Page 14 - Visini_de Nardis_Lavecchia_2010

S256 Int J Earth Sci (Geol Rundsch) (2010) 99 (Suppl 1):S243–S264

Table 2 continued
N Y M D Lon Lat Depth Strike-A Dip-A Rake-A Strike-B Dip-B Rake-B M w Ref
46* 2000 11 20 14.9410 37.8420 23.71 270 41 102 75 50 80 3.3 N05
Events with M w C4.0 and depths \40 km occurred since 1968
C73 Console et al. (1973), G85 Gasparini et al. (1985), A&J87 Anderson and Jackson (1987), G89 Gasparini et al. (1989), F&A Frepoli and
Amato (2000), P&AL Pondrelli et al. (2004), F&A97 Frepoli and Amato (1997), R89 Riguzzi et al. (1989), P02 Pondrelli et al. (2002), S03
Santini (2003), N05 Neri et al. (2005), P06 Pondrelli et al. (2006)

Magnitude–frequency distributions 2006, ±0.25 for 1911–1981, ±0.35 for 1500–1911 in
Jenny et al. 2006) and different completeness intervals in
In order to calculate a representative Gutenberg–Richter the calculated error ranges. We obtained values of b equal
(1944) magnitude–frequency distribution for each prov- to 1.09 ± 0.06 and of a equal to 4.56 ± 0.33 for the
ince, we have extracted from Table 1 only the earthquakes ABT1; of b equal to 1.10 ± 0.07 and of a equal to
falling within the time of completeness for magnitude 4.15 ± 0.34 for the SBT1; of b equal to 1.19 ± 0.11 and a
classes. Plots of the seismicity rate variation with time for equal to 4.83 ± 0.57 for the STC1. The computed ranges
magnitude classes allow us to identify time period of of uncertainties are graphically represented by the grey
completeness. The Marche–Adriatic dataset may be con- areas drawn above the GR slopes in Fig. 4. The G–R slopes
sidered complete since the year 1640 ± 200 for M w C5.5, ABT2, SBT2, STC2 and ABT3, SBT3, STC3, which were
since 1680 ± 80 for 5.0 B M w \ 5.5 and since 1860 ± 20 computed assuming the completeness intervals calculated
for 4.5 B M w \ 5.0; the Sicilian dataset would be com- by the Working group MPS (2004b) and by Pace et al.
plete since 1600 ± 200 for M w C5.5, since 1680 ± 100 for (2006), also fall within this uncertainty range.
5.0 B M w \ 5.5 and since 1820 ± 60 for 4.5 B M w \ 5.0
(Table 3 and reference therein). Substantially similar Other input data and associated error
completeness intervals are computed for the same areas by

_
other authors (Working group MPS 2004b; Pace et al. The computation of the scalar seismic moment rate M o ;
2006). In the case of seismic sequences (e.g. Ancona 1972, depends on the assumed a and b values of the G–R slope,
Belice 1968 and Palermo 2002), we have only considered the c and d constants of the moment–magnitude relation
the maximum magnitude event. and the value assumed for the maximum magnitude.
The G–R distributions (ABT1, SBT1 and STC1 slopes Values and standard errors in a and b have been discussed
in Fig. 4) are expressed as log N (M) = a - bM, where N above. The c and d values have been assumed equal to 1.5
is the number of events of a certain magnitude M per year, and 16.05 following Kanamori and Anderson (1975) with
a is the recurrence rate of the smallest events, and b rep- standard errors of 0.05 and 0.26, respectively, given by
resents the proportion of small to large magnitude earth- Papazachos and Kiratzi (1992). The standard error in M smax
quakes and determines the slope of the curve (Kanamori was taken from the Working group CPTI (2004a). A value
2
and Anderson 1975; Udias 1999; Kagan 2002a, b). Because of 3.0 9 10 10 N/m has been assumed for the rigidity
the G–R distribution is sensitive to completeness intervals modulus (Hunstad et al. 2003).
and to the selected bins of magnitude classes, for each
province we have also computed two other G–R slopes
(ABT2, ABT3, SBT2, SBT3 and STC2, STC3 in Fig. 4) Velocity tensor computations
applying completeness intervals calculated by other
authors (Working Group MPS 2004b; Pace et al. 2006) Applying the formulations Eq. 1–6 described in ‘‘Proce-
(Table 3) and using two bins of magnitude (DM w = 0.1 dure’’ and using the input data discussed in ‘‘Input data’’,
and DM w = 0.23). All the determined G–R slopes are with associated standard deviation errors, we have first

_
shown in Fig. 4. They appear substantially insensitive to computed the rate of seismic moment release M o and

the completeness assumptions and to the choice of the subsequently the strain rate _ e ij and velocity tensor (U ij )
magnitude bin. For this paper, in order to estimate quan- within the ABT, SBT and STC seismotectonic provinces
titatively the standard deviations of the ABT1 and SBT1 (Table 4; Fig. 5). The errors involved in the evaluation of
slopes, we have simulated 20,000 catalogues for each the crustal deformation using the above seismicity rela-
province (Rhoades 1996). The catalogues were generated tionships may be significant, owing to possible restrictions
assuming different uncertainties ranges in the magnitude on the validity of the adopted formulas, as well as to the
for different time periods (±0.2 for the time period 1981– uncertainties of the used parameters, both those depending

123