 |
| Titel |
Understanding the dynamics of a geyser using seismic ambient noise |
| VerfasserIn |
Estelle Cros, Philippe Roux, Jean Vandemeulebrouck, Sharon Kedar |
| Konferenz |
EGU General Assembly 2010
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 12 (2010) |
| Datensatznummer |
250042073
|
|
|
|
|
|
| Zusammenfassung |
| Old Faithful Geyser in Yellowstone National Park, Wyoming, is one of the most studied
geysers in the world. The predictability, the repeatability and the short time lag, ~1.5 hour,
between 2 eruptions make the study convenient. The surface expression of the geyser is a
4m high, 60m wide mound with an approximately 2m x 1m opening at the top,
which permits to deploy a dense network of sensors closed to the orifice. In 1992,
Sharon Kedar deployed 96 vertical geophones in a tight grid over the geyser’s dome.
The geophones recorded the ambient seismic noise during an entire eruptive cycle,
including a short period of quiet seismic activity. The survey was completed by
seven shots carried out with a sledge hammer. The signal consists in a series of
impulsive events, most likely due to bubble collapse in boiling water areas inside the
geyser’s plumbing system. The aim of this study is to locate the sources of these
events.
We revisited a 10 minutes-long data set from S. Kedar’s records and processed the signal
using a Matched Field Processing (MFP) algorithm derived from ocean acoustics. The
cross-correlation of the signals recorded by the 96 geophones showed a great level of
coherency between the sensors, which is a pre-requisite to use MFP. This method introduced
in geophysics by Capon is based on comparing forward modelling solutions of the wave
equation in a grid search with acquired data, measured on an array of motion sensors. The
process consists in placing a test source at each point of the grid search, computing the
acoustic field corresponding at all the elements of the array and then correlating this modelled
field with the data. The correlation is maximum when the candidate point source is
co-located with the true point source. We used both linear (Bartlett) and non linear
(MVDR : Minimum Variance Distorsionless) processors. The MFP processor was
performed either incoherently from the raw ambient noise data or coherently from the
cross-correlated traces between all pairs of seismic stations. The processing of the
10 minutes-long record using a velocity model adapted from S. Kedar’s results
gave a very stable source at 12 m below the orifice. This location is consistent with
in-situ observations using a remote camera made by Hutchinson et al. on 1992
and 1993. The MFP output is well focused, with an accuracy of about 2m. The
processing of the seismic signals recorded during an eruptive cycle should lead us to
monitor the rise of the boiling region and thus to better understand the geyser’s
dynamics. |
|
|
|
|
|
|