Post on 12-Jan-2020
PROBABILISTIC SEISMIC HAZARD ASSESSMENT
IN THE BLACK SEA AREA
I.A. MOLDOVAN, M. DIACONESCU, R. PARTHENIU, A.P. CONSTANTIN,
E. POPESCU, D. TOMA-DANILA
National Institute for Earth Physics, 12 Calugareni Street, Magurele, Ilfov, Romania
E-mail: irenutza_67@yahoo.com
Received August 18, 2016
Abstract. The paper has as final goal the probabilistic assessment of seismic
hazard in the Black Sea area as input for the tsunami hazard evaluation. Maximum
and most expected magnitudes and their recurrence periods have been computed for
all defined seismogenic sources from the marine area, and hazard curves have been
plotted.
Key words: Seismogenic zones, probabilistic seismic hazard assessment,
Black Sea.
1. INTRODUCTION
The Black Sea region is known to be an area of active tectonics and
moderate to high seismicity, that very rare triggers tsunami waves.
Black Sea Basin represents a back arc basin opened in the early Cretaceous-
Early Paleogene subduction of the Neotethys below the Balcanides-Pontides
volcanic arc and is surrounded by a system of Alpine orogenic chains, such as:
Balkanides-Pontides, Caucasus-Crimea system and North Dobrogea and Strandja-
Sakarya zones [1]. Deep seismic reflection studies demonstrate the existences of
two extensional sub-basins, one to the West, called Western Black Sea Basin, was
opened in Early Cretaceous, and another to the East, called Eastern Black Sea
Basin, which was opened in Eocene; these basins are separated by the continental
uplifted block called Mid-Black Sea Ridge (or Andrusov Ridge) [1].
The largest earthquake recorded in the Black Sea is the one from March 31,
1901, Mw = 7.2, and depth of 15 km, occurred near Balchik, Bulgaria and triggered
a 3–4 m tsunami waves, that hit the Bulgarian and Romania coast [2, 3].
Romanian Journal of Physics 62, 809 (2017)
Article no. 809 I.A. Moldovan et al. 2
From the seismotectonical point of view the earthquakes which are
responsible for tsunami are those associated with thrust faults (subduction zones),
normal and inverse faults and less strike slip faults (only if the oblique-slip and
deep slip components are predominant), with magnitude higher than 6.5 (even the
USGS cited tsunami at 5.1 magnitude) and depth, a shallow one, less than 20 km
depth. In order to delimit the seismic sources from Black Sea and to discrimate
among them the tsunamigenic ones, the following elements have to be taken into
account: – depth of the earthquakes foci, that allow separation of two major
categories: deeper than 40 km depth and crustal, normal, (less than 40 km deep); –
development of the earthquakes epicenters in the orogen zone or in zones with
active tectonics (fault systems); – establishment of the areas of active faults along
which the earthquakes epicenters are aligned; – the absence of a recent or actual
tectonic activity; – the epicenters recorded in these tectonically stable zones are
considered as the result of a diffuse, accidental seismicity.
The studies on active tectonics have clearly shown the position of the
seismic sources (connected to well define active fault) which do not interfere and
do not result in alternatives of other seismotectonic model constructions.
According to the distribution map of earthquakes and as well as to the map of the
areas with active tectonics, ten seismic sources were established [4, 5, 6].
In the present paper the maximum possible magnitude of each seismic
source was obtained through three aproaches: (i) using Gutenberg Richter’s [7] “a”
and “b” parameters; (ii) using Cornell [8] statistical distributions and (iii) using
extreme values Gumbel I [9] to model the seismogenic process for all the
earthquake sources from the Black Sea region. The advantage of the statistical
methods is the possibility to compute all the quantities used in probabilistic hazard
assessment, including recurrence times for different magnitudes.
Another important issue of the paper was to estimate the seismic hazard for
the Black Sea seismogenic sources using a probabilistic approach.
2. SEISMIC ZONATION
The seismic zonation of the Black Sea Area was obtained using the
distribution map of earthquakes and the map of the zones with active tectonics. We
took into consideration various past seismic zonation studies carried out in the
framework of different projects (SHARE project – www.share-eu.org,
MARINEGEOHAZARD project – www.geohazard-blacksea.eu, DARING project
– http://daring.infp.ro/ and ASTARTE RO project – http://astarte-ro.infp.ro/
3 Probabilistic seismic hazard assessment in the Black Sea area Article no. 809
BIGSEES project – infp.infp.ro/bigsees/default.htm,). The seismic source
configuration in Fig. 1 is a synthesis of all the previous approaches.
The present configuration of the potential seismic sources contains fifteen
crustal and one intermediate-depth seismic sources [4, 5]: Vrancea intermediate-
depth (VRI), Vrancea normal (VN), Barlad Depression (BD), Intramoesian Fault
(IMF), North Dobrogea (PD), North Dobrogea Black Sea (BS1), Central Dobrogea
(BS2), Shabla (BS3), Istanbul (BS4), North Anatolian Fault (BS5), Georgia (BS6),
Novorossjsk (BS7), Crimeea (BS8), West Black Sea (BS9) and Mid Black Sea
(BS10). Only five sources are inland, the rest being marine seismic sources.
Fig. 1 – The seismic zonation of the Eastern part of Romania and the Black Sea Area,
for earthquakes with Mw > 3.5 [6].
In order to have the most reliable and homogeneous seismic dataset, the
catalogues available at the European scale covering historical and modern
instrumental seismicity until present days (ANSS – Advanced National Seismic
System-USA, NEIC – National Earthquake Information Centre, World Data for
Seismology Denver-USA, ISC – International Seismological Centre-UK) and the
catalog of the National Institute for Earth Physics (Romplus catalogue, updated)
have been compiled. The parameters describing the seismic sources from Fig. 1 are
given in Table 1.
Article no. 809 I.A. Moldovan et al. 4
Table 1
Black Sea seismic sources (SeS) parameters [6]
SeS Coordinates h
km
Mmax
Mw
Seismic
activity rate
SeS Coordinates h
km
Mmax
Mw
Seismic
activity rate
BS1
45.11 30.55
3.5 3.0 0.386363 BS6
41.22 39.99
5.5 3.0 1.039215 44.56 30.36 43.17 40.01
44.9 29 42.92 41.83
45.55 29.6 40.93 41.69
BS2
44.24 28.22
5.0 3.0 0.118644 BS7
44.89 35.83
5.2
3.0
0.59091
44.9 29 45.40 36.70
44.48 30.69 43.46 40.24
43.77 30.57 42.96 39.52
43.32 29.56
BS3
44.24 28.22
7.2 3.0 0.165137 BS8
44.09 32.86
6.5 3.0 0.25301 43.32 29.56 45.32 32.63
43.03 29.39 44.83 35.65
43.42 28.05 43.72 35.06
BS4
41.19 28.07
6.7 3.0 0.47761 BS9
45.05 30.77
4.9 3.0 0.19512 42.28 28.72 45.69 30.94
41.89 31.52 45.62 31.71
40.94 31.82 44.98 31.47
BS5
40.93 31.82
6.1 3.0 0.740741 BS10
42.51 30.48
3.9 3.0 0.25581 41.89 31.52 44.54 31.26
42.77 34.17 44.30 32.48
40.97 40.92 42.40 31.84
3. SEISMIC HAZARD ASSESSMENT IN THE BLACK SEA AREAL
USING STATISTICAL TOOLS
In this chapter, statistical tools are applied for seismic hazard assessment.
We have applied a common statistical techniques to derive the required parameters
describing the rates at which each seismic source zone has generated earthquakes
of different magnitudes in the past, which are then taken as the expected
probabilities to generate future earthquakes for use in the assessment of hazard.
The key parameters – the activity rate, the b-value, and the maximum magnitude
Mmax have been assessed for offshore seismogenic sources. The regional earthquake
catalogues and the defined seismic source zone geometries have been used to
derive magnitude-dependent catalogue completeness, to de-cluster aftershocks, to
fix prior-distributions of maximum magnitudes and to evaluate statistical
uncertainties. Seismic activity ν0 is defined as the annual average number of
earthquakes with magnitude higher than m0 (Mw).
The parameters of the Gutenberg-Richter [7] distribution (a, b) have been
compiled for each source, and the b values have been mapped in Fig. 2 to
emphasize the zones with low and high stress, for 115 years.
5 Probabilistic seismic hazard assessment in the Black Sea area Article no. 809
Using a and b values we have computed some statistical parameters of BSi
zones:
a) maximum possible magnitude in T1 years (taken from the catalogue –
67 years):
Mmax
= a/b (1)
b) most probable magnitude in a return period of TR = 50 yr:
bT
TaMmp
R
/)log( 1 (2)
c) principal magnitude M that might appear annually (TR= 1):
Mp = (a–log T1)/b. (3)
In Table 2 are presented the statistical parameters obtained from Gutenberg
Richter (GR) relation of each source from the Black Sea (BSi). Because BS1 and
BS9 are very low risk seismic sources they will not be further analyzed.
With the input data set from Tables 1 and 2, we have applied the algorithm
of [8] and [10] to compute the seismic hazard parameters for BS1-S10 seismic
sources: the number of events with a given magnitude per year, the annual hazard,
the hazard for 50, 100, 475 and 1000 years, the return periods for different
magnitudes. Using numerical computations we have also obtained the maximum
possible magnitude for each zone (see columns 4 and 8 from Table 3).
Fig. 2 – b values for BSi sources.
Article no. 809 I.A. Moldovan et al. 6
Table 2
Statistical parameters obtained from GR relation
Seismic
Zone a b
Mmax
(comp)
Mmp
(Tr = 50y)
Mp
(1 year)
Mmax
(observed)
BS2 3.15 0.65 4.85 4.27 1.66 5
BS3 2.13 0.32 6.66 5.55 0.24 7.2
BS4 3.29 0.53 6.21 5.97 2.76 6.7
BS5 2.91 0.61 4.77 4.66 1.88 6.1
BS6 3.5 0.59 5.93 5.93 3.15 5.5
BS7 3.84 0.75 5.12 5.11 2.90 5.2
BS8 2.43 0.38 6.39 5.76 1.29 6.5
BS10 3.3 0.81 4.07 4.07 2.27 3.9
Table 3
Statistical parameters obtained using [8]
Seismic
Sources
Mmin
(Mw)
Mmax
(Mw)
Mmax
comp
(Mw)
Seismic
Sources
Mmin
(Mw)
Mmax
(Mw)
Mmax
comp
BS1 3.0 3.5 – BS6 3.0 5.5 5.7
BS2 3.0 5.0 5.6 BS7 3.0 5.2 5.8
BS3 3.0 7.2 7.5 BS8 3.0 6.5 6.8
BS4 3.0 6.7 7.2 BS9 3.0 4.9 –
BS5 3.0 6.1 6.3 BS10 3.0 3.9 –
Table 4
The return periods for Mw = 6 computed with [8]
Source BS2 BS3 BS4 BS5 BS6 BS7 BS8
Tr (years) >10000 1422 134 2778 >10000 >10000 3717
We observe that the computed values from Table 3 are different from those
from Table 2 and from the maximum observed magnitudes. The values from Table
2 are lower and the values from Table 3 are higher. Although, the differences do
not exceed 1.0 degrees of magnitude. The return periods seem to be very large
and far from those expected.
7 Probabilistic seismic hazard assessment in the Black Sea area Article no. 809
As an example in Table 4 are the return periods for Mw = 6 for sources
BS2-BS8.
In Figs. 3–5 we have represented the dependence of the expected magnitude
versus the return period (left panel) and the hazard curves for sources BS2 to BS8
(right pane l).
10 100 1000 10000 100000Tr (years)
3
4
5
6
Mw
S1
3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
Mw
0
0.2
0.4
0.6
0.8
1
PS
H
T=1Year
T=50Years
T=100Years
T=475Years
T=1000Years
S1
Fig. 3 – Return periods for earthquakes with different magnitudes (left) and the hazard curves
for different exposure periods (right) for seismic sources BS2.
BS2
BS2
Article no. 809 I.A. Moldovan et al. 8
1 10 100 1000 10000 100000Tr (years)
3
4
5
6
7
Mw
S2
3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
Mw
0
0.2
0.4
0.6
0.8
1
PS
H
T=1Year
T=50Years
T=100Years
T=475Years
T=1000YearsS2
1 10 100 1000 10000 100000Tr (years)
3
4
5
6
7
8
Mw
S3
3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
Mw
0
0.2
0.4
0.6
0.8
1P
SH
T=1Year
T=50Years
T=100Years
T=475Years
T=1000Years
S3
1 10 100 1000 10000 100000Tr (years)
3
4
5
6
7
Mw
S4
3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
Mw
0
0.2
0.4
0.6
0.8
1
PS
H
T=1Year
T=50Years
T=100Years
T=475Years
T=1000YearsS4
Fig. 4 – Return periods for earthquakes with different magnitudes (left) and the hazard curves
for different exposure periods (right) for seismic sources BS3-BS5.
BS3 BS3
BS4 BS4
BS5 BS5
9 Probabilistic seismic hazard assessment in the Black Sea area Article no. 809
10 100 1000 10000 100000Tr (years)
3
4
5
6
Mw
S5
3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
Mw
0
0.2
0.4
0.6
0.8
1
PS
H
T=1Year
T=50Years
T=100Years
T=475Years
T=1000YearsS5
1 10 100 1000 10000 100000Tr (years)
3
4
5
6
Mw
S6
3 3.5 4 4.5 5 5.5 6 6.5 7 7.5
Mw
0
0.2
0.4
0.6
0.8
1P
SH
T=1Year
T=50Years
T=100Years
T=475Years
T=1000Years
S6
Fig. 5 – Return periods for earthquakes with different magnitudes (left) and the hazard curves
for different exposure periods (right) for seismic source BS6-BS8.
BS6 BS6
BS7 BS7
Article no. 809 I.A. Moldovan et al. 10
4. EXTREME VALUES GUMBEL I (GI) STATISTICAL METHOD
USED FOR THE SEISMOGENIC PROCESS MODELING IN BLACK SEA
The first one to recognize the close relationship between the weakest
connection model and asymptotic theory of extreme values was Peirce – one of the
first authors of statistical models of parts. Gumbel's extreme value theory [9]
implies the existence of three types of asymptotic distributions of extreme values
(or cumulative distribution function) as the variable is unlimited, having lower and
upper limits respectively.
a b
Fig. 6 – a) Most probable and the expected magnitude as function of the return period (Tr)
for BS2 and BS3; b) Hazard curves for BS2 and BS3 (Shabla).
The extreme value theory applied to the occurrence of maximum magnitude earthquakes is based on the following hypotheses:
11 Probabilistic seismic hazard assessment in the Black Sea area Article no. 809
A. The occurrence of maximum magnitude earthquake in a seismic region in
a certain period of time is a random, independent event.
B. The behaviour of maximum magnitude earthquake in the future will be
similar to that of previous years of observation. This method is mainly used when
working with extreme values of statistical variables such as magnitude or
maximum ground acceleration. For the maximum magnitude earthquake
occurrence study were considered only the first and the third distribution.
In Figs. 6 and 7 are represented the return periods for most probable and
expected magnitudes (a) and the hazard curves (b) for different periods of time for
BS2-BS8.
Fig. 7
Article no. 809 I.A. Moldovan et al. 12
Fig. 7 (continued) – a) Most probable and the expected magnitude as function of the return period Tr
for BS5, BS7 and BS8; b) Hazard curves for BS5, BS7 and BS8.
5. CONCLUSIONS
During this study we have obtained the probabilistic seismic hazard curves
and the return periods of different magnitudes for the seismic sources from the
Black Sea Basin, using two analyzing methods [8, 9]. With the first analyzing
method [8] the return periods of different magnitudes seems to be large in
comparison with the return periods obtained using the second statistical processing
method [9].
As an example in Table 4 are the return periods for Mw = 6 for sources BS2-
BS8 for both analyzing methods. Comparing the results from Table 5 we can see
the huge differences in the values of the return period. A possible explanation is
given by the fact that the GI distribution is not limited in the superior part leading
to a very fast growth of the magnitudes in time.
Table 5
The return periods for Mw = 6 for sources BS2-BS8
Source BS2 BS3 BS4 BS5 BS6 BS7 BS8
Tr (years) Cornell >10000 1422 134 2778 >10000 >10000 3717
Tr (years) GI 250 90 – 95 – 130 55
Another credible explanation of this differences is given by the earthquake
catalogs for all this sources, catalogues that reveal the low earthquake potential of
the sources and also the bad coverage with recordings systems of the Black Sea
13 Probabilistic seismic hazard assessment in the Black Sea area Article no. 809
basin leading to inconsistent catalogues. The solution for this issue is a joint
seismic monitoring of the Black Sea basin, involving all countries around the
sea [11].
The maximum expected magnitude obtained with both methods for the
studied seismic sources are presented in Table 6 together with the maximum
observed magnitude.
Table 6
The return periods for maximum expected magnitudes in 1000 years for sources BS2-BS8
Source BS2 BS3 BS4 BS5 BS6 BS7 BS8
M (1000 years) Cornell [5] 5.0 5.9 6.7 5.7 5.1 5.4 5.5
M (1000 years) GI [7] 6.4 8.4 – 7.5 – 6.8 8.5
Mmax observed 4.6 7.2 6.7 6.1 5.5 5.2 6.5
As we expected, the magnitude values for the GI method are very high in
comparison with those obtained using the Cornell method and also with those
observed. The explanation is the same as for the low values of return periods, i.e.
the working hypothesis without upper limit for the magnitude.
That’s why the Gumbel III extreme values distribution [9] studies should be
needed for the sources from Black Sea Basin. Unfortunately the existing catalogues
does not permit this type of numerical statistical analysis.
The final conclusion is that for a reliable statistical seismic analysis of the
Black Sea areal is needed a common and uniform seismic monitoring for all states
along the sea coast.
Acknowledgements. This work was partially supported by the Partnership in Priority Areas
Program – PNII, under MEN-UEFISCDI, DARING Project no. 69/2014, and FP7 FP7-ENV2013 6.4-
3 Project number: 603839/2013, ASTARTE/PNII, Capacity Module III Project 268/2014 and Nucleu
Program PN 16 35 03 01 and PN 16 35 01 06.
REFERENCES
1. I. Munteanu, L. Matenco, C. Dinu, S. Cloetingh, Kinematics of back-arc inversion of the Western
Black Sea Basin, Tectonics, 30, TC5004 (2011).
2. A. Yalciner, E. Pelinovsky, T. Talipova, A. Kurkin, A. Kozelkov, A. Zaitsev, Tsunamis in the
Black Sea: Comparison of the historical, instrumental, and numerical data, J. Geophys. Res. 109,
C12023 (2004).
3. G.A. Papadopoulos, G. Diakogianni, A. Fokaefs, B. Ranguelov, Tsunami hazard in the Black Sea
and the Azov Sea: a new tsunami catalogue, Nat. Hazards Earth Syst. Sci. 11, 945 (2011). 4. A.P. Constantin, R. Partheniu, I.A. Moldovan, Macroseismic intensity distribution of some recent
Romanian earthquakes, Rom. Journ. Phys. 61 (5-6), 1120 (2016).
5. A. Bala, V. Raileanu, C. Dinu, M. Diaconescu, Crustal seismicity and active fault systems in
Romania, Rom. Rep. Phys., 67 (3), 1176 (2015).
Article no. 809 I.A. Moldovan et al. 14
6. I.A. Moldovan, M. Diaconescu, E. Popescu, M. Radulian, D. Toma-Danila, A.P. Constantin,
A.O. Placinta, Input parameters for the probabilistic seismic hazard assessment in the Eastern
part of Romania and Black Sea area, Rom. Journ. Phys. 61, 1412 (2016).
7. B. Gutenberg, C.F. Richter, Magnitude and energy of earthquakes, Ann. Geophys. 9, 1 (1956).
8. C.A. Cornell, Engineering seismic risk analysis, Bull. Seismol. Soc. Am. 58, 1583 (1968).
9. E.J. Gumbel, Statistics of extremes, Columbia University Press, New York (1958).
10. R.K. McGuire, EQRISK-Evaluation of Earthquake Risk to Sites, United States Department of the
Interior, USGS, Open File Report 76-67 (1976).
11. I.A. Moldovan, Modeling The Seismogenic Process of Earthquake Occurrence in the Black Sea
Region using Statistical Methods, Special volume in honor of Professor Emeritus Michael E.
Contadakis, Faculty of Rural and Surveying Engineering, Department of Geodesy and Surveying,
Aristotle University of Thessaloniki, Greece, 247 (2013).