Limits...
Reduction of randomness in seismic noise as a short-term precursor to a volcanic eruption

View Article: PubMed Central - PubMed

ABSTRACT

Ambient seismic noise is characterized by randomness incurred by the random position and strength of the noise sources as well as the heterogeneous properties of the medium through which it propagates. Here we use ambient noise data recorded prior to the 1996 Gjálp eruption in Iceland in order to show that a reduction of noise randomness can be a clear short-term precursor to volcanic activity. The eruption was preceded on 29 September 1996 by a Mw ~5.6 earthquake that occurred in the caldera rim of the Bárdarbunga volcano. A significant reduction of randomness started occurring 8 days before the earthquake and 10 days before the onset of the eruption. This reduction was observed even at stations more than 100 km away from the eruption site. Randomness increased to its previous levels 160 minutes after the Bárdarbunga earthquake, during which time aftershocks migrated from the Bárdarbunga caldera to a site near the Gjálp eruption fissure. We attribute this precursory reduction of randomness to the lack of higher frequencies (>1 Hz) in the noise wavefield caused by high absorption losses as hot magma ascended in the upper crust.

No MeSH data available.


Related in: MedlinePlus

Map of Iceland showing the locations of HOTSPOT stations included in this study.The rose diagrams represent the azimuth distributions obtained from the polarization analysis of seismic noise at the period when PE exhibited minimum values. The double arrows indicate the prevailing direction of the azimuth for each station, as determined by the application of the Rayleigh test (see text for more details). The map was created using GMT version 4.5.11 (http://gmt.soest.hawaii.edu/) the rose diagrams were created using MATLAB R2012 (http://www.mathworks.com/products/matlab/) and combined together using Adobe illustrator CS6 (http://www.adobe.com/products/illustrator.html).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5121890&req=5

f5: Map of Iceland showing the locations of HOTSPOT stations included in this study.The rose diagrams represent the azimuth distributions obtained from the polarization analysis of seismic noise at the period when PE exhibited minimum values. The double arrows indicate the prevailing direction of the azimuth for each station, as determined by the application of the Rayleigh test (see text for more details). The map was created using GMT version 4.5.11 (http://gmt.soest.hawaii.edu/) the rose diagrams were created using MATLAB R2012 (http://www.mathworks.com/products/matlab/) and combined together using Adobe illustrator CS6 (http://www.adobe.com/products/illustrator.html).

Mentions: During the time when the PE reached its minimum value and before the onset of high-frequency volcanic tremor (26–27 September), we also conducted a polarization analysis at all stations (see Methods section). The spatial variation in polarization azimuth for all stations can be seen in Fig. 5, while Supplementary Figure S4 contains the results of polarization azimuth and dip. In order to place some quantitative bounds on our azimuth polarization results we conducted two statistical tests and the results of these tests are summarized in Supplementary Tables S1–S3 for each station. First we performed a χ2–test employing the hypothesis that the azimuth data follow a uniform distribution (i.e. there is no preferential direction in the data). We find that the hypothesis can be rejected with a confidence level of 95% for all stations. The second test we conducted is called a Rayleigh test and employs the hypothesis that there is no single prevailing direction in the azimuth data19. The results show that the hypothesis can be rejected at a confidence level of 95% and that all stations exhibit NNE-SSW or NNW-SSE prevailing directions in azimuth (cf. Fig. 5). This observation combined with the dominant frequencies (0.01–0.25 Hz) suggest that the noise wavefield was dominated by sources originating along the southern or northern coast of Iceland, caused by the coupling of atmospheric-oceanic disturbances with the solid Earth20. From the polarization results we can also conclude that there is no evidence of very-long-period (VLP) tremor (~0.1–0.5 Hz) occurring prior to the Bárdarbunga earthquake as has been observed in other volcanoes2122. If this was the case, we would expect the prevailing direction at stations HOT23, HOT25 and HOT14 to point radially towards the Bárdarbunga caldera. This precludes the possibility that the reduction of randomness in ambient noise prior to the Gjálp eruption is due to the presence and overlap of VLP tremor with the ambient noise wavefield.


Reduction of randomness in seismic noise as a short-term precursor to a volcanic eruption
Map of Iceland showing the locations of HOTSPOT stations included in this study.The rose diagrams represent the azimuth distributions obtained from the polarization analysis of seismic noise at the period when PE exhibited minimum values. The double arrows indicate the prevailing direction of the azimuth for each station, as determined by the application of the Rayleigh test (see text for more details). The map was created using GMT version 4.5.11 (http://gmt.soest.hawaii.edu/) the rose diagrams were created using MATLAB R2012 (http://www.mathworks.com/products/matlab/) and combined together using Adobe illustrator CS6 (http://www.adobe.com/products/illustrator.html).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5121890&req=5

f5: Map of Iceland showing the locations of HOTSPOT stations included in this study.The rose diagrams represent the azimuth distributions obtained from the polarization analysis of seismic noise at the period when PE exhibited minimum values. The double arrows indicate the prevailing direction of the azimuth for each station, as determined by the application of the Rayleigh test (see text for more details). The map was created using GMT version 4.5.11 (http://gmt.soest.hawaii.edu/) the rose diagrams were created using MATLAB R2012 (http://www.mathworks.com/products/matlab/) and combined together using Adobe illustrator CS6 (http://www.adobe.com/products/illustrator.html).
Mentions: During the time when the PE reached its minimum value and before the onset of high-frequency volcanic tremor (26–27 September), we also conducted a polarization analysis at all stations (see Methods section). The spatial variation in polarization azimuth for all stations can be seen in Fig. 5, while Supplementary Figure S4 contains the results of polarization azimuth and dip. In order to place some quantitative bounds on our azimuth polarization results we conducted two statistical tests and the results of these tests are summarized in Supplementary Tables S1–S3 for each station. First we performed a χ2–test employing the hypothesis that the azimuth data follow a uniform distribution (i.e. there is no preferential direction in the data). We find that the hypothesis can be rejected with a confidence level of 95% for all stations. The second test we conducted is called a Rayleigh test and employs the hypothesis that there is no single prevailing direction in the azimuth data19. The results show that the hypothesis can be rejected at a confidence level of 95% and that all stations exhibit NNE-SSW or NNW-SSE prevailing directions in azimuth (cf. Fig. 5). This observation combined with the dominant frequencies (0.01–0.25 Hz) suggest that the noise wavefield was dominated by sources originating along the southern or northern coast of Iceland, caused by the coupling of atmospheric-oceanic disturbances with the solid Earth20. From the polarization results we can also conclude that there is no evidence of very-long-period (VLP) tremor (~0.1–0.5 Hz) occurring prior to the Bárdarbunga earthquake as has been observed in other volcanoes2122. If this was the case, we would expect the prevailing direction at stations HOT23, HOT25 and HOT14 to point radially towards the Bárdarbunga caldera. This precludes the possibility that the reduction of randomness in ambient noise prior to the Gjálp eruption is due to the presence and overlap of VLP tremor with the ambient noise wavefield.

View Article: PubMed Central - PubMed

ABSTRACT

Ambient seismic noise is characterized by randomness incurred by the random position and strength of the noise sources as well as the heterogeneous properties of the medium through which it propagates. Here we use ambient noise data recorded prior to the 1996 Gjálp eruption in Iceland in order to show that a reduction of noise randomness can be a clear short-term precursor to volcanic activity. The eruption was preceded on 29 September 1996 by a Mw ~5.6 earthquake that occurred in the caldera rim of the Bárdarbunga volcano. A significant reduction of randomness started occurring 8 days before the earthquake and 10 days before the onset of the eruption. This reduction was observed even at stations more than 100 km away from the eruption site. Randomness increased to its previous levels 160 minutes after the Bárdarbunga earthquake, during which time aftershocks migrated from the Bárdarbunga caldera to a site near the Gjálp eruption fissure. We attribute this precursory reduction of randomness to the lack of higher frequencies (>1 Hz) in the noise wavefield caused by high absorption losses as hot magma ascended in the upper crust.

No MeSH data available.


Related in: MedlinePlus