Main

Thermomechanical processes caused by plume–lithosphere interactions have been studied for several decades. Various models have been proposed to explain the generating mechanism of the Hawaiian swell: (1) a model in which the dynamic pressure of the plume material flowing in the asthenosphere away from the plume centre supports the uplift of the sea floor4,6; (2) the thermal resetting (rejuvenation) model, in which the oceanic lithosphere is heated when it passes the hotspot, and thinning of the lithosphere causes isostatic uplift of the Hawaiian swell5; (3) a model in which extracted basaltic melt of the plume material underplates the base of the lithosphere. This residue of the plume material is compositionally lighter and has high seismic velocity7,8,9. Different models predict different seismic thicknesses of the lithosphere. In the first model, the lithospheric thickness along the swell is relatively stable. The second model (thermal rejuvenation) predicts a thinner lithosphere when it passes the hotspot. The lithospheric thickness in the third model will be larger than normal.

Until now, only the following observations of the lithospheric thickness by means of different seismic data have been available. In studies of surface-wave dispersion between Big Island and Oahu, a thickness of 88 km was suggested for a lithosphere 80–90 million years (Myr) old10. Between Oahu and Midway, 2,000 km northwest of Oahu, the average lithospheric thickness is estimated to be 100 km for a 50–110-Myr-old lithosphere11,12. In studies using converted waves, the thickness beneath Oahu was suggested to be 65–80 km (refs 13, 14), and 80 km southwest of Oahu15. From surface-wave dispersion data of an OBS (ocean bottom seismometer) array southwest of Oahu and Big Island, the average seismic structure of a 100-Myr-old oceanic lithosphere was derived and a significant velocity reduction at asthenospheric depths closer than 300 km to the island chain was found16. These facts are not sufficient to distinguish between competing models of lithosphere–plume interaction.

We have applied seismic techniques (S-receiver functions13,17,18) developed for P-receiver function studies to S-to-P converted waves (see Supplementary Information for a comparison of both techniques). S-receiver functions have been calculated from data of the permanent broadband stations POHA (IRIS/USGS), MAUI (GEOFON/ GeoForschungsZentrum, Potsdam) and KIP (IRIS/GEOSCOPE), which are located on the islands of Hawaii, Maui and Oahu, respectively. The locations of the stations POHA, MAUI and KIP and the piercing points of the S-to-P conversions at 100 km depth are shown in Fig. 1. Figure 2 shows the stacked S-receiver functions for a section NW–SE along the island chain (Fig. 2a) and a section SW–NE perpendicular to it (Fig. 2b). The sections are formed by stacking individual traces with piercing points at depths of 100 km within a window of length 30 km. Then the window is moved by 20 km for a new summation trace. Before summation, all traces were corrected for epicentral distance or slowness. A reference slowness of 6.4 s deg-1 has been chosen, similar to that in many P-receiver function studies, making P-to-S and S-to-P differential times directly comparable.

Figure 1: Location map of permanent broadband stations KIP (IRIS/GEOSCOPE), MAUI (GEOFON) and POHA (IRIS/USGS).
figure 1

Piercing points of S-receiver functions at 100 km depth are shown (same symbols as stations). Two lines labelled NW–SE and SW–NE indicate the two sections shown in Fig. 2 onto which the S-receiver functions are projected. Section NW–SE is along the island chain while section SW–NE is perpendicular to the chain.

Full size image
Figure 2: Stacked S-receiver functions of the three stations projected on the two profiles of Fig. 1 according to their piercing points at 100 km depth.
figure 2

a, Along the island chain (NW–SE); b, perpendicular to it (SW–NE). Zero time corresponds to the S-wave arrival. The timescale is reversed. The stack bin is 30 km with 10 km overlap. Slowness (or distance) move-out corrections are applied (reference slowness of 6.4 s deg-1) before stacking. In section NW–SE (a) all S-to-P data are used that cover an up to 200-km broad sampling region at 100 km depth on each side of the section. In section SW–NE (b) only S–P data southeast of KIP (Oahu) are used. A 7–20 s bandpass filter is applied to suppress the noise. LAB indicates the lithosphere–asthenosphere boundary, which is a precursor of the S wave.

Full size image

To suppress oceanic noise, a 7–20 s bandpass filter is applied. We reversed the polarity of amplitudes to compare the data directly with P-receiver functions. A clear negative phase (LAB, lithosphere–asthenosphere boundary) is identified in Fig. 2a, b. Negative (shaded) amplitudes are caused by velocity reduction downwards. To confirm that this phase is caused by the LAB, we compared the S-receiver functions with conventional P-receiver functions at the same stations and found good agreements (see Supplementary Information). The data in Fig. 2 are in the time domain. These delay times (in seconds) can roughly be converted into depth (in kilometres) by multiplying all times with a factor of 9, from the IASP91 model.

Figure 2a shows clearly that along the island chain the LAB thins from Big Island (12 s or 108 km depth at profile kilometre 500) to Kauai (6 s or 54 km depth at profile kilometre 0). Perpendicular to the island chain on the SW–NE section (Fig. 2b), the LAB is at 12 s delay time (108 km deep) at profile kilometres -150 to -100 and again at profile kilometre 200, northeast of the chain. It thins to 7–8 s (63–72 km deep) at profile kilometres 50–100 which is directly beneath the island chain along the profile (Fig. 1). The NW–SE section (Fig. 2a) contains all available S-to-P data. The perpendicular SW–NE section (Fig. 2b) contains only records with piercing points southeast of Oahu, which have similar minimum LAB depths. The two profiles in Fig. 2 indicate thinning of the lithosphere from Big Island to Oahu and Kauai as imaged in Fig. 3. The end of the thinning in the northwest direction is not visible. A station on Kauai or OBS stations northwest of Kauai would be necessary to follow the continuation of the lithospheric thinning.

Figure 3
figure 3

Sketch of the obtained lithosphere–asthenosphere boundary below the Hawaiian island chain.

Full size image

We obtain from our observations the following picture of the plume–lithosphere interaction. Whereas the Pacific plate overrides the Hawaiian plume with a velocity of 7–8 cm yr-1, plume material hits the lithosphere and is dragged off. It spreads mainly in the direction of the plate motion and causes a long tail along the Hawaiian island chain. The bottom of the lithosphere is heated by the plume material from below and the increased temperature will reduce the velocity of elastic waves. In the vicinity of the present plume centre, only the lowest part of the lithosphere is heated and the lithospheric thickness does not change much with respect to that of a 90–100-Myr-old oceanic plate. As time increases, more and more material from the lower part of the lithosphere is warmed up and the seismic LAB becomes shallower. After 3–4 Myr of heating time the thinnest lithosphere of 50–60 km is achieved beneath Oahu and Kauai—a location about 400 km away from the current plume centre10. Further away, at Midway Island, which passed the plume centre about 28 Myr ago, the plume material begins to lose its temperature and the lithosphere thickens again11,12.

We conclude that the Hawaiian plume has gradually reheated up to 50 km of the lower lithosphere within the last 3–4 Myr in an area of 500 × 300 km along the island chain. This means that a substantial amount of heat is trapped below the lithosphere. The lateral extent of the asthenospheric updoming is significantly smaller than the extent of the Hawaiian topographic swell. A consequence of our result could be that the entire topographic swell is caused by, for example, the dynamic support of the plume, and so rejuvenation would be contributing only in the central part. Data from longer-term broadband stations on Kauai, and on the ocean bottom northwest of Kauai, could permit imaging of the ‘re-ageing’ of the rejuvenated Hawaiian lithosphere.