Fig. 1 Ice mass balance and measured velocity variations. (A) Map of the ice mass changes over southwest Greenland in 2012. The stations used in this study are indicated by inverted triangles; the other Greenland stations of the Greenland Ice Sheet Monitoring Network (GLISN) are indicated by red dots. The colored lines between the stations show the dv/v averaged over the summer months June, July, and August (JJA). The black contours show the ice thickness at a 1000-m increment; the dashed black curve shows the contour used to integrate the ice mass changes from GRACE data. (B) Same as (A) for 2013. (C) Time series of the relative velocity variations. (D) Time series of the relative velocity variation uncertainties. The summer months are between dashed vertical lines.
Fig. 2 Velocity variation modeling. The thick red curve is the ice mass variation corrected from its quadratic trend and filtered in the 4- to 17-month period band. The green curve is the poroelastic modeling of dv/v based on the red curve. The black curve is the result of the viscoelastic modeling. The blue curve is the dv/v measurements averaged over all station pairs and filtered in the 4- to 17-month period, whereas the pale blue dots and the thin pale blue curves are the raw average dv/v measurements and the corresponding average uncertainties, respectively.
Fig. 3 Pore pressure diffusion in the Greenland crust through a till layer. The straight arrow indicates surface pressure and the wavy arrow indicates pore pressure diffusion; the thickness of the arrows indicates the amplitude of the pressure change. (A) Maximum pressure at Earth’s surface due to snow accumulation. The pore pressure change at depth is delayed by the bedrock and the till layer. The change is felt in the near surface and then starts to be felt at depth. (B) The maximum pore pressure change has reached the depth sampled by the seismic waves, and the relative velocity variation is therefore minimum. Meanwhile, surface pressure decreases as a result of ice melting and atmospheric pressure variations. (C) The surface pressure change reaches its minimum, followed by the pore pressure in the near surface. At depth, the pore pressure continues to decrease, whereas the velocity continues to increase. (D) Surface pressure starts to increase again. At depth, pore pressure is minimum, and relative velocity variation is maximum. Sections are not to scale.
- Table 1 Parameters used in the dv/v modeling.
Parameter Symbol Value Reference Glaciostatic pressure Pg 1600 Pa From data Ice area Si 6.5 × 1011 m2 From data Gravitational acceleration g 9.81 m/s2 Pressure field wave number k 2π/(60 km) Jiang et al. (11) Depth of investigation z 5 km From data S-wave velocity Vs 3300 m/s Kumar et al. (56) Vp/Vs ratio Vp/Vs 1.8 Kumar et al. (56) Upper-crust density ρc 2700 kg/m3 Schmidt-Aursch and Jokat (57) P-wave velocity Vp = Vs(Vp/Vs) 5940 m/s Poisson’s ratio ν 0.2768 Young’s modulus E 7.5 × 1010 Pa Mantle viscosity η 1021 Pa⋅s Viscoelastic relaxation time T 1011 s Lamé’s first parameter λ 3.65 × 1010 Pa Shear modulus μ 2.94 × 1010 Pa Murnaghan constant m −2.77 × 1016 Pa From inversion Distance from the ice x 12.5 km Biot’s coefficient α 0.7 Tsai (34) Hydraulic diffusivity of the crust Kc 0.5 m2/s Shapiro et al. (58) Angular frequency ω 2π/(365 days) Till layer thickness zt 2.85 m From inversion Hydraulic diffusivity of till Kt 5 × 10− 6 m2/s Iverson et al. (38)
Supplementary Materials
Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/2/5/e1501538/DC1
fig. S1. SNR of the reference correlations.
fig. S2. Seismic noise spectrograms.
fig. S3. Characteristics of the analysis window used to measure the velocity variations for each station pair.
fig. S4. Viscoelastic modeling: Vertical distribution of stress due to the ice sheet load.
fig. S5. Estimation of zt and m/μ.
fig. S6. Influence of the number of correlations stacked.
fig. S7. Influence of the symmetrization of the correlation on the dv/v measurements (ILULI-SFJ).
fig. S8. Influence of the analysis-window length on the dv/v uncertainties.
fig. S9. Effect of the analysis-window start time for the pair ILULI-NUUG (60-day stack, 0.1- to 0.3-Hz band, 300-s window).
fig. S10. Example of doublet measurements for NRS-IVI and comparison with the stretching method.
fig. S11. Example of doublet measurements for ILULI-SFJ and comparison with the stretching method.
Additional Files
Supplementary Materials
This PDF file includes:
- fig. S1. SNR of the reference correlations.
- fig. S2. Seismic noise spectrograms.
- fig. S3. Characteristics of the analysis window used to measure the velocity variations
for each station pair.
- fig. S4. Viscoelastic modeling: Vertical distribution of stress due to the ice sheet
load.
- fig. S5. Estimation of zt and m/μ.
- fig. S6. Influence of the number of correlations stacked.
- fig. S7. Influence of the symmetrization of the correlation on the dv/v measurements (ILULI-SFJ).
- fig. S8. Influence of the analysis-window length on the dv/v uncertainties.
- fig. S9. Effect of the analysis-window start time for the pair ILULI-NUUG (60-day
stack, 0.1- to 0.3-Hz band, 300-s window).
- fig. S10. Example of doublet measurements for NRS-IVI and comparison with the stretching
method.
- fig. S11. Example of doublet measurements for ILULI-SFJ and comparison with the stretching method.
Files in this Data Supplement:
- fig. S1. SNR of the reference correlations.
