Content uploaded by François Thouvenot
Author content
All content in this area was uploaded by François Thouvenot on Aug 08, 2019
Content may be subject to copyright.
200-m-deep earthquake swarm in Tricastin (lower Rho
ˆne Valley,
France) accounts for noisy seismicity over past centuries
Franc¸ois Thouvenot, Liliane Jenatton and Jean-Pierre Gratier
Laboratoire de ge
´ophysique interne et tectonophysique (CNRS ⁄UJF), Observatoire de Grenoble, France
Introduction
Between the French Massif Central to
the west and the Alps to the east, the
ÔSillon Rhoˆ danienÕ(Fig. 1) is the
southern branch of the European
Cainozoic Rift System that dislocated
western Europe from the North Sea to
the Mediterranean (De
`zes et al.,
2004). In its middle part, midway
between Valence and Avignon, the
Tricastin area has long been recog-
nized as the seat of long-lasting earth-
quake swarms: besides the classical
way of releasing seismic energy
through mainshock–aftershock se-
quences, earthquake swarms are char-
acterized by long series of large and
small shocks, with no outstand-
ing principal event. The term
ÔSchwarmbebenÕ(i.e. Ôswarm quakeÕ)
was first used by Knett (1899) to
describe the random seismic activity
observed in the border region between
Germany and the Czech Republic
(Vogtland ⁄NW Bohemia), where this
phenomenon is frequent.
Swarms are common in volcanic
regions such as Japan, Central Italy,
Afar or oceanic ridges where they
occur before and during eruptions.
They are also observed in zones of
Quaternary volcanism such as Vogt- land ⁄NW Bohemia, where fluid
migration in a magmatic environment
can be invoked (e.g. Hainzl and
Fischer, 2002). In intraplate regions
(S
ˇpic
ˇa
´k, 2000) or orogenic belts, for
instance in the western Alps (Jenatton
et al., 2007), the dynamic evolution of
earthquake swarms remains more
mysterious, even if fluid migration is
a likely regulating factor (Daniel
et al., 2009).
Although hydrothermal sources are
documented, Tricastin is clearly not a
volcanic region. The boundary be-
tween Eurasia and the colliding Adri-
atic microplate, usually likened to the
ABSTRACT
In the lower Rho
ˆne Valley (France), the Tricastin area was struck
in 2002–2003 by an earthquake swarm with a maximum
M
L
-
magnitude of 1.7. These shocks would have gone unnoticed if
they had not occurred beneath habitations and close to the
surface, some events being only 200-m deep. A several monthsÕ
monitoring of the seismic activity by a 16-station mobile
network showed that earthquakes clustered along a N–S-
trending, at least 5-km long, shallow rupture zone, with no
corresponding fault mapped in the surface. Half of the seismic
events occurred in a massive,
c.
250-m-thick, Lower Cretaceous
limestone slab that outcrops near by. Since the late eighteenth
century, several much more severe earthquake swarms have
struck Tricastin. The 1772–1773 and 1933–1936 swarms were
prolific and protracted, with reports of numerous detonations
and even damage. Obviously, the abnormal noises that caused
panic in the past centuries can be explained by the shallowness
of the phenomena, a 200-m focal depth being perhaps a record
value for tectonic earthquakes.
Terra Nova, 21, 203–210, 2009
Correspondence: Franc¸ ois Thouvenot,
Laboratoire de ge
´ophysique interne et
tectonophysique, Maison des ge
´osciences,
BP 53, 38041 Grenoble Cedex, France.
Tel.: +33 (0) 4 76 63 51 50; fax: +33 (0) 4
76 63 52 52; e-mail: thouve@ujf-grenoble.fr
Fig. 1 Simplified map of the southern Rhoˆ ne Valley, with main geological contours
after Service de la Carte ge
´ologique de la France (1969): cross pattern, French Massif
Central; shaded, Mesozoic; blank, Cainozoic and Quaternary; barbed lines, main
thrusts; other faults mainly involve strike-slip motion. Dash-dotted lines in the
eastern part of the map: fold axes from Gratier et al. (1989). Box: study area of Fig. 2.
2009 Blackwell Publishing Ltd 203
doi: 10.1111/j.1365-3121.2009.00875.x
Piedmont seismic arc in Italy, is
located 200 km to the east (Thouve-
not and Fre
´chet, 2006). The Provenc¸ al
domain of the Alpine belt begins just
to the east of Tricastin, but the most
active part of the orogen in terms of
deformation and seismicity is located
150 km farther east. Thus, although
Tricastin is sited in the Sillon Rhoˆ da-
nien, which indeed has suffered exten-
sion since the Cainozoic, one may
rather classify it as an intraplate
region.
Two of the earthquake swarms that
struck Tricastin in 1772–1773 and
1933–1936 have been described in
testimonies as accompanied by explo-
sion noises similar to cannonades.
Rothe
´(1936) first evoked shallow
seismicity to explain these auditory
phenomena, but he lacked reliable
seismic data to argue this point, and
anyway shallow focal depths were
considered with scepticism at that
time when earthquakes were believed
to be usually seated much deeper in
the crust.
What seismologists now consider as
Ôshallow seismicityÕis of course a
matter of scale. At a global scale,
ÔshallowÕearthquakes are those that
occur in the first 40 km of the EarthÕs
interior. Man-made or volcanic seis-
micity documents events much closer
to the surface. In mines or gas fields,
seismicity usually occurs between the
surface and the depleted layers, and
focal depth values of a few hundred
metres are common. Earthquakes ow-
ing to dam filling usually occur much
deeper in the crust beneath the reser-
voir. Dry-rock experiments show that
fracturing occurs in a several hundred-
metre zone over and below the
injection point, which is usually sev-
eral-kilometre deep (e.g. Phillips,
2000). However, a thorough search
in the literature for shallow tectonic
earthquakes does not provide any
evidence for foci shallower than 1 or
2 km. In the following, the term Ôultra-
shallow seismicityÕwill be used to refer
to events with focal depths shallower
than 1 km.
The Tricastin 2002–2003 earth-
quake swarm revived memories of
the conflagration-like noises heard
in the previous episodes. When we
became aware of this phenomenon, we
judged that it offered a rare opportu-
nity to understand its origin, all the
more so as it could evolve into a
sequence of destructive events, in the
close vicinity (10 km) to major nuclear
installations on the western bank of
the Rhoˆ ne river. This article details
information gained by the deploy-
ment of a temporary seismic net-
work, and insists on the fact that
observed events were indeed ultra-
shallow.
Stratigraphy and tectonics
Our study area is 4420.5¢N–4426¢N
and 444¢E–450.5¢E (Box, Fig. 1).
Between the widely stretched-out
Rhoˆ ne Valley to the west and the
Visan Miocene Basin to the east,
Tricastin emerges as a region of Cre-
taceous and Cainozoic jagged hills,
with outcropping stratigraphy ranging
from Barremian (118–106 Ma) to
Burdigalian (20–15 Ma). The Upper
Barremian stage is characterized, as
elsewhere in southeast France, by the
Urgonian facies. This massive, hard,
white, reef limestone formation has an
approximate thickness of 250 m (Ser-
vice de la Carte ge
´ologique de la
France, 1975). In the study area, it
outcrops only in a limited flat zone, to
the southeast of La Garde-Adhe
´mar
(Fig. 2). Seismic and electric explora-
tion in the whole Tricastin area indi-
cates that the top of the Urgonian
slab, often only a few tens of metres
deep, is slashed by a complex fault
network (Service de la Carte ge
´ologi-
que de la France, 1964).
Major faults, west of the Rhoˆ ne
River, involve mostly strike slip along
the N40–N60E Hercynian direction
(Fig. 1). Some probably extend far-
ther to the east, where they meet the
Alpine domain, here characterized by
north- and south-verging thrusts, and
N–S-striking faults. Most probably
because of their poor preservation
within loose sedimentation, few tec-
tonic fractures are known in the study
area. However, seismic exploration
recognized several Upper Miocene
N–S-striking normal faults in the
Rhoˆ ne Valley. With throws reaching
several hundred metres, their juxtapo-
Fig. 2 Temporary seismological stations (triangles) and 38
HYPODD
relocated earth-
quakes, with a lighter shade for events shallower than 200 m. Symbol size is
proportional to the magnitude. Station CLAN marked by a white dot; Chabrelet is
the hypothetical epicentre of the 1936 earthquake swarm. Topographical contours are
at 10-m vertical intervals.
200-m-deep earthquake swarm in Tricastin (France) •F. Thouvenot
et al.
Terra Nova, Vol 21, No. 3, 203–210
.............................................................................................................................................................
204 2009 Blackwell Publishing Ltd
sition makes this zone a real rift
(Service de la Carte ge
´ologique de la
France, 1964). In the study area, other
minor features affect Aquitanian lime-
stones to the northeast of La Garde-
Adhe
´mar, where several conjugate
faults striking NW–SE and SW–NE
are documented.
Historical seismicity
The earthquake swarm that visited
Tricastin between June 1772 and
December 1773 is particularly well
documented by a contemporaneous
four-page report (Revol, 1773) and a
geological investigation (Faujas de
Saint-Fond, 1781). It affected the
whole study area shown in Fig. 2
(Clansayes, Sole
´rieux, Chantemerle,
Valaurie, Les Granges-Gontardes),
with more than 60 felt events (Boisse,
1936; Rothe
´, 1941). The old village of
Clansayes, perched on an outlier, had
its church tower knocked down by the
strongest event of the sequence (23
January 1773, maximum MSK inten-
sity I
max
= VII–VIII). According to
Revol (1773), the epicentral area
seems to have migrated afterwards,
and houses at Saint-Raphae
¨l (Sol-
e
´rieux), to the southeast of Clansayes,
suffered cracking damage from sub-
sequent events. Faujas de Saint-Fond
(1781) conversely states that, at the
end of the swarm in 1773, earthquakes
were more felt in villages to the
northwest of Clansayes. Throughout
the 19 months of the swarm, under-
ground noises similar to cannon
explosions were reported, whereas
earth vibrations did not seem to be
systematically noticeable.
In 1933–1936, another swarm vis-
ited the same area. This time most of
the underground noises were reported
in the northern villages of Les
Granges-Gontardes and La Garde-
Adhe
´mar (Rothe
´, 1936) – although
this statement might be biased by the
detailed observations left by Abbe
´
Boisse who precisely exercised his
priesthood at Les Granges-Gontardes.
The swarm was active between Octo-
ber 1933 and December 1934. After
10 months of quiescence, the activity
burst again in October 1935 till
August 1936. The total swarm activity
amounted to 24 months, with a climax
being reached by mid-May 1934 when
shocks were reported to be felt every
minute during the night of the 11–12
May 1934 (Boisse, 1936). A few hours
later, a stronger shock damaged sev-
eral churches and houses at Vallaurie,
Roussas and La Garde-Adhe
´mar,
with chimneys and one church
tower knocked down (12 May 1934,
I
max
= VII). Further slight damage
was reported at Clansayes on 11
January 1936 (I
max
= V), and at La
Garde-Adhe
´mar and Les Granges-
Gontardes on 13 February 1936 (I
max
= VI; Rothe
´, 1939a).
From the report of underground
noises, Rothe
´(1936) estimated that
the epicentre was situated at Chabr-
elet, southeast of Les Granges-Gon-
tardes (Fig. 2). He first tried to
determine the focal depth of an event
that occurred during the night of the
11–12 May 1934, and which was
particularly well recorded by four
seismological observatories (Clermont
and Strasbourg in France; Neuchaˆ tel
and Zurich in Switzerland). The clos-
est instrument (Clermont) being
200 km away, this attempt was
doomed to failure. However, Rothe
´
believed a 0-km focal depth better
fitted observed arrival times. He also
tried (Rothe
´, 1939b) to use isoseismal
curves observed for the 13 February
1936 event, but the various empiric
relations he used provided scattered
values (between 4 and 18 km). He
judged them unrealistic. In one last
attempt, Rothe
´(1939b) made use of
records provided for the 1936 active
period by two horizontal Mainka-
SOM seismographs, which had been
installed at Les Granges-Gontardes in
July 1934. From the average S–P
interval of 1.2 s, and taking into
account that the station was 2.5 km
away from his preferred epicentral
zone, he computed a focal depth of
3 km. However, when we scrutinized
original seismograms recorded on
smoked paper with a drum speed of
0.25 mm s
)1
, we found such minute
S–P intervals hardly discernable.
From all these attempts, one con-
cludes that the shocks were shallow,
even if one cannot prove that they
were ultra-shallow.
The 2002–2003 earthquake swarm
The 2002–2003 earthquake swarm ini-
tiated at the beginning of Decem-
ber 2002 by shocks perceived as
explosions by the inhabitants of a
c. 20-house hamlet close to Clansayes.
These abnormal sounds were not at
once identified as earthquakes by the
inhabitants because local earthquakes
are inexistent in the inter-swarm qui-
escence periods, and – to our knowl-
edge – the latest felt swarm dates back
to 1933–1936. A temporary velocimet-
ric station (CLAN) was installed in
the basement of one of the houses at
the end of December 2002.
On several seismograms recorded
by this station, we observed events
with an S–P interval of only 45 ms
(Fig. 3), which implies a very shallow
focus. In the first minutes of the New
YearÕs Day, 2003 (31 December 2002
UTC), two stronger (and felt) earth-
quakes occurred at 23:19 (M
L
= 1.3)
and 23:20 (M
L
= 1.7), both with S–P
intervals of about 100 ms. It
prompted us to install another 15
mobile stations (Fig. 2): 11 were fitted
with velocimeters and 4 with acceler-
ometers from the French mobile ac-
celerometric network. Although this
network has been operated for
8 months, seismic events were de-
tected only during the first three
months. During this period (10 Janu-
ary–7 April 2003), we located 51
events with magnitudes ranging from
Fig. 3 Tricastin swarm earthquake
recorded by three-component station
CLAN in the epicentral area (see posi-
tion in Fig. 2). Seismometers have a
2-Hz natural frequency. Amplitude win-
dow for each component (vertical, N–S
and E–W) is ±100 lms
)1
. Three-sec-
ond time scale; sampling frequency is
200 Hz. The minute 45-ms S–P interval
is clearer on the E–W component, bot-
tom signal.
Terra Nova, Vol 21, No. 3, 203–210 F. Thouvenot
et al.
•200-m-deep earthquake swarm in Tricastin (France)
.............................................................................................................................................................
2009 Blackwell Publishing Ltd 205
)0.7 to 1.4. For the present study,
earthquakes were first picked and
located using the
PICKEV2000
software
(Fre
´chet and Thouvenot, 2000), which
enables an interactive control of picks.
We then used
HYPREF2005
(Fre
´chet,
2005
)
, a modified version of the
HYPO71
programme (Lee and Lahr,
1975), with a one-dimensional velocity
model consisting of two 100-m thick
layers, with P-wave velocities of 2 and
4kms
)1
, and a 5.3 km s
)1
medium
underneath (S-wave velocities were
derived by assuming a V
P
⁄V
S
ratio
of 1.71). This model was built on
velocity measurements obtained for
similar sedimentary series at the Eguil-
les borehole, to the south of Avignon
(Mari, 1977). We eventually formed
travel time differences from P- and
S-picks and used the
HYPODD
pro-
gramme (Waldhauser and Ellsworth,
2000; Waldhauser, 2001) to improve
location precision.
Out of the initial 51 events, 38 only
were relocated (Fig. 2) because relo-
cation demands a higher data quality,
and events recorded by too few
stations are excluded. We prefer
relocated events which, although few-
er, are more reliable (Jenatton et al.,
2007). Although relocation involves
relative positioning, we have a good
control here on how the centroid of
the relocated swarm is positioned: the
largest-magnitude event (1.4) that
occurred on 26 January 2003 (Fig. 4)
is relocated right beneath the station
CLAN (triangle with white dot in
Fig. 2), in accordance with what could
be ascertained from a P-wave almost
exclusively recorded on the vertical
component at that station. However,
we note a slight vertical discrepancy
between the 400-m relocated depth
(relative to sea level), and the 45-ms
S–P interval observed at that station,
which rather corresponds to a 200-m
focal depth (relative to the sea level,
with a mean surface elevation of
100 m). We have not attempted to
correct this, which means that depth
values used in the following might
be overestimated (but by 200 m at
most).
Although activity was maximal
right beneath CLAN (Fig. 2), other
shocks were detected along a N–S-
trending zone whose length reaches at
least 5 km, even if one excludes the few
events that occurred at Roussas and
Sole
´rieux. Half of the events occurred
in the 0–200-m depth range (Fig. 5);
half of the remaining foci clustered in
the 400–600-m depth range; few others
occurred at a depth of about 1000 m.
Migration of epicentres with time is
uneasy to detect. However, earlier
events in the series were deeper and
more clustered in the central part of
the active zone (Fig. 6). While becom-
ing shallower, activity has migrated
southwards and northwards since the
beginning of February.
Figure 7 shows the complete time
series for the earthquake swarm.
Station CLAN, installed in December
2002, recorded many small-magni-
tude shocks whose epicentres cannot
be located. For most of them, P- and
S-wave arrivals can be read on
seismograms. We considered that an
event belonged to the earthquake
swarm whenever the observed S–P
interval was smaller than 500 ms.
These earthquakes daily recorded at
CLAN provide an estimate of the
swarm activity. For 79 events that
could not be located, we estimated
the M
L
magnitude by assuming that
earthquakes were beneath CLAN.
Figure 7 also includes magnitudes
for the 51 earthquakes located by
the temporary network, so that the
magnitude series totals 130 events,
with magnitude ranging from )1.3 to
1.7. Activity was variable in the
course of the 4 monthsÕperiod. The
two New YearÕs Day shocks in-
disputably generated aftershocks,
whereas other ÔlargeÕshocks in Janu-
ary and March did not. For located
events, we observed variable intervals
between consecutive events, ranging
from less than 1.5 s to more than
10 days.
Fig. 4 Example of normalized amplitude signals recorded by stations of the
temporary network for the M
L
-1.4 26 January 2003 earthquake. Thirteen-second
time scale; epicentral distances range from 0.2 km (top) to 2 km (bottom). Amplitude
windows range from ±300 lms
)1
(top) to ± 9 lms
)1
(bottom).
200-m-deep earthquake swarm in Tricastin (France) •F. Thouvenot
et al.
Terra Nova, Vol 21, No. 3, 203–210
.............................................................................................................................................................
206 2009 Blackwell Publishing Ltd
Earthquake populations are classi-
cally characterized by the Gutenberg–
Richter law (Gutenberg and Richter,
1956)
log10 N¼abM ;
where Nis the number of earthquakes
with magnitudes larger than or equal
to M. Figure 8 shows the frequency–
magnitude distribution for the 130
events of the magnitude series. The
deviation from the Gutenberg–Rich-
ter law for negative magnitudes obvi-
ously results from our catalogue being
incomplete for small-magnitude earth-
quakes. The bvalue of 1.0 ±
0.3, a figure similar to that found for
the western Alps as a whole (0.95 ±
0.03), was estimated by a maximum-
likelihood analysis (Aki, 1965; Utsu,
1966). The large uncertainty for b
partly results from the Tricastin series
being limited in size, but also from
significant variation with time: if we
split the time series into two and
analyse its two halves, bdecreases
from 1.2 ± 0.4 at the beginning of the
swarm (ÔdeepÕevents) to 0.8 ± 0.3 at
its termination (events shallower than
0.2 km).
Discussion and conclusions
No fault has ever been mapped at the
surface where the approximately N–S-
trending, c. 5-km long, 0- to 1-km
deep rupture zone imaged by the 38
relocated events was identified. One
reason is that most brittle Lower
Cretaceous series that could be used
as tectonic markers are obliterated in
the study area by looser sediments.
However, we mentioned that seismic
exploration recognized several Upper
Miocene N–S-striking faults in the
Rhoˆ ne Valley and that electric explo-
ration reveals extensive fracturing of
the top of the Urgonian slab. N–S-
oriented topography in the central
part of Fig. 2 between Clansayes and
Valaurie can also be noticed. The
identification of such a widespread
rupture zone instead of a pinpointed
focal zone can explain the impression
of migrating events reported by inhab-
itants during episodes of the past
centuries.
The main peculiarity of the Trica-
stin swarm is its ultra-shallowness,
other swarms being usually deeper-
seated (Table 1). For station CLAN,
we indeed observed S–P intervals of
only 45 ms. If we assign a P-wave
velocity of 5.3 km s
)1
to the massive
Urgonian limestone slab that outcrops
1.5 km from CLAN, and if we use a
V
P
⁄V
S
ratio of 1.71, the hypocen-
tral distance would be c. 300 m.
This would be the focal depth value
(relative to the surface) for a focus
right beneath the station; a still
shallower value would be obtained
otherwise.
Half of the events occurred in the
0–200-m depth range (Fig. 5), which
very likely corresponds to the 250-m
thick Urgonian slab that outcrops
nearby. Seismic foci also cluster
400 m below in the 400–600-m depth
range, in the midst of the Lower
Cretaceous sediments where relatively
(a) (b) (c)
Fig. 5 Depth histograms for the 38 relocated earthquakes. (a) All events; (b) events
before 1 February 2003; (c) events after 2 February 2003.
Fig. 6 As a function of time (vertical
axis), position of epicentres along the N–
S axis (km 0 is station CLAN, dotted
triangle in Fig. 2). Each event is repre-
sented by a circle with radius propor-
tional to the magnitude. Solid circles
represent events with focal depth larger
than 200 m and open circles indicate
events with focal depth shallower than
200 m.
Fig. 7 Full-time series for the 2002–2003
Tricastin earthquake swarm. Histogram
shows daily number of earthquakes
detected by station CLAN. An M
L
magnitude was computed for events that
could be located (plotted as circles with
radii proportional to the magnitude);
some smaller events recorded at CLAN
were also assigned a magnitude under
the assumption that the focus was right
beneath the station. Note that, although
monitoring continued till August 2003,
no event could be located later than
April.
Fig. 8 Cumulated frequency–magnitude
distribution for the 130 events of
the series yields a Gutenberg–Richter
b-value of 1.0 ± 0.3.
Terra Nova, Vol 21, No. 3, 203–210 F. Thouvenot
et al.
•200-m-deep earthquake swarm in Tricastin (France)
.............................................................................................................................................................
2009 Blackwell Publishing Ltd 207
rigid limestones alternate with marls.
A third cluster is sited at a depth of
about 1000 m. This is precisely where
another rigid limestone slab can be
expected in the stratigraphy (the so-
called Tithonian facies characteristic
of Upper Jurassic series in southeast
France). Thus, the upper and lower
clusters occurred in brittle limestone
slabs and the intermediate cluster in a
relatively rigid part of the series.
We could not evidence any migra-
tion of epicentres with time for the
beginning of the swarm (before 1
February 2003), when only the central
part of the rupture zone was active
(Fig. 6). After that date, activity seems
to have spread southwards and north-
wards by several kilometres. We prob-
ably lack a detailed description of the
swarm at earlier times (i.e. in Decem-
ber 2002) to understand this pheno-
menon. As a result of its suddenness,
it cannot be attributed to fluid diffu-
sion from a single source point, as is
sometimes observed for other deeper
swarms (see, e.g. Hainzl and Fischer,
2002; Daniel et al., 2009).
Figure 6 also shows that, before 1
February 2003, the swarm involved
ÔdeepÕearthquakes (with focal depths
larger than 200 m). In contrast, all
later earthquakes but three clustered
in the first 200 m. Thus, the extension
of seismic activity we observed along
the rupture zone in the last part of the
swarm series was accompanied by the
upward migration of seismic foci. This
difference between the two halves of
the swarm series can also be evidenced
with regard to the bvalue, which
decreased from 1.2 (deeper, earlier
events) to 0.8 (shallower, later events).
Hainzl and Fischer (2002), when anal-
ysing the Vogtland ⁄NW Bohemia
swarm, that is in contrast much deeper
(8.5 km), also observed similar b-
value variations.
As past swarm episodes that were
marked by detonation-like sounds, the
2002–2003 Tricastin earthquake
swarm was noticed for its many felt
events. This can be surprising for
magnitudes that did not exceed 1.7.
However, a recent study in the south-
ern French Jura (Thouvenot and Bou-
chon, 2008) showed that earthquakes
sited at a depth of c. 900 m could be
felt even for negative magnitude val-
ues (down to magnitude )0.7). In
Tricastin, as all relocated events but
two have magnitudes larger than )0.7,
and as focal depths are much shal-
lower, practically all shocks could
have been felt or – more probably –
heard.
Finally, as with all earthquake
swarms, the most puzzling problem
remains that of the initiation and
duration of the phenomenon. In Tri-
castin, just like in Ubaye, there is an
interesting common belief that earth-
quake swarms are often the conse-
quence of flooding. Seasonal
groundwater recharge and rainfall
have effectively been described as
triggering agents for some swarms
(e.g. Saar and Manga, 2003; Kraft
et al., 2006; Husen et al., 2007). Miller
(2008) theorized that Ôunambiguous
rain-triggered seismicity will only oc-
cur in karst regionsÕ. This hypothesis
is appealing in our particular case
because, where exposed in southeast
France, the Urgonian slab indeed
presents karstic features. However,
daily rainfall at the nearby Monte
´li-
mar weather station (Fig. 9) does not
reveal any exceptional variations prior
to the observed swarm activity. In
September 2002, the catastrophic
storm that swept the Avignon region,
50 km to the south of Tricastin, and
whose rainfall is held by Rigo et al.
(2008) as responsible for an increase in
seismicity rate on faults between
Nıˆ mes and Avignon cannot be traced
in Fig. 9, probably because of its local
– although devastating – character.
Unless we hypothesize that the weath-
er station used here and situated
20 km to the north missed a similar
heavy rain episode in Tricastin, we
Table 1 Mean depth for a selection of instrumentally studied earthquakes swarms.
Depth is referred to the surface where the source text is explicit about surface
elevation, assumed to be referred to the surface otherwise.
Location Date
Mean
depth
(km) Reference
Imperial Valley, California (USA) 1975 6 Johnson and Hadley (1976)
Reykjanes Peninsula (IS) 1972 3.5 Klein
et al.
(1977)
Arkansas (USA) 1982 5.5 Chiu
et al.
(1984)
Remiremont, Vosges (F) 1984–1985 7 Haessler and Hoang-Trong (1985)
Mammoth Mt., California (USA) 1989 7.5 Hill
et al.
(1990)
Steigen, Nordland (N) 1992 6.5 Atakan
et al.
(1994)
Crested Butte, Colorado (USA) 1986 6.5 Bott and Wong (1995)
Izu Peninsula, Honshu (J) 1997 4.5 Aoki
et al.
(1999)
Manchester (GB) 2002 2.5 Baptie and Ottemoeller (2003)
Colfiorito, Umbria-Marche (I) 1997 6 Chiaraluce
et al.
(2003)
Vogtland ⁄NW Bohemia (D ⁄CZ) 1985–2001 8.5 Fischer and Hora
´lek (2003)
Mt. Hood, Oregon (USA) 1980–2002 4.5 Saar and Manga (2003)
Campi Flegrei, Campania (I) 2000 2 Bianco
et al.
(2004)
Usu Volcano, Hokkaido (J) 2000 5.5 Zobin
et al.
(2005)
Mt. Hochstaufen, Bavaria (D) 2002 2 Kraft
et al.
(2006)
Ubaye, western Alps (F) 2003–2004 7 Jenatton
et al.
(2007)
Obsidian Buttes, California (USA) 2005 5 Lohman and McGuire (2007)
Tricastin, Rho
ˆne Valley (F) 2002–2003 0.5* This study
*With 50% of events in the 100–300-m depth range.
CZ, Czech Republic; D, Germany; F, France; GB, Great Britain, I, Italy; IS, Iceland; J, Japan; N, Norway; USA,
United States of America.
Fig. 9 Daily rainfall at Monte
´limar
weather station (Me
´te
´o-France), 20 km
north of Tricastin, over a 10-year period
(1994–2003). Daily rainfall exceeded
100 mm several times over this period
(1994, 1998, 1999, 2000 and 2003), with
a maximum in autumn 1999
(c. 220 mm), but no triggered seismic
activity was ever detected in Tricastin.
The 2002–2003 earthquake swarm fol-
lows a rainy episode very similar to
other autumnal heavy rain events.
200-m-deep earthquake swarm in Tricastin (France) •F. Thouvenot
et al.
Terra Nova, Vol 21, No. 3, 203–210
.............................................................................................................................................................
208 2009 Blackwell Publishing Ltd
still lack any clear triggering phenom-
enon that could explain why Upper
Jurassic and Lower Cretaceous series
can be healed for so many tens of
years before suddenly bursting out in
a veritable cannonade.
Acknowledgements
The Conseil Ge
´ne
´ral de lÕIse
`re, the De
´le
´-
gation aux Risques Majeurs (French Min-
istry of the Environment), the Institut
National des Sciences de lÕUnivers (CNRS)
and the Conseil Re
´gional Rhoˆ ne-Alpes
funded the Sismalp network. The Bureau
Central Sismologique Franc¸ ais, the Obser-
vatoire de Grenoble and several Conseils
Ge
´ne
´raux (Ise
`re, Alpes-de-Haute-Prov-
ence, Haute-Savoie, Ain and Savoie)
supported its running costs. The Conseil
Ge
´ne
´ral de la Droˆ me granted special funds
for studying the earthquake swarm. The
authors are indebted to M. Garin, Mayor
of Clansayes, and to the inhabitants of
Clansayes and nearby villages who facili-
tated field work between December 2002
and August 2003. Robert Guiguet was
involved in station maintenance and data
processing; accelerometric stations were
installed by several other colleagues from
LGIT whose help is acknowledged. Julien
Fre
´chet (IPG Strasbourg) provided the
authors with the 1936 records at Les
Granges-Gontardes. Figures of this article
were drawn by using the GMT software
(Wessel and Smith, 1998). Three anony-
mous reviewers who provided helpful com-
ments are also acknowledged.
References
Aki, K., 1965. Maximum likelihood esti-
mate of bin the formula log N=a)bM
and its confidence limits. Bull. Earth-
quake Res. Inst. Univ. Tokyo,43,
237–239.
Aoki, Y., Segall, P., Kato, T., Cervelli, P.
and Shimada, S., 1999. Imaging magma
transport during the 1997 seismic swarm
off the Izu Peninsula, Japan. Science,
286, 927–929.
Atakan, K., Lindholm, C.D. and Havskov,
J., 1994. Earthquake swarm in Steigen,
northern Norway: an unusual example
of intraplate seismicity. Terra Nova,
6, 180–194; doi:10.1111/j.1365-3121.
1994.tb00652.x.
Baptie, R. and Ottemoeller, L., 2003. The
Manchester earthquake swarm of Octo-
ber 2002. Geophys. Res. Abstr.,5, 10286.
Bianco, F., Del Pezzo, E., Saccorotti, G.
and Ventura, G., 2004. The role of
hydrothermal fluids in triggering the
July–August 2000 seismic swarm at
Campi Flegrei, Italy: evidence from seis-
mological and mesostructural data. J.
Volcanol. Geotherm. Res.,133, 229–246.
Boisse, L., 1936. Les tremblements de terre
dans la Droˆ me et spe
´cialement dans le
Tricastin. Bur. Centr. Se
´ismol. Int., Ser.
B, Monogr., Fasc.,6, 1–33.
Bott, J.D.J. and Wong, I.G., 1995. The
1986 Crested Butte earthquake swarm
and its implications for seismogenesis in
Colorado. Bull. Seismol. Soc. Am.,85,
1495–1500.
Chiaraluce, L., Ellsworth, W.L., Chiarab-
ba, C. and Cocco, M., 2003. Imaging the
complexity of an active normal fault
system: the 1997 Colfiorito (central Italy)
case study. J. Geophys. Res.,108, 2294;
doi:10.1029/2002JB002166.
Chiu, J.-M., Johnston, A.C., Metzger,
A.G., Haar, L. and Fletcher, J., 1984.
Analysis of analog and digital records of
the 1982 Arkansas earthquake swarm.
Bull. Seismol. Soc. Am.,74, 1721–1742.
Daniel, G., Renard, F., Thouvenot, F.,
Jenatton, L., Helmstetter, A., Hainzl, S.,
Marsam, D. and Guiguet, R., 2009.
Diffusion property and spatio-temporal
evolution of the 2003–2004 Ubaye,
French Alps, earthquake swarm. Sub-
mitted to Geophys. J. Int.
De
`zes, P., Schmidt, S.M. and Ziegler, P.A.,
2004. Evolution of the European Ceno-
zoic Rift System: interaction of the
Alpine and Pyrenean orogens with their
foreland lithosphere. Tectonophysics,
389, 1–33.
Faujas de Saint-Fond, B., 1781. Histoire
naturelle de la province de Dauphine
´, Vol.
1. Chez la Vve Giroud, Grenoble.
Fischer, T. and Hora
´lek, J., 2003. Space-
time distribution of earthquake swarms
in the principal focal zone of the NW
Bohemia ⁄Vogtland seismoactive region:
period 1985–2001. J. Geodyn.,35, 125–
144.
Fre
´chet, J., 2005.
HYPREF
2005; http://sis-
malp.obs.ujf-grenoble.fr/ftp-sismalp/
unix/.
Fre
´chet, J. and Thouvenot, F., 2000.
PIC-
KEV
2000; http://sismalp.obs.ujf-greno-
ble.fr/ftp-sismalp/msdos/.
Gratier, J.-P., Me
´nard, G. and Arpin, R.,
1989. Strain-displacement compatibility
and restoration of the Chaıˆ nes Subal-
pines of the western Alps. In: Alpine
Tectonics (M.P. Coward, D. Dietrich
and R.G. Park, eds). J. Geol. Soc. Lond.
Spec. Publ.,45, 65–81.
Gutenberg, B. and Richter, C.F., 1956.
Earthquake magnitude, intensity, energy
and acceleration. Bull. Seismol. Soc.
Am.,46, 105–145.
Haessler, H. and Hoang-Trong, P., 1985.
La crise sismique de Remiremont
(Vosges) de de
´cembre 1984: implications
tectoniques re
´gionales. C. R. Acad. Sci.
Paris,II-14, 671–675.
Hainzl, S. and Fischer, T., 2002. Indica-
tions for a successively triggered rupture
growth underlying the 2000 earthquake
swarm in Vogtland ⁄NW Bohemia.
J. Geophys. Res.,107, B122338;
doi:10.1029/2002JB001865.
Hill, D.P., Ellsworth, W.L., Johnston,
M.J.S., Langbein, J.O., Oppenheimer,
D.H., Pitt, A.M., Reasenberg, P.A.,
Sorey, M.L. and McNutt, S.R., 1990.
The 1989 earthquake swarm beneath
Mammoth Mountain, California: an
initial look at the 4 May through 30
September activity. Bull. Seismol. Soc.
Am.,80, 325–339.
Husen, S., Bachmann, C. and Giardini, D.,
2007. Locally triggered seismicity in the
central Swiss Alps following the large
rainfall event of August 2005. Geophys.
J. Int.,171, 1126–1134.
Jenatton, L., Guiguet, R., Thouvenot, F.
and Daix, N., 2007. The 16,000-event
2003–2004 earthquake swarm in Ubaye
(French Alps). J. Geophys. Res.,112,
B111304; doi:10.1029/2006JB004878.
Johnson, C.E. and Hadley, D.M., 1976.
Tectonic implications of the Brawley
earthquake swarm, Imperial Valley,
California, January 1975. Bull. Seismol.
Soc. Am.,66, 1133–1144.
Klein, F.W., Einarsson, P. and Wyss, M.,
1977. Reykjanes Peninsula, Iceland,
earthquake swarm of September 1972
and its tectonic significance. J. Geophys.
Res.,82, 865–888.
Knett, J., 1899. Das Erzgebirgische
Schwarmbeben zu Hartenberg vom 1.
Ja
¨nner bis Feber 1824. Sitzungsber.
Deutsch. Naturwiss.-Med. Ver. Bo
¨hmen,
Lotos Prag N.F.,19, 167–191.
Kraft, T., Wassermann, J., Schmedes, E.
and Igel, H., 2006. Meteorological trig-
gering of earthquake swarms at Mt.
Hochstaufen, SE-Germany. Tectono-
physics,424, 245–258.
Lee, W.H.K. and Lahr, J.E., 1975.
HYPO
71:
a computer program for determining
hypocenter, magnitude, and first-motion
pattern of local earthquakes. U.S. Geol.
Surv. Open-File Rept., 75-331,
p. 110.
Lohman, R.B. and McGuire, J.J., 2007.
Earthquake swarms driven by aseismic
creep in the Salton Trough, California.
J. Geophys. Res.,112, B04405;
doi:10.1029/2006JB004596.
Mari, J.-L., 1977. Sismique Hyperprofonde.
Me
´m. Dipl. Ing.-Ge
´ophys., Inst. Phys.,
Globe Strasbourg.
Miller, S.A., 2008. Note on rain-triggered
earthquakes and their dependence on
karst geology. Geophys. J. Int.,173,
334–338.
Phillips, W.S., 2000. Precise microearth-
quake locations and fluid flow in the
geothermal reservoir at Soultz-Sous-
Foreˆ ts, France. Bull. Seismol. Soc. Am.,
90, 212–228.
Revol, A., 1773. Relation des Tremble-
ments de Terre Principalement Ressentis
a
`Clansayes. Arch. Mun. Clansayes.,10
Mai 1773.
Terra Nova, Vol 21, No. 3, 203–210 F. Thouvenot
et al.
•200-m-deep earthquake swarm in Tricastin (France)
.............................................................................................................................................................
2009 Blackwell Publishing Ltd 209
Rigo, A., Be
´thoux, N., Masson, F. and
Ritz, J.-F., 2008. Seismicity rate and
wave–velocity variations as conse-
quences of rainfall: the case of the
catastrophic storm of September 2002
in the Nıˆ mes Fault region (Gard,
France). Geophys. J. Int.,173,
473–482.
Rothe
´, J.-P., 1936. Les tremblements de
terre en France en 1934. Ann. Inst. Phys.
Globe 1934,2, 88–110.
Rothe
´, J.-P., 1939a. Les tremblements de
terre en France en 1935 (suite) et en
1936. Ann. Inst. Phys. Globe 1936,2,
84–123.
Rothe
´, J.-P., 1939b. Les secousses se
´ismi-
ques du Tricastin. Sur les me
´thodes de
de
´termination de la profondeur du foyer.
Ann. Inst. Phys. Globe 1936,1-2,
134–141.
Rothe
´, J.-P., 1941. Les se
´ismes des Alpes
franc¸ aises en 1938 et la se
´ismicite
´des
Alpes Occidentales. Ann. Inst. Phys.
Globe 1938,3-3, 1–100.
Saar, M.O. and Manga, M., 2003.
Seismicity induced by seasonal
groundwater recharge at Mt. Hood,
Oregon. Earth Planet. Sci. Lett.,214,
605–618.
Service de la Carte ge
´ologique de la
France, 1964. Carte Ge
´ologique au 1 ⁄50
000e Valre
´as (XXX-39). Bur. Rech.
Ge
´ol. Minie
`res, Orle
´ans.
Service de la Carte ge
´ologique de la
France, 1969. Carte Ge
´ologique de la
France au 1 ⁄1 000 000 (2 sheets). Bur.
Rech. Ge
´ol. Minie
`res, Orle
´ans.
Service de la Carte ge
´ologique de la
France, 1975. Carte Ge
´ologique au 1 ⁄50
000e Nyons (XXXI-39), Bur. Rech.
Ge
´ol. Minie
`res, Orle
´ans.
S
ˇpic
ˇa
´k, A., 2000. Earthquake swarms and
accompanying phenomena in intraplate
regions: a review. Studia Geoph. Geod.,
44, 89–106.
Thouvenot, F. and Bouchon, M., 2008.
What is the lowest magnitude threshold
at which an earthquake can be felt or
heard, or objects thrown into the air? In:
Historical Seismology: Interdisciplinary
Studies of Past and Recent Earthquakes
(J. Fre
´chet, M. Meghraoui and M.
Stucchi, eds), pp. 313–326. Springer,
Dordrecht.
Thouvenot, F. and Fre
´chet, J., 2006.
Seismicity along the north-western edge
of the Adria microplate. In: The Adria
Microplate: GPS Geodesy, Tectonics, and
Hazards (N. Pinter, G. Grenerczy,
J. Weber, S. Stein and D. Medak, eds),
pp. 335–349. Springer, Dordrecht.
Utsu, T., 1966. A statistical significance test
of the difference in b-value between two
earthquake groups. J. Phys. Earth,14,
37–40.
Waldhauser, F., 2001.
HYPODD
– a program
to compute double-difference hypocenter
locations. U.S. Geol. Surv. Open-File
Rept., 01-113, p. 25.
Waldhauser, F. and Ellsworth, W.L., 2000.
A double-difference earthquake location
algorithm: method and application to the
Northern Hayward Fault, California.
Bull. Seismol. Soc. Am.,90, 1353–1368.
Wessel, P. and Smith, W.H.F., 1998. New,
improved version of Generic Mapping
Tools released. EOS Trans AGU,79,
579.
Zobin, V.M., Nishimura, Y. and Miyam-
ura, J., 2005. The nature of volcanic
earthquake swarm preceding the 2000
flank eruption at Usu Volcano,
Hokkaido, Japan. Geophys. J. Int.,163,
265–275.
Received 11 July 2008; revised version
accepted 16 March 2009
200-m-deep earthquake swarm in Tricastin (France) •F. Thouvenot
et al.
Terra Nova, Vol 21, No. 3, 203–210
.............................................................................................................................................................
210 2009 Blackwell Publishing Ltd
