Research ArticleGEOPHYSICS

Crustal seismic velocity responds to a magmatic intrusion and seasonal loading in Iceland’s Northern Volcanic Zone

See allHide authors and affiliations

Science Advances  27 Nov 2019:
Vol. 5, no. 11, eaax6642
DOI: 10.1126/sciadv.aax6642
  • Fig. 1 Station map.

    Seismic stations used in this study: University of Cambridge stations are shown as red triangles, University College Dublin stations as yellow stars, IMO stations as inverted blue triangles, and the British Geological Survey (BGS) station as a green circle. Borehole B5704 is shown by the yellow circle. Central volcanoes are delineated, with their calderas shown by ticked lines; earthquakes associated with the Bárðarbunga-Holuhraun dike intrusion are shown as black dots; and the erupted lava is in dark gray. Stations discussed throughout the text and the Northern Volcanic Zone (NVZ) and Eastern Volcanic Zone (EVZ) are labeled. Inset: rift segments are shown in orange.

  • Fig. 2 Example of single-station cross-component NCFs.

    (A) NCFs between the horizontal components at station UTYR in the frequency band 0.4–1.0 Hz, here stacked over 10-day windows. The reference function, a stack of all NCFs up to 15 August 2014, is shown to the right. (B) Correlation coefficient of the NCFs shown in (A) with the reference function, between ±120 s. The decreases in correlation coefficient in April 2010, May 2011, and August 2014 to February 2015 correspond to eruptions of the Eyjafjallajökull, Grimsvötn, and Bárðarbunga volcanoes, respectively. The durations of the eruptions are grayed out, indicating that dv/v measurements are rejected during these times.

  • Fig. 3 Strain modeling of dike intrusion and comparison with dv/v response at individual stations.

    (A) dv/v in the 0.4–1.0 Hz band for a selection of stations, 30-day stacks. The zero line is solid gray for each station; each horizontal dashed line is 0.25%. Fit of the time series according to Eq. 1 is shown in black. (B) Model of the volumetric strain field caused by the 2014 dike intrusion (details in Results). Negative strain (blue) is compression; positive strain (red) is dilatation. (C) Coefficient of the step in dv/v from before to after the rifting event [from (A)] against the modeled volumetric strain [from (B)] at each station; color codes are the same as in (A).

  • Fig. 4 Continuous network-averaged dv/v measurements across five frequency bands.

    (A) dv/v measured from single-station cross-component NCFs with the stretching technique and averaged over the network. dv/v is measured by comparing 30-day stacks to a single reference function for the lowest two frequency bands, and between pairs of 5-day stacks offset by one day—with the earlier stack acting as a moving reference function—for the three higher frequency bands (see the Supplementary Materials for further details). The correlation coefficient is measured between windows of the stretched current NCF and the reference NCF; the time windows depend on the frequency band but are always within the NCF coda. (B) dv/v for each year is shown in gray, and the average is overlain in color. (C) Snow thickness (gray), atmospheric pressure (dark blue), and GWL (light blue) are averaged yearly over the same time period as the dv/v measurements. Elastic loading from snow and atmospheric pressure shown as water equivalent. GWL shown as depth below surface, i.e., a higher water level is plotted downward. Snow and atmospheric pressure are from the IMO’s meteorological model and GWL is modeled; see Results for details. Note that the y scales vary between the five panels of (A), but are consistent in the lowest panels (B and C).

  • Fig. 5 Model of seasonal variations in dv/v controlled by elastic loading and pore pressure changes.

    (A) Weather from IMO’s meteorological model averaged across the Northern Volcanic Zone (between stations KODA, FLUR, KRE, and HELI). Snow thickness as snow water equivalent (SWE) shown in gray. Rainfall, at the same scale, is in black. Atmospheric pressure is in green (1 hPa = 100 Pa). A 5-day rolling mean is applied to weather data for comparison with the stacked dv/v. (B) Modeled groundwater depth shown as depth below surface (see Discussion for details) shown in light blue. Modeled total load (combining snow, atmospheric pressure, and GWL; see Discussion for details) is shown in dark blue at the same scale. (C) Comparison of total load with dv/v measured in the 0.1−0.4 Hz band. A 30-day rolling mean is applied to the load because dv/v is stacked over 30 days in this frequency band. Gray bars show volcanic eruptions, as in Figs. 2 and 4. (D) Network-average of dv/v measured in the 2−4 Hz band (5-day stack, average of components NE, NZ, and EZ) is shown in red. Modeled dv/v is in black (see Discussion for details).

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/11/eaax6642/DC1

    Section S1. Robustness of dv/v results

    Section S2. Choice of reference functions

    Section S3. Lateral sensitivity of NCF coda waves

    Section S4. Spatial variations in dv/v

    Section S5. Comparison of measured and modeled GWL

    Section S6. Forward model of changes in dv/v from Rayleigh wave phase velocities

    Section S7. Modeling pore pressure variations

    Section S8. Seasonal variation in dv/v and frost

    Fig. S1. Depth sensitivity kernels.

    Fig. S2. Analysis of changes in dv/v before and after the dike intrusion across different frequency bands.

    Fig. S3. Comparison of MWCS and stretching dv/v results.

    Fig. S4. Comparison of dv/v results from station pairs and single-station cross-components.

    Fig. S5. Comparison of dv/v results from different time lags in the NCFs.

    Fig. S6. Comparison of dv/v results with the frequency content and amplitude of the noise source.

    Fig. S7. Comparison of dv/v measurements using static references and moving references.

    Fig. S8. Frequency content with lag time of an NCF at FLUR in the frequency band 0.4−1.0 Hz.

    Fig. S9. Spatial variations in dv/v at 0.4−1.0 Hz.

    Fig. S10. Comparison of measured and modeled GWL.

    Fig. S11. Model of seasonal variations in dv/v at station SVA.

    Fig. S12. Comparison of pore pressure and GWL models.

    Fig. S13. Comparison of dv/v and temperature data.

    Table S1. MSNoise parameters

    References (5358)

  • Supplementary Materials

    This PDF file includes:

    • Section S1. Robustness of dv/v results
    • Section S2. Choice of reference functions
    • Section S3. Lateral sensitivity of NCF coda waves
    • Section S4. Spatial variations in dv/v
    • Section S5. Comparison of measured and modeled GWL
    • Section S6. Forward model of changes in dv/v from Rayleigh wave phase velocities
    • Section S7. Modeling pore pressure variations
    • Section S8. Seasonal variation in dv/v and frost
    • Fig. S1. Depth sensitivity kernels.
    • Fig. S2. Analysis of changes in dv/v before and after the dike intrusion across different frequency bands.
    • Fig. S3. Comparison of MWCS and stretching dv/v results.
    • Fig. S4. Comparison of dv/v results from station pairs and single-station cross-components.
    • Fig. S5. Comparison of dv/v results from different time lags in the NCFs.
    • Fig. S6. Comparison of dv/v results with the frequency content and amplitude of the noise source.
    • Fig. S7. Comparison of dv/v measurements using static references and moving references.
    • Fig. S8. Frequency content with lag time of an NCF at FLUR in the frequency band 0.4–1.0 Hz.
    • Fig. S9. Spatial variations in dv/v at 0.4–1.0 Hz.
    • Fig. S10. Comparison of measured and modeled GWL.
    • Fig. S11. Model of seasonal variations in dv/v at station SVA.
    • Fig. S12. Comparison of pore pressure and GWL models.
    • Fig. S13. Comparison of dv/v and temperature data.
    • Table S1. MSNoise parameters
    • References (5358)

    Download PDF

    Files in this Data Supplement:

Stay Connected to Science Advances

Navigate This Article