Slide 7
Slide 7 text
CORE Axe 1 : Sismob (+RLBP)
148
Figure 9.3: Statistics and 2D representation of the MFP output
(a) Distribution of the average number of events located per d
as a function of the MFP output. Green shaded areas show t
Expérience RESOLVE (Glacier
d’Argentière) 2018
Nanni et al., 2021
Noubiat et al., en révision
Expérience ALPARRAY
15/11/2021 RST Résif-Epos, Obernai, 15-18 Nov. 20211 7
(b)
wing the regions, plates and main geological boundaries discussed in the text (modified from Faccenna et al., 2014). Black
ain thrusts and subduction zones. Black lines with squares represent extensional faults associated with development of the
ystem. (b) Location map of the broadband seismic networks used in this study. The white frame indicates the main focus
the Western Alps.
ROUP-VELOCITY MAPS
babilistic Vs
model of the Alpine
st step towards this goal, we com-
nd associated uncertainties using a
screte periods from 4 s to 150 s.
y maps are derived by exploring
ferent parameterizations using the
Monte-Carlo method (rj-McMC)
) first applied in a seismic tomog-
012). The parameterization of the
nversion without any explicit regu-
resolution to self-adapt to the path
he information contained in group-
odel complexity required to fit the
vel, which is treated as an extra pa-
rmined within a hierarchical Bayes
., 2004).
e crust is strongly heterogeneous
city contrasts, the non-linearity of
ed for by iteratively updating the
t Marching Eikonal solver (FMM,
.
ing to Bayes’s theorem (Bodin et al., 2012)
p(m|d) / p(d|m)p(m) (1)
where p(m) is the a priori probability density of the model param-
eters m, i.e. what we know about the velocity field independently
of the data. The term p(d|m) is the likelihood function and repre-
sents the probability of observing d given a model m, and given
the statistics of data errors. Assuming normally distributed uncor-
related data errors, p(d|m) can be expressed with the general Gaus-
sian form
p(d|m) =
1
N
Q
i=1
p
2⇡ di
⇥ exp
✓
(m)
2
◆
(2)
where di
is the standard deviation of data errors on the ith ob-
servation, N is the number of observations, and (m) is the misfit
function for the model m
(m) =
N
X
i=1
(g(m) d)2
2
d i
(3)
The term gi(m) represents data computed by the forward problem,
i.e. the travel time of the ith ray predicted by the model m, and
computed from
gi(m) =
n
X
Lij
vj
(4)
A new crustal Vs model of the Alps 11
PB
PB
PB
PB
SFB
SFB
SFB
SFB
IB
IB
IB
IB
W E
Transect Cifalps