Research ArticleSeismology

Relative seismic velocity variations correlate with deformation at Kīlauea volcano

See allHide authors and affiliations

Science Advances  28 Jun 2017:
Vol. 3, no. 6, e1700219
DOI: 10.1126/sciadv.1700219
  • Fig. 1 Maps of Kīlauea volcano.

    Seismic stations are shown as red triangles, and strainmeters are shown as black diamonds. Seismic stations enclosed in the red circle, centered on the eruptive vent, are used to average dv/v. Seismic instrumentation details are given in Materials and Methods. Tilt data in this study are from a tiltmeter located at UWE (zoomed-in map). The lava lake in the eruptive vent is located within Halema’uma’u caldera, which is itself located within the larger Kīlauea caldera. PAUD, Pauahi, Hawaii Digital; RIMD, Caldera Rim, Hawaii Digital; KWL, Keller Well; MLS, Mauna Loa Strip Road.

  • Fig. 2 Volcanic tremor due to spatter.

    The lava lake fluctuates between spattering (white) and nonspattering regimes (gray). (A) Visible spatter source at the lake surface. (B) No spattering visible. (C) UWE amplitude spectrogram, low-pass–filtered at 10 Hz. The tremor associated with spatter has greatest amplitude just above 0.5 Hz. (D) Real-time seismic amplitude measurement (RSAM) at station UWE calculated by bandpass filtering between 0.33 and 5.0 Hz and then averaging seismic amplitude in 1-min intervals. (E) UWE seismic trace, bandpass-filtered between 0.33 and 5.0 Hz. (F) Lava lake elevation, generally observed to be a few meters higher during nonspattering compared to spattering regimes. (A), (B), and (F) reproduced with permission from Patrick et al. (21). GMT, Greenwich mean time.

  • Fig. 3 Stability of the noise source location.

    (A) Amplitude spectrogram for seismometer UWE (see Fig. 1 for location). Amplitude spectrograms are calculated in 10-min windows after decimating and low-pass filtering at 4 Hz. The median value is used for each day. The frequency band used in this study is bounded by black lines. White bands are data gaps. (B) NCFs stacked over 3-day moving windows for station pair PAUD-RIMD (see Fig. 1 for locations). The white band is a data gap.

  • Fig. 4 Tremor source location.

    See text for methodology. The most likely source location is shown by the small red square, which is also the closest grid point to the lava lake. The larger red square shows the results of a jackknife test: Of 1000 best locations, calculated with only half of the network (randomly chosen), 90% fall in this box.

  • Fig. 5 Results of dv/v and its relationship with radial tilt.

    (A) Raw relative velocity variations, dv/v (blue). Light blue shading indicates the error in the measurement, calculated from the linear regression of dt against t. Raw radial tilt measured at UWE (gray). The radial component of tilt is calculated with respect to the eruptive vent in Halema’uma’u caldera. (B) Short-term dv/v and radial tilt, estimated by linearly detrending the raw series. The gray-green bar shows the correlation coefficient between dv/v and radial tilt in 30-day moving windows with an overlap of 6 days. Times highlighted in yellow correspond to (from left to right) a breakout eruption at Puʻu ʻŌʻō (episode 60), a large deflation and an “anomalous” deflation event (see text), and an overflow of the lava lake onto Halema’uma’u caldera floor. (C to G) Enlargements of time periods in (B). (E) is a large DI event. (H) Coherence (correlation coefficient) between current 3-day moving window NCF and reference NCF.

  • Fig. 6 Evidence for the positive correlation between dv/v and radial tilt.

    (A) Example of period in 2014 when radial tilt and dv/v track each other closely (gray-green bar as in Fig. 5B). (B) Histogram of the correlation coefficients calculated between radial tilt and dv/v for 30-day moving windows with an overlap of 6 days over the whole time period, as shown in Fig. 5B.

  • Fig. 7 Model of volumetric strain due to inflation of point sources at 1-km depth (Kīlauea, left) and 2.6-km depth (Piton de la Fournaise, right).

    (A) Map view of volumetric strain model at Kīlauea for a deformation point source at 1-km depth below the surface. Seismic stations are depicted as black triangles, and strainmeters are depicted as yellow diamonds. Positive strain is extensional; negative strain is compressional. Note that black and white values lie below and above the limits of the color scale, respectively. (B) Map view of volumetric strain model at Piton de la Fournaise for a deformation point source at 2.6-km depth below the surface. Seismic stations (black triangles) were those used in the study by Brenguier et al. (1). (C) Cross section of strain model for the dashed line shown in (A). Surface topography is marked on, but the model was calculated for an elastic half-space. (D) Rayleigh wave depth sensitivity kernels at Kīlauea for frequencies between 0.33 and 1.0 Hz (periods, 1 to 3 s every 0.25 s). (E) Same as in (C) but corresponding to Piton de la Fournaise and (B). (F) Same as in (D) but for Piton de la Fournaise.

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/3/6/e1700219/DC1

    Differential interstation distance from source

    Coda arrivals in NCFs

    Two reference functions

    Expected change in dv/v from strain data

    Frequency variations in the volcanic tremor source

    dv/v measured between 0.1 and 0.3 Hz

    Robustness of positive correlation between radial tilt and dv/v and comparison with meteorological effects and seismicity

    fig. S1. Explanation of differential interstation distance from source.

    fig. S2. Decay of coherent coda wave arrivals in the NCFs.

    fig. S3. Results when using two reference functions.

    fig. S4. Detailed view of frequency content of the volcanic tremor source and dv/v.

    fig. S5. Comparison of dv/v with 0.33- to 1-Hz and 0.1- to 0.3-Hz filters.

    fig. S6. Radial tilt-dv/v correlation and its association with meteorological effects and seismicity.

  • Supplementary Materials

    This PDF file includes:

    • Differential interstation distance from source
    • Coda arrivals in NCFs
    • Two reference functions
    • Expected change in dv/v from strain data
    • Frequency variations in the volcanic tremor source
    • dv/v measured between 0.1 and 0.3 Hz
    • Robustness of positive correlation between radial tilt and dv/v and comparison with meteorological effects and seismicity
    • fig. S1. Explanation of differential interstation distance from source.
    • fig. S2. Decay of coherent coda wave arrivals in the NCFs.
    • fig. S3. Results when using two reference functions.
    • fig. S4. Detailed view of frequency content of the volcanic tremor source and dv/v.
    • fig. S5. Comparison of dv/v with 0.33- to 1-Hz and 0.1- to 0.3-Hz filters.
    • fig. S6. Radial tilt-dv/v correlation and its association with meteorological effects and seismicity.

    Download PDF

    Files in this Data Supplement:

Navigate This Article