P-wave velocity structure of the crust and uppermost mantle beneath Iceland from local earthquake tomography

Transcription

1 Earth and Planetary Science Letters 235 (2005) P-wave velocity structure of the crust and uppermost mantle beneath Iceland from local earthquake tomography Ting Yang*, Yang Shen Graduate School of Oceanography, University of Rhode Island, South Ferry Road, Narragansett, RI 02882, USA Received 31 December 2004; received in revised form 25 April 2005; accepted 2 May 2005 Available online 14 June 2005 Editor: S. King Abstract The structure of the crust and uppermost mantle beneath Iceland, the keys to understanding the magma plumbing system of the hotspot and hotspot ridge interaction, was poorly constrained in previous seismological investigations. Here we develop a three-dimensional P-wave velocity model of the Icelandic crust and uppermost mantle from tomographic inversion of over 3500 first-arrivals from local earthquakes recorded in Iceland. The model shows a broad low-velocity anomaly in the middle and lower crust underlying a high velocity body in the shallow crust in central Iceland. With seismic rays traversing below the crust, the inversion also reveals a pronounced P-wave velocity reduction, about 5% or times that in the km depth range imaged by teleseismic tomography, in the uppermost mantle beneath central Iceland. The large velocity reduction requires an excess temperature of up to 500 degrees or, more likely, a combination of excess temperature and partial melt. The localized nature of the region of low velocity beneath central Iceland and the lack of comparable velocity reduction along the volcanic zones suggest a relatively focused melt supply of the hotspot. D 2005 Elsevier B.V. All rights reserved. Keywords: Iceland; crust; uppermost mantle; local earthquake tomography; hotspot; mantle plume; hotspot ridge interaction 1. Introduction * Corresponding author. Tel.: ; fax: address: tyang@gso.uri.edu (T. Yang). The Iceland hotspot has been a site of extensive seismological investigations. Tomographic inversions of travel times of teleseismic body waves recorded by the seismic arrays in the ICEMELT and HOTSPOT experiments (Fig. 1) have imaged a columnar lowvelocity anomaly in the upper mantle beneath the hotspot [1 5]. These studies, however, did not resolve the crustal and uppermost mantle structures above ~100 km depth because of the lack of crossing teleseismic ray paths in that depth range [6,7]. Independent seismic constraints from wide-angle reflection and refraction [8 12], surface wave studies [13,14] and receiver function analysis [15,16] provided evidence of a crust up to km in thickness beneath Iceland, though controversies remain. The shear velocity crustal model constructed from surface X/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi: /j.epsl

2 598 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) o -22 o -20 o -18 o -14 o 67 o 67 o TFZ o Ak 18 Snaef MNZ 19 2 B 65 o 1 28 WNZ o 64 SISZ Hk o RR WFj RP ENZ Gr Kr Ba NNZ Ka 100 km 63 o 63 o -24 o -22 o -20 o -18 o -14 o Fig. 1. Map of Iceland, showing the major geological features (RR: Reykjanes Ridge; RP: Reykjanes Peninsula; SISZ: Southern Iceland Seismic Zone; WNZ: Western Neovolcanic Zone; ENZ: Eastern Neovolcanic Zone; MNZ: Middle Neovolcanic Zone; NNZ: Northern Neovolcanic Zone; TFZ: Tjornes Fracture Zone; Snaef: Snæfellsnes; WFj: West Fjords ), volcanoes (Ka: Katla; Hk: Hekla; Gr: Grímsfjall; Ba: Bárðarbunga; Ak: Askja; Kr: Krafla) and glaciers in Iceland, stations of the HOTSPOT (solid circles with number) and ICEMELT (large triangles) experiments, ICEMELT profile s shots (stars) and stations (small triangles), and earthquakes used in this study (small solid circles). waves [13] shows a broad low-velocity zone in the upper crust (0 15 km depth) along the Icelandic neovolcanic zones, and a low velocity anomaly confined to a circular region in the lower crust (N15 km depth) beneath central Iceland. These features are suggestive of broad regions of high temperature and perhaps the presence of partial melt in the crust and lateral transport of magma along the volcanic zones from the presumed plume center. However, an extensive presence of partial melt in the crust contradicts several other lines of evidence suggesting that the entire Icelandic crust is subsolidus except for small volumes beneath central volcanoes [12,17 19]. In an effort to estimate the seismic velocity of the uppermost mantle, Allen et al. [4] removed the crustal signal from teleseismic body wave travel times using the crustal model constructed from surface waves and found a relatively high velocity anomaly beneath central Iceland extending from the crust mantle boundary (Moho) to ~100 km depth in their tomographic models. This relatively high velocity anomaly is above the tomographically imaged low-velocity anomaly at greater depth [2 4] and has been interpreted as the effect of a very highly depleted mantle over the plume conduit compared to the surrounding mantle [4]. Since this high relative-velocity feature is not present in other tomographic models without direct a priori corrections for crustal heterogeneity [2,3,5], it depends strongly on the correction for teleseismic travel time delays caused by crustal velocity anomalies. Because corrections for teleseismic P waves are determined from the S-velocity crustal model using a simple linear V p /V s scaling [13], the high P-and S-velocity anomalies in the shallow mantle are not independent. This limits our ability to assess this high relative-velocity feature in the shallow upper mantle and to understand its cause. Resolving the P-wave crust and uppermost mantle structure is thus important to understanding the mechanisms of magma transfer at the hotspot and hotspot ridge interaction.

3 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) Several studies used active seismicity in Iceland to obtain high-resolution, three-dimensional (3D) tomographic images of the Iceland crust, but resolutions were limited to small areas and in the upper and middle crust due to limited lateral station coverage in those studies [20 22]. In this paper, we develop a threedimensional P-wave velocity model of the Icelandic crust and uppermost mantle from tomographic inversions of over 3500 first-arrival P waves ( P g /P n ) from local earthquakes recorded by the ICEMELT and HOTSPOT stations in Iceland. With short wavelengths, body waves from local earthquakes offer potentially higher resolutions in the crust and uppermost mantle than surface waves and teleseismic body-wave arrivals with steep incidence angles in shallow depth. 2. Data and method We selected 200 earthquakes with magnitude greater than 2.5 from local earthquakes during the period of the HOTSPOT experiment [13]. We also included 13 earthquakes (m b N3.7) recorded by the ICEMELT stations [1,2] and the active-source ICE- MELT profile data [12]. The initial earthquake locations and origin times were obtained from the SIL seismic network [23]. Fig. 1 shows the distributions of the events and seismic stations. In order to minimize the uncertainty associated with phase identification in the construction of this preliminary P-wave model, we used only the clearly identifiable, firstarrival phase ( P g or P n depending on the epicentral distance) and picked the arrival time by hand. The uncertainty of arrival pickings is estimated to be less than 0.3 s. We obtained a total number of 3521 arrival times. Fig. 2 shows a map view of the event-station paths. The notation P n is used in this paper to denote a P arrival having a turning point in the uppermost mantle [24,25]. We note that the observed uppermost mantle P phase does not necessitate a positive mantle velocity gradient. Numerical simulations of wave propagation show that a 0.5 1% velocity perturbation superimposed on a negative velocity gradient is sufficient to -24 o -22 o -20 o -18 o -14 o 67 o 67 o o o B o 64 o o 63 o -24 o -22 o -20 o -18 o -14 o Fig. 2. Map view of the ray path coverage.

4 600 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) km 30 km 50 km Fig. 3. Three-dimensional view of the ray paths of one earthquake. The source is located beneath southwestern Iceland. Rectangles on the surface represent the HOTSPOT stations, and the mesh is the Moho derived from the S-velocity crustal model [13] with point constraints from previous receiver function and refraction studies. 16 Reduced Travel Time (s) Distance (km) Fig. 4. Reduced travel times for arrivals from events located at the TFZ, RP and SISZ. The reduction velocity is 7.85 km/s. Most of these arrivals with epicentral distance greater than 200 km traverse through the central Iceland.

5 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) o -20 o -24 o -20 o -24 o -20 o 64 o 64 o 64 o Depth: 12 Km Depth: 24 Km Uppermost Mantle >60 Hitcount Fig. 5. Maps of hit count (the number of ray paths in each grid cell) at the depth of 12, 24 km, and immediately beneath the Moho. generate P n phases [25]. Because of the large variation in the crustal thickness in Iceland, most P n waves here are the P arrivals turning in the uppermost mantle (Fig. 3). Ray paths that yielded the smallest travel times from source to receiver were associated with the observed first arrivals. Fig. 4 shows the reduced travel times versus distances (the reduction velocity is 7.85 km/s) for the events located at the TFZ, RP and SISZ. Most of the rays with epicentral distance greater than 200 km traverse through central Iceland. The flat distribution of the reduced travel times at distances (a) (b) 8 greater than 175 km indicates that those arrivals have a mantle velocity. An earlier study has also identified clear P n arrivals traveling beneath central Iceland (Fig. 6b in[9]). We used the tomographic inversion method based on Benz et al. [26], in which the velocity structure and source parameters were solved jointly through the technique of parameter separation [27,28]. We adapted the method of Zhao et al. [29,30], which was developed to deal with large 3-D velocity variation and complex discontinuities, for ray tracing and 2.5 Norm of velocity model Traveltime residual (s) Variance of traveltime residuals (s 2 ) Distance (km) Fig. 6. (a) Tradeoff curve between the variance of travel time residuals and the L 2 norm of the velocity model. Each point presents a damping factor used in the inversion. The solid square marks our preferred model. (b) Travel time residuals for all the arrivals before (open circles) and after inversion (solid circles).

6 602 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) travel time calculations. The velocity structure was parameterized by a 3-D grid with a grid spacing of 18 and 3 km in horizontal and vertical directions, respectively. The velocity at any point in the model was determined from a linear interpolation of velocities at surrounding grid nodes. The travel time calculation and ray tracing were performed on a scale finer (1 km) than the node interval. Fig. 5 shows the number of ray paths in each grid cell (hit count) at the depth of 12 km, 24 km, and immediately beneath the Moho. An iterative, damped least squares algorithm [31] was used to perform the inversion. Ray paths were updated and hypocenter parameters were adjusted after each iteration. A P-wave crustal velocity model constructed from the surface-wave derived S- velocity model using a linear V p /V s ratio [13] was used as our initial velocity model. The depth of the Moho was derived from the same S-velocity model with point Moho-depth constraints from previous receiver function studies and seismic refraction data [13]. The starting velocity below the Moho was set to 8.0 km/s, increasing with depth with a small gradient (0.015 km/s/km). The damping factor was determined from the trade-off between the variance of travel time residuals and model roughness [32]. With the chosen damping factor (Fig. 6a), we reduced the variance of travel time residuals by 80.2%, and the root-mean-square (rms) travel time residual of our final model is 0.37 s. Fig. 6b shows the travel time residuals versus epicentral distances before and after the inversion. 3. Tomographic results and resolution The crustal and uppermost mantle velocity structure beneath Iceland from our local earthquake inversion is shown in Fig. 7. In the crust, the P-wave velocity model shows some features similar to those in previous studies, e.g., low velocity anomalies beneath active volcanic regions such as Hekla, Katla in the upper crust (Fig. 7c) [22,33] and beneath central Iceland in the middle and lower crust [13,34], a band of high velocity anomaly between the western fjords and the rest of Iceland [13], and a high velocity dome beneath Northern Volcanic Zone [35]. Our inversion also reveals anomalies not seen in previous studies. Beneath central Iceland, a high velocity body (~0.25 km/s in magnitude) is found in the upper crust above the board low velocity anomaly in the middle and lower crust. In the uppermost mantle, a pronounced low velocity anomaly ( 0.4 km/s or ~5% below the average of the surrounding resolved nodes at the same depths) is found beneath the Moho in central Iceland (Fig. 7d). The location of this low velocity body coincides with the region with the thickest crust and high 3 He / 4 He [36] and above the columnar low-velocity anomaly in the deeper mantle imaged by previous teleseismic tomography [2 5]. Additionally, a large velocity reduction was imaged beneath Snæfellsnes and the region between the western fjords and the rest of Iceland. To assess the resolving ability of our inversion, we conducted a series of resolution tests. Fig. 8 shows the resolution in the crust beneath central Iceland. The input is a high velocity anomaly cylinder (~60 km in radius, from surface to 15 km depth) above a broader low velocity anomaly cylinder (~75 km in radius, extending from 18 km depth to the Moho) beneath central Iceland. The input anomaly can be resolved down to the middle crust, although the resolutions of the very shallow and deep crust (N30 km depth) are relatively poor. Fig. 9 shows the results of three tests to understand the resolution in the uppermost mantle. The input model of the first test is sinusoidal checkerboard velocity perturbations beneath the Moho with a block size of 72 by 72 by 12 km. The second test has a square low-velocity anomaly beneath central Iceland with a width of ~130 km. The input model for the third test is a low-velocity band beneath the neovolcanic zones. The results indicate that our data set is sufficient to detect a low-velocity anomaly in the uppermost mantle with lateral dimensions of the features imaged beneath central and western Iceland. The magnitude and location of the input perturbation are best recovered beneath western Iceland and the resolution degrades east of central Iceland and along the coast. We used the Moho depth in the S-velocity model constructed from surface wave tomography [13] as a prescribed parameter in our inversion. The crust mantle boundary affects ray tracing, and thus the resultant velocity model. In order to evaluate the sensitivity of

7 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) Fig. 7. The P-wave velocity model from local earthquake tomography. Vaules in the cross-sections are deviations in wave speed (km/s) from the average velocity for the crust or the uppermost mantle at each depth, which is shown in (b). Grid cells with hit counts less than 5 were not used in the inversion and the calculation of the average velocity. (a) Vertical cross-sections of the P-wave velocity model through central Iceland; Black lines indicate the Moho in Allen et al. [13]. (c) The velocity anomaly at 12 km depth. (d) Map view of the velocity anomaly beneath the Moho. the inversion results to the Moho depth, we changed the Moho depth systematically and randomly by up to F8 km, and used an alternative crustal thickness model [34]. The inversions consistently show a lowvelocity anomaly at the uppermost mantle depths beneath central Iceland, and generally similar struc-

8 604 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) km 9 km km 24 km Velocity perturbation (%) Fig. 8. Results of a test showing resolution in the crust beneath central Iceland. The input is a sinusoidal high velocity anomaly cylinder (~60 km in radius, from the surface to 15 km depth), underlain by a broader low velocity anomaly cylinder (~75 km in radius, extending from 18 km depth to the Moho). The maximum of the velocity perturbation is 5%. A Gaussian noise with a standard error of 0.2 s was added in the synthetic travel times. tures beneath other parts of Iceland. The low-velocity anomaly in the uppermost mantle beneath central Iceland is broader and stronger in the tests with a thinner crust. Because the surface-wave derived crustal thickness beneath central Iceland (46 km) [13] is considerably larger than other estimates (38 42 km [12,14,34]), the low-velocity anomaly in the uppermost mantle beneath central Iceland in Fig. 7 represents a conservative estimate. We also evaluated the effect of the initial crustal velocities on the inversion. Instead of the 3-D velocity model, we used the1-d average velocity as the starting velocity model. The inversion also yielded the low velocity anomalies in the uppermost mantle beneath central Iceland and Snæfellsnes and the high velocity feature in the upper crust beneath central Iceland. Compared to the result with the 3-D crustal velocity as the initial model, the distribution and magnitude of the anomalies in the uppermost mantle are similar, but the high velocity body in the upper crust is broader and deeper. To assess the effect of mislocation of earthquakes, we conducted synthetic tests with known perturbation in source parameters. We found that the latitude and longitude locations of the sources can be well resolved, but there is a strong tradeoff between the earthquake depth and origin time. A similar velocity model was obtained from the inversion without hypocenter relocation, with the rms travel time residual of 0.40 s, slightly larger than that from our final model. 4. Discussions 4.1. The crustal structure in central Iceland Our inversion revealed a crustal structure generally similar to that in the previous studies except that we found a high velocity anomaly in the upper crust beneath the central Icelandic volcanic zone (Bárðarbunga and Grímsfjall). High velocity anomalies above

9 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) (a) (b) (c) Input Input Input Output Output Output Velocity pertubation (%) Fig. 9. Map views of the input and output velocity perturbations to evaluate the resolution in the uppermost mantle. A Gaussian noise with a standard error of 0.2 s was added in the synthetic travel times for the three models. (a) Sinusoidal checkerboard velocity perturbations with the maximum amplitude of 5% in the layer directly beneath the Moho (block size: 7272 km). (b) A square low-velocity perturbation with a width of 130 km beneath central Iceland. (c) A low-velocity band beneath the neovolcanic zones. volcanic zones are a common feature of tomographic images in volcanic areas. They have also been found beneath Krafla in NVZ [37], at Yellowstone [38], Mt. St. Helens [39], Kilauea [40], Newberry volcano [41], Mt. Tsukuba [42] and other volcanic regions. These high-velocity anomalies are located over the magma chamber or in the surrounding walls, and they have usually been interpreted as the dense, igneous intrusive materials [43]. We note that this result is significantly different from the crust model constructed from surface waves [13], in which a low S-velocity was found at the same location. Fig. 10 shows a comparison of the observed P-wave arrival times along the ICEMELT refraction profile [12], the predicted first arrival times using the P velocity calculated from the S-velocity model with a linear V p /V s ratio [13] and the predicted first arrival times based on our P-velocity model. Our model significantly improves the fit of arrival times. Since the source of the profile was a dynamite explosion with a known location and origin time and these rays sampled the crust shallower than 20 km depth, this comparison indicates clearly that the P-wave velocity of the upper crust beneath central Iceland is significantly higher than that calculated from the S-velocity model with a linear V p /V s ratio. The discrepancy could be caused by the differences in resolution, errors and lateral variations in the V p /V s ratio, and seismic anisotropy, which may affect surface (Love) waves and P waves differently. This high velocity body in our P-wave model is also consistent with the upper crust structure in central Iceland constrained by receiver function studies [16,34]. One unexpected result of the mantle tomographic study from the HOTSPOT project was the inference of relatively high velocities (up to +2% for S and % for P waves) in the shallow upper mantle (Moho to ~100 km depth) beneath central Iceland [4], which were attributed to the effect of the presence of highly depleted mantle over the plume conduit compared to the surrounding mantle. For scale, a 1.75% high velocity in the km depth range would

10 606 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) (a) T0 T2 64Ê -24Ê -24 o sp15 sp08 sp12 sp04 va04 va03 va06 va02 va01 T0 T2-20 o -20Ê T0 T2 T0 T2 T2 64 o T0 T0 T2 T2T0 T2 T Travel time (s) T0 T2 T2T0 T2T0 T0T2 T0T2 va va va va va va sp sp sp sp sp sp sp (b) (c) -24 o -20 o Depth (km) Initial model Final model 64 o o o -20 o P velocity (km/s) Fig. 10. (a) A comparison of the observed P-wave arrival times (T0) along the ICEMELT refraction profile [12], and the predicted first arrival times () using the P-velocity calculated from the S-velocity model with a linear V p /V s ratio[13], and the predicted first arrival times (T2) based on our P-velocity model. The inset shows the locations of the shot (star) and stations (triangles) in seismograms. The epicentral distance in kilometer for each station is shown beneath the station name. (b) A comparison of the crustal P-velocity in central Iceland ( W, N) between our initial [13] and final models. (c) A comparison of the crustal corrections (in second) based on the initial model [13] (top) and the final model (bottom) for vertical rays at the HOTSPOT and ICEMELT stations in central Iceland.

11 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) lead to ~0.1 s travel time difference. As shown in Fig. 10c, the high velocity in the upper crust beneath central Iceland in our model results in a comparable difference in crustal corrections for vertical rays ( s) at the HOTSPOT and ICEMELT stations. Because the velocity of the shallow mantle depends strongly on crustal correction and the P velocity beneath central Iceland from our local earthquake tomography is significantly higher than that calculated from the S-velocity model, we suggest that those relatively high velocity features in the shallow mantle [13], especially the P-wave model, need to be carefully reevaluated The uppermost mantle As shown in Fig. 7d, velocities in the uppermost mantle beneath central Iceland are ~5% lower than the average of the resolved nodes at the same depths. If this velocity reduction is caused by a purely thermal anomaly, it would require a ~500 degree excess temperature [44], an anomaly much higher than the estimates based on the geochemistry of the Icelandic basalts [45,46] and the transition zone thickness anomaly [47]. The low velocity can also be attributed to the presence of up to 1.5% partial melt [48]. Because of the horizontal deflection of a hot material driven by buoyant upwelling at depth and plate spreading, thermal anomalies are expected to spread horizontally, and thus not consistent with the localized nature of the low-velocity anomaly beneath central Iceland. It is likely that the velocity reduction in the uppermost mantle beneath central Iceland is caused by a combination of high temperature and the presence of partial melt. The magnitude of the velocity reduction beneath central Iceland, which is underestimated as shown by resolution tests (Fig. 9), is much larger than that in the km depth range imaged by teleseismic tomography ( 2% 3.5% for P wave) [2,4,5]. This may indicate a higher melt content in the uppermost mantle than at depth. The low velocity in the uppermost mantle does not follow the distribution of fissure swarms (Fig. 7d), the surface manifestation of the rifting zones. In particular, there is no significant low velocity anomaly in the uppermost mantle beneath the MNZ and the northern part of the WVZ, where our dataset has the best resolution. This observation, along with the low-velocity anomaly beneath central Iceland, suggests that the melt supply to the hotspot is primarily focused in central Iceland at the uppermost mantle depths. We note that the low-velocity anomaly beneath the volcanically active Snæfellsnes [49] and the area between the western fjords and the rest of Iceland is associated with a relatively thin crust (15 25 km) compared to elsewhere in Iceland [13,50,51]. From the mean velocity of the uppermost mantle at that depth (~7.92 km/s, Fig. 7b), we calculate that the P wave velocity in the uppermost mantle in this region is ~7.5 km/s. Like the low-velocity anomaly beneath central Iceland, the large velocity reduction in western Iceland also suggests the presence of partial melt. This low-velocity anomaly may indicate the residual mantle upwelling beneath a failed spreading center that was centered on the hotspot from 15 to ~6.5 Ma [52,53]. Assuming a thin lithosphere along the hotspot track from the failed spreading center to the present location of the hotspot, we speculate that the relatively shallow crust mantle boundary in this region may provide an upward slope for the buoyant mantle flow along the hotspot track and the associated melt generation. 5. Conclusions We have constructed the crustal and uppermost mantle structure beneath Iceland from 3521 first arrival P waves from local earthquakes. With seismic rays traversing below the crust, our model shows a pronounced P-wave velocity reduction, about 5% or ~1.4 2 times that in the km depth range imaged by teleseismic tomography, in the uppermost mantle beneath central Iceland. This low velocity can be attributed to a 500-degree excess temperature or, more likely, a combination of a thermal anomaly and the presence of partial melt. The model also shows a low-velocity anomaly in the uppermost mantle beneath Snæfellsnes and the area between the western fjords and the rest of Iceland. We also found a relatively high velocity in the upper crust beneath central Iceland, which can be interpreted as dense, igneous intrusive materials associated with magmatic activities. This high velocity anomaly, which is not seen in the crustal model constructed from surface waves, suggests that the relatively high P-velocity feature in

12 608 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) the shallow mantle (from the Moho to ~100 km depth) in previous teleseismic tomography [4] may result from over-corrected crustal signals. Acknowledgements We thank all participants of the HOTSPOT and ICEMELT experiments, in particular Fiona Darbyshire for providing the ICEMELT profile data. Dr. Dapeng Zhao provided the original tomographic code. We acknowledge constructive comments by three anonymous reviewers. This project is support by U.S. National Science Foundation. References [1] I.Th. Bjarnason, C.J. Wolfe, S.C. Solomon, G. Gudmundson, Initial results from the ICEMELT experiment: body-wave delay times and shear-wave splitting across Iceland, Geophys. Res. Lett. 23 (1996) [2] C.J. Wolfe, I.Th. Bjarnason, J.C. VanDecar, S.C. Solomon, Seismic structure of the Iceland mantle plume, Nature 385 (1997) [3] G.R. Foulger, M.J. Pritchard, B.R. Julian, J.R. Evans, R.M. Allen, G. Nolet, W.J. Morgan, B.H. Bergsson, P. Erlendsson, S. Jakobsdottir, S. Ragnarsson, R. Stefansson, K. Vogfjord, The seismic anomaly beneath Iceland extends down to the mantle transition zone and no deeper, Geophys. J. Int. 142 (2000) F1 F5. [4] R.M. Allen, G. Nolet, W.J. Morgan, K. Vogfjord, B.H. Bergsson, P. Erlendsson, G.R. Foulger, S. Jakobsdottir, B.R. Julian, M. Pritchard, S. Ragnarsson, R. Stefansson, Imaging the mantle beneath Iceland using integrated seismological techniques, J. Geophys. Res. 107 (B12) (2002) 2325, doi: /2001jb [5] S.H. Hung, Y. Shen, L.Y. Chiao, Imaging seismic velocity structure beneath the Iceland hot spot: a finite frequency approach, J. Geophys. Res. 109 (2004), doi: / 2003JB [6] W. Keller, D. Anderson, R. Clayton, Resolution of tomographic models of the mantle beneath Iceland, Geophys. Res. Lett. 27 (2000) [7] C.J. Wolfe, I.Th. Bjarnason, J.C. VanDecar, S.C. Solomon, Assessing the depth resolution of tomographic models of upper mantle structure beneath Iceland, Geophys. Res. Lett. 29 (2002), doi: /20001gl [8] I.Th. Bjarnason, W. Menke, O.G. Flovenz, D. Caress, Tomographic image of the Mid-Atlantic plate boundary in southwestern Iceland, J. Geophys. Res. 98 (1993) [9] W. Menke, B. Brandsdottir, P. Einarsson, I.Th. Bjarnason, Reinterpretation of the RRISP-77 Iceland shear-wave profiles, Geophys. J. Int. 122 (1996) [10] R.K. Staples, R.S. White, B. Brandsdottir, W. Menke, P.K.H. Maguire, J.H. McBride, Faroe-Iceland ridge experiment I. Crustal structure of northeastern Iceland, J. Geophys. Res. 102 (1997) [11] N.R.W. Weir, R.S. White, B. Brandsdottir, P. Einarsson, H. Shmanura, H. Shiobara, Crustal structure of the northern Reykjanes ridge and Reykjanes peninsula, southwest Iceland, J. Geophys. Res. 106 (2001) [12] F.A. Darbyshire, I.Th. Bjarnason, R.S. White, O.G. Flovenz, Crustal structure above the Iceland mantle plume imaged by the ICEMELT refraction profile, Geophys. J. Int. 135 (1998) [13] R.M. Allen, G. Nolet, W.J. Morgan, K. Vogfjord, M. Nettles, G. Ekstrom, B.H. Bergsson, P. Erlendsson, G.R. Foulger, S. Jakobsdottir, B.R. Julian, M. Pritchard, S. Ragnarsson, R. Stefansson, Plume driven plumbing and crustal formation in Iceland, J. Geophys. Res. 107 (2002), doi: /2001jb [14] A. Li, Seismic structure of Iceland: evidence for a depleted hotspot root, submitted for publication. [15] F.A. Darbyshire, K.F. Priestley, R.S. White, R. Stefansson, G.B. Gudmundsson, S. Jakobsdottir, Crustal structure of central and northern Iceland from analysis of teleseismic receiver functions, Geophys. J. Int. 143 (2000) [16] Z. Du, G.R. Foulger, Variation in the crustal structure across central Iceland, Geophys. J. Int. 145 (2001) [17] W. Menke, V. Levin, Cold crust in a hot spot, Geophys. Res. Lett. 21 (1994) [18] W. Menke, V. Levin, R. Sethi, Seismic attenuation in the crust at the mid-atlantic plate boundary in southwest Iceland, Geophys. J. Int. 122 (1995) [19] S. Olafsson, R. Sigbjornsson, P. Einarsson, Estimation of source parameters and Q from acceleration recorded in the Vatnafojll earthquake in south Iceland, Bull. Seismol. Soc. Am. 88 (1998) [20] D.R. Toomey, G.R. Foulger, Tomographic inversion of local earthquake data from the Hengill Grensdalur central volcano complex, Iceland, J. Geophys. Res. 94 (1989) [21] G.R. Foulger, A.D. Miller, B.R. Julian, J.R. Evans, Threedimensional V p and V p /V s structure of the Hengill triple junction and geothermal area, Iceland, and the repeatability of tomographic inversion, Geophys. Res. Lett. 22 (1995) [22] A. Tryggvason, S.Th. Rognvaldsson, O. Flovenz, Three-dimensional imaging of the P- and S- wave velocity structure and earthquake locations beneath southwest Iceland, Geophys. J. Int. 151 (2002) [23] R. Böðvarsson, S.Th. Rögnvaldsson, R. Slunga, E. Kjartansson, The SIL data acquisition system at present and beyond year 2000, Phys. Earth Planet. Inter. 113 (1999) [24] G. Nolet, C. Coutlee, R. Clouser, S n velocities in western and eastern North America, Geophys. Res. Lett. 25 (1998) [25] M. Tittgemeyer, F. Wenzel, K. Fuchs, On the nature of P n,j. Geophys. Res. 105 (2000) 16,173 16,180. [26] H.M. Benz, B.A. Chouet, P.B. Dawson, J.C. Lahr, R.A. Page, J.A. Hole, Three-dimensional P and S wave velocity structure

13 T. Yang, Y. Shen / Earth and Planetary Science Letters 235 (2005) of Redoubt Volcano, Alaska, J. Geophys. Res. 101 (1996) [27] G.L. Pavlis, J.R. Booker, The mixed discrete-continuous inverse problem: application to the simultaneous determination of earthquake hypocenters and velocity structure, J. Geophys. Res. 85 (1980) [28] C. Spencer, D. Gubbins, Traveltime inversion for simultaneous earthquake location and velocity structure determination in laterally varying media, Geophys. J.R. Astron. Soc. 63 (1980) [29] D. Zhao, A. Hasegawa, S. Horiuchi, Tomographic imaging of P and S wave velocity structure beneath northeastern Japan, J. Geophys. Res. 97 (1992) 19,909 19,928. [30] D. Zhao, A. Hasegawa, H. Kanamori, Deep structure of Japan subduction zone as derived from local, regional and teleseismic events, J. Geophys. Res. 99 (1994) 22,313 22,329. [31] C.C. Paige, M.A. Sanders, LSQR: an algorithm for sparse linear equations and sparse least squares, ACM Trans. Math. Softw. 8 (1982) [32] W. Menke, Geophysical Data Analysis: Discrete Inverse Theory, Revised Edition, Academic Press, INC., [33] O. Gudmundsson, B. Brandsdottir, W. Menke, G.E. Sigvaldason, The crustal magma chamber of the Katla Volcano in south Iceland revealed by 2-D seismic undershooting, Geophys. J. Int. 119 (1994) [34] G.R. Foulger, Z.J. Du, B.R. Julian, Icelandic type crust, Geophys. J. Int. 155 (2003) [35] W. Menke, M. West, B. Brandsdottir, D. Spartks, Compressional and shear velocity structure of the lithosphere in northern Iceland, Bull. Seismol. Soc. Am. 88 (1998) [36] K. Breddam, M.D. Kurz, M. Storey, Mapping out the conduit of the Iceland mantle plume with helium isotopes, Earth Planet. Sci. Lett. 176 (2000) [37] B. Brandsdottir, W. Menke, P. Einarsson, R.S. White, R.K. Staples, Faroe Iceland ridge experiment 2. Crustal structure of the Krafla central volcano, J. Geophys. Res. 102 (1997) [38] H.M. Benz, R.B. Simth, Simultaneous inversion for lateral velocity variations and hypocenters in the Yellowstone region using earthquake and refraction data, J. Geophys. Res. 89 (1984) [39] J.M. Lees, R.S. Crosson, Tomographic inversion for threedimensional velocity structure at Mount St. Helens using earthquake data, J. Geophys. Res. 94 (1989) [40] C.H. Thurber, Seismic detection of the summit magma complex of Kilauea volcano, Hawaii, Science 223 (1984) [41] J.R. Evans, J.J. Zucca, Active high resolution seismic tomography of compressional wave velocity and attenuation structure at Medicine Lake Volcano, northern California Cascade range, J. Geophys. Res. 93 (1988) [42] M. Ishido, A.H. Hasemi, Three-dimensional fine velocity structure and hypocentral distribution of earthquakes beneath the Kanto Tokai district, Japan, J. Geophys. Res. 93 (1988) [43] G.R. Foulger, S.K. Arnott, Local tomography: volcanoes and the accretionary plate boundary in Iceland, Seismic Tomography: Theory and Practice, Chapman & Hall, London, 1993, pp [44] S. Karato, Importance of anelasticity in the interpretation of seismic tomography, Geophys. Res. Lett. 20 (1993) [45] J.G. Schilling, Fluxes and excess temperatures of mantle plumes inferred from their interaction with migrating midocean ridges, Nature 325 (1991) [46] R.S. White, J.W. Bown, J.R. Smallwood, The temperature of the Iceland plume and origin of outward-propagating v-shaped ridges, J. Geol. Soc. Lond. 152 (1995) [47] Y. Shen, S.C. Solomon, I.Th. Bjarnason, C.J. Wolfe, Seismic evidence for a lower-mantle origin of the Iceland plume, Nature 395 (1998) [48] W.C. Hammond, E.D. Humphreys, Upper mantle seismic wave velocity: effects of realistic partial melt geometries, J. Geophys. Res. 105 (2000) 10,975 10,980. [49] A. Gudmundsson, Dynamics of volcanic systems in Iceland: example of tectonism and volcanism at juxtaposed hot spot and mid-ocean ridge system, Annu. Rev. Earth Planet. Sci. 28 (2000) [50] Z. Du, G.R. Foulger, The crustal structure beneath the northwest fjords, Iceland, from receiver functions and surface waves, Geophys. J. Int. 139 (1999) [51] Z. Du, G.R. Foulger, Crustal structure beneath the western and eastern Iceland from surface waves and receiver functions, Geophys. J. Int. 149 (2002) [52] B.S. Hardarson, J.G. Fitton, R.M. Ellam, M.S. Pringle, Rift relocation a geochemical and geochronological investigation of a paleo-rift in northwest Iceland, Earth Planet. Sci. Lett. 153 (1997) [53] J. Helgason, Shifts of the plate boundary in Iceland some aspects of tertiary volcanism, J. Geophys. Res. 90 (1985)