Attenuation of Intensity for the Zemmouri Earthquake of 21 May 2003 (Mw 6.8): Insights for the Seismic Hazard and Historical Earthquake Sources in Northern Algeria - Historical Seismology: Interdisciplinary Studies of Past and Recent Earthquakes - page 323

value of zero for 50% probability that the predicted parameter a (ground accelera-
tion) will exceed the real (observed) value and a value of 1 for 84% probability.
Recently, Laouami et al. (2006) published for northern Algeria a new attenuation
law (Equation (3)) derived from the strong motion dataset of four moderate Alge-
rian earthquakes (Constantine, 1985; Mont Chenoua, 1989; Mascara, 1994 and Ain
Benian, 1996).

32927 M s ) D 0 . 29202 557574 − 1 . 537231 − 0 . 27024 R s 2 )

a ( m / = 0 . 38778 exp(0 . + 1 . + 0 . 03

(3)

√ R 2

Where D d 2 is the hypocentral distance, the constants are obtained by

fitting the experimental maximum acceleration of the three components at each dis-
tance the least square technique (Laouami et al. 2006).
The 2003 Zemmouri earthquake occurred in an area for which no other reliable
and complete isoseismal maps exist for previous seismic events. It is important,
therefore, to analyse the attenuation for this single event by taking into considera-
tion the local lithological characteristics. Based on the analysis of the attenuation of
intensities during an earthquake, the variation of the intensity is assumed to be pri-
marily related to the surface wave propagation, which is controlled by the change of
soil conditions. The Zemmouri earthquake represents a good example for assessing
the intensity attenuation with distance in several ways. The approach we followed
here consists in performing a regression analysis using an equation of the form:

= +

b ∗ D c ∗ log(D) I = I 0 + a + + (4)

for all the data set (Douglas 2001). For all computations I 0 is 10, D is the macroseis-
mic epicentral distance in km and the coefficients a, b and c are given in Table 2 for
various cases. For practical reasons and since the instrumental epicentre is coastal
and the activated fault emerged at about 8 km offshore, we used in our computations
the macroseismic epicentre estimated at the city of Zemmouri. From the isoseis-
mal map, the intensity decreases gradually until 3 within a radius of 500 km to

the SE as well as in the NE-SW direction parallel to the fault rupture strike. The
intensity-distance relationship for all the combined data is shown on Fig. 4a. We
calculate this attenuation using two directions: to the SE (perpendicularly to the
fault azimuth) and to the SW (parallel to the fault azimuth) (Fig. 4b and 4c). The
regression is performed using average values of D (from isoseismals) and also all
data in the two directions with respect to the macroseismic epicentre (Fig. 4d). For
soil classification, Fig. 4e shows that the intensity decreases clearly at the rock soil

∼

Ta b l e 2 Regression coefficients for the used equation

FA+ 90 ◦ Parameters All data Fault azimuth (FA) Isoseismals

FA FA+90 ◦

A 1 . 31 0 . 137 0 . 38 1 . 22 0 . 159

B − 0 . 001 − 0 . 0032 − 0 . 0004 − 0 . 01 − 0 . 0004

C − 2 . 95 − 1 . 193 − 1 . 167 − 1 . 820 − 2 . 325