHood(H), MountJefferson (J),andSouthSister(S). algorithm,shouldnot be takenliterally. Because ourdataare impreciseandfinitely sampled,andbecausethe propagation of electromagnetic fieldsin theEarthis diffusive,thedatacan only resolvesmoothresistivityvariationsin the subsurface. We presumethat the media in our finite elementmodel approximatelyrepresentlocal averagesof conductivity (inverse of resistivity). possess resistivities of the orderof 100 ohmm or a bit less. Belowthemat a depthof about2 km lies a low resistivity unit which is essentiallyhorizontal under most of this area. Conductive marine sediments and alluvium of the Willamette Basindeepensteadilyto the eastand reacha thickness of 3.75 km nominallyby y = 75 km (distanceeast of the coast). Fartherinland,the thicknessdecreases to only about 2 km by y = 100 km, near the town of Stayton,beyond which there is a second, lesser depth maximum. Low Model ResistivityCross Section The two-dimensional finite element resistivity model is resistivity extendsabout 20 km east of the Willamette Valley-WesternCascades physiographic boundary. Farther east, the Western Cascadesblock contains high presented in Figure3 andPlate1 in slightlysimplitledform. extendingcloseto the surface(Figure3). Around (Plate 1 can be found in the separatecolor sectionin this resistivities issue.) Specifically,the resistivityvalueshavebeengrouped y = 145 km, a newlow-resistivity layerappears at a depth intorangesof one-halfdecadein orderto emphasize themajor of only500 m or sobelowmainlybasalticrocksof the High structural members of the model. Minor variations within Cascadesextrusiveplatform [Priestet al., 1983]. The layer each groupoccurpredominantly in the upper2 km of the deepensonly slightly eastwardbut, somewherebelow the axis(abouty = 170km), undergoes an model and simulatelocal, short-periodanomalies.Note the presentHigh Cascades abrupt fivefold increasein conductance(conductivityverticalscalechangesat 5 and 150 km. thickness product)anddeepensto about600-800m beneath The landwardportionof the LincolnLine model(Figure3) resemblesthat of EMSLAB Group [1988] but is refined somewhat. Marine basaltic rocks of the Siletz River formation [Shayely, 1987] outcroppingin the Coast Range extensive Pliocene basalt flows overlying the Deschutes Basin. All the conductiveupper crustalunits just described reside on a basement and middle crust which is resistive WANNAMAKER ET At,.: R•smT•vrryCROSS SECIION,JUANDE FUCASYSTEM CB 150 NB 100 50 CR c WB 50 WC lOO 14,131 HC 150 DB 200 km 4:1 50 1:1 100 ß . . 1:4 200 400 km p Fig. 3. East-west resistivitycrosssectionderivedfromforwardmodelingof MT dataalongthe LincolnLine. Section preserves detailsof thefiniteelementmeshgeometry butresistivities havebeengrouped in half-decade intervalsfor grayscaledisplay. Seafloorportionof the model is shownonly to 175 km offshore,beyondwhich variationsare minor. Note changes in verticalexaggeration, labelledon the left, at 5 and 150 km depth. Importantphysiographic regionscrossed includethe Cascadia Basin(CB),NewportBasin(NB), CoastRange(CR), WillametteBasin0NB), WesternCascades 0NC), HighCascades (HC),andtheDeschutes Basin(DB). A colorversion of thiscrosssection appears asPlate1 in theseparate color section of this issue. throughout (hundreds of ohm-m). It is difficult to resolve variationswithin the basement,but the highest resistivities appearto exist under the Western Cascades. about 6. Its top is somewhat shallower, and its bottom is substantiallydeeper. However, a precisedeterminationof its depth extent is made difficult by a strong regional threedimensionaleffect east of the High Cascades,as described Perhapsthe bestdeterminedelementof the deeperstructure is the subhorizontal conducting layerbelow 25 km extending later. eastfrom abouty = 60 km. The conductance of this layer is Anotherimportantmemberof the deeperstructureis a low600 S, and it is responsiblefor the strongridge in the TM resistivitylayer beneaththe Coast Range. This layer dips phase •y,, at 30-100s andthe distinctminimumin eastward about 20ø from a depth near 12 km under the describedin the data paper [Wannamakeret al., this issue]. coastline to 45 km under the west-central Willamette The bottom of the conductor is somewhat shallower Our data suggeststhe layer is thin relative to its depth such than in the preliminary model presentedby EMSLAB Group [1988] and fits the TM phasebetter below the Western Cascadesand easternWillamette Basin at periods from 300-3000 s. This conductorappearsto be resolved as well by the detailed geomagnetic measurements of Hermance et al. [1989, in press] in a profile some 30 km to the south of the Lincoln Line. Just west of the present High Cascadeschain, a relatively narrow conductive axis extends up from the subhorizontal conductor to about 5 km from the surface. East of the High Cascadesbeneaththe DeschutesBasin, the subhorizontal deepconductoris shownthickerby a factorof Basin. that only its conductancecan be resolved. The conductanceof this dipping unit decreaseseastward from a maximum near 1000 S underthe coastto lessthan 100 S by y = 60 km. Its exact geometric relation to the subhorizontal conductor discussedpreviouslycannotbe determinedfrom our data. In contrast to the model of EMSLAB Group [1988], the two conductivelayers are not connectednear y = 60 km, but instead a small resistive gap is shown which negligibly affects the response. The data neither require nor forbid an extensionof the dipping, low-conductancelayer under the subhorizontal conductor above it. 14,132 WANNAMAKER ET AL.: RES•STtVrrv CROSS SECI'ION, JUANDE FUCASYSTEM The ocean bottom portion of the finite element model is drawn only as far as y =-175 km in Figure 3 and Plate 1. Beyond this, the model remains essentially the same apart from a gradual decrease in the conductance of the lowresistivity sedimentary layer of the abyssal Cascadia Basin from values around2500 S beneathsite SF3 to essentially zero at SF8 near the ridge. The resistivitiesof the sediments of this basin and of the near-shoreNewport Basin are very low, in the range of 1-2 ohm m. However, the resistivityof the Miocene and older melange and broken formation [Snavely, 1987], which protrude to shallow depths beneath the geomagneticvariation sites SF1 and SF2, is much higher with a value of nearly 30 ohm m in our model. Extending down the trench and connecting to the dipping conductor beneath the Coast Range is a strip of low resistivity whose COMPUTED w E CR WV WC HC DB , i i i i i iii i i i i i i i i i i i • i I i i i ii ii i i i i i • +• o•+2 Pyx +3 ,4 I • I•::•1 I I III -2 I I ?ø,x_ -1 conductance, however, is lower than that of the Cascadia Basin sediments nearby. The continuity of this strip is tenuous and resolved best by the westernmost land MT stations. Shallow water MT measurements would •+• - ½yx be necessaryto resolve trench structure with more confidence. These are logistically very difficult becausethe continental shelf is swept continuouslyby fishing nets. Below the conductive seafloor sedimentarylayer resides a resistive (3000 ohm m) lithosphere a few tens of km thick (Figure 3). This unit is resolvedbest by the westerntwo MT sites where the seafloor sedimentsare relatively thin and do not screen the responseof the resistive layer below. At depths near 35 km, we see a dramatic fall in resistivity to around20 ohm m from 3000 ohm m. This conductiveregion is responsiblefor the steepdrop in apparentresistivitiesand the high impedancephasesat the shortestperiodsat SF7 and SF8. In the model, the resistivity of this mantle region is essentially constantto a nominal depth of 215 km. Models of geomagneticvariationsby Law and Greenhouse[ 1981] and DeLaurier et al. [1983] suggest that resistivity increases again below about 100 km, but the deeper reaches of these models are relatively poorly resolved [Oldenburg et al., 1984]. At depthsbeyond215 km in the oceanicmantle, even lower resistivities (about 1 ohm m) are shown. However, the resistivities at this depth and beyond are difficult to resolve. This difficulty stems from an inconsistencybetween the apparent resistivity and impedance phase of both modes at +4 •40 _ i i 0 50 Fig. 4. i 100 km Pseudosectionsof calculated transversemagnetic apparent resistivity Pyxandimpedance phasetpy x alongthelandward portion of theLincolnLine. Contour values of Pyxarein ohmm andof tpy x arein degrees. These and subsequentpseudosections shouldbe comparedto observedpseudosections presentedby Wannamakeret al. [thisissue]to assessgoodnessof fit. Color versionsof thesepseudosections appear in Plate 2 in the separatecolor sectionof this issue. COMPUTED W E SF8 SF7 SF5 SF4 SF3 +2 _ 13yx periods of 10ns and more, as discussedlater with the calculated response curves. The eastward extent of the oceanicmantle conductoris not well definedbut extendingit eastof thecoasthelpssimulatetheabruptfalloffin Mzy at periods below about 500 s. The conductivitybelow 215 km in fact could be drawn +5 +2 across the entire section because the land data consideredextendsonly to 10,000 s period. ComputedResponsePseudosections Next we show the calculatedMT responses of the model for comparisonwith the observationsin Wannamakeret al. [this •+3 (I)yx o +4 +5 issue].Firstwill be theTM modequantitiesPyxand •yx since 0 100 I I 200 km they have been emphasizedin the modeling. As seen in Figure4 and Plate 2, the agreementon land of the calculations with observationsis very good in both apparentresistivity and impedancephase. The fit of the model to the observed Fig. 5. py• and <py•ontheseafloor is shownin Figure5 andPlate3 resistivity Pyxandimpedance phasedpy x forthefiveMT soundings on I Pseudosections of calculatedtransversemagnetic apparent and is similarly good with one exception. (Plates2 and 3 can be found in the separatecolor sectionof this issue.) At the the oceanbottom segmentof the Lincoln Line. Contoursof apparent resistivity are in ohm m and of impedancephase in degrees. Color versionsof thesepseudosections appearin Plate 3 in the separatecolor very longperiods,greaterthan 10• s, computedpy• is section of this issue. WANNAMAKER ET AL.: RESIST[VrrV CROSSSECTION, JUANDE FUCASYSTEM 14,133 somewhat smallerthanobserved althoughthe fit to ½yxis The high calculated ½xy resultsfrom the very low good. This discrepancyis even worse for the TE quantities describedsubsequentlyand reflects an inconsistencyin the very long-periodseafloordata to which we will return. resistivities deep in the oceanic mantle, which must not be affectingthe land datain a two-dimensionalmannerfor the TE mode. Finally, the three-dimensional nature of deep Thecalculated TE pseudosections (p,, and½•) fortheland segmentof the Lincoln Line appear in Figure 6 and Plate 4. (Plate 4 can be found in the separatecolor section of this issue.) While they resemblethe data of Wannamaker et al. [this issue] to some extent, the assumption of a twodimensional structure for the TE mode fails seriously in several areas of the response. Due to the variety of threedimensionaleffectsinterpretedto exist in the TE data on land, these data have not been utilized in determiningthe structure of our model. Most obviously,the two-dimensionalTE model calculations showa very smoothlateralvariationin p, at periods longer than about 10 s. In contrast, the data pseudosectionsretain a strong vertical orientation to the apparent resistivity contours. This fundamental difference results from numerous finite strike length effects in the measuredTE data over upper crustal conductive structures under our profile. There are at least three other discrepancies between calculated and measured TE data that are due to three- dimensionaleffects. First, no two-dimensionalgeometrycan reproduce the very long-period anomalies in ½,, andthe slopeof p,, versus T, of narrowlateralextentaroundy = 30, 40 and 135 km [Wannamaker et al., this issue]. We must appeal to some three-dimensionalconnection or coupling between upper crustal structures local to the measurement sites and large-scale bodies off-line to our profile for an explanation.Next, approaching 104s, calculated •.• on resistivitystructurebelow the Deschutes Basinaffects p•y and ½,,• stronglyhere at long periods. The calculated responseof the preferredmodel in Figure 6, is far too low for p, andtoohighfor ½, at periodslongerthanabout10 s. An E-W boundary a few tens of kilometers south of our profile, with deep conductive material on its farther side, explains the occurrence[cf. Wannarnakeret al., 1984]. Such a geometry has been indicated by Gough et al. [this issue] from their geomagnetic array measurements. They have named the resistive northern region the Blue Mountains Resistive Block. The conductiveBasin and Range province is to the south. This east-west boundary provides an example, on the Lincoln Line, where a purely two-dimensional TM mode analysis may be inaccuratedue to "sideswipe"by a nearby, off-line structure of regional extent. The TM data alone suggest low resistivity to depths of at least 100 km here while the verticalmagneticfield, whichis measuredto l0 n s, implies that conductordepth extent is limited to only 65 km, as shown in Figure 3. The true upper mantle structurebelow the Deschutes Basin will not be resolved more precisely without three-dimensionalsimulations and probably further N-S MT profiling in this region. On the seafloor, as stated earlier, the preferred twodimensional model fits both TM and TE mode impedance responses almostequallywell, at leastdownto 10ns(Figure7 and Plate 5). (Plate 5 can be found in the separatecolor averageover the whole land segmentbecomeshigher than observed, whilethefall in p, vs.periodbecomes toosteep. COMPUTED W w COMPUTED SF8 E SF7 I_^ 0 -1 SF5 I SF4 SF3 I II •' +3- •' +1 Pxy Pxy • +4--• --•+2 '' +3 -- +5 -I I I I ! i i i +4 i i -2 • -1 •'+1 +3 +5 -I +4 I 0 0 •'• _ 403 '-0,•...• q) xy • +4•67•' +2 Fig. 6. 0 •+3--• I 50 I 100km Pseudosectionsof calculated transverse electric apparent Fig. 7. I I 0 100 I I xy I I 200 krn I Pseudosectionsof calculated transverseelectric apparent resistivity Px• andimpedance phaseqx• alongthelandward portion of resistivity Px• andimpedance phaseqx• alongtheseafloor portion of the Lincoln Line. Color versionsof thesepseudosections appear in Plate 4 in the separatecolor sectionof this issue. the Lincoln Line. Color versions of these pseudosectionsappear in Plate 5 in the separatecolor sectionof this issue. 14,134 WANNAMAKER ET AL.: RESISTIVITY CROSSSEC•ON,JUANDE FUC^ SYSTEM COMPUTED section of this issue.) One notable example of the twodimensionalnatureof the TE mode here at moderateperiodsis the fact that p• w below about3000 s appearslaterally uniform across the whole seafloor segment of the Lincoln Line. The inconsistency between the observed apparent resistivityand phaseat periodsgreaterthan 104s, noted +2 earlier for the TM mode, is even more obvious in the TE. • +3 Specifically, we canfit •p,qwiththemodelof Figure3 butnot p•. This difficultyis discussed furtherin the upcoming • +4 SF8 SF7 I I SF5 SF4 SF3 SF2 SF1 I I I i i I I I I I I ! i i i i i i i sensitivity tests. The verticalmagneticfield function M,y particularlyon +5 land has been of greater use in the interpretation of the LincolnLinethanhavetheTE impedance quantities p,q and •p,q.The areasof greatest valueincludetheresistivities of the +2 _ • -- /-.2• sedimentary sections offshore and the conductive axis associated with the High Cascades volcanic arc. The calculated response of M,• appears in Figure8 andPlate6 and its agreement with the observationsof Wannamaker et al. [this issue] is good overall. (Plate 6 can be found in the separate color section of this issue.) One minor area of disagreement occurs near the westernmost edge of the Willamette Basin sediments (y ,-- 30 km). The measured responsehere is weaker probably becausethe sedimentsare areally very patchy, i.e., have poor N-S continuity. The computedpseudosections of M,y for the seafloorstations appearin Figure 9 and Plate 7 and show good agreementwith the data of Wannamaker et al. [this issue] as well. (Plate 7 can be found in the separatecolor sectionof this issue.) That the vertical magnetic field is so much more consistentwith the TM mode impedance functions than with the TE mode impedance implies a greater continuity along strike of the current-densityvariations, which determine H,, relative to the electricfield variations,which affect the TE impedance. COMPUTED W E CR WB WC HC DB +5 Fig. 9. •, I I ] iiii iii i i i i i i i i i i i o i-- +1 • +2 +3 +4 I I 1 i i i ! ] i iii i i i i i i i i I I I 0 100 [ i I I I I 200 km i Pseudosections of calculatedvertical magneticfield transfer function M2y for the seafloorportionof the LincolnLine. Color versionsof thesepseudosections appearin Plate 7 in the separatecolor section of this issue. Model-Response Sensitivity Tests Any interpretationof incomplete, inaccurategeophysical data involves nonuniqueness[Parker, 1977]. We have presenteda model that fits the data very well. Other models undoubtedlyexist. One approachto a nonuniqueproblemis to look for modelswith minimumstructurerequiredto explain the data. We have undertakenthis in an informal way by adding deep structure only when the data could not be explained by shallow structure. Jiracek et al. [this issue] implementsuchan approachin a formal way by minimizing a mathematicalmeasureof structuresubjectto a certainlevel of data misfit. Another approach is to perturb structural elements I _ and examine the effect on the data. This is not a complete method of model appraisal becauseone limits the searchfor alternatemodelsto thosethat are fairly close to the starting model. However, it allows one to directly test a model against important competing hypotheses [Parker, 19771. We will follow the latter approach. Our initial tests concern the subhorizontaldeep conductorbelow the central Willamette Basin to the High Cascades. Figure 10 displays the observedTM and TE quantitiesfor a site 123 km inland from the coast. The goodnessof fit of the preferredmodel of Figure3 andPlate1 is apparent (solidcurvesof P•x and •x). Note that the broadbandand long-periodsites are not exactly coincident in geographic location so that the long-period apparentresistivitieshave a small static shift relative to the Fig. 8. Pseudosections of calculatedvertical magneticfield transfer broadband. It is only important in this case that the model functionM,y alongthelandward portionof theLincolnLine. Color versionsof thesepseudosections appearin Plate 6 in the separatecolor responsebe parallel to the long-period data in the log-log section of this issue. plot and that the phasebe fit. 0 50 100 km WANNAMAKER ET AL.: R•SIS•IViTYCROSSSECttON,JUANDE Fuc^ SYSTEM •- +3 xu i i ux •- i EMSL i 123 i i - 14,135 ,-, 0.75 ' •- =u •- zx N "• 0.00 E +2 _. I 1, .......... o -0.75 , +1 +0 I "• 90 '• I I I I , , , 0 , , EMSL, 147, I ' -0.75 313 I -2 I -1 I 0 I +1 I +2 +3 -1 I I +1 I +2 I +3 I +4 I +4 log T ($) 1 og T Fig. 11. Real and imaginary observedresponsecurves for elements M2xandM2ymeasured by University of Oregon andGSCabout147km Fig. 10. Broadband and long-period MT data collected by the Universityof Oregonand the GeologicalSurveyof Canada(GSC) about inland on the Lincoln Line. Computedresponsesinclude that of the model of Figure 3 (solid curves), that with the midcrustal conductive axis removed (dashes), and that with the mantle conductor below the 123 km inland on the Lincoln Line. Error bars are one standard DeschutesBasin increasedto infinite depths(dots). deviationplotted where larger than data symbol. The solid curve is the calculatedTM mode responsefor the preferredmodel of Figure 3. A test caserejectedfor its misfit is that of removingthe strong,subhorizontal conductoreast from the central Willamette Basin to the High Cascades both the vertical field and TM responseimplies that the and instead extendingthe Coast Range conductordowndip to 85 km depthbeneaththe High Cascades(dashedcurves). A secondalternative conductiveaxis has substantialN-S continuity. To the east, rejectedhashigh conductivityfrom 25 km depththroughto the downdip the TM response requires the subhorizontal conductor extensionof the CoastRange conductor(dottedcurve). extendingto at least 100 km depth and essentiallycan detect no bottom. However, an infinite depthextentdisagreeswith observedM,• at periodslongerthanabout1000 s (dotted As a first test model, the subhorizontal conductor is removed and the dipping conductorunder the Coast Range simply extended downdip to 85 km depth below the High Cascades (Figure 10). The conductance of the layer is increaseddowndip also to preservethe amplitudeof the peak in •y,•;for example,thelayerconductance at 65 km depthfor this test is about 104 S. One sees from the dashed curves of py,• and •,• thatsuchan alternative is untenablesincethe periodrangeof boththepeakin •,• andtheminimum in is clearly too low. As a secondtest model, we add lowresistivity material from the top of the subhorizontal conductorat 25 km depth all the way down to the extended dipping conductorjust considered. The responseof this structure (dotted curves) also is inconsistent with the observationsas it extendsthe phase peak to periods which are too long and reducesboth the amplitudeand slope of the apparentresistivity below what were measured.However, the observationsdo allow an extension of the dipping Coast Range conductor below the subhorizontal layer at 25 km depth as long as the conductanceof the former is limited, e.g., to no more than about 1000 S at 65 km depth. The next tests, illustrated in Figure 11, concern the pronounced, uplifted conductive axis beneath the High Cascades and the depth to the bottom of the deep subhorizontal conductor under the Deschutes Basin. Removing the upliftedaxisviolatesRe(M,•) in theperiod range from 100 to 1000 s around y = 147 km (dashedcurves). It also destroysthe closure of the negative contoursin the pseudosection of M,y in thesameperiodrange(Figure8 and Plate 6). That a single two-dimensionalstructureexplains curves). We suspectthat the TM responseis sensinghigher conductivityof the Basin and Range to the south and that a bottomof nominally65 km consistent with M,• is more likely. Again, additional N-S profiling and threedimensionalmodeling are the proper meansof resolving the deep structurehere. The sensitivityof the MT responseto the geometry of the dipping thin conductorbelow the Coast Range is assessedin Figure 12. Here we presentthe TM mode impedancephase valuesmeasuredin the periodrange3-300 s at sims1, 48, and 71 km inland from the coast. In the top panel, the phases from all three sites are superimposed to show that the peaks in •,• aredistinctbothin amplitude andperiodrangewithin the scatterof the data points.In the lower panels, the phases are plotted as one standarddeviation error bars centeredon the measuredvalues to demonstratethe good data quality. The solid curveslabeled"1", "a", and "A" indicatethe goodnessof fit of ourmodelto themeasured<py,•. Thefit to p•,• at these three sites is equally good, as exemplified for the site at 48 km in the EMSLAB Group [1988], and as can be deducedfrom the colorpseudosections.Curvesof dashesand dotsin Figure 12 are responsesof alternatemodels discussednext. That the Coast Range conductivelayer dips inland appears to be resolvedby our MT measurements independentlyand is not imposed from external tectonic considerations. In the first test, we make this layer flat and extend it westward to infinity with a constantdepthto top of 30 krn. The calculated ½•,,response at 48 km (notshown)is essentially unchanged. However, at 1 km (dashedcurve "2" in Figure 12) the phase anomaly has significantly lower amplitude and peaks at 14,136 WANNAMAKER ET At..: R•s•vrry I I 5O ß - vv • ß v• v• • lkm ß ß ß ß v v 48 km ß ?l km 20 I 58 Cnoss S•nON, JUANDE FUCASYSTEM I . 4O 1 km i. 'o 28 "J "> ......... 7.?.? 3 .= 2 I I 58 4O rule out some electrical . so 48 km i 20 50 - 40 71 km 30 2O I +1 connection. These errors result from strong cultural EM interference in this area. Since such a connectionwould have profound tectonic implications, it is important to remeasureseveral sites in this period range using a fully remotereferenceMT system. Resolutionof resistivity structureoffshoreby its effect on the verticalmagneticfield is examinedin Figure 13. The data qualityat T < 20 s is not the best,but fortunatelymostof the responseoccurs at longer periods. The fit of the preferred model to the data is reasonablygood. Note that both data and calculationsshow slight inflectionsin Re(M,y), with complementary behavior in Im(M,y), around1000s period. -- I low conductance(<100 S) portion of the dipping conductor. This is demonstrated by removingit east of abouty = 20 km and replacing it with background material of 350 ohm m. Suchan alternativefails to reproducethe subtlephasepeak in this area (dashedresponse"b"). A deep conductive root connecting the base of the Willamette Basin sedimentsto the western portion of the deep subhorizontal conductor appears in the minimumstructureresistivity crosssectionof Jiracek et al. [this issue]. We examine this possibility, which is in contrast to our model, in the lower panel of Figure 12. The computed responseof our model (solid curve "A") is comparedto the computedresponseof Jiracek et al. (dashedcurve "B"). Our model appearsto fit the data somewhatbetter. However, the error barsbelow 10 s are quite large and it seemsprematureto +2 When the sedimentarydistributionsof both the near-offshore NewportBasinandthe abyssalCascadiaBasinareremovedby 1 og T (s] setting them to 300 ohm m, we obtain the responseof the seawater alone (dotted curves). The seawater is quite insufficient by itself to explain the measuredvertical field. Fig. 12. Transverse magnetic impedance phase •x• in theperiod The dashedresponseoccurswhen the abyssalsedimentsare range3-300 s approximatelyat siteswhich are 1, 48, and 71 km inland from the coast. Phases in the lower panels are representedby one presentand only the nearby Newport Basin is removed.We standard deviation error bars centered on the estimated values. Solid see that this basin is very important in determining the curvesare the responseof the model of Figure 3 at eachsite. Alternative vertical field on land near the coast. If the oceanic upper model responsesplotted with dashesand dots are discussedin the text. mantle conductorin Figure 3 is not extendedeast of the coast, Data were collectedusingthe UURI broadbandsystem. the misfit is about 0.05 worse below 500 s, with the computed valuesof Re(M:•) beingtoolarge. longerperiodsthanobserved.The calculated<p•,,of thistest is in fact much weaker now at 1 km than that observed at 48 km, which is in opposition to the measured case. These lower phasesat 1 km are a result of close proximity to the ocean. Being able to specifya priori the seawaterresponsein the finite element model increasesour resolving capability under the CoastRange. The next test fixes the layer depth at the coastline and reducesthe dip inland to 10ø suchthat the layer depthat y = 48 km is about 20 km rather than 32 km as in Figure 3. Resultingis the dottedcurve "c" in Figure 12 which peaks at somewhat shorter periods than observed. We therefore concludethat the dip exceeds10ø and the preferredvalue is approximately20ø. The dottedcurve "3" at y = 1 km in Figure 12 demonstrates the effect of electrically disconnectingthe dipping Coast Range conductor from the trench offshore. This disconnectionwas made by settingthe resistivityof the layer ,-., 0.75 •- =u •,- =x f ! i i i i I EN$L 7•' ii i I I i i I O O O •--ZTll.... o .............. • ß • [ bgh..3.__k.]blmmlmmnj. -0.75 I 0.75 0.00 -0.75 I I I I I I - - I I -2 -1 ! 13 I +1 1og T I +2 I +3 I +4 (s) from the trench to about 15 km offshore to 300 ohm m, the valueof thematerialaboveandbelowit. Thecalculated<p•,, again falls far below that observed and cannot be reconciled by making the remainderof the Coast Range conductorless resistive. The final Coast Range test is for existenceof the Fig. 13. Real and imaginary observedresponsecurves for elements M:xandM:y measured byUniversity of Oregon andGSCabout 7 km inland on the Lincoln Line. Computedresponsesinclude that of the best model (solid), that with low-resistivitysedimentsof the Newport Basinremoved(dashes),and that with only the seawaterpresent(dots). WANNAMAKER ET AL.: RESISqlVlTY CROSSSECTION, JUANDE FUCASYSTEM The vertical magneticfield measuredon the seafloorat site SF1 is comparedto that due to the model of Figure 3, where the agreementis good,plus thoseof two alternativestructures (Figure 14). The first alternativecontainsvery conductive(2 ohm m) sedimentarymaterial throughoutthe whole melange and down the trench to a distance of about 15 km from the shoreline. Such a structure drastically curtails the peak responseat SF1, and even more at SF2, and movesthe peak to longerperiods(dashedcurves).Second,removal of all low resistivity material except the seawater results in a real componentof vertical field at SF1 that increasesto large values at periodsmuch lessthan the shortestmeasured(dotted curves in Figure 14). Disagreementbetween observedand computed Im(Mo) is largein thiscasealso. Negative values of Re(M,x ) develop beyond 104 s in Figure 14 and all other seafloor sites on the Lincoln Line. This behavior implies higher resistivitiesof regional extent southof our oceanicprofile. A similar characterto Mzx is observed on the land also. longer than 3000 s (dashed curves). The latter model has resistivitiesnear 20 ohm m extendingall the way to 400 km depth. This datainconsistencyexistsfor all five seafloorMT stations on the Lincoln 0 re' 0.75 ,."'-'" i ]; 4 - $F1 i i i ,_, 0.75 •- '" and remains a serious issue for America. One alternativemodel that seemsclearly inconsistentwith the data has resistivity in the oceanic mantle conductor falling steadilywith depthfrom 30 ohm m to 10 ohm m at 90 o.oop -0.75 Line the deep oceanic structure. Non-planar source fields which have escapedrobust processing are a possible explanation [but see Bahr and Filloux this issue], or perhapssome other form of bias error is present. Our measurementsappear only weakly sensitive to the resistivity of the oceanic lithosphere,in part becausethe data do not extend to short enough periods but also due to the nature of the ocean-continent resistivity transition. Ranganayaki and Madden [1980] show that the high ocean lithosphere resistivity, if contiguous with high resistivity under the continents,can strongly depresselectric fields and thus apparentresistivitieson the seafloor at long periods by restrictingcurrent flow in the overlying seawater. However, in the Juan de Fuca region, low resistivitymaterial extending down the subductiondecollement as in Figure 3 or beneath Vancouver Island [Kurtz et al., 1986] provides an electrical pathway between the ocean and low regional resistivity beneath North ,-, 14,137 0.00 km andto 3 ohmm at 215km. Thecalculated p,• fallstoo uniformlyand the calculated•,,• remainstoo constantto agree with the measurements(dotted curves). Note that the differencesbetweenthe variousmodel responsesin Figure 15 are significantlygreaterfor the TE mode than the TM. This could be expectedbecausewe are testingdeepmodel features which terminatelaterally near the coast. Limiting the lateral extent of the deep structurethis close to the seafloor profile reduces the TM mode response more than the TE. We therefore are presuming that off-line (three-dimensional) effectsare not influencingthe seafloorTE data as much as the range of our test computations. TECTONIC IMPLICATIONS -0.75 +2 +3 log +4 T +5 [s} OF RESISTIVITY STRUCTURE The electrical resistivity structure of the Juan de Fuca subductionsystemis a function of the chemical composition and physicalstateof that system. However, a particularvalue of resistivity can result from a plurality of mechanisms[see reviews by Brace, 1971; Olhoeft, 1981; Shankland et al., 1981; Shankland and Ander, 1983; Wannarnaker, 1986; Hyndman, 1988]. Therefore it is usually not possible to Fig. 14. Realandimaginary responses forMo andM:y observed at produce a unique interpretation of resistivity without oceanbottom site SF1. Error bars are one standarddeviationwhere they independentgeophysicaland geological constraints. There exceedthe size of the plotting symbols. Computedresponsesinclude thoseof the model of Figure 3 (solid), that with the melangeand trench exist a good many constraintsfor EMSLAB on upper crustal sedimentsgiven a resistivity of 2 ohm m (dashes),and that with only structure,seismicvelocity, the geotherm,and subductedslab the seawaterpresent(dots). mineralogy. Within this framework, electrical resistivity may augment our understandingof volcanic structure and history, sediments and pore water during subduction, The fit of the model to the measurements in the vicinity of evolution and migration of fluids during deep metamorphism, the Juande Fuca ridge is examinedmore closelyin Figure 15. and the distributionof melt in the upper mantle. Note that the agreementof the preferredmodel calculations (solidcurves)with both ½,, and ½yxis goodat almostall periodsbut thatthecalculated apparent resistivitiesp,,• and py,,arelowerthanobserved forperiods greater than104s. If the apparentresistivitiesare fit at the longestperiods, then the calculated phases are lower than observed at periods Upper CrustalStructureon Land Model resistivitiesin Figure 3 of about 100 ohm m or less for the Siletz River marine volcanicsof the Coast Range are representative of basalt formations near-surface. However, 14,138 WANNAMAKER ET AL.: RESIST•VrrY CROSSSF.CqION,JUANDE FUCASYSTEM I +3 I I ' I I I $F7 I $F 7 • - I I yX +2 .o o +1 IIIII--• +0 o90 --- I I I I I I I I I I I I I I I I i i i i i i i i i i i i +2 +3 +4 log T +5 +2 (s) +3 log +4 +5 T Fig. 15. Apparent resistivity and impedance phase data at ocean bottom site SF7 comparedto the computedresponse of the model of Figure 3 (solid curves, both modes). Error bars are one standard deviationplotted where larger than datasymbol. Notetheinconsistency in apparent resistivity andimpedance phase for T > 104 s, wherewe cannot fit the former without giving misfit to the latter. the conductive layer around 2 km depth below this unit suggestsa changein lithology. The SiJetzRiver basaltswere extruded over preexisting oceanic crust starting in the Paleocene [Snavely, 1987]. The conductor may represent sediments deposited below the Siletz River formation or within it during a hiatus in igneousactivity. Later intrusive igneouseventsduring the Miocene [Baldwin, 1981] may have further reducedresistivity by hydrothermalalteration. The large volume of sedimentarymaterialof the Willamette Basin constitutesthe most Lmportantupper crustal resistivity inhomogeneityon the landwardsegmentof the Lincoln Line. Marine units of Eocene throughOligocene age comprisethe volumetrically greatestproportion of sedimentsespecially of the westernportion of the basin [Wells and Peck, 1964] and possessresistivities in the range of 10-15 olhm m in the model. Toward the east, pyroclasticvolcanicsof Oligocene and Miocene age of Western Cascadesaffinity dominate the stratigraphy [Verplanck and Duncan, 1987] and also are in the range 10-15 ohm m. The relatively restrictedoccurrence of Columbia River basalts near the center of the basin probably is representedby members of the Grande Ronde Formation and locally the WanapumBasalt [Baldwin, 1981]. The variation in the basementprofile of our resistivitymodel in Figure 3, particularly the horst-like feature near y = 100k•n, is suggested also in the cross-sections of Couch and Riddihough [1989, in press] derived from gravity and drilling and of Hermanceet al. [1989, in press]from their profile of geomagneticvariation observations. Beneath the easternportion of the Western Cascades,the volcanic strata were extensively intruded by the Halls Diorite Porphyry of early Miocene age [Baldwin, 1981] which probably contributes to the high model resistivities at shallow depths here. The shallow conducting layer commencingnear Detroit Lake (y = 150 km in Figure 3) and continuing eastward is tentatively correlated with unwelded tuffaceous units of the Breitenbush formation [Priest et al., 1983, 1987]. The overlying more resistive unit is mainly middle to late Miocene lavas of late Western Cascades association.The westernboundaryof the conductorappears to coincidewith a major N-NE trendinggrabenfault, known as the Hoover fault in this area [Priest et al., 1987], which is part of the regional systemseparatingthe downdroppedHigh Cascades graben from the relatively uplifted Western Cascadesblock [Priest et al., 1983; Smith and Taylor, 1983]. The abrupt increase in conductance of the shallow conductivelayer east of the High Cascadesaxis suggeststhat different or additional lithologies are present. The MT stationseastof Mount Jeffersonlie a good 10 km southof the WANNAMAKER ET AL.: RESISTIVITY CROSSSECTION, JUANDE Fuc^ SYSTEM peak andcrossan areaof anomalously low Bouguerresidual gravity [Couchet al., 1982;Smith and Taylor, 1983;Smith et al., 1987]. The low resistivity both within the eastern portionof the High Cascades grabenandeastof the bounding Green Ridge fault in the DeschutesBasin thus may represent an accumulationof tuffaceous material from ancestral High Cascadesvolcanic centersnearby or from an early Pliocene silicic volcanic "highland"some50 km to the south [Smith and Taylor, 1983]. The conductoris coverednow by resistive basalticrocks of the upper DeschutesFormationof the early High Cascadesvolcanicepisode[Priest et al., 1983]. That the conductivelayer doesnot appearsignificantlydeeperwest of Green Ridge fault than to the east indicatesthat late High Cascadesvolcanicseruptedafter formationof the grabenare not very thick in this area. We believe that the MT data have not resolved significant lateral variations within the middle crust underlying the land segmentof the Lincoln Line. Thus no new light has been shed on the boundary between Mesozoic and older rocks comprisingthe deepbasementof centralOregonandthe early Cenozoicoceanicrocks presumablyforming the deepercrust of western Oregon [e.g., Couch and œiddihough,1989, in press]. Resistivityvaluesof hundredsof ohm m are typicalof a wide range of lithologies with water-saturated, interconnectedporosities below 1%. Delineation of any terraneboundaryis hamperedbothby limitationson the twodimensional assumptionand by limitations on data quality especiallyin the culturallydevelopedWillametteBasin. DeepCrustand UpperMantle onLand Hyndman [1988], prograde metamorphic conditions which the subducting slab encountersfavor the transition from blueschist or zeolite to greenschist facies with attendant releaseof water. Most of this dewateringmay occur west of y = 20 km, where the more conductiveportion of the Coast Range conductorresides,althoughthis latter feature may in part also represent a greater abundance of subducted sediments. Further updip under the melange and Newport Basin area, breakdown of clays might be another source of water. The dipping conductivelayers inferred beneaththe Oregon Coast Range and beneath Vancouver Island [Kurtz et al., 1986] have been interpretednot to exist preciselyat the plate interfacebut insteadto representfluids risen buoyantly from the plate surfaceand trappedbelow an impermeableboundary above [Hyndman, 1988]. This boundarywould be formed by hydration of minerals by the fluids or by precipitation of silica from the fluids as they cool. The dip of the conductor under Vancouver Island has been correlated with dip of the isotherms, in particular the 450øC isotherm, in accordance with decreasingheat flow eastward. In contrast,dip of the isothermsis not resolved by the sparseheat flow data from western Oregon [Blackwell and Steele, 1983; Couch and Riddihough, 1989, in press] while inland dip of the Coast Range conductor evidently is. Also, based on COCORP seismicreflection profiling about 30 km southof the Lincoln Line, Keach et al. [1989] interpretthe subductiondecollement of the Juande Fuca plate to lie at a depthnear 35 km some50 km inland. This is essentially coincident with our dipping conductivelayer at a correspondingdistancefrom the coast, at least Discussionin this sectionfocuseson the deep conducting layer a few tensof kilometersbelow nearly all of the Lincoln Line on land. Under the Coast Range and westernmost Willamette Basin, the thin conductor is presumed to represent,at least in part, pore water or sedimentsalong or near the interfacebetweenthe subductingJuan de Fuca plate and the overridingNorth American continent. This parallels the interpretationof Kurtz et al. [1986] on VancouverIsland where a similar thin conductor was resolved using MT. In their study, interconnected porositiesof one to a few percent were inferred by assumingfluids of seawater salinity or greater. A similar inferencecanbe made for the layer beneath the CoastRange (Figure 3). Interconnectedporositiesof 12% would result for the interval y = 20-60 km where the conductanceis less than 100 S but might reach 10% for the portionwest of y = 20 km whoseconductancemay approach 1000 S. Alternatively, this higher conductancein part may representa greaterlayer thickness. In addition to possiblesubductedsedimentsor pore waters, fluids releasedduring mineral devolatilizationreactionsmay reduce resistivity here. Constraints on the geotherm and alteration mineralogy of the subductedslab are needed to decide which reactions could be taking place [e.g., Wannamaker, 1986; Hyndman, 1988]. Heat flow and crustal heat productionof western Oregon are similar to those of southwestern British Columbia [Blackwell and Steele, 1983; Lewis et al., 1988; Couch and Riddihough,1989, in press]so that somecalculationsperformedfor the latter region can be made use of in EMSLAB. Accordingly, temperaturesin the range of 250-500øC appearappropriatefor the depth interval 12-30 km of the dippingCoastRangeconductor.As notedby 14,139 within the uncertainties involved. The EMSLAB results therefore do not unequivocally supportthe notion of rising fluids beneath the Coast Range trapped at a dipping boundarysignificantly above the subductionslip plane. Fluids are offered also as an explanation for the subhorizontalzone of high conductanceextendingeast from y = 60 km to the High Cascadesaxis (Figure 3). We suggest that this pronounced conductor represents primarily a concentrationof water-rich fluids which have risen buoyantly from the surface of the subducted plate where further dehydrationreactions are taking place. The upper limit of this fluid migration, i.e., the top of the conductor, we interpret to be controlledby temperature. Heat flow data in Oregon [Couch and Riddihough, 1989, in press] suggesta temperaturearound450øCfor the top of the conductor.Sucha temperatureappearsappropriatefor the mineral hydration and silica precipitation mechanisms of permeability sealing mentionedpreviously. Furthermore,the rheology of the host rock becomesincreasinglystiff toward shallower levels due to lower temperatures[Kirby, 1983]. The temperaturesover most of the conductor appear too low to support hydrous melting, but more accurateheat flow data to characterizethe geothermin this area are highly desirable. To understandwhich hydrous minerals may be breaking down to contribute the fluid, we need again to constrain temperatureat the subductedplate surface. However, it is hazardousto extrapolate a conductivegeothermbeyond the strong conductor being considered. Heat absorbed in dehydrationat the slab,hot fluids migratingupward, and heat producinghydrationreactionsat higher levels and within the conductor may constitute significant advective heat flow [Anderson et al., 1976; Lewis et al., 1988]. This would tend 14,140 WANNAMAKER ET AL.: RESlSTIVrrY CROSSSECTION, JUANDE FUCASYSTEM to reduce the geothermal gradient below the electrical conductor. At the easternmostend of the dipping weaker conductorof the Coast Range, i.e., just before the start of the low-resistivity region under discussion, a conductive geotherm suggests temperatures of 550-600øC. If this is subductionreviewed by Wyllie [1988], whereinoceaniccrust whose surface is cooled by endothermic dehydration is subducted below a mantle wedge warmed by induced convection. greenschist facies minerals as the cause of the strong conductorat least near its western end around y = 60 km. This interpretationof the fluid source is consistentwith the observation that greenschist minerals, especially chlorite, carry the bulk of the chemically bound water in subducted oceanic crust [Humphris and Thompson, 1978; Ito et al., Next we considerthe conductiveaxis extendingup from the east end of the strong, horizontal conductorjust described around y = 160 km. Its depth to top, about 5 km, may representthe upper limit of developmentof andesiticmagmas within the crust accordingto Anderson[ 1982]. We doubtthat this midcrustalconductorresultsprimarily from the presence of silicatemelt in large volume but rather from hydrothermal fluids exsolvedduring repeatedintrusion and crystallization. 1983]. It is not clear what structural Breakdownof hydrousmineralsgenerallyis incongruentso that fluid release occurs over a finite temperatureinterval [e.g., Etheridge et al., 1983]. The interval of breakdownof greenschistfacies may be broadenedfurther if serpentineand epidote, commonly observed also, are included. A finite interval of breakdown temperature translatesto a region of fluid release and upwelling which is of finite lateral extent, especially over a plate whose angle of subductionis not very steep. The lateral extentof fluid releasemay be increasedyet more through the buffering of temperature along the slab surfaceby the dehydrationreactionsthemselves[Andersonet be centeredjust west of the presentandesiticarc insteadof directly beneath it. However, numerous basaltic vents and vent lineamentsalsohave developedthis far west of the arc in the last 5 m.y. in response to a relaxation of regional compressiveforces[Hughesand Taylor, 1986]. The presence approximately representative, it impliesdehydration of al., 1976]. As we look farther downdip, more refractory hydrate minerals of upper greenschistand amphibolite facies should not breakdown to release H20 fluid directly but instead produce water-undersaturated melts through vapor-absent fusion [Burnham, 1979; Wyllie, 1979; Wannamaker, 1986]. The water-bearing melts themselves may rise through the overlying mantle wedge and crystallize as they lose heat or encounterthe H20-saturatedsolidusat lower pressures.This process, in essence, merely complicates the migration of fluid to higher levels. Nevertheless,melts are believed to be more effective metasomaticagentsthan fluids [Eggler, 1987] so that more substantial upward movement of alkalis and incompatible elements may be occurring here. The strong horizontal conductor and its interpretation may have an analog in the model of Kurtz et al. [1986] where a thick conductive region at a similar depth lies below mainland British Columbia. It may also be manifest seismicallyin the low velocity zone interpretedbeneath the Willamette Basin region by Langston [1977]. The model of hydrousmineral breakdownand fluid release presentedhere helps explain the imbalancenoted by Ito et al. [1983] between the amount of H20 contained in altered oceanic crust and the much lesseramountreleasedduring arc magmatism. Our interpretationis that most of the water in the altered ocean crust is liberated well before the volcanic facies breakdown boundaries described here axis has been detected to also to the north of our line near BreitenbushHot Springs and to the south near the latitude of Mount Washingtonfrom the MT profiling of Stanley et al. [1989]. While the depth of this zone is somewhat beyond that currently considered viable for geothermal resources, its substantial size may warrant evaluationfor future energyneeds. As noted already, a broad three-dimensional effect has complicated the determination of deep resistivity structure under the Deschutes Basin using a two-dimensional interpretationprocedure. The nature of the three-dimensional effect suggestsa roughly E-W transitionto lower resistivity south of the Lincoln Line. A similar structureis implied in the geomagneticarray studyof Gough et al. [this issue]where resistivelithosphereof the Blue Mountainsblock, including the Deschutes Basin, lies north of the more conductive Basin and Range province [Gough, 1983; Wannamaker, 1983]. This transitionmay be marked at the surfaceby the Brothers fault zone and a coincidenttrend of Late Pliocene to present volcanic vents oriented ESE [Guffanti and Weaver, 1988]. The magnetotelluric and geomagneticdata together suggest that the Brothers fault zone may be a better choice for the northern limit of recent Basin and Range activity than is the SW-NE trending Klamath-Blue Mountains lineament [cf. Eaton et al., 1978; Riddihough et al., 1986]. Further threedimensionalanalysisand data acquisitionis sought to better define the back-arc electrical asthenosphereas teleseismic results imply low velocity in this area extending to 200 km depth [Rasmussenand Humphries, 1988]. Implications of Seafloor ResistivityStructure arc is reached. Also, the model of this paper is similar to that of Fyfe and McBirney [1975] but with one importantdifference. The of the conductive controls cause this conductor are a full metamorphicgrade lower than those of Fyfe and McBirney [1975] at correspondingpositions between the trench and arc. Significant uplift due to volumetric expansion from hydration reactionsbetweenreleasedfluid and mafic minerals overlying the subducted plate is interpreted to affect the Western Cascadesand easternWillamette Basin by Fyfe and McBirney [1975]. An unknownportion of the releasedfluid may not be absorbedin hydrationbut perhapsreachescrustal levels and eventually the hydrosphere. Our resistivity cross section more closely ressembles a petrological model of The resistivities of the thick sedimentarylayering of the abyssalplain (Cascadia Basin) and on the continental shelf (Newport Basin) are strikingly low, 1-2 ohm m (Figure 3). These values are in keeping, however, with very high porosities observed in drilling of accretionary complexes which range from 25 to 75% in the upper 2 km [Kulm et al., 1973; Bray and Karig, 1985]. In contrast, such low resistivities cannot exist throughout the melange and subducted trench sediments without causing a vertical magneticfield responsemuch smaller than observed(Figure 14). Our magnetotelluricresultsthereforeappearto sensethe off-scrapingof seafloor sedimentsor the porosityreductions attending sediment dewatering during subduction. The WANNAMAKER ET AL.: RESISTIVITY CROSSSECTION, JUANDE FUCASYSTEM primary pathways for dewatering appear to be along the decollementbetween the subductedplate and the overlying accretionarywedge as well as up thrust faults in the wedge [Kulm et al., 1986; Moore et al., 1987]. The final resistivity from the imbrication and dewatering processas represented by the Miocene melangeand broken formation is in the tens of ohm m at least and suggestsbulk porosity reductionsof a factor approaching10. The nominal value of 3000 ohm m for seafloorlithospheric resistivity in Figure 3 may be a minimum. Resistivitiesin the thousandsof ohm m have been inferred for depthsgreater 14,141 distanceinland agreeswith the P-wave tomographicimage of Rasmussenand Humphries [1988] which implies a transition to a deep, high-velocity root in the mantle from below the centralWillamette Basin to the High Cascades. The higher resistivities of regional extent south of the Lincoln Line on the seafloorimplied by the vertical magnetic field may reflect lower temperaturesat depth south of the Blanco fracture zone. These in turn are suggestedfrom the greaterbathymetricdepthsin the vicinity of the Gorda Ridge. An explanationfor higher regionalresistivityto the southon the landwardportionof our profile is unclearat present. than 1 km below the seafloor from active source A physical explanation for very low model resistivity electromagneticsurveying as well [Young and Cox, 1981]. below 215 km in the oceanicupper mantle is problematic. Traces of water may be responsiblefor these values in the While its mere existence is questionable given the data middle oceanic crust, while thermally activated rock inconsistencydiscussedpreviously, very low resistivities at semiconductionis probably the mechanismof resistivity at depths less than 300 km have also been interpretedbeneath greaterdepths. It was noted earlier that the resistivityof the the East Pacific Rise spreadingcenter [Filloux, 1982] and ocean lithosphereis hard to resolvehere in large part due to beneathHawaii, a site of presumedhot spot activity [Larsen, low-resistivity material extending down the subduction 1975]. These depthsappeartoo shallowto be correlatedwith decollement beneath the continent. Subduction surrounds the olivene-spinel phase transition [Akimoto and Fujisawa, almost the entire Pacific Ocean so, to obtain better estimates 1965] occurringnear 400 km. One possibilityfor thesevery of ocean lithosphere resistivity using MT, a profile of low resistivitiesis a deep region of additionalmelting in an soundingsshouldbe carried out acrossa continentalmargin olivene eclogite source composition [Anderson, 1987; in the Atlantic. Wyllie, 1988]. Added thermal impetus for deep melting The low resistivity beyond about 35 km depth below the below the Juan de Fuca spreadingcentermay come from its seafloor (Figure 3) is proposed to represent partial melt proximity to the Cobb hot spot [Karsten and Delaney, 1989]. beneaththe Juande Fucaplate. To relate valuesof resistivity If mantle resistivitiesdown to 400 km depthremain in the to degreeof partial melting, we assumefirst that temperature range 15-20 ohm m however, as the apparent resistivity within the melt zone is buffered near the volatile-free alone suggests,then no such deepmelting is implied. These peridotitc solidusby the processof adiabaticupwelling and higher resistivities would be consistent with solid-state fusion [Mysen and Kushiro, 1977; Wyllie, 1979, 1988; semiconduction in olivene along a near-adiabatictemperature Oxburgh, 1980]. An estimateof melt fraction can be made profile of 1700-1800øCfrom 200-400 km depth. This is an thereby using the tholeiite liquid conductivitymeasurements importantalternatemodel of the deep physicochemicalstate of Tyburczy and Waft [1983] together with an appropriate in regionsof mantle upwelling and fusion [Wyllie, 1988] that mixing law relatingbulk conductivityto the conductivitiesof our magnetotelluricmeasurementscannot clearly distinguish the melt and hostmineral phases. The mixing law employed at present. This underscoresa need for obtaining reliable here is the grain edgetubulemodel discussedby Shanklandet planewave impedanceestimatesto very longperiods,3x104 al. [1981] andvon Bargertand Waft [1986] which represents s at least. the equilibrium texture of melt distribution in olivene-basalt CONCLUSIONS melt systems.The olivenedata of Duba et al. [1974] are used for the solid-stateconductivityof the mineralphase. Near the top of the low resistivityregion in the oceanic The inferencesdrawn from interpretingthe Lincoln Line upper manfie, a melt fraction of about 7% is implied from a observationsin EMSLAB fall into two broad categories; thoseon the physicochemical stateof the young Juande Fuca resistivity value of 20 ohm m. As depth increases,this fractionfalls due mainly to increasingmelt conductivitywith plate and its subduction beneath Oregon, and those increasingsolidustemperature but also to increasingolivene concerning the two-dimensional modeling of MT data in conductivityespeciallybelow 100 km depth. The computed environments which depart somewhat from the twodimensional assumption. Regarding the former, melt fractionis about3% at 75 km depth,just under1% at 125 km, andbecomeszero by 200 km depthwhereolivenesolid- measurementson the seafloor indicate very low resistivity in state resistivity at 1700øC has fallen to 20 ohm m. A the eastwardthickening sedimentarywedge of the abyssal reasonablygood comparisonexistsbetweenour conductivity CascadiaBasin provided its depth extent is constrainedby model and seismic models of young (0-20 m.y.) oceanic marine seismicsections. The majority of thesesedimentsare plates,of which the Juande Fuca beneathour profile is one. inferred to be off-scraped or lose most of their interstitial Sato et al. [1988], in their compilationof shearwave models, water before being carried more than ten or so kilometers suggesta few percentpartial melt to depthsof about 150 km. down the trench. In the oceanicuppermantle, of the Juande Laboratory and theoreticalinvestigationsof partial melting Fuca plate, moderatelylow resistivifiespresumablyreflect up also suggest that melting should have similar effects on to severalpercent partial melt attendingregional upwelling conductivityand velocity [Shanklandet al., 1981]. Our data in the vicinity of the ridge. Very low resistivitiesmodeled do weigh against resistivity decreasingsteadily in the below 215 km depthin the oceanicmanfie are hard to explain oceanicasthenosphere (Figure8) which wouldhave suggested physically but problems with the very long-period data hamper the modeling here. an interconnected melt fraction of 1% persistingfrom 75 to 3vet 300 km depth [cf. Anderson, 1987]. We note in addition On land, upper crustal resistivity structure indicates the that the termination of this asthenospherea moderate cumulative and ongoing effects of plate convergenceand 14,142 WANNAMAKER ET AL.: RESlST•TrYCROSSSECI•ON,JUANDE Fuc^ SYSTEM volcanism. Residual sediments, pore waters, possibly sulfides,and dehydratingalteredoceancrustcarrieddown the subductionzone appear to produce a low resistivity upper mantle layer, albeit a subtle one, dipping about20ø eastward beneath westernmost Oregon. The position of this layer appears consistent with the Juan de Fuca plate subduction decollement. Farther inland and downdip at appropriately high P-T conditions, a massive breakdown of greenschist minerals is hypothesizedto liberate most of the water carried in the ocean crust alteration well before the volcanic arc is reached. This fluid releaseproducesa strong,subhorizontal conductorwith a depthto top of about25 km. The resistivity manifestationof the arc magmatismitself is not strong,but consistsof a low resistivity axis in the middle crust below and to the west of the arc at the surface. It presumably representsmainly fluids releasedfrom crystallizingmagmas and should be considered for its geothermal energy significance. North central Oregon is resistive at depth comparedto the crustand uppermanfie to the southand it is suggestedthat the Brothersfault zone marksan electrical,and thus possibly a thermal and magmatic, boundary with the northernBasin and Range. Regarding the derivation of the resistivity cross section from the observations,the strongN-S preferredorientationof most structures in the region has allowed successfultwodimensionalmodeling of data approximatingthe TM mode. A fixed north-orientedx axis was essentialin clarifying the subtle anomalies due to subduction structure. The vertical magnetic field, however, confirmed many structuresderived using the TM impedancefunctionsand also was sensitiveto narrow conductors such as the midcrustal axis beneath the High Cascades. It was also important in establishingthe resistivity of the sedimentary sections offshore and the conductanceof the oceanic asthenosphere.On the land, the data correspondingto the TE mode exhibit a variety of threedimensional effects including shallow statics, finite strike length of regional features, and some strongly threedimensionalbehavior at very long periods which is difficult to explain. A fruitful area for three-dimensionalmodeling could be the DeschutesBasin region of the Blue Mountains resistive block to better establish the depth extent of the backarc electrical asthenosphere.On the seafloor, the much greaterlateraluniformityof structurenear-surfacehas allowed reasonablygood fits to both modes of impedanceplus the verticalfield downto 104s (-200 km depth)butnotmore. The unprecedenteddata quality achievedin the EMSLAB project has allowed a geophysical interpretation which is similarly unparalleled in scope and detail from induction studies. Nevertheless, several data collection and interpretation needs for the future of MT have been highlighted by our experiment. Strong cultural noise and near-field effects often correlated over several kilometers compromised resolution beneath the Willamette Basin. These can only be overcome using a remote reference at distancesexceeding the correlation length of the noise. A more flexible approachto electric field collection, or even continuous E field profiling (e.g., EMAP) locally, would efficiently meet data sampling requirementsespecially over volcanic rocks. Many more details of trench structure beneath the near-offshore continental shelf could be revealed if the logistics of shallow water MT were worked out. Characterizationof the oceanic lithospherewould be greatly enhancedby extendingthe seafloordata bandwidthto shorter periods using more sensitive magnetometers and precise timing control. Better processingto removesourceeffectsor bias may be neededto achieve good data at periods beyond ! 04s and resolvestructurein the 200-500 km depthrange. Finally, addition of independentgeological knowledge can substantially increase the information that can be extracted from induction data. For example, incorporatingthe ocean bathymetry and resistivity permitted determining the dip of the conductor under the Coast Range. Thus inversion algorithms that in the future may replace the laborious modeling processcarried out in this paper should allow for such independentlydefined constraints. Acknowledgments. The modelingand interpretationof this paper were supportedby the U.S. National ScienceFoundationunder grants EAR-8410638, 8512895 and 8708382. Many valuable discussions were held with RichardW. Couchand GeorgeR. Jiracek Financial assistancefor publicationcostswas provided by AT & T Bell Laboratories,GeologicalSurvey of Canada,and University of Washington. Philip E. Wannamakeris grateful to the San Diego Supercomputing Center and the University of Utah for accessto the SDSC Cray XM-P under the University Block Grantsprogram. This is GeologicalSurvey of CanadaContribution23489. REFERENCES Akimoto, S.-I., and H. Fujisawa, Demonstration of the electrical conductivityjump producedby the olivene-spineltransition, J. Geophys.Res., 70, 443-449, 1965. Anderson,A. T., Jr., Parentalbasaltsin subductionzones:Implications for continentalevolution,J. Geophys.Res., 87, 7047-7060, 1982. Anderson, D. L., The depths of mantle reservoirs, in Magmatic Processes:Physicochemical Principles,editedby B. O. Mysen, Spec. Publ. 1, pp. 3-12, The Geochemical Society, Dayton,Ohio, 1987. Anderson, R. N., S. Uyeda, and A. Miyashiro, Geophysical and geochemical constraints at converging plate boundaries, part I, Dehydration in the downgoing slab, Geophys.J. R. Astron. Soc., 44, 333-357, 1976. Bahr, K., andJ. H. Filloux, Local Sq responsefunctionsfrom EMSLAB data,J. Geophys.Res., this issue. Baldwin, E. M., Geology of Oregon, 3rd ed., 170 pp., Kendall/Hunt, Dubuque,Iowa, 1981. Banks, R. J., The overall conductivity distribution of the Earth, J. Geomagn.Geoelectr.,23, 337-351, 1972. Blackwell, D. D., and J. L. Steele,A summaryof heat flow studiesin the Cascade Range, Geotherm. Resour. Counc. Trans., 7, 233-236, 1983. Bostick, F. X., Jr., Electromagneticarray profiling (EMAP), extended abstractof paperpresentedat the 56th annualmeetingof the Soc. of Explor. Geophys.,Houston,Tex., 1986. Brace, W. F., Resistivityof saturatedcrustalrocksto 40 km basedon laboratorymeasurements, in The Structureand PhysicalPropertiesof the Earth's Crust, Geophys.Monogr. Ser., vol. 14, editedby J. G. Heacock,pp. 243-256, AGU, Washington,D.C., 1971. Bray, C. J., and D. E. Karig, Porosity of sedimentsin accretionary prismsand someimplicationsfor dewateringprocesses, J. Geophys. Res., 90, 768-779, 1985. Burnham, C. W., Magmas and hydrothermalfluids, in Geochemistryof HydrothermalOre Deposits,2nd ed., editedby H. L. Barnes,pp. 17136, JohnWiley, New York, 1979. Couch,R. W., G. S. Pitts, M. Gemperle,D. E. Braman,and C. A. Veen, Gravity anomaliesin the CascadeRange in Oregon:Structuraland thermal implications, OF Rep. 0-82-9, Oreg. Dep. of Geol. and Miner. Ind., 43 pp., Portland, 1982. Couch, R. W., and R. P. Riddihough,The westerncontinentalmargin of N. America: Baja California to the Queen Charlotte Islands, in GeophysicalFramework of the United States, GSA Monogr. Ser., vol. X, editedby W. Mooney and L. Palriser,GeologicalSocietyof America, Boulder,Colo., in press,1989. DeLaurier, J. M., D. R. Auld, and L. K. Law, The geomagneticresponse acrossthe continentalmargin off VancouverIsland: Comparisonof results from numerical modeling and field data, J. Geomagn. Geoelectr., 35, 517-528, 1983. WANNAMAKER ET AL.: RESIS• CROSSSECTION, JUANDE FUCASYSTEM 14,143 Duba, A., H. C. Heard, and R. N. Schock, Electrical conductivity of olivene at high pressureand under controlled oxygen fugacity, J. Geophys.Res., 79, 1667-1673, 1974. Law, L. K., and J.P. Greenhouse,Geomagneticvariationsoundingof the asthenosphere beneaththe Juande Fuca ridge,J. Geophys.Res., 86, Eaton, G. P., R. R. Wahl, H. J. Prostka,D. R. Mabey, and M.D. Lewis, T. J., W. H. Bentkowski, E. E. Davis, R. D. Hyndman, J. G. Souther, and J. A. Wright, Subductionof the Juan de Fuca plate: Thermal consequences,J. Geophys.Res., 93, 15,207-15,225, 1988. Livelybrooks,D. W., W. W. Clingman, J. T. Rygh, S. A. Urquhart, and H. S. Waft, A magnetotelluricstudyof the High Cascadesgrabenin central Oregon, J. Geophys.Res., this issue. Moore, J. C., et al., Expulsionof fluids from depth along a subduction- K!einkopf, Regional gravity and tectonic patterns: their relation to late Cenozoic epeirogeny and lateral spreading in the western Cordillera, in Cenozoic Tectonicsand Regional Geophysicsof the Western Cordillera, edited by R. B. Smith and G. P. Eaton, Mere. Geol. Soc. Am., 152, 51-92, 1978. Eggler, D. H., Solubility of major and trace elements in mantle metasomatic fluids: experimental constraints, in Mantle Metasomatism,editedby M. A. Menzies and C. J. Hawkesworth,pp. 21-41, Academic, San Diego, Calif., 1987. EMSLAB Group, The EMSLAB electromagneticsoundingexperiment, Eos Trans. AGU, 69, (7), 89, 98-99, 1988. Etheridge,M. A., V. J. Wall, and R. H. Vernon, The role of the fluid phase during regional metamorphism and deformation, J. Metamorph. Geol., 1, 205-226, 1983. Filloux, J. H., Magnetotelluric experiment over the ROSE area, J. Geophys.Res., 87, 8364-8378, 1982. Filloux, J. H., Instrumentationand experimentalmethodsfor oceanic studies,in Geomagnetism,1, editedby J. A. Jacobs,pp. 143-248, Academic,San Diego, Calif., 1987. Fyfe, W. S., and A. R. McBirney, Subduction and the structure of andesitic volcanic belts, Am. J. Sci.,275-A, 285-297, 1975. Gough, D. I., Electromagnetic geophysicsand global tectonics, J. Geophys.Res., 88, 3367-3378, 1983. Gough,D. I., D. McA. McKirdy, D. V. Woods,andH. Geiger,Conductive structuresand tectonicsbeneaththe EMSLAB land array,J. Geophys. Res., this issue. Guffanti, M. and C. S. Weaver, Distribution of late Cenozoic volcanic vents in the CascadeRange: Volcanic arc segmentationand regional tectonic considerations,J. Geophys.Res., 93, 6513-6529, 1988. Hermance, J. F., S. Lusi, W. Slocum, G. A. Neumann, and A. W. Green, Jr., A high densityremotereferencemagneticvariationprofile in the Pacific Northwest of North America, Phys. Earth Planet. Inter., in press, 1989. Hughes, S.S., and E. M. Taylor, Geochemistry, petrogenesis,and tectonicimplicationsof centralHigh Cascadesmafic platform lavas, Geol. Soc. Am. Bull., 97, 1024-1036, 1986. Humphris,S. E., and G. Thompson,Hydrothermalalterationof oceanic basalts by seawater, Geochim. Cosmochim. Acta, 42, 107-125, 1978. Hyndman, R. D., Dipping seismic reflectors, electrically conductive zones, and trapped water in the crust over a subductingplate, J. Geophys.Res., 93, 13,391-13,405,1988. Ito, E., D. M. Harris, and A. T. Anderson,Jr., Alteration of oceanic crust and geologic cycling of chlorine and water, Geochim. Cosmochim. Acta, 47, 1613-1624, 1983. Jiracek, G. R., J. H. Curtis, J. Ramirez, M. Martinez, and J. Romo, Two- dimensionalmagnetotelluricinversion modeling of the EMSLAB Lincoln Line, J. Geophys.Res., this issue. Jones, A. G., The problem of current-channeling:a critical review, Geophys.Surv., 6, 79-122, 1983. Kariya, K .A., Limitations of the magnetotelluricmethod as applied to the Pioche-Marysvaletrend, Utah, M. S. thesis, 56 pp., University of Utah, Salt Lake City, 1986. Karsten,J. L., and J. R. Delaney,Hot spot-ridgecrestconvergence in the northeastPacific, J. Geophys.Res., 94, 700-712, 1989. Keach, R. W., II, Oliver, L. D. Brown, and S. Kaufman, Cenozoic active marginand shallowCascadesstructure:COCORP resultsfrom westem Oregon, Geol. Soc. Am. Bull., 101,783-794, 1989. Kirby, S. H., Rheology of the lithosphere,Rev. Geophys.,21, 14581487, 1983. Kulm, L. D., R. von Huene,et al., Sites174 and 175, Initial. Rep. Deep Sea Drill. Proj., 18, 7-212, 1973. Kulm, L. D., et al., Oregon subductionzone: venting, fauna, and carbonates,Science, 231, 561-566, 1986. Kurtz, R. D., J. M. DeLaurier, and J. C. Gupta, A magnetotelluric soundingacrossVancouverIsland detectsthe subducting Juande Fuca plate, Nature, 321, 596-599, 1986. Langston,C. A., Corvallis, Oregon,crustaland upper mantle receiver structure from teleseismic P and S waves, Bull. Seismol. Soc. Am., 67, 713-724, 1977. Larsen,J. C., Low frequency(0.1-6.0 cpd) electromagneticstudyof the deep mantle electrical conductivitybeneath the Hawaiian Islands, Geophys.J. R. Astron. Soc.,43, 14-56, 1975. 967-978, 1981. zone decollement horizon, Nature, 326, 785-788, 1987. Mysen, B. O., and I. Kushiro, Compositional variations of coexisting phaseswith degreeof melting of peridotitein the upper mantle, Am. Mineral., 62, 843-865, 1977. Oldenburg,D. W., K. P. Whittall, and R. L. Parker, Inversion of ocean bottom magnetotelluricdata revisited, J. Geophys.Res., 89, 18291833, 1984. Olhoeft, G. R., Electricalpropertiesof rocks,in Physical Properties of Rocks and Minerals, McGraw-Hill/CINDAS Data Ser. on Mater. Properties, vol. II-2, editedby Y. S. Touloukianand C. Y. Ho, pp. 57-297, McGraw-Hill, New York, 1981. Oxburgh,E. R., Heat flow and magma genesis,in Physics of Magmatic Processes, pp. 161-200, edited by R. B. Hargraves, Princeton University Press, Princeton, N.J., 1980. Parker, R. L., Understandinginverse theory, Annu. Rev. Earth Planet. Sci., 5, 35-64, 1977. Priest, G. R., N.M. Woller, G. L. Black, and S. H. Evans, Overview of the geologyof the centralOregon CascadesRange, in Geology and GeothermalResourcesof the Central OregonCascadeRange, edited by G. R. Priest and B. F. Vogt, Oreg. Dep. of Geol. Miner. Ind. Spec. Pap., 14, 2-28, 1983. Priest, G. R., N.M. Woller, and M. L. Ferns, Geologic map of the BreitenbushRiver area, Linn and Marion Counties, Oregon, Map GMS-46, Oreg. Dep. of Geol. Miner. Ind., Portland, 1987. Ranganayaki,R. P., and T. R. Madden, Generalizedthin sheetanalysisin magnetotellurics:An extensionof Price's analysis, Geophys.J. R. Astron. $oc., 60, 445-457, 1980. Rasmussen,J., and G. Humphties, Tomographicimage of the Juan de Fuca plate beneathWashingtonand westernOregonusingteleseismic P-wave travel times, Geophys.Res. Lett., 15, 1417-1420, 1988. Riddihough,R. P., and D. A. Seeman,Gravity anomalymap of the Juan de Fuca plate, Map JFP-8, Can. Dep. of Energy, Mines and Resour., Ottawa, 1982. Riddihough, R. P., C. Finn, and R. Couch, Klamath-Blue Mountain lineament,Oregon, Geology, 14, 528-531, 1986. Roberts,R. G., The deepelectricalstructureof the Earth, Geophys.J. R. Astron. Soc., 85, 683-700, 1986. Sato, H., I. S. Sacks, T. Murase, and C. M. Scarfe, Thermal structure of the low velocity zone derived from laboratory and seismic investigations,Geophys.Res. Lett., 15, 1227-1230, 1988. Shankland,T. J., and M. E. Ander., Electricalconductivity,temperatures, and fluids in the lower crust, J. Geophys. Res., 88, 9475-9484, 1983. Shankland, T. J., R. J. O'Connell, and H. S. Waff, Geophysical constraintson partial melt in the upper mantle, Rev. Geophys.,198, 394-406, 1981. Smith, G. A., and E. M. Taylor, The Central Oregon High Cascades graben:What? Where? When?, Geotherm.Resour. Counc. Trans., 7, 275-278, 1983. Smith, G. A., L. W. Snee, and E. M. Taylor, Stratigraphic, sedimentological,and petrologicalrecordof late Miocene subsidence of the centralOregonHigh Cascades,Geology,15, 389-392, 1987. Snavely, P. D., Jr., Tertiary geologic framework, neotectonics,and petroleum potential of the Oregon-Washingtoncontinental margin, in Geology and ResourcePotential of the Continental Margin of WesternNorth Americaand AdjacentOceanBasins- BeaufortSea to Baja California, Earth Sci. Ser., vol. 6, editedby D. W. Scholl, A. Grantz, and J. G. Vedder, pp. 305-335, Circum-PacificCouncil for Energy and Mineral Resources vol. 6, Houston,Tex., 1987. Stanley,W. D., G. S. Fuis, and W. D. Mooney, Details of crustalstructure in the Cascade Range and surroundingregions from seismic and magnetotelluric data, in Geological, Geophysical and Tectonic Settingof the CascadeRange, U.S. Geol. Surv. Open File Rep. 89178, pp. 31-73, editedby L. J.P. Muffler, C. S. Weaver, and D. D. Blackwell, 1989. Tyburczy,J. A., and H. S. Waft, Electricalconductivityof molten basalt and andesiteto 25 kilobars pressure:Geophysicalsignificanceand 14,144 WANNAMAKER ET AL.: RESIS• CROSS SECTION, JUANDE FUCASYSTEM implications for charge transport and melt structure:J. Geophys. Res., 88, 2413-2430, 1983. Verplanck, E. P., and R. A. Duncan, Temporal variations in plate convergenceand eruption rates in the Westem Cascades,Oregon, Tectonics, 6, 197-209, 1987. yon Bargen,N., and H. S. Waff, 1986, Permeabilities,interfacialareas, and curvaturesof partially molten systems:Results of numerical computationsof equilibrium microstructures,J. Geophys.Res., 91, 9261-9276, Young, P. D., and C. S. Cox, Electromagnetic active sourcesounding near the East Pacific Rise, Geophys. Res. Lett., 8, 1043-1046, 1981. Young,C. T., andM. R. Kitchen,A magnetotelluric profile in the Coast Range of Oregon,J. Geophys.Res., this issue. Young, C. T., J. R. Booker,R. Femandez,G. R. Jiracek,M. Martinez, J. C. Rodgers, J. A. Stodt, H. S. Waff, and P. E. Wannamaker, Verification of five magnetotelluric systems in themini-EMSLAB experiment, Geophysics,53, 553-557, 1988. 1986. Waft, H. S., J. T. Rygh, D. W. Livelybrooks,and W. W. Clingman, Resultsof a magnetotellurictraverseacrosswestem Oregon:Crustal resistivitystructureand the structureof the Juande Fuca plate, Earth Planet. Sci. Lett., 87, 313-324, 1988. Wannamaker, P. E., Resistivity structure of the Northern Basin and Range, in The Role of Heat in the Developmentof Energy and Mineral Resources in the Northern Basin and Range Province, Geotherm.Resour. Counc. Spec.Rep. 13, 345-362, 1983. Wannamaker, P. E., Electrical conductivity of water-undersaturated crustalmelting, J. Geophys.Res., 91, 6321-6327, 1986. Wannamaker,P. E., On thin-layer telluric modelingof magnetotelluric responses,Geophysics, in press, 1989. Wannamaker, P. E., G. W. Hohmann, and S. H. Ward, Magnetotelluric responsesof three-dimensional bodiesin layered earths,Geophysics, 49, 1517-1533, 1984. Wannamaker, P. E., J. A. Stodt, and L. Rijo, Two-dimensional topographic responsesin magnetotellurics modeled using finite elements, Geophysics,51, 2131-2144, 1986. Wannamaker,P. E., J. A. Stodt, and L. Rijo, A stable finite element solutionfor two-dimensionalmagnetotelluricmodeling,Geophys.J. R. Astron. Soc., 88, 277-296, 1987. Wannamaker, et al., Magnetotelluric observationsacrossthe Juan de Fuca subductionsystemin the EMSLAB project, J. Geophys.Res., this issue. Wells, F. G., and D. L. Peck, Geologicmap of Oregonwest of the 121st meridian, U.S. Geol. Surv. Map 1-325, 1964. Wyllie, P. J., Magmas and volatile components,Am. Mineral., 64, 469500, 1979. Wyllie, P. J., Magma genesis, plate tectonics, and chemical differentiationof the earth, Rev. Geophys.,26, 370-404, 1988. J. R. Booker, GeophysicsDept., AK-50, Universityof Washington,Seattle,WA 98195. A.D. Chave, AT&T Bell Laboratories,Rm 1E-444, 600 Mountain Ave., Murray Hill, NJ 07974. J. H. Filloux, ScrippsInstitutionof Oceanography, MS A030, University of California at San Diego, La Jolla, CA 92093. A. G. Jones,LithosphericGeophysics,GeologicalSurvey of Canada,1 ObservatoryCrescent,Ottawa, Ontario,Canada, K1A 0Y3. L. K. Law, Pacific GeoscienceCenter,GeologicalSurveyof Canada,P.O. Box 6000, Sidney,British Columbia,Canada V8L 4BZ. H. S. Waft, Departmentof Geology,Universityof Oregon, EUgene,OR 97403. P. E. Wannamaker,Universityof Utah ResearchInstitute, 391-C ChipetaWay, Salt Lake City, UT 84108. (Received October 28, 1988; revised March 29, 1989; acceptedApril 3, 1989.) WANNAMAKER ET AL.: RESISTIVITY CROSSSEC•ON,JUANDE FUCASYSTEM NB CB 100 50 CR c WB 50 WCHC 100 150 14,277 DB 200 km , 25 50 l:l 1,:50 200 1:4 300 300 p <10 3 >103 m) Plate 1. [Wannamakeret al.]. East-westresistivitycrosssectionderivedfrom forward modeling of MT data along the Lincoln Line. Section preservesdetails of the finite elementmesh geometrybut resistivitieshave been groupedin halfdecade intervals for color display. Seafloor portion of the model is shown only to 175 km offshore, beyond which variationsare minor. Note changesin vecdcalexaggeration,labelled on the left, at 5 km and 150 km depth. Important physiographic regionscrossedincludethe CascadiaBasin (CB), NewportBasin (NB), CoastRange(CR), Willameue Basin (WB), WesternCascades(WC), High Cascades(HC), andthe DeschutesBasin(DB). km 14,278 W•,•LaJ•R ET AL: Rm•s• CROSS Sr.CT•ON, JUA• DE FUCASYSTma COMPUTED CR ' i ! I i i--iii WB Ii i iii ! i i i WC i i I i!1 ii HC ! ili iii!1 i DB I ! I -2 -1 o +1 Pyx +2 +3 +4 ___1 I I I I III .11 I I1.1,,, I I I I ....1. I I...1.11 _. I I I II i I..1111,, .1, ,I 1ooo i._1 I I I I"11iII I' I'-('i"i'! I"1"1'"i! I'"i'i'"'l • I I'1f"i"i'• I"I•Ai:I• 1•'•'•I = -2 (deg) - ß -I 7.5 - 65 qbyx +1 +2 +3 +4•, -15 I III .lJ I 111 I I I I .,1 I .1.. I 11. I I _.1.II I I II!.1 I I O 50 ............... 1. I I, I__ I 100 km W E Plate2. [Wannamaker etal. ]. Pseudosections oftransverse magnetic apparent resistivity Pyxandimpedance phase computedalongthe landwardportionof the Lincoln Line. WANNAMAKER ET AL: RESISTIVITY CROSSSECrIoN,JUANDE FUCASYSTEM 14,279 COMPUTED SF8 S F7 SF5 4 3 2 1 (D, -m} +2 *3 p yx ,4 -1ooo I ....................... (deg) i ............ i ...... I ........i ......i----l - 7• +3 .... :;:7 - 4 5 !.... "*';'"•:" - :35 *4' -2.5 -15 *5 Ir 0 W Plate3. ..... I I00 ....... 1 200 km E [Wannamaker et al.] . Pseudosections of transverse magnetic apparent resistivity Pyxandimpedance phase computed alongthe seafloorportionof the LincolnLine. Note differences in horizontalscaleandperiodrangebetweenland and seafloor results. 14,280 WA.NNAMAi(•R ET AL.: RBSISTIVITY CROSS SECTION, JUANDE FUCASYSTEM COMPUTED CR WB WC HC DB (•-m) -2 - 1.o -1 o 1o +I 1oo +2 Pxy +3 +4 ___1 I I I I 1 I I I Ill I !_1 1.11 I I [__1.... 1_.1 I 1.1.1 .1 I I I! !.l 111 I I Illl iii I I"1 I'1'1¾'11 1 i 1! I 11'1 I"1 i l I I i I'! I 1.... i. I 1ooo ._ I (deg) -2 - 7.5 . -1 - 65 55 +1 xy 35 +2 '1 +3 +4 - 25 -15 .....1_1 I I J 0 1.....111 I I I 111 I I I I_ I 1....I. ' " 1...... .50 !.1.1 I I I I1 !._1.1111 l.... i I_ :l...... 1 I00 km W E Plate4. [Wannamaker et al.].Pseudosections of transverse electric apparent resistivity Px•andimpedance phase computedalong the landwardportion of the Lincoln Line. WANNAMAKER ET At,.: RF,S•S• CROSSS•rION, JUANDE FUCAS¾s'r• 14,281 COMPUTED SF8 I SF7 SF54 3 21 (g.m) I -1o p xy -IOOO ß.I .....I........I ! ...... I ............ I *2 i 1__ (deg) 1....... • -7.5 *3 (:/:)xy "4 -25 *5 i i00 w Plate5. _1 1. i L_ -15 200km E [Wannarnaker et al.]. Pseudosections of transverse electricapparent resistivityPx•andimpedance phase•xy computedalongthe seafloorportionof the LincolnLine. Note differences in horizontalscaleandperiodrangebetweenland and seafloor results. 14,282 WANNAMAKER ET AL.: RESISTIVr•CROSSSECTION, JUANDE FUCASYSTEM COMPUTED CR " WB WC HC DB I ! I• I I'' II1!"1I ................. i"'1 i I I' I i '1 '1'I- '1I'i i""! ...... I I I I !'•II '1• .......... I I I'1 (%) -2 -1 - 60 - 4O o +1 +2 +3 +4 - -60 ___l I I I I III I I I i11. I I I 1.....I I I 111 I 1_ I 1.1...1 1.11.11 .l.. J _1 j,, = I I I--I I' I i'J'"-!--I '1"1I I I'"'i'"i'"i'1 i I III I! ! I'!-"l-""l' l' l'i I i I I I (%) -2 - 60 - 4o -1 +1 Z'ZE'Z +2 •. +3 '•- -".40 +4 - -60 _ I 0 I I I . I III ! I ! III 50 !. I I I I I I ! I I.I ! I.I1 I I IIII. i ...1.... I_ I 100 km w Plate6. - -20 E [Wannamaker et al.]. Pseudosections of verticalmagnetic fieldtransfer element Mzy,realandimaginary parts, computedalong the landwardportion of the Lincoln Line. At..:RESISTIVrrY CROSS SF. eHON,JUANDEFUCASYSTEM 14,283 COMPUTED SF8 ß . SF7 SF54 321 . ............... I .................. i - 'i •3 +4 -90 +5 ,, +2 I I I i ! I •3 •5 __1 I O I IOO w I ..... I -90 I ' 200 k m E Plate 7. [Wannamaker etal.].Pseudosections of vertical magnetic fieldtransfer element MR, realandimaginary parts, computed alongthe seafloorportionof the LincolnLine. Notedifferences in horizontal scale,periodrange,andcontour imerval between land and seafloor results.