Limits...
Prompt gravity signal induced by the 2011 Tohoku-Oki earthquake

View Article: PubMed Central - PubMed

ABSTRACT

Transient gravity changes are expected to occur at all distances during an earthquake rupture, even before the arrival of seismic waves. Here we report on the search of such a prompt gravity signal in data recorded by a superconducting gravimeter and broadband seismometers during the 2011 Mw 9.0 Tohoku-Oki earthquake. During the earthquake rupture, a signal exceeding the background noise is observed with a statistical significance higher than 99% and an amplitude of a fraction of μGal, consistent in sign and order of magnitude with theoretical predictions from a first-order model. While prompt gravity signal detection with state-of-the-art gravimeters and seismometers is challenged by background seismic noise, its robust detection with gravity gradiometers under development could open new directions in earthquake seismology, and overcome fundamental limitations of current earthquake early-warning systems imposed by the propagation speed of seismic waves.

No MeSH data available.


Statistical analysis of gravity and seismic data.Distribution of the background reduced gravity signals  for (a) the superconducting gravimeter record only (d=2, T=690 s) and (b) the weighted stack of gravimeter and broadband seismometers records (d=1, T=1,900 s). The dashed red vertical lines represent the reduced gravity signal for the event . Note that the weighted stack is dimensionless. Empirical cumulative probability function (probability for a signal to exceed ) (blue curve) for (c) the gravimeter record only and for (d) the weighted stack record. The probability to obtain a signal larger than  (dashed vertical line) is 1.6% for (c) and 0.8% for (d). In all plots, the black curve corresponds to the best fitting Gaussian distribution.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5121411&req=5

f3: Statistical analysis of gravity and seismic data.Distribution of the background reduced gravity signals for (a) the superconducting gravimeter record only (d=2, T=690 s) and (b) the weighted stack of gravimeter and broadband seismometers records (d=1, T=1,900 s). The dashed red vertical lines represent the reduced gravity signal for the event . Note that the weighted stack is dimensionless. Empirical cumulative probability function (probability for a signal to exceed ) (blue curve) for (c) the gravimeter record only and for (d) the weighted stack record. The probability to obtain a signal larger than (dashed vertical line) is 1.6% for (c) and 0.8% for (d). In all plots, the black curve corresponds to the best fitting Gaussian distribution.

Mentions: We computed the reduced gravity signal of all selected background time intervals for a range of values of parameters d and T. The parameter values providing the best fit, that is, the smallest variance of , are d=2 and T=690 s. The reduced gravity signal time series for the Tohoku-Oki data with this optimal parameter setting are shown in Fig. 2. The resulting gravity signal strength is  μGal. It is compared with the background distribution in Fig. 3a. The cumulative probability, that is, the probability for a signal strength to exceed , is displayed in Fig. 3c. We note that the tails of the background distribution are not Gaussian. The fraction of background intervals with a larger gives the statistical significance P of the result. The reduced gravity signal equals or exceeds for 2,061 out of 127,885 background intervals, hence the probability that the signal strength in the Tohoku interval arises from the background fluctuations is P=2,061/127,885=1.6% (dashed vertical line in Fig. 3c). In other words, we can reject the hypothesis that the observed Kamioka signal is due simply to background fluctuations with 98.4% confidence.


Prompt gravity signal induced by the 2011 Tohoku-Oki earthquake
Statistical analysis of gravity and seismic data.Distribution of the background reduced gravity signals  for (a) the superconducting gravimeter record only (d=2, T=690 s) and (b) the weighted stack of gravimeter and broadband seismometers records (d=1, T=1,900 s). The dashed red vertical lines represent the reduced gravity signal for the event . Note that the weighted stack is dimensionless. Empirical cumulative probability function (probability for a signal to exceed ) (blue curve) for (c) the gravimeter record only and for (d) the weighted stack record. The probability to obtain a signal larger than  (dashed vertical line) is 1.6% for (c) and 0.8% for (d). In all plots, the black curve corresponds to the best fitting Gaussian distribution.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5121411&req=5

f3: Statistical analysis of gravity and seismic data.Distribution of the background reduced gravity signals for (a) the superconducting gravimeter record only (d=2, T=690 s) and (b) the weighted stack of gravimeter and broadband seismometers records (d=1, T=1,900 s). The dashed red vertical lines represent the reduced gravity signal for the event . Note that the weighted stack is dimensionless. Empirical cumulative probability function (probability for a signal to exceed ) (blue curve) for (c) the gravimeter record only and for (d) the weighted stack record. The probability to obtain a signal larger than (dashed vertical line) is 1.6% for (c) and 0.8% for (d). In all plots, the black curve corresponds to the best fitting Gaussian distribution.
Mentions: We computed the reduced gravity signal of all selected background time intervals for a range of values of parameters d and T. The parameter values providing the best fit, that is, the smallest variance of , are d=2 and T=690 s. The reduced gravity signal time series for the Tohoku-Oki data with this optimal parameter setting are shown in Fig. 2. The resulting gravity signal strength is  μGal. It is compared with the background distribution in Fig. 3a. The cumulative probability, that is, the probability for a signal strength to exceed , is displayed in Fig. 3c. We note that the tails of the background distribution are not Gaussian. The fraction of background intervals with a larger gives the statistical significance P of the result. The reduced gravity signal equals or exceeds for 2,061 out of 127,885 background intervals, hence the probability that the signal strength in the Tohoku interval arises from the background fluctuations is P=2,061/127,885=1.6% (dashed vertical line in Fig. 3c). In other words, we can reject the hypothesis that the observed Kamioka signal is due simply to background fluctuations with 98.4% confidence.

View Article: PubMed Central - PubMed

ABSTRACT

Transient gravity changes are expected to occur at all distances during an earthquake rupture, even before the arrival of seismic waves. Here we report on the search of such a prompt gravity signal in data recorded by a superconducting gravimeter and broadband seismometers during the 2011 Mw 9.0 Tohoku-Oki earthquake. During the earthquake rupture, a signal exceeding the background noise is observed with a statistical significance higher than 99% and an amplitude of a fraction of μGal, consistent in sign and order of magnitude with theoretical predictions from a first-order model. While prompt gravity signal detection with state-of-the-art gravimeters and seismometers is challenged by background seismic noise, its robust detection with gravity gradiometers under development could open new directions in earthquake seismology, and overcome fundamental limitations of current earthquake early-warning systems imposed by the propagation speed of seismic waves.

No MeSH data available.