Paper published in Nature: The anatomy of a mantle plume: Seismic structure of the Iceland hotspot

• S-wave Anomaly
• P-wave Anomaly
• Figure 1
• Figure 1. Location of ICEMELT seismic stations (filled triangles) within Iceland. We have also used data from the Global Seismographic Network station BORG in western Iceland (filled circle). The regional setting is illustrated by the satellite-derived gravity field of the North Atlantic; yellow-to-red colors indicate more positive gravity values, while green-to-blue colors indicate more negative values. Positive gravity anomalies generally coincide with shallower than normal seafloor. Icelandic neovolcanic zones are shaded in red. The two rectangles represent the exterior boundary, in map view, of the seismic velocity grid used for delay-time inversion and the interior boundary where the best resolution is obtained and within which we confine our interpretations. The model grid extends from 0 to 1000 km in depth.
• Figure 2
• Figure 2. Cross sections through the P-wave (left-hand column) and S-wave (right-hand column) models. Within this region, velocity nodes are spaced at 25 km in depth, 0.5¡ in longitude, and 0.25¡ in latitude. Velocity perturbation models are shown in map view at depths of 125 km (top row) and 300 km (middle row); vertical cross sections (bottom row) are along the white line in map views. Yellow-to-red colours represent low-velocity anomalies; regions with sparse coverage are faded to black. Station locations are shown as white squares in the middle row. Square and triangular symbols in the top row indicate the magnitude and sign of the station terms. The inversions solve simultaneously for independent (undamped) station terms to absorb as much shallow structure as possible into these terms and to avoid projecting such structure into the underlying velocity model. These station-term times represent the vertical travel-time of waves associated with each station and thus account in part for the integrated differences in mantle structure above 100 km depth, combined with the effects of crustal thickness variations and differences in station elevation. The Icelandic coastline and neovolcanic zone boundaries are shown as black lines.
• Figure S1
• Figure S1. Location of earthquakes at distances of 30-100 degrees from the Iceland seismic network that yielded travel-time residuals employed in this study. We also used observations from earthquakes at distances > 100 degrees with turning points in the Earth's core (not shown).
• Figure S2
• Figure S2. A resolution test for P waves. Left column shows input synthetic model; right column shows output after nonlinear inversion (note change in anomaly scale). Synthetic delay-time data are calculated by three- dimensional ray-tracing through the synthetic model and adding gaussian noise with 50-ms rms errors. The inversion procedure is identical to that described in the text for the observed delay-time data. We recover the input model reasonably well, except that the magnitude of the anomaly is weaker and the structure of the anomaly is broadened. See Figure 2 for further information.
• Figure S3
• Figure S3. A resolution test for S waves. This test is similar to that of Figure S2, except that the added noise has an rms error of 200 ms.
• Figure S4
• Figure S4. A resolution test for P waves for a synthetic model of a plume exending only to 200 km depth. The inversion solution does not produce a significant low- velocity anomaly at depths greater than 200 km. See Figure S2 for further information.
• Figure S5
• Figure S5. A resolution test for S waves for a synthetic model of a plume exending only to 200 km depth. See Figure S4 for further information.
• Figure S6
• Figure S6. A resolution test for P waves for a synthetic model of a broad plume with a radius of 300 km and a gaussian-shaped temperature contrast in cross section. The inversion solution does not produce a narrow velocity anomaly. See Figure S2 for further information.
• Figure S7
• Figure S7. A resolution test for S waves for a synthetic model of a broad plume. See Figure S6 for further information.