Melt distribution beneath a young continental rift: The Taupo Volcanic Zone, New Zealand

Melt distribution beneath a young continental rift: The Taupo Volcanic Zone, New Zealand
of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Melt distribution beneath a young continental rift: The TaupoVolcanic Zone, New Zealand Wiebke Heise, 1 Hugh M. Bibby, 1 T. Grant Caldwell, 1 Stephen C. Bannister, 1 Yasuo Ogawa, 2 Shinichi Takakura, 3 and Toshihiro Uchida 3 Received 8 February 2007; revised 13 June 2007; accepted 2 July 2007; published 25 July 2007. [ 1 ] Taupo Volcanic Zone (TVZ) is a zone of intensevolcanism and rifting associated with the subduction of thePacific Plate beneath the continental crust of New Zealand’s North Island. An image of the conductivity structure beneath the central part of the TVZ has been constructedusing 2-D inverse modeling of magnetotelluric data. A rapidincrease in conductivity at a depth of 10 km beneath theTVZ,  3 km beneath the base of the seismogenic zone but well above the base of the quartzo-feldspathic crust (  16 km), is interpreted to mark the presence of aninterconnected melt fraction (<4%) within the lower crust.Beneath the quartzo-feldspathic crust the model shows azone of increased conductivity on the eastern side of theTVZ consistent with an increased concentration of melt. At deeper levels the Pacific Plate is resistive compared with theoverlying mantle.  Citation:  Heise, W., H. M. Bibby, T. G.Caldwell, S. C. Bannister, Y. Ogawa, S. Takakura, and T. Uchida(2007), Melt distribution beneath a young continental rift: TheTaupo Volcanic Zone, New Zealand,  Geophys. Res. Lett. ,  34 ,L14313, doi:10.1029/2007GL029629. 1. Introduction [ 2 ] New Zealand’s Taupo Volcanic Zone (TVZ) is aregion of intense volcanism and rifting associated with thesubduction of the Pacific Plate beneath the continental crust of the North Island (Figure 1). The central section of theTVZ is the world’s most productive rhyolitic system [ Wilsonet al. , 1995], with a mean eruption rate of 0.3 m 3 /s over thelast 1.6 Ma. This part of the TVZ is also characterized by anextremely high natural heat output (4200 ± 500 MW). Theheat is transported convectively and is discharged via23 high-temperature hydrothermal systems [  Bibby et al. ,1995] (Figure 1). The output is equivalent to the heat obtained from cooling   2 m 3 /s of magma from its solidus(  600  C) to a point where a quartzo-feldspathic rock  becomes brittle (  350  C). This suggests a much greater amount of magma is cooled within the crust to supply thegeothermal heat flux than is (on average) erupted at thesurface. Despite this evidence for the presence of magma,little is known about the deeper magmatic system itself. In particular, there is little geophysical evidence as to wherethe magma resides in the crust.[ 3 ] Electrical exploration methods within the TVZ have proved to be highly effective for delineating the near surfaceextent of the liquid-dominated high-temperature hydrother-mal systems [  Bibby , 1988] and have been used for investi-gating the caldera-related structures that characterize thetectonics of the upper few kilometers of the crust [e.g.,  Risk et al. , 1999]. Widely spaced magnetotelluric (MT) datafrom the TVZ [ Ogawa et al. , 1999;  Ingham , 2005], dem-onstrated that conductivities in the lower crust are anoma-lously high. However, the measurement spacing limits theresolution of these surveys. In this paper we report theresults from an MT profile (Figure 1) across the TVZconsisting of 28 broad-band (0.003-2000 s period) and3 long period (10-10000 s) soundings. 2. MT Data [ 4 ] MT information on the subsurface conductivity struc-ture is contained in the spatial and period dependence of the‘impedance tensor’ ( Z ) and ‘induction vector’ ( K  ) defined by (linear) frequency-domain relationships between thesurface horizontal magnetic field components ( H ) and theelectric field vector ( E ) and the vertical magnetic fieldcomponent (H z ), respectively. The phase informationcontained in the impedance is a tensor defined by the matrixequation  F  =  X  1 Y  where  X  and  Y  are the real andimaginary parts of   Z  [ Caldwell et al. , 2004]. The sensitivityof the MT response to variations in the conductivity at depthis contained in F ; the sensitivity depth being determined bythe overlying resistivity and period. As the period increases,the phase becomes more sensitive to structure at greater depth and less sensitive to structure at shallow depth. Incontrast, the effect produced by the near surface conductiv-ity structure on the amplitude response or ‘apparent resis-tivity’ is present at all periods. While the phase response issensitive to changes in conductivity, information on theapparent resistivity is required before the geometry of theconductivity structure can be determined. In this paper, wewill use the phase tensor to visualize the data and2-dimensional (2-D) inverse modeling [  Rodi and Mackie ,2001] of both the apparent resistivity and phase to deter-mine the conductivity structure.[ 5 ] The phase tensor is represented graphically as anellipse, the major and minor axes ( F max ,  F min ) showingthe maximum and minimum phase difference (coordinateinvariants) between the magnetic and electric fields. If theconductivity structure is 3-D, a third coordinate invariant  parameter, the skew angle  b  , is needed to represent theasymmetry in the phase response. This parameter provides acoordinate invariant measure of the significance of 3-Deffects in the MT phase response. Values of   b   (not shown) GEOPHYSICAL RESEARCH LETTERS, VOL. 34, L14313, doi:10.1029/2007GL029629, 2007 1 GNS Science, Lower Hutt, New Zealand. 2 Volcanic Fluid Research Center, Tokyo Institute of Technology, Tokyo,Japan. 3 Geological Survey of Japan, National Institute of Advanced IndustrialScience and Technology, Tsukuba, Japan.Copyright 2007 by the American Geophysical Union.0094-8276/07/2007GL029629 L14313  1 of 6  are generally low (<3  ) for periods <30 s but increase to  6   at long period (  350 s) in the TVZ.[ 6 ] Figure 2a shows the phase tensor ellipses for profile 1as a pseudo-section. The orientation of the ellipse axesindicates the direction of the steepest conductivity gradient (or its normal) in a depth range determined by the periodand overlying resistivity structure. The value of the min-imum phase, shown by the color filling the ellipses,indicates whether the conductivity is increasing or decreas-ing with depth; minimum phases >45   indicating increasingconductivity.[ 7 ] At short periods in Figure 2a, high phase values( F min  > 60  ) within the TVZ are a consequence of thedecrease in resistivity with depth produced by low temper-ature alteration of the rhyolitic volcanics [  Bibby et al. , 1998]that in fill the rift basins and collapse calderas that charac-terize the near surface structure of the TVZ. At the TVZ’ssouth-eastern (SE) margin (i.e. the SE extent of rifting andassociated down-faulting of the basement greywacke) the phase-tensor ellipses change orientation by 90  . Comparedwith ellipses outside the TVZ this change is ‘delayed’ by theoverlying conductive volcanics.[ 8 ] At the western margin of the TVZ, a similar 90  change in orientation occurs. However, the major axes of the tensor ellipses at longer periods tend to become alignedE-W, indicating that significant 3-D effects are present [ Caldwell et al. , 2004]. In the centre of the TVZ,  F min values reach a peak at periods of    30 s (Figure 2a),indicating that the lower crust is conductive beneath thisregion. 3. Results of 2-D Modeling [ 9 ] Modeling of the ocean shows that 3-D effects pro-duced by the conductive seawater surrounding North Islandare insignificant for periods <350 s. These calculations, thesmall  b   values and the nearly 2-D behavior of the phasetensors (Figure 2a) and induction vectors (Figure 1) at themargins of the TVZ suggest that although 3-D effects are present in the data, a 2-D approach to the analysis is justified.[ 10 ] Results of 2-D inverse modeling [  Rodi and Mackie ,2001] using data from 31 soundings for periods <350 salong profile 1 are shown in Figure 3 with the correspondingcalculated phase tensors shown in Figure 2b. The modelassumes that the strike of the conductivity structure is N45  E, consistent with the phase tensor analysis and other geophysical data [ Ogawa et al. , 1999;  Bibby et al. , 1998].Data used for the modeling were the impedance tensor  phases and apparent resistivities rotated to N45  E and theinduction arrows projected onto the profile direction(N135  E). The starting model used for the inversion con-sisted of a 100  W m half space with conductors (0.3  W m) at each side of the model to represent the effect of the ocean(not shown in Figure 3). In addition, the conductive rocks of the accretionary prism in the SE were modeled by a 8  W m body (Figure 3).[ 11 ] Prior to inversion, the effects of galvanic distortionwere removed using the method of   Bibby et al.  [2005]. At asmall number of locations, where the MT apparent resistiv-ity was significantly different from that at adjacent mea-surement sites and from the mapped DC resistivity value(Figure 1), the apparent resistivity has been shifted toremove the discordance. For depths down to   0.5 km themodel agrees well with the DC apparent resistivity data(Figure 1) suggesting that residual galvanic distortioneffects in the model are small.[ 12 ] In a 2-D earth, MT data separate into two indepen-dent modes, TM and TE, corresponding to electric current flow perpendicular and parallel to the strike directionrespectively. These modes have different sensitivities tooff profile (3-D) structure. Since the TE mode is moresensitive to such structure,  Wannamaker et al.  [2002]advocated down-weighting the TE mode data in 2-D inver-sions to avoid introducing artifacts into the model.[ 13 ] The approach taken here has been to down-weight the TE apparent resistivity by doubling the acceptable error floor compared with the value (5%) used for the TM dataand TE phase data, slightly greater that the mean error of thedata (  4%). Inversions with and without the TE apparent resistivity and phase included give similar conductivitystructures and suggest that 3-D effects are small. However,the error floor chosen for the induction vectors had aconsiderable influence on the modeling results. As theinduction vectors are strongly influenced by 3-D structure,a large error floor (0.08, absolute value) was chosen toavoid over-fitting the induction vectors.[ 14 ] A large number of models were calculated usingdifferent error floors for apparent resistivity, phase andinduction vector data. Models were also calculated for  Figure 1.  Locations and tectonic setting of the TVZ.Colored background shows DC apparent resistivitiesobtained from Schlumberger array measurements after   Bibby  [1988]. Conductive areas (<30  W m), shown in red,mark the geothermal systems. Dashed lines mark the outlineof the TVZ; the red line shows the profile analyzed in this paper. The locations of the MT sites are shown by circles.The arrows show the real part of the induction vectors at 42.7 s. With the convention used in this paper, the inductionvectors point in the direction of increasing conductance. Note the large induction vectors at both margins of the TVZ point inwards, consistent with higher conductance withinthe TVZ. L14313  HEISE ET AL.: MELT DISTRIBUTION OF TAUPO VOLCANIC ZONE  L14313 2 of 6  Figure 2.  (a) Phase tensor ellipse data for profile 1. The ellipse sizes have been normalized by their major axis ( F max ). Theellipse color shows F min . The area of minimum phase >30   marked by the grey area indicates the part of the phase response produced by the deep conductor beneath the central part of the TVZ. (b) Phase tensor ellipses calculated using the modelshown in Figure 3. (c) Tensor misfit  D calculated from the observed and calculated phase tensors shown in Figures 2a and2b. The color used to fill the ellipses shows the arithmetic mean of the maximum and minimum misfit: ( j D max j + j D min j )/2;small ellipses imply that the misfit is small. Note the departure from the 2-dimensionality at the northwestern end of the profile (sites 212, 600, 575) at long periods which is shown by the systematic NW-SE orientation of the misfit tensor major axes. L14313  HEISE ET AL.: MELT DISTRIBUTION OF TAUPO VOLCANIC ZONE  L14313 3 of 6  different values of the regularization parameter   t  , whichcontrols the relation between model roughness and misfit.These trials showed that the main features of the model(Figure 3) are robust to changes in the inversion parameters.The actual inversion parameters used are specified in thecaption of the Figure 3.[ 15 ] The model shown in Figure 3 has a normalized RMSerror of 2.64. An alternative method of assessing the misfit,which allows its spatial dependence to be assessed, is tocalculate a tensor misfit for the phase data (Figure 2c) wherethe misfit tensor used here is given by: D ¼ I     1obs  mod þ  mod   1obs   = 2 ; where I is the identity matrix and  F mod  and  F obs  are thecalculated (Figure 2b) and observed (Figure 2a) phasetensors, respectively. This method of visualizing the misfit has the advantage that it includes all components of the phase data, including the off diagonal phases not included inthe inversion, and depicts the departure of the 2-D modelresponses due to 3-D structure. The magnitude of the misfit is indicated by the size of the ellipse while the ellipseorientation indicates the magnetic field polarization direc-tion in which the maximum phase error occurs. If theorientation of the misfit ellipses is spatially coherent then asystematic mismatch of the regional conductivity structureis implied; e.g. the orientation of the conductivity structureassumed is incorrect. This effect can be seen in the north-western (NW) area of the model (Figure 2c). Apart from thisarea, the phase misfit ellipses are generally small and showno preferred orientation, suggesting that the 2-D model hasotherwise captured the main features of the data.[ 16 ] MT data resolves the total conductance rather thanthe resistivity. Hypothesis tests were carried out in order totest the sensitivity of the data to the location and resistivityof the conductor below 10 km depth (Figure 3). For example, a model with the conductor constrained to liewithin a block of 10  W m at 10–16 km depth beneath theTVZ gives a similar RMS misfit as the inversion model inFigure 3. These tests also show that removing the conduc-tive zone resulted in a significantly worse fit to the data. In particular, a conductor at depth   10 km is necessary to Figure 3.  Depth section along profile 1 (Figure 1) showing the 2-D inversion model obtained using the code of   Rodi and  Mackie  [2001]. Earthquake hypocenters (white circles) mark the slab of the Pacific Plate. Precisely located earthquakehypocenters (white crosses) using the double-difference tomography algorithm (tomoDD) of   Zhang and Thurber   [2003]show the depth of the seismogenic zone beneath the TVZ. The model was calculated minimizing 2nd derivatives usingequal grid Laplacian, and a regularization parameter   t   = 5. Note the rapid change in resistivity that occurs beneath the TVZat 10 km,  3 km beneath the bottom of the seismogenic zone and the high resistivity zone in the SW that correlates with thetop of the Pacific Plate. L14313  HEISE ET AL.: MELT DISTRIBUTION OF TAUPO VOLCANIC ZONE  L14313 4 of 6   produce the peak in  F min  observed at about 30 s period inthe central part of the TVZ. The concentration of conduc-tance in the east of the conductor represented in Figure 3 bythe red spot centered   20 km beneath the eastern TVZ, isalso required by the data. However, the depth extent of theconductor below its top is not well constrained. 4. Discussion [ 17 ] Except for a narrow region along the eastern marginof the TVZ, a resistive layer (resistivities 500–1500  W m)underlies a surficial cover made up of young resistivevolcanics and older conductive volcaniclastics [  Bibby et al. , 1998;  Risk et al. , 1999]. This resistive layer thins from  20 km beneath the Kaingaroa Plateau to  10 km beneaththe TVZ, thickening again to the west. East and west of theTVZ and along the SE margin of the TVZ, this resistivelayer can be identified as greywacke basement [  Bibby et al. ,1995]. Although the basement appears to be continuous beneath the TVZ, along the SE margin, where approximate-ly 50% of the TVZ’s total geothermal output is discharged[  Bibby et al. , 1995] it is significantly more conductive.The combination of high geothermal heat and gas flux[ Giggenbach , 1995] appears to have reduced the resistivityof the greywacke; most plausibly by altering the mineralogyof the rock.[ 18 ] The resistivity beneath the TVZ drops very rapidly at a depth of 10 km; about 3 km beneath the seismogenic zone(Figure 3).  Bibby et al.  [1995] interpreted the bottom of theseismogenic zone as marking the brittle-ductile transitionand the lower extent (6–7 km depth) of the TVZ’s convec-tive geothermal systems [  Bryan et al. , 1999;  Fournier  ,1999]. Within the convective regime the pressure regimeis dominantly hydrostatic and heat is transported by con-vection of water that is discharged through the geothermalfields. Further,  Bibby et al.  [1995] argued that once therocks become ductile and the pressure regime lithostatic, it would not longer be possible to maintain the permeable paths necessary for convective flow and the dominant mechanism of heat transport would become conductiveaugmented by sporadic magma advection. To maintain theTVZ’s 700 mW/m 2 average heat flux by conduction alonerequires a temperature gradient in excess of 200  C/km. Theonset of ductile conditions in quartz rich rocks occurs at about 350  C [  Hill  , 1992;  Smith and Braile , 1994].Thus,temperatures would be expected to exceed those inferred for TVZ magmas; 730  C, [  Nairn et al. , 2004] and 820–850  C,[ Sutton et al. , 2000] within 3 km of the brittle-ductiletransition. Thus the decrease in resistivity at 10 km isconsistent with the presence of a connected melt fractionas suggested by  Bibby et al.  [1995]. A melt fraction of <4%is sufficient to explain the modeled low resistivities assum-ing interconnectivity of the melt [ Schilling et al. , 1997]. The presence of a melt fraction is supported by the seismicreceiver function analysis of   Bannister et al.  [2004], whointerpreted a low S-wave velocity layer at 10 km depth to becaused by the presence of melt.[ 19 ] At a depth of about 16 km seismic reflection studiesshow a change in P-wave velocity inferred to be the base of the quartzo-feldspathic crust. Beneath this, underplating is believed to occur, although its depth extent is controversial[ Stratford and Stern , 2006;  Harrison and White , 2006]. Theresistivity model of Figure 3 suggests that the low resistivityzone extends to depths as great as 30 km, although it must  be emphasized that the MT modeling is insensitive to thevertical extent of the conductive zone. Model tests showthat the conductive zone must extend below the velocitychange at 16 km for reasonable values of partial melt conductivity. The zone of greatest conductivity in our modeloccurs in the region where underplating is occurring. This isconsistent with greater proportions of melt expected at thesedepths [ Charlier et al. , 2005].[ 20 ] AtdeeperlevelsintheSE,thetopofanorth-westwarddipping high resistivity layer in Figure 3 correlates wellwith the Wadati-Benioff Zone marked by earthquake hypo-centers [  Reyners et al. , 2006]. This correlation suggests that the model resistivities in the mantle wedge and crust abovethe subducted plate provide a good estimate of the actualresistivities present. Thus the mantle wedge beneath theTVZ appears to be anomalously conductive compared withPacific Plate Lithosphere beneath New Zealand’s SouthIsland [ Wannamaker et al. , 2002], suggesting the presenceof a partially interconnected melt fraction in the mantlewedge above the plate as suggested by seismological data[  Reyners et al. , 2006]. 5. Conclusions [ 21 ] The increased resolution of the 2-D conductivitymodel presented here compared with the earlier model of  Ogawa et al.  [1999] provides a geologically plausible picture of a zone of melt accumulation beneath the thinnedcrust of the TVZ and partially interconnected melt in themantle wedge above the subducted plate. The depth to thetop of the melt zone in the crust beneath the TVZ, about 3 km beneath the base of the seismogenic zone, is consistent with the model of heat transport presented by  Bibby et al. [1995] for the hydrothermal systems. The melt fraction or melt interconnectivity is greater along the SE margin of theTVZ where concentration of geothermal flux is greatest.[ 22 ]  Acknowledgments.  We thank J.R. Booker and C.J.N. Wilson for their reviews and Stewart Bennie, Nobuo Matsushima, Yuji Nishi andToshiyuki Tosha for assistance in data collection. This project was fundedthe New Zealand Foundation for Research Science and Technology and theJapanese Science and Technology Agency. References Bannister, S. C., C. J. Bryan, and H. M. Bibby (2004), Shear wave velocityvariation across the Taupo Volcanic Zone, New Zealand, from receiver function inversion,  Geophys. J. Int. ,  159 , 291–310.Bibby, H. M. (1988), Electrical resistivity mapping in the central volcanicregion of New Zealand,  N. Z. J. Geol. Geophys. ,  31 , 259–274.Bibby, H. M., T. G. Caldwell, F. J. Davey, and T. H. Webb (1995),Geophysical evidence on the structure of the Taupo Volcanic Zone andits hydrothermal circulation,  J. Volcanol. Geotherm. Res. ,  68 , 29–58.Bibby, H. M., T. G. Caldwell, and G. F. Risk (1998), Electrical resistivityimage of the upper crust within the Taupo Volcanic Zone, New Zealand,  J. Geophys. Res. ,  103 , 9665–9680.Bibby, H. M., T. G. Caldwell, and C. Brown (2005), Determinable andnon-determinable parameters of galvanic distortion in magnetotellurics, Geophys. J. Int. ,  163 , 915–930.Bryan, C. J., S. Sherburn, H. M. Bibby, S. C. Bannister, and A. W. Hurst (1999), Shallow seismicity of the central Taupo Volcanic Zone, NewZealand: Its distribution and nature,  N. Z. J. Geol. Geophys. ,  42 , 533– 542.Caldwell, T. G., H. M. Bibby, and C. Brown (2004), The magnetotelluric phase tensor,  Geophys. J. Int. ,  158 , 457–469.Charlier, B. L. A., C. J. N. Wilson, J. B. Lowenstern, S. Blake, P. W. vanCalsteren, and J. P. Davidson (2005), Magma generation at a large, L14313  HEISE ET AL.: MELT DISTRIBUTION OF TAUPO VOLCANIC ZONE  L14313 5 of 6
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks