You are on page 1of 27

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/258512500

Updated Seismic Hazard Assessment of Tunisia

Article  in  Bulletin of Earthquake Engineering · October 2013


DOI: 10.1007/s10518-013-9548-y

CITATIONS READS

9 3,354

2 authors:

Ahmed Ksentini Najla Bouden Romdhane


École Nationale d'Ingénieurs de Tunis University of Tunis El Manar
7 PUBLICATIONS   76 CITATIONS    15 PUBLICATIONS   198 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Fractured Rock Mass Characterization and Applications View project

All content following this page was uploaded by Ahmed Ksentini on 13 July 2015.

The user has requested enhancement of the downloaded file.


Updated seismic hazard assessment of
Tunisia

Ahmed Ksentini & Najla Bouden


Romdhane

Bulletin of Earthquake Engineering


Official Publication of the European
Association for Earthquake Engineering

ISSN 1570-761X
Volume 12
Number 2

Bull Earthquake Eng (2014) 12:647-670


DOI 10.1007/s10518-013-9548-y

1 23
Your article is protected by copyright and all
rights are held exclusively by Springer Science
+Business Media Dordrecht. This e-offprint
is for personal use only and shall not be self-
archived in electronic repositories. If you wish
to self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.

1 23
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670
DOI 10.1007/s10518-013-9548-y

ORIGINAL RESEARCH PAPER

Updated seismic hazard assessment of Tunisia

Ahmed Ksentini · Najla Bouden Romdhane

Received: 22 September 2012 / Accepted: 29 October 2013 / Published online: 7 November 2013
© Springer Science+Business Media Dordrecht 2013

Abstract Modern earthquake loss models make use of earthquake catalogs relevant to the
seismic hazard assessment upon seismicity and seismotectonic analysis. The main objective
of this paper is to investigate a recently compiled catalog (National Institute of Meteorology or
INM catalog: 412-2011) and to generate seismic hazard maps through classical probabilistic
seismic hazard assessment (PSHA) and smoothed-gridded seismicity models for Tunisia. It
is now established with the local earthquake bulletin that the recent seismicity of Tunisia
is sparse and moderate. Therefore, efforts must be undertaken to elaborate a robust hazard
analysis for risk assessment and seismic design purposes. These recommendations follow
the recently published reports by the World Bank that describe the seismic risk in Tunis City
as being beyond a tolerable level with an MSK intensity level of VII. Some attempts were
made during the past two decades to assess the seismic hazard for Tunisia and they have
mostly failed to properly investigate the historical and instrumental seismicity catalog. This
limitation also exists for the key aspect of epistemic and random uncertainties impact on the
final seismic hazard assessment. This study also investigates new ground motion prediction
equations suitable for use in Tunisia. The methodology applied herein uses, for the first time
in PSHA of Tunisia, seismicity parameters integrated in logic tree framework to capture
epistemic uncertainties through three different seismic source models. It also makes use of
the recently released version of OpenQuake engine; an open-source tool for seismic hazard
and risk assessment developed in the framework of the Global Earthquake Model.

Keywords PSHA · Smoothed seismicity · Uncertainties · Seismic catalog · Logic tree ·


Ground motion maps

A. Ksentini (B) · N. B. Romdhane


Département de génie civil, Ecole Nationale d’Ingénieurs de Tunis, Université Tunis El Manar,
BP 37 le Belvédère, Tunis, Tunisia
e-mail: ahmedksentini@yahoo.fr

123
Author's personal copy
648 Bull Earthquake Eng (2014) 12:647–670

Abbreviations

EMSC Euro-Mediterranean Seismological Centre


FMD Frequency magnitude distribution
GEM Global Earthquake Model
GMPE Ground motion predictive equation
GR Gutenberg Richter
GSHAP Global seismic hazard assessment program
INM National Institute of Meteorology
ISC International Seismological Centre
NGA Next generation of attenuation
PSA Pseudo spectral acceleration
PSHA Probabilistic seismic hazard assessment

1 Introduction

The seismicity of Tunisia is considered moderate when compared to the neighbour countries
(e.g., Algeria, Italy). Most instrumental magnitudes fall into the range between 2.0 and 5.5
according to the local seismic bulletin. Several major earthquakes occurred in Tunisia in the
past with epicentral intensities of VIII to X on the MSK scale. Those events were described by
several authors (Sieberg 1932; Ambraseys 1962; Rothé 1970; Vogt 1993; Mejri et al. 2010).
The most important ones occurred in Utique region AD 412 (with cracks in the ground and
aftershocks for one week), Kairouan AD 854 (13 villages destroyed), in 856 in Tunis (45,000
victims) and in 1758 in Tunis (great number of houses destroyed, thousands of victims under
debris).
Accounting for this seismicity level, the first seismic hazard assessment of Tunisia was
initiated in early 1990 (Ben Ayed and Zargouni 1990). In parallel, the seismic network of the
National Institute of Meteorology (INM) was in phase of update. First drafts of the proba-
bilistic seismic hazard calculation and mapping (PSHA) were elaborated by the INM-IZIIS
(The Institute of Earthquake Engineering and Engineering Seismology, Skopje) working
group “Investigation for elaboration of seismotectonic map and draft seismic design code
of Tunisia” (Jordanovsky 1991). Other studies have been developed across the larger North
Africa and Ibero-Maghreb regions (e.g. Benouar et al. 1996 and Jiménez et al. 2001). Early
PSHA results were combined with other studies related to the seismic hazard and seismotec-
tonic contexts in the framework of the Global Seismic Hazard Assessment Program (GSHAP)
(1999); see also H’faiedh and Allouche (1993); Ben Ayed (1993); Vogt (1993), and the INM
catalog (412-1991) to calculate the Peak Ground Acceleration (PGA) on bedrock for dif-
ferent return periods. Since then, few projects have been undertaken to collect new data,
analyze and map the seismic hazard in Tunisia (Ksentini 2004; Dlala and Kacem 2009). The
most recent study to assess the seismic hazard and risk in Tunis City was conducted by the
World Bank (World Bank/Egis Bceom International/IAU-IDF/BRGM 2011). In this study,
the seismic hazard is evaluated through geological and geotechnical investigations and the
intensity is estimated to VII on MSK scale over 475 years return period.
However, previous studies to assess the seismic hazard for Tunisia were mostly remote,
with undefined origin and unclear database for deliverables. No unifying project and studies
helped to provide homogeneous and harmonized results to assess the seismic risk level, and
all proposed probabilistic models have been relatively “limited”, especially for the seismic
source models, ground motions prediction equations, site conditions and related uncertainties.

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 649

In order to develop a realistic and sound probabilistic seismic hazard and risk assessment
in Tunisia, for about ten years, the civil engineering department of the National School
of Engineers of Tunis, in collaboration with the Tunisian Enterprise of Petroleum Activities
(ETAP), the French Central Laboratory of Roads and Bridges (LCPC), the Center of Technical
Studies and Equipment of Nice (CETE, France), the department of Applied Geophysics (Paris
VI, France) and the North African Group for Earthquake and Tsunami studies (NAGET),
undertakes a hazard and risk assessment project named “RISKTUN”. Through this project,
different components of the seismic risk are investigated such as regional hazard, site effect,
liquefaction, building stock vulnerability, etc.; in this context, the basin of Tunis was retained
as a pilot region mainly due to its specific geological and geotechnical aspects leading to a
significant level of amplification in addition to its urban situation (Romdhane et al. 2000;
Romdhane and Mechler 2002).
In this paper, new investigations are presented in the frame of the pilot project. They include
the analysis of a newly compiled earthquake catalog, a detailed characterization of seismic
source models and a selection of appropriate ground motion prediction equations (hereafter
referred to as GMPEs). In this work, the aim is also to estimate the lack of data necessary for
the re-evaluation of the seismic hazard for risk analysis through the key approach of logic
trees and epistemic and random uncertainties treatment (SSHAC 1997).

2 Methodology

Recent developments in the seismic hazard assessment field were applied to provide spectral
accelerations for short and long shaking periods (0–3 s according to the buildings height
range in Tunisia). In particular, the methodology applied in the US-National Seismic Hazard
Mapping Project (NSHMP) (Petersen et al. 2008) is selected here, and a classical approach
is adopted to compile, homogenize and decluster a local earthquake catalog covering the
region 32.5–37.5◦ N, 7.5–12◦ E. In addition, random errors (b values analysis, magnitudes of
complete reporting, time spans, seismicity rate changes) and epistemic uncertainties (from
magnitude conversion, smoothed-gridded seismicity models, recent GMPEs and previously
elaborated uniform source model) are estimated. The adopted source model was proposed
by the INM-IZIIS working group (Jordanovsky 1991). It is characterized by seven areal
sources which have been delineated according to seismotectonic considerations. This model
is compared to smoothed gridded seismicity model to meet the variability in ground motion
levels and spatial distribution. The second model allows characterizing annual activity of
different magnitudes (Mwmin = 4.0–Mwmax = 6.5) based on earthquake catalog analysis
and periods of complete reporting. For ground motion calculation, two recent ground motion
prediction equations (derived for shallow crustal seismicity) are adopted to calculate the
acceleration attenuation by magnitude and distances and for the adopted spectral periods.
Finally, related uncertainties are integrated in the probabilistic seismic hazard assessment
(PSHA) using a formulated logic tree to give a weighted hazard level in agreement with
the studied parameters. This approach is original since it is the primary study in which the
seismic hazard assessment of Tunisia is based on a logic tree framework which accounts for
three seismic source models.

3 Major seismotectonic structures of Tunisia

The tectonic map of Tunisia (inspired from Bouaziz et al. 2002) is shown in Fig. 1. With
respect to its pattern, the Tunisia region is divided into four major geographic zones taking

123
Author's personal copy
650 Bull Earthquake Eng (2014) 12:647–670

Fig. 1 Structural map of Tunisia (inspired from Bouaziz et al. 2002) and major geographic domains of Tunisia
(Peña and Abdelsalam 2006)

into account their seismotectonic characteristics (ETAP 2003; Peña and Abdelsalam 2006).
A brief description of these regions and the related major structures is given below in this
section.
The Tell province (northern Tunisia) results from the Cenozoic Alpine tectonic phase and
is composed of NE-SW trending fold and thrust structures with diapir zones (Bouaziz et al.
2002). This region is characterized by neogene and Quaternary conjugated EW and NS strike
slip faults associated with compressive and extensive structures (Ben Ayed 1993; Amiri et
al. 2011). The Central Region shows NE-SW major normal faults structures (Bouaziz et al.
2002). The major North-South Axis is a major tectonic structure that crosses Tunisia along

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 651

100 km from the Tunis region to the “Chotts”, the desert salt lakes to the south (Bouaziz
et al. 2002), separating the Central Region and western folded domain (Qafsah range) from
the Eastern Sahel Region (Bouaziz et al. 2002; Dhahri and Boukadi 2010). The Eastern
Region encompasses the Sahel region and nearby offshore domain and shows NW-SE and
E-W normal faults (Bouaziz et al. 2002). The Southern Region is part of the stable Sahara
Platform that includes the Dahar Plateau and the Jeffara Plain to the east (NE-SW network
of normal faults, Castany 1954).

4 Seismicity catalog

4.1 Local agency catalog

The earthquake catalog of Tunisia [AD 412-2011] is here compiled from previous works
(Jordanovsky 1991; Vogt 1993) and elaborated since 1922 by the National Meteorology
Institute (INM). The historical and instrumental seismicity distribution covers all Tunisia
except the Sahara Platform. Magnitudes of historical events (pre 1976) are derived from the
INM catalog and related regression equation M = 0.6 ∗ I0 + 0.78 (H’faiedh and Allouche
1993), where I0 is the epicentral intensity and M is the reported magnitude. However, the
magnitude type, magnitude standard deviation and uncertainties of earthquake location are
not specified in the INM catalog.
From 1926 until 1976, the INM used a single station with two horizontal components
to monitor the instrumental seismicity. The seismic network was subsequently developed
with six telemetred stations from 1976 to 1985 (Attafi et al. 1997). During the latter period,
the magnitude was first manually calculated according to seismogram duration amplitude.
Later on, duration magnitude (MD) was automatically calculated through USGS location
methods and magnitudes processing programs (i.e., Hypoinverse suite). The last upgrade
of the network was in 2006. Magnitudes and location are transmitted in real time with fur-
ther processing in collaboration with the Euro-Mediterranean Seismological Centre database
(EMSC; http://www.emsc-csem.org). The duration magnitude for local events was calibrated
through the classical formula MD = logt + a + c.log D, where t is the duration of the signal
in seconds, D is the epicentral distance in km and a and c are calibrating parameters that
have been derived for every station. This duration (or coda) magnitude is the most common
magnitude used by short period networks in the Euro-Mediterranean area and it can be con-
sidered for compilation into uniform size databases for seismic hazard assessment (Gardini
et al. 1997). Presently, the Tunisian SBS seismic network (cf: Table 1; Fig. 2) operated by
INM, is composed of 13 seismological mono-component, short period stations (analogic
transmission UHF) and 5 digital (3 components) seismic stations around Thala subcentre
(EMSC Newsletters, May 2007, No 22).

4.2 ISC and EMSC catalogs

In addition to the INM catalog (Fig. 3a), two other datasets are taken into account to compile
the final earthquake catalog used for ground motion computation and within the geographical
window Longitude: 6.5◦ E–13.0◦ E, Latitude: 32◦ N–38.5◦ N; Fig. 3b, c, covering the whole
seismic study area: the Tunisian territory and sites at 300 km around. The first dataset was
obtained from the International Seismological Centre (ISC) bulletin [from 1965 to 2011]
which was downloaded from the ISC website. The second catalog was provided for this
study by the Euro Mediterranean Seismological Centre (EMSC) [from 1998 to 2011]. The

123
Author's personal copy
652 Bull Earthquake Eng (2014) 12:647–670

Table 1 Tunisian SBS seismic network stations


Code Latitude Longitude Elevation Station name

BERT 34.2410 9.0118 320.0 Berda


BLIT 36.7130 8.9528 225.0 Baltah
BTHT 35.1323 10.2852 240.0 Jabal bu Thady
GAF 34.4167 8.8000 314.0 Gafsa (1976)
GFA 34.3382 9.7265 250.0 Gafsa (1989–1999)
GHAT 36.4958 8.3048 400.0 Ghardimaou
KCHT 37.1100 9.9400 565.0 Kechabta
KRIT 36.3383 9.0748 640.0 Krib
MART 33.5768 10.2495 140.0 Marith
MBZ 36.6800 10.6700 220.0 Menzel Bouzelfa
MEDT 34.1082 9.9203 90.0 Meda
OAR 34.5317 8.3968 475.0 Oum el Arais
SBS 36.8700 10.3500 125.0 Sidi-Bou-Said
SGNT 33.6608 9.8702 90.0 Sidi Gnaou
SYA 34.7418 8.8070 550.0 Sidi Yaiche
THA 35.5603 8.6997 1080.0 Thala (1977)
TROT 35.5617 9.6002 900.0 Trozza
ZGN 36.3700 10.1000 720.0 Zaghouan
http://seisan.ird.nc/USGS/mirror/neic.usgs.gov/neis/station_book/SBS_NETWORK.html

reported magnitudes in both catalogs are: mL, mb, MS, Mw , and MD, as reported by two or
more agencies (mainly: United States Geological Survey: USGS, International Seismological
Centre: ISC, Euro-Mediterranean Seismological Centre: EMSC, National Earthquake Infor-
mation Center: NEIC, Italian Rome station: ROM, French LDG network, CRAAG, Algeria);
they provide in most cases more than one magnitude type for the same events (about 400
events are reported by more than two agencies and with more than two magnitude types).
Additional earthquake parameters like relocated epicenters and depths are also provided by
ISC and EMSC catalogs.
Time-Magnitude distribution is provided for every catalog in Fig. 4 (9 sparse events from
410 to 1850 in the INM bulletin in Fig. 4a are not represented for more clarity in figures). It
shows that INM dataset (2,324 events) contains sparse events until 1976, which is in accor-
dance with the network update that was performed in the same date as stated in the foregoing.
Until this date, the INM catalog reported the epicentral intensity and the derived magnitude
for historical events. The intensity level was included in the catalog through previous study
(Vogt 1993) performed by INM. Since 1976, the network allowed detecting low magnitudes
(lower than 2.5) and since 2006, the second update of the network, the magnitude distribution
over different magnitudes levels has become more precise and homogeneous. Since 1965,
the ISC catalog (Fig. 4b) has contained only instrumental magnitudes (358 events) rising
from M3.0 to M6.0. This catalog gives more than one magnitude type for all events and
it has been found that mb is the most frequent magnitude for the Tunisian territory in the
ISC catalog. The third catalog, EMSC [1998–2012], was provided within the context of this
study (Fig. 4c). Complementary events were downloaded from the EMSC website in order
to compile the full EMSC catalog to cover the seismic study area. The reported magnitude

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 653

Fig. 2 Seismic network for Tunisia

in this dataset is almost MD given the fact that it has been transmitted from INM since 2006.
111 events are reported in this bulletin with magnitude rising from M3.0 to M5.0.

4.3 Magnitude homogenization and catalog declustering

The use of catalogs in seismic hazard assessment needs an essential task dealing with mag-
nitude homogenization in the compiled dataset. In this case, the moment magnitude Mw is
the most preferred earthquake parameter scale in hazard assessment practice because it does
not saturate for large events (Hanks and Kanamori 1979).
While this paper focuses on the PSHA for Tunisia, it also uses recent ground motion
prediction equations which were calibrated using Mw (Akkar and Bommer 2010; Chiou
and Youngs 2008). In this section, an attempt for deriving a magnitude conversion equa-
tion is presented through linear fitting between ISC mb reported magnitudes and INM MD
magnitudes (Fig. 5). This work is based on a detailed manual investigation on the common
events between the two catalogs. Every earthquake in the ISC catalog is searched in the INM
database (according to the occurrence date, the magnitude and the location). 199 events are
consequently used to fit the data. The resulting equation to convert MD into mb is given
below:
mb (ISC) = 1.92 (±0.25) + 0.56 (±0.055) ∗ MD(INM),
Pear son  s R = 0.58, Ad justed R-Squar e = 0.34

123
Author's personal copy
654 Bull Earthquake Eng (2014) 12:647–670

Fig. 3 a INM catalog [412-2011], b ISC earthquake catalog [1965–2011], c EMSC earthquake catalog
[1998–2011]

This equation is characterized by a low R-square value showing that it does not fit enough
the data used for the regression. The resulting regular residuals reach 1 degree. Consequently,
the uncertainty is higher especially when considering the aggravating factor of a double stage
conversion through MD-mb and mb-Mw relationships. This led to the consideration of eight
other peer-reviewed equations derived by many authors from countries around the Euro-
Mediterranean region (Yenier et al. 2008; Ulusay et al. 2004; Grünthal and Wahlström 2012;
Deniz 2006; Popescu et al. 2003; Hamdache et al. 2010; Hussein et al. 2008; Pelaez et
al. 2007). Those equations consider one or two staged conversions from MD to Mw . The
difference among them is represented in Table 2 and Fig. 6.
The weak correlation factors in the regression made between INM and ISC catalog
is shown by Fig. 6; it is biased vis-à-vis the range of other relationships. According to
this figure, the regression overestimates and underestimates magnitudes below and above
M5.0, respectively. At this phase of the study, it was judged that considering only one
of these conversion relationships (based on one or two stages) would be related to high

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 655

Fig. 4 Time-magnitude distribution for: a INM original catalog [410-2011], 2324 events (magnitude type:
MD), b ISC catalog [1965–2012], 358 events (magnitude type: mb) and c EMSC catalog [1998–2012], 111
events (magnitude type: mb and ML)

Fig. 5 Magnitude fitting of ISC catalog with INM dataset. Left linear fitting, right regular residuals

123
Author's personal copy
656 Bull Earthquake Eng (2014) 12:647–670

Table 2 Peer-reviewed MD-Mw conversion relationships for the Mediterranean region

Authors Equations Region of interest Number of


stages

Yenier et al. (2008) Mw = 0.764 ∗ MD + 1.379 Turkey 1


Ulusay et al. (2004) Mw = 0.9594 ∗ MD + 0.4181 Turkey 1
Grünthal and Wahlström Mw = 1.472 ∗ MD − 1.49 Euro-Mediterranean 1
(2012)
Deniz (2006) Mw = 1.3 ∗ MD − 1.3 Turkey 1
Popescu et al. (2003) Mw = 0.8 ∗ MD + 0.3 Romania 1
Hamdache et al. (2010) Mw = MD Algeria 1
Hussein et al. (2008) Mw = 0.667 ∗ M0 − 10.7; M0 = Egypt 2
1.45 ∗ MD + 16.3
Pelaez et al. (2007) Mw = 0.311+0.637∗mblg+0.061∗ Morocco 2
mblg2 mblg = 0.91 ∗ MD + 0.32

Fig. 6 Duration magnitude-Moment magnitude conversion relationships (M3.0–M6.0)

uncertainty. The procedure followed was to consider different conversion relationships (this
regression equation, Yenier et al. 2008, one stage and Hamdache et al. 2010, one stage)
in a logic tree framework to capture this uncertainty. Three catalogs were then prepared
using each relationship to convert the MD magnitude into Mw . These catalogs were used
to derive the correspondent seismicity parameters which will be related to the area source
model.
As for the intensity-magnitude conversion, the relationship of Mezcua (2002) was adopted:
Mw = 0.6.I0 +0.96. This formula was calibrated through data from the Ibero-Maghreb region
and was widely adopted in the region (Hamdache et al. 2010; Pelaez et al. 2007; Giardini
et al. 1999). Finally, the adopted catalog (Fig. 7a) is compiled considering the following
steps:

(1) Common events between the three catalogs are manually investigated considering that
they occur in the same day, with a magnitude difference of one degree and falling into a
geographic radius of 0.5 decimal degrees.
(2) The main catalog is subsequently done by (a) ISC events after removing double events
from EMSC and INM catalogs, (b) ISC additional events, (c) EMSC events after remov-

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 657

Fig. 7 Instrumental and historical seismic events of Tunisia: a compiled catalog, b declustered catalog using
Gardner and Knopoff (1974)

ing double events from INM catalog, (d) EMSC additional events and (e) INM additional
evens.
(3) Duration magnitudes of INM catalog are converted into moment magnitudes through the
three retained conversion relationships.
(4) Magnitudes from ISC and EMSC catalogs are converted into Mw through global regres-
sion relationships of Scordilis (2006).
(5) INM epicentral intensity-moment magnitude conversion is made through Mezcua (2002)
relationship.
Another important consideration in PSHA is the assumption of Poissonian process for tem-
poral distribution of seismicity. Consequently, we used two standard programs of Reasenberg
(1985), and Gardner and Knopoff (1974) to decluster the catalog. More than 600 events (out
of 2,600 events of the total catalog) were identified as dependent events in clusters and were
removed from the final catalog (Fig. 7b). In this study, different time-space windows para-
meters were adopted to decluster the final catalog: Gardner and Knopoff (1974, California),
Grünthal (personal communication, Europe) and Uhrhammer (1986).
In order to consider an appropriate declustering algorithm, the two programs were applied
to the INM original catalog in order to see their behavior. Table 3 shows the number of clusters
as identified by Reasenberg (1985) and Gardner and Knopoff (1974) algorithms. These
numbers are compared to the real number of clusters as identified by INM investigations and
to main shock magnitudes greater than MD 4.5. It was established that the Reasenberg (1985)
algorithm, when applied with default parameters, resulted in less identified aftershocks.
This was observed for all retained events. For Gardner and Knopoff (1974) algorithm, all
aftershocks were systematically removed except for one event. At this stage, the latter is
adopted to decluster the compiled catalog.

5 Data quality in the compiled catalog

A careful estimate of the data quality is performed in this study in order to describe the
annual probabilities of occurrence and to verify the catalog completeness. Figure 8 shows

123
Author's personal copy
658 Bull Earthquake Eng (2014) 12:647–670

Table 3 Main shocks and aftershocks as identified by Reasenberg (1985) and Gardner and Knopoff (1974)

N Month(s)/year Zone Period Magnitude Known Algorithms identified clusters


(days) (MD) clusters
Reasenberg Gardner and Knopoff
(1985) (1974)

1 01/1977 Dkhila 10 5.0 13 8 13


2 02/1977 Menzel Nour 3 5.0 12 9 12
3 04/1979 Ras Jbal 15 5.0 9 4 9
4 06–07/1987 Mjez El Bab Sparse 5.0 6 0 6
5 01/1989 Offshore 11 4.5 15 7 15
6 06/1992 Mdhila 2 4.9 7 4 7
7 09/1994 Thibar 2 4.7 6 3 6
8 01/2002 Chebba 1 4.7 4 1 4
9 06/2002 Kairouan 1 4.6 2 1 1

Fig. 8 Seismicity rate changes by magnitude classes [Mw 3.5–6.5]: The catalog reports in 1976 an increasing
seismicity due to its updates (Attafi et al. 1997). After 1975, only events with M < 5.0 present a clear rate
changes. Dashed lines represent the last slope in rate changes that describes the completeness period for
magnitude intervals

the cumulative number of the events distribution by time. This plot is given for different
magnitude classes (±0.25). It allows studying the complete time periods above certain mag-
nitudes. These intervals are identified when linear behavior is observed at the end of every
curve. This study clearly reveals rate changes especially for low magnitudes in the period
from 1976 to 2011.
Another method was adopted to verify the completeness of the catalog: (Stepp 1972).
Here, the standard deviation σλ in the mean rate of occurrence of earthquake λ is cal-
culated and plotted for each magnitude class and for the corresponding time subinter-
vals T (see Fig. 9). This calculation was made with the Openquake-Hazard modeler’s
toolkit (Weatherill et al. 2012) which is a part of the OpenQuake engine (Silva et
al. 2013) developed in the framework of the Global Earthquake Model √ (GEM) (Pinho
2012). In this method, the stationarity expects that σλ behaves as 1/ T (which is

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 659

Fig. 9 Standard deviation in the mean rate of occurrence by time period subintervals and by magnitude classes

Table 4 Completeness periods


Magnitude class Complete since (years)
by magnitude classes
3.51–4.0 25
4.01–4.5 25
4.51–5.0 25
5.01–5.5 50
5.51–6.0 65
6.01–6.5 75

represented by the constant slopes√ in Fig. 9) in each time period subinterval. The
observed departure of σλ from 1/ T explains then the incompleteness of earthquake
reporting. Very similar results were carried out regarding the presented completeness
periods by magnitude classes in Fig. 8. The catalog completeness is summarized in
Table 4.
The annual rate of occurrence for earthquakes is described herein by the widely applied
GR model and the maximum likelihood solution (Gutenberg and Richter 1954) that present
the magnitude distribution per seismic source. In order to perform such representations, all
magnitudes are reported and found valid for the whole catalog. The threshold magnitude
for completed reports is Mc = 4.1 and the b value is 0.81 (±0.03 bootstrapped uncertainty,
see Fig. 10).
Since the catalog is not homogeneous over the whole Tunisian territory, the completeness
analysis was made for each source. For every catalog (depending on magnitude conversion
relationships), the magnitude of complete reporting, b value and a value are computed and
integrated within the area source model which was defined under Openquake Engine (Silva
et al. 2013) to prepare for the logic tree computation as described in Sect. 8. The b values
were calculated for every source using the Maximum Likelihood Estimation (MLE) such
as described in (Bollinger et al. 1993) and (Weichert 1980). The calculation was made with
Z-MAP software (Wiemer and Wyss 1994).

123
Author's personal copy
660 Bull Earthquake Eng (2014) 12:647–670

Fig. 10 Doubly truncated Gutenberg Richter curve for the compiled catalog. Mc is threshold magnitude
for completed reports and studies. Triangles represent the number of earthquakes and squares represent the
cumulative number

Fig. 11 Seismic models for Tunisia: a simplified zonal model of Jordanovsky (1991); b Smoothed-gridded
seismicity: activity rates ∗ 10−3 for Mw 5.0 normalizing magnitude and 100 years normalizing time period
over the total length of the catalog

6 Seismic source models

In this section, the created area based model (Jordanovsky 1991) and the smoothed seismicity
model (Frankel 1995; Woo 1996a; Lapajne et al. 2003) are described. The first seismic zoning
shown in Fig. 11a was obtained in the framework of the analysis performed by the INM-IZIIS
working group (Jordanovsky 1991). It was updated in 1999 by Giardini et al. (1999), in the
Global Seismic Hazard assessment Program (GSHAP), in which the Ibero-Maghreb was a
pilot region; a more recent update was presented in the Global Earthquake Model GEM1
technical reports. Our work addresses important aspects for every source through GR curves,
added to the estimation of Mc and b values for each source zone (Table 5). In order to generate
the hazard maps according to this area source model, the OpenQuake engine (V1.0 Beta)
was used through the Classical probabilistic seismic hazard analysis (PSHA). This engine
helped defining a logic tree with three source models and different seismicity parameters

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 661

Table 5 Seismic parameters for adopted source models

Source model (magnitude conversion equation of Yenier et al. 2008)

ZONE No a value b value Mc Mw max Number of events used for the regression by
(annual) magnitude classes, by sources

[Mc-5.0] [5.01–5.5] [5.51–6.0] [6.01-Mmax] Sum (N)

1 3.45 0.9 4.1 7.0 6 4 4 2 16


2 4.3 1.01 4.1 6.5 34 3 3 0 40
3 4.61 1.04 4.2 7.0 24 4 1 1 30
4 3.97 0.90 4.5 6.5 17 9 0 0 26
5 5.69 1.24 4.5 7.0 24 12 0 1 37
6 3.99 0.98 4.5 6.0 5 1 0 0 6
7 6.13 1.59 4.3 5.5 7 0 0 0 7

(depending on the magnitude conversion relationships) to provide hazard maps that follow
the classical integration procedure of (Cornell 1968).
The diffuse and moderate seismicity of Tunisia were the most important reasons that
led us to use the smoothed-gridded seismicity as a second source model (Fig. 11b). This
model limits the use of expert judgment about zonal sources and provides more automated
computation of the seismicity rate. In order to account for the shallow seismicity of Tunisia
(Benouar 1994; Dlala and Kacem 2009; Mejri et al. 2010; Klett 2001), a two-dimensional
Kernel Smoothing method (Woo 1996a) is also used. In this approach, the spatial probability
distribution is calculated for event occurrences using the kernel smoothing technique through
Kergrid program, a generalized version of Kerfract (Woo 1996a,b). The resulting gridded
seismicity and related source model associate large events (M > 5.5) to smaller earthquakes
as reported in the compiled catalog with completeness levels derived from time distribution of
events (also described in Sect. 5). The adopted correlation distance depends on the uncertainty
in the epicenter location. For this task, Fig. 12 shows distance errors between epicenters of
the same events reported in both INM and ISC catalogs. Visual inspection of this map shows
that the average distance error is about E = 75 km. This led to consider a correlation distance
c = E/3 = 25 km as recommended by (Frankel 1995).
Another important conclusion was made about the source zone delineation. When com-
paring the distance errors to the source geometry, it was noted that the major part of uncertain
epicenters still belongs to the same area sources independently of the adopted location except
for eight event for which the distance error reaches more than 100 km. Accounting for this,
no measures were made for the impact of this uncertainty on the final estimates of the hazard
according to the area source model. These errors are contrastingly integrated in this work
through the smoothed gridded seismicity model and the correlation distance c as described
in the original method (Frankel 1995; Woo 1996a).
The third model adopted in this study consists in earthquake frequency computed through
energy release, assuming a log-linear distribution relation to magnitudes (Lapajne et al. 2003).
The expected seismicity rates are computed in this model (using OHAZ 6.0, Zabukovec et
al. 2007) only through major earthquakes (Mw > 6.0) of the compiled catalog to account for
uncertainties in reported magnitudes and locations of major earthquakes. The epicentral area
of major earthquakes is used to associate the released energy area using 50 km correlation
distance. This technique is suitable for handling old strong historical earthquakes since their

123
Author's personal copy
662 Bull Earthquake Eng (2014) 12:647–670

Fig. 12 Epicenter uncertainties described by distance error (km); according to their lengths, colored bars
represent links between epicenters of the same event reported in both the ISC and INM catalog (these events
occurring on the same day, with a separating distance <200 km and for magnitude difference <1.0). Underlined
values are event magnitudes, italic values are occurring event years

activity rate is not available through the catalog complete time span (Lapajne et al. 2003) as
in the case of Tunisia.
The maximum magnitude is a crucial parameter that can bias the hazard estimates and its
estimation is important in seismic hazard assessment. It truncates the upper bound of mag-
nitude classe that is used to carry out the seismicity parameters for every source. It can be
defined through two methodologies according (1) to the rupture length: deterministic or (2) to
the seismicity data through appropriate statistical estimation procedures: probabilistic (Kijko
2004). Given that the number of earthquakes used for the regression as presented in Table 5 is
moderate and that the regressions are done by sources, the procedure followed here is deter-
ministic. According to the fault length carried out from the structural map of Tunisia (see
Fig. 1), the maximum magnitude is calculated through the relationships of Wells and Copper-
smith (1994). This procedure takes into account the fault geological site and the fault type. In
this study, when the fault is located in a quaternary site, the maximum magnitude is calculated
through appropriate equation of Wells and Coppersmith (1994) according to the fault surface
length which is assumed in this case to be equal to the rupture length. In the other cases, the
maximum magnitude taken here is equal to the maximum observed magnitude plus 0.5◦ .

7 Ground motion predictive equations GMPEs

In this study, ground shaking is characterized by empirical ground motion models. Actu-
ally, more than 250 ground motion prediction equations (Douglas 2011) have been derived

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 663

Fig. 13 Spectral acceleration for three magnitudes (5, 6 and 7) for spectral periods from 0.0303 (PGA) to
3 s and source-site distance from 0 to 100 km according to the Akkar and Bommer (2010) ground motion
predictive equation

and updated by regression from worldwide datasets in accordance with the regional seis-
motectonic regime (i.e., stable continental, subduction zones, shallow crustal seismicity,
volcanic zones and deep focus non-subduction earthquakes). For Tunisia, no robust GMPE
was derived to characterize an instrumental-based ground motion model due to lack of strong
motion records of large earthquakes, especially for different vibration frequencies.
For this reason, it was decided here to use one New Generation of Attenuation; NGA
models (Chiou and Youngs 2008) selected on reasonable grounds and relevant for the crustal
seismicity (Stafford et al. 2008). Another important consideration taken here was to include
a recently proposed equation for Europe, Middle East and Maghreb regions (Akkar and
Bommer 2010):

log (P S A) = b1 + b2 M + b3 M 2 + (b4 + b5 M) log R 2jb + b62
+ b7 Ss + b8 S A + b9 FN + b10 FR + εσ
where SS and SA take into account the soil type (rock, stiff soil, soft soil) that is defined
through Vs30 velocity, FN and FR represent fault type, ε is the number of standard deviations,
σ (inter and intra events) is the value of standard deviation and bi are correlation constants with
different sets of constants for different vibration periods. Figure 13 highlights the variation
in the expected spectral accelerations (in units of g) for three scenarios (Mw = 5, 6 and 7)
according to several spectral periods and source-site distances.
The above equation is compared in Fig. 14 to the NGA equation and shows prediction
of median pseudo-spectral acceleration (PSA) by periods for a Mw 5.5 event (fault type:
normal, distance to rupture = 10 km and for bedrock, one standard deviation in intensity was
accounted for here). The approach taken herein is to combine this equation with the NGA
equation in a logic tree framework (as recommended in Akkar and Bommer 2010), which is
discussed in the next section. The selection of those GMPEs follows the recently published
study for Seismic Hazard Harmonization for Europe: the SHARE Project. In this project,
a selection and ranking study was performed to find the best available GMPEs for use in
Euro-Mediterranean region and for different seismotectonic regimes (Delavaud et al. 2012).
This selection is based on expert judgments, data testing and sensitivity analysis. The two
first preferred equations for Active shallow crustal regions based on non-European database
according to the SHARE project are those selected in this paper.
The afore-discussed sections in this paper allow analyzing in details several uncertainties
that arise from the seismic hazard assessment for Tunisia. To go forward, many methodologies
and softwares that are proposed to perform PSHA are investigated. In this work, three seismic

123
Author's personal copy
664 Bull Earthquake Eng (2014) 12:647–670

Fig. 14 Ground Motion Predictive equations: mean acceleration and standard deviations in Chiou and Youngs
(2008) and Akkar and Bommer (2010) (Mw = 5.5, D = 10 km, normal fault and for rock site)

source models (areal sources, smoothed gridded seismicity and energy based seismicity) are
adopted to account for epistemic uncertainties related to the source activities and depths,
spatial distribution of major structures, data quality in the compiled catalog, magnitudes
scales and completeness periods, and prediction of ground motion. At this stage, a logic
tree approach is, as described in the next section, adopted so as to capture epistemic errors
accounting for many reliable scenarios.

8 Mapping probabilities of exceedance for PGA using logic tree approach

A practical tool in seismic hazard assessment is the logic tree (Kulkarni et al. 1984) that
incorporates the computed epistemic potential errors, especially for the basic components of
seismicity model and ground motion prediction (Fig. 15). Implementing a probability model
of uncertainties implies that each branch of the logic tree is investigated to infer a reliable
model with different degrees of confidence. Some important considerations in associating
weights to the tree branches are addressed hereafter:
– The seismic source models: After selecting a clustered or declustered homogeneous cata-
log, the second node of the tree splits into three retained seismic models discussed in Sect. 6.
An equal weight of 40 % was considered for the source zone model of Jordanovsky (1991)
and the smoothed-gridded seismicity, and 20 % for the energy based model. The adopted
weights are selected according to the reliability of the source model. For the energy based
model, the computed acceleration accounts only for some major earthquakes (which epi-
centers are uncertain) that cannot represent the whole seismicity of Tunisia, and therefore
results in a lower probability value than other models.
– The magnitude homogenization of the catalog: as described in Sect. 4, three conversion
methods were adopted to generate the seismicity parameters. This led to obtain three
catalogs with different completeness periods, maximum magnitude, b value and a-value.
In order to account for this uncertainty, the three obtained catalogs are applied to the three
source models (zone model, smoothed seismicity model, and energy model) in the logic
tree under OpenQuake engine with different weights: 20 % for the regression equation

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 665

Fig. 15 Formulated PSHA logic tree

proposed in this study, 60 % for the equation of Yenier et al. (2008) and 20 % for the
simplest model; Mw = MD.
– The ground motion predictive equation: the used equations are those of Akkar and Bommer
(2010) and Chiou and Youngs (2008). An equal weight (0.50) was assigned to the two
GMPEs. The first equation accounts for data from Maghreb region, the second model is
used herein given its validity at long distances (more than 100 km).

Using the above described logic tree, hazard calculation was carried out to produce maps
for Peak Ground Acceleration (PGA), Peak Ground Velocity (PGV) and Spectral Accelera-
tions at 0.3 and 1.0 s (SA0.3 and SA1.0). In Fig. 16a–c are presented the PGA maps according
to three source models, and in Fig. 16d is presented the combined PGA, for a return period
of T = 475 years. The PGA maps are fundamental in the seismic hazard module of modern
earthquake risk assessment methodologies. This allows the definition of the elastic spectrum
at any site for a region wide scenario such as recommended in Hazus’99 (National Institute
of building sciences 1999). The combination of these maps with local soil maps results in
spectral acceleration at the surface for use in seismic risk assessment for the Tunis basin in
further studies.

9 Discussion

The three adopted methodologies to assess the seismic hazard of Tunisia resulted in dif-
ferent Peak Ground Acceleration maps according to three different seismic source models.
A maximum PGA of 0.36g resulted in these models almost related to the northern region.
Quantitative comparison between the three PGA maps shows that the seismicity is moderate,
with acceleration levels ranging from 0.10 g to 0.36 g. These levels are related to the major
seismotectonic regions described in the foregoing; mainly the northern, central and east-
ern regions and the Qafsah region in the south-west. The comparison also shows that those

123
Author's personal copy
666 Bull Earthquake Eng (2014) 12:647–670

Fig. 16 Peak Ground Accelerations for T = 475 years (in g’s): a Cornell (1968) classical zonal method, b
Woo (1996a) smoothed-gridded seismicity model, c Lapajne et al. (2003) energy based model, d the logic tree
PGA map

models are complementary in the way that every one accounts for a specific seismo-tectonic
parameters and makes use of the seismicity in a specific manner.
This study shows also that the spatial distribution of these levels of acceleration depends
on the source model used for ground motion computation. While for the energy based model,
the acceleration is concentrated around major earthquake epicenters, the two other models
resulted in a more uniform distribution of accelerations over the mentioned faults in the
Fig. 1. Here, a visual inspection of Figs. 1 and 16b allows considering that the smoothed-
gridded seismicity based model is in quite similarity with the seismotectonic context of
Tunisia. This is particularly noticed for the south eastern region of Qafsah (0.28 g) and the
central region (0.16 g); the latter is characterized by major normal faults. The major north
south axis structure is characterized by an acceleration level of 0.2 g in the area source model
(Mwmax = 7.0). The smoothed-gridded seismicity model resulted in 0.22–0.26 g for the same
structure.

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 667

The difference between the three maps is clear. For the energy based model, the comparison
is not possible since it accounts only for major earthquakes to integrate their effects in the
final combined map. In contrast, the two first maps (Fig. 16a, b) are different. This difference
is mainly due to the source model. For the area source model, it accounts for the activity of
all events inside the sources, and considers that the occurrence rate is uniform over the area
giving a uniform acceleration level. This is true when the seismotectonic properties of the
source are homogenous (fault type, direction, etc.). As for the second model, the delineation
of the seismic source is removed and the occurrence rate is only based on the seismicity data
(the earthquake catalog). Subsequently, this model gives results that follow the seismicity
distribution over the whole seismic study region (Tunisia and sites at 300 km around). The
weakness in this latter model is related to the choice of the correlation parameters which
depends on the uncertainty in the earthquakes location. This model leads to high or low
levels of acceleration for low and high correlation distances, respectively.
Hence, the choice between one of those two models is not a straightforward task. They
both represent the seismic hazard of Tunisia for a return period of 475 years, with a probability
of exceedance of 10 % in 50 years. Their combination in the logic tree provided the final map
(Fig. 16d) which allows capturing epistemic errors, especially those related to the adopted
source models.

10 Conclusion

In this study, the primary objective was to elaborate a new seismic hazard module for Tunisia.
The compiled parametric earthquake catalog was carefully investigated using the ISC, CSEM
and INM of Tunisia catalogs, and checking time and space distribution and magnitude
frequencies to meet the key parameters of uncertainties in the analysis of seismic hazard
assessment. Different approaches were employed to homogenize, decluster and check data
quality (in magnitude and distance) reported in the compiled catalog. The previously elabo-
rated seismic source model of INM-IZIIS (1991) was investigated in order to weight it with
smoothed-gridded (Woo 1996a) and energy based (Lapajne et al. 2003) seismicity models
by means of a logic tree framework to capture epistemic errors. This logic tree accounted
for two applicable empirical ground motion models calibrated for shallow crustal seismicity
and three magnitude homogenization equations. In this model, it was attempted to analyze
the probability function of uncertainties through various cross-checking with available geo-
logical and seismotectonic data. As a result, peak ground acceleration and spectral ordinates
maps were generated in generic rock sites for T = 475 years return period. It was con-
cluded that peak ground acceleration does not exceed 0.36 g in the tunisian territory which
is in accordance with the moderate seismicity of Tunisia. The level of acceleration of 0.3 g’s
places the site of Tunis in a non tolerable level of seismic risk, especially when combining
it with the local building scheme of the old city of Tunis and even old buildings of the 19th
century in the modern city. In addition, accounting for the specific lithological context for
Tunis City, we expect to reach considerable level of amplified acceleration and consequently
higher seismic losses.

Acknowledgments We would like to thank Dr. Gordon Woo from RMS for providing us with “KERGRID”
program and also for his precious comments on the use of the smoothed-gridded seismicity approach in
the context of moderate seismicity regions such as the case of Tunisia. We are grateful to Dr. Barbara Šket
Motnikar from the Environmental Agency of Slovenia who accepted to provide us with OHAZ program and
who helped us with her valuable recommendations and support in the use of this program. We wish to express
our special thanks to Dr. Mustapha Meghraoui from University of Strasbourg, and Dr. Vunganai Midzi, from

123
Author's personal copy
668 Bull Earthquake Eng (2014) 12:647–670

the Council of Geosciences of South Africa, for their wide contribution to the revision of this manuscript. We
also gratefully acknowledge constructive and detailed reviews made by Dr. Marco Pagani and Dr. Graeme
Weatherill from GEM Technical Staff. Particular thanks are addressed to Dr. Rui Pinho for his assistance in
improving this work in collaboration with the GEM technical team by using the OpenQuake engine and hazard
modeling tools, as well as to Dr. Dario Slejko and three other anonymous experts whose detailed reviews led
to improvements of previous versions of this manuscript.

References

Akkar S, Bommer JJ (2010) Empirical equations for the prediction of PGA, PGV, and spectral accelerations
in Europe, the Mediterranean region, and the Middle East. Seismol Res Lett 81(2):195–206
Ambraseys NN (1962). The seismicity of Tunis. Ann Geofis 15:233–244
Amiri A, Chaqui A, Nasr IH, Inoubli MH, Ben Ayed N, Tlig S (2011) Role of preexisting faults in the geody-
namic evolution of Northern Tunisia, insights from gravity data from the Medjerda valley. Tectonophysics
506:1–10. doi:10.1016/j.tecto.2011.03.004
Attafi A (1997) Generality about seismicity and the seismological network of Tunisia. Bull Int Inst Seismol
Earthq Eng 3:357–389
Ben Ayed N, Zargouni F (1990) Carte sismotectonique de la Tunisie. Tunis: ministère de l’Éducation de
l’Enseignement supérieur et de la Recherche scientifique (in French)
Ben Ayed N (1993) Evolution tectonique de l’Avant-pays de la chaîne alpine de Tunisie du début du Mésozoïque
à l’Actuel, Annale des Mines et de la Géologie de Tunisie 32, 286 pp (in French)
Benouar D (1994) Geographic distribution of earthquakes in the Maghreb. Annali di Geofisica XXXVII(4)
Benouar D, Molas GL, Yamakazi F (1996) Earthquake hazard mapping in the Maghreb countries: Algeria,
Morocco, Tunisia. Earthq Eng Struct Dyn 25(10):1151–1164
Bollinger GA et al (1993) A comparison of earthquake damage areas as a function of magnitude across the
United States. Bull Seismol Soc Am 83:1064–1080
Bouaziz S, Barrier E, Soussi M, Turki MM, Zouari H (2002) Tectonic evolution of the northern African margin
in Tunisia from paleostress data and sedimentary record. Tectonophysics 357:227–253
Castany G (1954) L’accident Sud-tunisien, son âge et ses relations avec l’accident Sudatlasique d’Algérie.
Comptes Rendus de l’Académie des Sciences de Paris, T.238, pp 916–918 (in French)
Chiou BS-J, Youngs RR (2008) A NGA model for the average horizontal component of peak ground motion and
response spectra. Earthq Spectr 24(1):173–215. February 2008; © 2008, Earthquake Engineering Research
Institute
Cornell CA (1968) Engineering seismic risk analysis. Bull Seismol Soc Am 58:1583–1606
Delavaud E, Cotton F, Akkar S, Scherbaum F, Danciu L, Beauval C, Drouet S, Douglas J, Basili R, Abdullah
Sandikkaya M, Segout M, Faccioli E, Theodoulidis N (2012) Toward a ground-motion logic tree for
probabilistic seismic hazard assessment in Europe. J Seismol 16:451–473. doi:10.1007/s10950-012-9281-z
Deniz A (2006) Estimation of earthquake insurance premium rates based on stochastic methods, M.Sc. Thesis,
Civil Engineering Department, Middle East Technical University, Ankara
Dhahri F, Boukadi N (2010) The evolution of pre-existing structures during the tectonic inversion process of
the Atlas chain of Tunisia. J Afr Earth Sci 56:139–148
Dlala M, Kacem J (2009) Aléa sismique régional du grand Tunis et ses environs (Tunisie nord orientale),
Annales du ministère de l’Equipement, de l’Habitat et de l’Aménagement du Territoire (in French)
Douglas J (2011) Ground-motion prediction equations 1964–2010, PEER 2011/102, April 2011, 455 pp
EMSC Newsletter Number 22 (2007) Special issue funded by EERWEM EC project (INCO-CT-2005-015107),
No ISSN: 1607–1980
ETAP, Enterprise Tunisienne d’Arctivités Pétrolières ETAP (2003) Tunisia open acreage. Conception and
Printing SIMPACT Publishing Co., 83 p (in French)
Frankel A (1995) Mapping seismic hazard in the central eastern United States. Seismol Res Lett 66(4):8–21
Gardini D, di Donato M, Boshi E (1997) Calibration of magnitude scales for earthquakes of the Mediterranean.
J Seismol 1:161–180
Gardner JK, Knopoff L (1974) is the sequence of earthquakes in Southern California, with aftershocks removed,
Poissonian? Bull Seismol Soc Am 64(5):1363–1367
Giardini D, Grünthal G, Shedlock KM, Zhang P (1999) The GSHAP global seismic hazard map. Ann Geophys
42:1225–1230
Grünthal G, Wahlström R (2012) The European-Mediterranean Earthquake Catalogue (EMEC) for the last
millennium. J Seismol 16(3):535–570. doi:10.1007/s10950-012-9302-y
Gutenberg B, Richter F (1954) Frequency of earthquakes in California. Bull Seismol Soc Am 34:185–188

123
Author's personal copy
Bull Earthquake Eng (2014) 12:647–670 669

Hanks TC, Kanamori H (1979) A moment magnitude scale. J Geophys Res 84(B5):2348–2350
Hamdache M, Pelaez JA, Talbi A, Lopez Casado C (2010) A unified catalog of main earthquakes for northern
algeria from a.d. 856 to, 2008. Seismol Res Lett 81(5) doi:10.1785/gssrl.81.5.732
Hussein HM, Elenean KMA, Marzouk IA, Peresan A, Korrat IM, Abu El-Nader E, Panza GF, El-Gabry MN
(2008) Integration and magnitude homogenization of the Egyptian earthquake catalogue. Nat Hazards
47:525–546. doi:10.1007/s11069-008-9237-3
H’faiedh M, Allouche M (1993) seismic hazard in Tunisia; the practice of seismic earthquake hazard assess-
ment, International Association of Seismology and Physics of the Earth’s Interior (IASPEI) and European
Seismological Commission (ESC)
Jordanovsky (1991) INM-IZIIS working group, Report 500-91-61 (1991), Investigation for elaboration of
seismotectonic map and draft seismic design code of Tunisia, Volume II: Seismic hazard analysis and
seismic zoning map of Tunisia
Jiménez MJ, Giardini D, Grünthal G (2001) Unified seismic hazard modelling throughout the Mediterranean
region. Bollettino di geofisica teorica ed applicata 42(1–2):3–18
Kijko A (2004) Estimation of the maximum earthquake magnitude, mmax , Pure. Appl Geophys 161:1655–1681
0033-4553/04/081655-27 doi:10.1007/s00024-004-2531-4
Klett KR (2001) Total petroleum systems of the Pelagian Province, Tunisia, Libya, Italy, and Malta–The Bou
Dabbous-Tertiary and Jurassic-Cretaceous Composite, U.S. Geological Survey Bulletin 2202-D
Ksentini A (2004) Apport des systèmes d’information géographique dans l’analyse et la gestion du risqué
sismique, mémoire de mastère à l’ENIT, 91 p (in French)
Kulkarni RB, Youngs RR, Coppersmith KJ (1984) Assessment of confidence intervals for results of seismic
hazard analysis. In Proceedings, eighth world conference on earthquake engineering, San Francisco, vol 1,
pp 263–270
Lapajne JK, Motnikar BS, Zupancic P (2003) Probabilistic seismic hazard assessment methodology for dis-
tributed seismicity. Bull Seismol Soc Am 93(6):2502–2515
Mejri L, Regard V, Carretier S, Brusset S, Dlala M (2010) Evidence of Quaternary active folding near Utique
(Northeast Tunisia) from tectonic observations and a seismic profile. Comptes Rendus Geoscience. 342(11),
864, # of pages: 9, 01 November 2010. ISSN:16310713
Mezcua J (2002) Seism Eng Course. Universidad Politécnica de Madrid report, Madrid (in Spanish)
National Institute of Building Sciences (NIBS) (1999) Hazus: earthquake loss estimation methodology, Tech-
nical manuals I, II & III, prepared for the Federal Emergency Management Agency. Washington, DC
Pelaez JA, Chourak M, Tadili BA, Aït Brahim L, Hamdache M, Lopez Casado C, Martinez Solares JM (2007)
A catalog of main Moroccan earthquakes from 1045 to 2005. Seismol Res Lett 78(6). November/December
2007
Peña SA, Abdelsalam MG (2006) Orbital remote sensing for geological mapping in southern Tunisia: impli-
cation for oil and gas exploration. J Afr Earth Sci 44:203–219
Petersen MD, Frankel AD, Harmsen SC, Mueller CS, Haller KM, Wheeler RL, Wesson RL, Zeng YB, Oliver
S, Perkins DM, Luco N, Field EH, Wills CJ, Rukstales KS (2008) Documentation for the 2008 Update of the
United States National Seismic Hazard Maps, Open-file report. U.S. Geological Survey. No. 2008–1128,
61 pp
Pinho R (2012) GEM: a participatory framework for open, state-of-the-art models and tools for earthquake risk
assessment worldwide. In: Proceedings of the 15th world conference on earthquake engineering, Lisbon,
Portugal
Popescu E, Grecu B, Popa M, Rizescu M, Radulian M (2003) Seismic source properties: indications of
lithosphere irregular structure on depth beneath Vrancea region. Romanian Rep Phys 55(3):303–321
Reasenberg PA (1985) Second-order moment of central California seismicity, 1969–1982. J Geophys Res
90:5479–5495
Romdhane BN, Mechler P, Duval AM, Meneroud P, Vidal S (2000) Microzoning the city of Tunis both
background noise and weak motion. In: Proceedings of 12th world conference on earthquake engineering,
Auckland, New Zealand, 2000, No. 0343, pp 8
Romdhane BN, Mechler P (2002) Seismic site effect, evaluation methods: application to the city of Tunis.
Bull Eng Geol Environ 61(3):269–281
Rothé JP (1970) Mission d’Information Géologique, Moyen Orient et Afrique du Nord, UNSECO, No série :
1759/BMS. RD/SCE, Paris, février 1970
Scordilis EM (2006) Empirical global relations converting MS and mb to moment magnitude. J Seismol
doi:10.1007/s10950006-9013-3
Sieberg A (1932) Erdbebengeographie. Hanbuch der geophysik, Bd. IV, Abshnitt VI, Berlin 1932, pp 872
Silva V, Crowley H, Pagani M, Monelli D, Pinho R (2013) Development of the OpenQuake engine, the Global
Earthquake Model’s open-source software for seismic risk assessment, Natural Hazards, March 2013.
doi:10.1007/s11069-013-0618-x

123
Author's personal copy
670 Bull Earthquake Eng (2014) 12:647–670

SSHAC (1997) NUREG/CR-6372 UCRL-ID-122160 Vol 1 (1997) Recommendations for probabilistic seismic
hazard analysis: guidance on uncertainty and use of experts, senior seismic hazard analysis committee
(SSHAC)
Stafford PJ, Strasser FO, Bommer JJ (2008) An evaluation of the applicability of the NGA models to
ground-motion prediction in the Euro-Mediterranean region. Bull Earthq Eng 6:149–177. doi:10.1007/
s10518-007-9053-2
Stepp JC (1972) Analysis of completeness of the earthquake sample in the Puget Sound area and its effect on
statistical estimates of earthquake hazard. In: Proceedings of Microzonation Conference, Seattle, WA, pp
897–909
Ulusay R, Tuncay E, Sonmez H, Gokceoglu C (2004) An attenuation relationship based on Turkish strong
motion data and iso-acceleration map of Turkey. Eng Geol 74:265–291
Uhrhammer RA (1986) Characteristics of northern and central California seismicity. Earthquake, Notes 57(1),
21(abstract)
Vogt J (1993) Further research on the historical seismicity of Tunisia. Terra Nova 5(5):475–476
Weatherill G, Pagani M, Monelli D (2012) Openquake modeler: a collection of tools for developing PSHA
input models, 15th World conference on earthquake engineering, Lisbon, Portugal, 24th–28th September
2012
Weichert DH (1980) Estimation of the earthquake recurrence parameters for unequal observation periods for
different magnitudes. Bull Seismol Soc Am 70:1337–1346
Wells DL, Coppersmith KJ (1994) New empirical relationships among magnitude, rupture length, rupture
width, rupture area, and surface displacement. Bull Seismol Soc Am 84:974–1002
Wiemer S, Wyss M (1994) Seismic quiescence before the Landers (M=7.5) and Big Bare (M=6.5) 1992
earthquakes. Bull Seismol Soc Am 84:900–916
World Bank, Egis BCEOM International/IAU-IDF/BRGM (2011) Adaptation au changement climatique et
aux désastres naturels des villes côtières d’Afrique du Nord, Phase 1: Évaluation des risques en situation
actuelle et à l’horizon 2030 pour la ville de Tunis, GED 80823T (in French)
Woo G (1996a) Kernel estimation methods for seismic hazard area source modelling. Bull Seismol Soc Am
86(2):353–362
Woo G (1996b) Seismic hazard program: KERFRACT, program documentation
Yenier E, Erdogan Ö, Akkar S (2008) Empirical relationships for magnitude and source-to-distance conversions
using recently compiled Turkish strong-ground motion database. The 14th world conference on earthquake
engineering, Beijing, China, 12–17 October 2008
Zabukovec B, Kuka N, Sostaric M, Motnikar BS, Suler T (2007) OHAZ: computer program for seismic hazard
calculation. User Manual, Environmental agency of Slovenia and Institute of Seismology of Albania, 65 p

123
View publication stats

You might also like