Jökull - 01.01.2010, Page 96

Ólafur Guðmundsson and Bryndís Brandsdóttir
-2 -1 0 1 2 3 4
-2
-1
0
1
distance [km]
lo
g[
am
pl
itu
de
]
Öl
ke
ldu
há
ls
ge
oth
erm
al
are
a
A
B
C
Profile AProfile B
Figure 7. Log amplitude as a function of distance from Ölkelduháls. Amplitude has been integrated over the
frequency range from 3 to 7 Hz and computed as a root-mean-squared average. Deviations from a monotonic
amplitude decay from Ölkelduháls are labeled A, B, and C (see text). – Lógaritmi mældra útslagsrófa sem fall
af fjarlægð frá jarðhitasvæðinu á Ölkelduhálsi. Útslagsrófið hefur verið tegrað á milli 3 og 7 Hz. Frávik frá
einfaldri hnignun með fjarlægð frá Ölkelduhálsi aru merkt með A, B og C (sjá texta).
have plotted ad hoc samples of particle motion at a
few of the stations in Figure 8. The three chosen sta-
tions are LA0, LA4 and LA9. Three columns are
shown for each station showing the particle motion
in the vertical N-Z and E-Z planes and in the horizon-
tal E-N plane respectively from left to right. These
columns are labeled Y/X where Y, X = Z, N, E. In
each case the particle motion in the frequency range
between 3 and 7 Hz is plotted with component Y on
the vertical axis and component X on the horizontal.
Each diagram shows one second of data. The 1 second
windows are chosen three minutes apart during a ran-
domly chosen 30 minute period during low wind and
devoid of seismic events or other noise sources. This
random sample is characteristic for the data in gen-
eral. It is clear that the particle motion is generally
elliptical, suggesting a dominance of surface waves in
the wave field. We have also used the modeling tools
of Roberts and Christoffersson (1990) to examine dif-
ferent wave-type models for the noise. A Rayleigh-
wave model is clearly the most successful model in
general, although requiring significant averaging to
stabilize. Bursts in the data can be modeled as body
waves. A comprehensive stochastic analysis of po-
larization is under way but beyond the scope of this
paper. The conclusion for now is that the wave field
consists primarily of surface waves, but may contain
significant amounts of body-wave energy.
CORRELATION
If the source of the noise were simple and clearly lo-
calized one might be able to extract information about
location and slowness from correlation analysis. If
the noise field were diffuse and stochastic we might
be able to extract a part of the intra-station Green’s
function, and thus constrain velocity or slowness. We,
96 JÖKULL No. 60