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

net/publication/307890701

On the Relationship between Mw and ML for Small Earthquakes

Article  in  Bulletin of the Seismological Society of America · September 2016

DOI: 10.1785/0120160130

CITATIONS READS
22 1,085

3 authors, including:

Irene Munafò Lauro Chiaraluce

National Institute of Geophysics and Volcanology National Institute of Geophysics and Volcanology
13 PUBLICATIONS   293 CITATIONS    111 PUBLICATIONS   4,041 CITATIONS   

SEE PROFILE SEE PROFILE

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

Earthquake rupture initiation processes and early warning View project

All content following this page was uploaded by Irene Munafò on 03 April 2018.

The user has requested enhancement of the downloaded file.

 Bulletin of the Seismological Society of America, Vol. 106, No. 5, pp. 2402–2408, October 2016, doi: 10.1785/0120160130

Short Note
On the Relationship between Mw and M L for Small Earthquakes
by Irene Munafò, L. Malagnini, and L. Chiaraluce

Abstract Estimating the moment magnitudes (Mw ) of a small earthquake is a chal-

lenging task. One viable option to measure its size is to calculate its local magnitude
(M L ) and convert it to the physically based M w . Unfortunately, to correctly perform such
a conversion is not easy; moreover, even though many studies demonstrate that the equiv-
alence between ML and M w is incorrect for small events, these two parameters are some-
times thought to be strictly equivalent, regardless of the earthquake’s size. Using random
vibration theory, we show that, below M w ∼ 4, the M L of a small earthquake is propor-
tional to the logarithm of its seismic moment, and the following relationship holds:

2
EQ-TARGET;temp:intralink-;;55;536 M wSMALL
M LSMALL  C′ :
3

We test our findings on a high-quality data set in the Upper Tiber Valley (northern Apen-
nines, Italy), composed of events in the range of 0 ≤ M L ≤ 3:8, for which we compute
accurate estimates of M L and Mw .

Online Material: Details of the processing procedure, figures of the empirical

regional attenuation functional, and source terms of 1191 events from the Alto Tiberina
fault (ATF) data set and earthquake catalog.

Introduction
Earthquakes span an exceptionally wide range of seismic catalogs of earthquake sizes to be used for purposes such as
moments, making it very difficult to consistently measure their seismic-hazard assessment.
sizes, from the tiniest fracture responsible for a single acoustic Originally, earthquake magnitudes were based on the
emission in a laboratory sample, to the gigantic event that rup- peak amplitudes of ground displacement, as they were mea-
tured the Chilean megathrust in 1960 (Mw 9.5). Several scales sured by standard seismometers. The best-known one is the
are used for defining the size (magnitude) of an earthquake, Richter magnitude (or local magnitude M L ), defined for earth-
most of them being based on amplitude measurements of spe- quakes at local and regional distances (within 600 km)
cific types of seismic waves, in specified frequency ranges, recorded in southern California by a network of short-period
sometimes recorded with a specific instrument. Wood–Anderson seismometers:
Because at different distances we rely on different waves
to estimate the earthquake’s magnitude, dissimilar estimates
EQ-TARGET;temp:intralink-;df1;313;240 M L
log10 A − log10 A0 δ
log10 A=A0 δ 1
of size for the same earthquake cannot precisely agree (e.g.,
(Richter, 1935), in which A is the peak-to-peak maximum
Mw , moment magnitude; mb , body-wave magnitude; or M L, excursion observed on a Wood–Anderson seismogram for
local magnitude). For example, deeper earthquakes do not adjacent peaks, δ is the station-epicenter distance, and A0 δ
efficiently generate surface waves as shallower earthquakes is an empirical function depending on the crustal attenuation
do, and surface-wave-based magnitude (M s ) estimates may in the region. Because of the short-period nature of the
be biased low for deeper events if equations calibrated for Wood–Anderson instruments, and of the spectral character-
shallower events are used. istics of the seismic radiation, the ML scale saturates beyond
As a result, it is possible for an individual earthquake to ML 7, giving magnitude estimates that are too small when
have different estimates of its magnitude, depending on the compared with the increment in the earthquakes’ radiated en-
applied method, and they may significantly differ between ergies (Kanamori, 1977). In addition, the distance corrections
each other. It is clear that the calibration of a magnitude con- should depend on crustal rocks and structures, and each region
version tool is very important, in order to obtain homogeneous should have a different definition of local magnitude.

2402
Short Note 2403

The moment magnitude M w (Kanamori, 1977; Hanks events (roughly below Mw 4 in our data set). In order to do
and Kanamori, 1979) is fundamentally different from the ear- that, we use a tool based on the random vibration theory
lier estimates of an earthquake’s size, because it is somehow (RVT; Cartwright and Longuet-Higgins, 1956) and test our
related to the radiated seismic energy. Rather than relying on findings on a high-quality data set composed of 1191 well-
peak values recorded by specific instruments, the M w is in relocated small earthquakes recorded in the northern Appen-
fact tied to the seismic moment (M0 ) of an earthquake, that nines (Italy), in the 0 ≤ ML ≤ 3:8 magnitude range. Although
is, to the low-frequency plateau of the moment-rate spectrum confirming the results of Deichmann (2006), our study pro-
(Aki and Richards, 2002). The seismic moment is defined as vides them with the necessary generality, theoretical back-
ground, and correct physical interpretation, as well as with an
EQ-TARGET;temp:intralink-;df2;55;637 M0
μAδ; 2 estimate of the upper bound for their validity.
in which μ is the shear modulus of the crustal rocks
(3 × 1010 Pa), A is the rupture area, and δ is the slip length. Data Set: Seismograms from Earthquakes of the Alto
Moment magnitude is calculated from seismic moment using Tiberina Fault
the relation of Hanks and Kanamori (1979): The Alto Tiberina fault (ATF) in the Upper Tiber valley of
the northern Apennines of Italy represents a well-documented
2
EQ-TARGET;temp:intralink-;df3;55;561 M w
log10 M 0  − 6:07; 3 example of an active low-angle normal fault (Barchi et al.,
3 1998; Boncio et al., 2000), east-dipping ∼20° toward the Adri-
in which M 0 is in newton meters. atic Sea. The ATF geometry has been imaged by analyzing
Unlike other empirical scales, moment magnitude does not active-source seismic reflection profiles as well as the persis-
saturate for large earthquakes, because it is tied to the radiated tent background seismicity (Mirabella et al., 2011).
seismic energy, and there is no upper limit to measurable mag- Up to now, only microearthquakes have been observed
nitudes, other than the physical dimension of the largest pos- to continuously nucleate along the ATF plane (Chiaraluce
sible rupture of the Earth’s lithosphere. Being somehow linked et al., 2007; De Luca et al., 2009). None of the moderate
to radiated seismic energy, moment magnitude should be the earthquakes occurred in the area in the past 20 years (Gubbio
seismologist’s preferred measurement of an earthquake’s size. 1984, M w 5.1; Umbria–Marche 1997–1998, Mw 6.0) nucleated
An issue that has been debated in many studies in the on the ATF itself. All these mainshocks activated antithetic
past years, and it is still debated today, is the relation between high-angle southwest-dipping normal faults located within the
ML and M w , or equivalently, between ML and M 0 . For ex- ATF hanging wall (Chiaraluce, Amato, et al., 2014).
ample, with reference to a data set from southern California, Here, we use a data set of 1191 well-relocated small earth-
Hanks and Kanamori (1979) concluded that for magnitudes quakes (Valoroso et al., 2014) in the range of 0 ≤ M L ≤ 3:8,
larger than 3, ML can be equated to M w , but they left open for a total of about 30,000 visually selected waveforms of good
the question as to whether this would also be valid for smaller quality with no spurious signals and excellent signal-to-noise ra-
events. More recently, Bethmann et al. (2011) argued that in tio (SNR). Among the earthquakes of our data set there are about
the magnitude range between 3 and 5, ML and M w scale 1∶1, 620 repeating events, grouped in 300 clusters, rupturing the same
whereas for smaller magnitudes the 1:1 scaling breaks down. fault patch (generally doublets), within a short time interval (less
Their numerical simulations indicated that the transition than 24 hrs; Chiaraluce et al., 2007; Valoroso et al., 2014).
from a 1∶1 to 1∶1.5 (ML ≈ 1:5 M w  C) for magnitudes be- Our waveforms are high-resolution broadband record-
low 2 could be attributed to anelastic attenuation. Many other ings acquired between 2010 and 2014 by 30 permanent seis-
studies in diverse tectonic study regions find similar values mic stations equipped with three-component seismometers,
over various magnitude ranges (see Hanks and Boore, 1984; deployed in the framework of The AltotiBerina near fault
Jost et al., 1998; Deichmann, 2006). As pointed out by Observatory project (TABOO; Chiaraluce, Amato, et al.,
Deichmann (2006), a deviation from the 1∶1 scaling for 2014), augmented with three nearby shallow-borehole sta-
earthquakes with magnitudes below about 3 can be due to tions (BAT1, BAT2, and BAT3; Chiaraluce, Collettini, et al.,
frequency-dependent attenuation along the propagation path. 2014), with three-component short-period seismometers
Variations of crustal attenuation property were considered (2 Hz) that sample the borehole displacement at 500 samples
responsible for similar deviations in volcanic environments. per second. A map view of our data set is given in Figure 1,
Petrosino et al. (2008) worked with local data of the Campi in which the cross sections show the earthquake hypocenters
Flegrei (Italy) area obtaining ML
1:23 M w − 0:12. The lo- within the investigated crustal volume.
cal and moment magnitude scales discussed in their study
were derived by correcting the amplitudes of the seismic sig- Spectral Seismological Measurements and the
nals for path and site effects, in order to obtain quantification Combined Earth-Wood–Anderson Filter
of the earthquake energy. In any case, the cited studies as-
sumed that for all practical uses, M L ≠ Mw for small events. Because in this study we deal with very small earthquakes,
In this work, we provide a general relationship between we maximize the SNR by using a technique based on the analy-
ML and M0 based on theoretical ground and valid for small sis of peak values of band-pass-filtered time histories (Malag-
2404 Short Note

(a)
12˚00' 12˚24' 12˚48'
(b)
km
A A’
CPGN 0 10 0

depth (km)
10
FSSB
PE3
20
A B’
PARC NARO MPAG

43˚36' SSP9 43˚36' 30

APEC
Sansepolcro
PIEI −30 −20 −10 0 10 20 30
BADI FRON
ATPC ATBU
CDCA
ATPI B B’
Citta di 0
ATMC ATSC SSFR
Castello

depth (km)
BAT3
ATVO ATFO 10
BAT1
BAT2 Gubbio
20°
ATMI ATLO
FOSV 20

ATVA
B MURB 30

43˚12' 43˚12'
ATTE
ATCC
−30 −20 −10 0 10 20 30
A’ distance (km)
Perugia

12˚00' 12˚24' 12˚48'

Figure 1. (a) Map view and seismicity distribution of 1191 well-located earthquakes (0 < ML < 3:8) recorded between 2010 and 2014
by The AltotiBerina near fault Observatory seismic network in the Alto Tiberina fault area. Black dots show the locations of the earthquakes.
Gray triangles indicate permanent seismic stations, white triangles show the location of the three stations (BAT) installed within shallow
boreholes (180–250 m), and black squares indicate the main cities of the area. (b) Two cross sections showing the earthquake distribution at
depth of our data set. Cross-section location traces are reported in (a).

nini and Dreger, 2016). Given a stationary, random time history spectrum is bandlimited (i.e., it is significantly different from
of length T, RVT may be used to estimate its peak value zero only between the frequencies f1 and f2 ), then

Peakat ≈ ηarms ; 4 r

RT rR
∞ ^
0 jatj dt −∞ jafj
EQ-TARGET;temp:intralink-;df4;55;339

2 2 df
EQ-TARGET;temp:intralink-;df7;313;339 arms
≈
in which arms is its root mean square (rms) average calculated T T
s
R f2
over T, and η
ηm0 ; m2 ; m4  is a function of the specified
2 f1 jafj ^ 2
df
spectral moments of the filtered time history
: 7
Z T
1 ∞ n
mn
ω jaωj
^ 2
dω; 5 We use the approximation sign because of the distortion of the
π 0
EQ-TARGET;temp:intralink-;df5;55;273

seismic spectrum due to the time-domain windowing. In prac-

in which ω is the angular frequency. tical applications, however, we compute the Fourier spectrum
The Parseval’s equality relates the time-domain integral over the same finite window (0; T) of the rms average in time
of the square of a time series to the frequency-domain inte- domain, and for this reason the equal sign may be used.
gral of the square of its Fourier amplitude spectrum Using equation (4), the peak value of the time history
Z ∞ Z ∞ will be
jatj2 dt
jafj
^ 22
df:
EQ-TARGET;temp:intralink-;df6;55;186 6 s
R
−∞ −∞ 2 ff12 jafj^ 2 df

We can now express the rms average of a windowed time apeak
η : 8 EQ-TARGET;temp:intralink-;df8;313;174

T
series using its Fourier amplitude spectrum and equation (4)
to directly relate the observed Fourier amplitudes to Figure 2a shows moment-rate spectra for magnitudes
the peak value of a seismogram in the time domain. If the −1 ≤ Mw ≤ 7. Up to M w 4, their stress drops are calibrated
time history is significantly different from zero only between over our data set; for larger M w we use the stress drops cal-
t
0 (the direct S-wave arrival time) and t
T (the effective culated for the L’Aquila earthquake sequence (Malagnini
duration of the signal that follows the direct S waves) and its et al., 2011).
Short Note 2405

Equation (8) demonstrates that, for effectively bandlimited ment spectra after the filtering action of the Wood–Anderson
recordings like the ones for which spectra are plotted in Fig- instrument, combined with that due to the anelastic attenu-
ure 2b, the peak values of the filtered time histories are carried ation of the Earth. We earlier explained how the peak values
by relatively similar dominant frequencies. Moreover, in what of the Wood–Anderson recordings are carried by their dom-
follows we will show that, in case of seismograms recorded by inant frequencies, and the dashed line plotted in Figure 2c
Wood–Anderson instruments, there are no contributions to the represents the dominant frequency of a typical M w 4 earth-
peak values from spectral components that are not in the quake of our data set (the source and attenuation parameters
immediate vicinity of the instrument’s natural frequency have been calibrated over our waveforms, as explained in
(f n
1:25 Hz). Consequently, equation (4) tells us that the detail in the Ⓔ electronic supplement to this article). Note
ML of a small earthquake, being the log of a normalized that such frequency (∼1:3 Hz) is almost equivalent to the
peak-to-peak maximum value on the Wood–Anderson seis- Wood–Anderson natural frequency (1.25 Hz) and still sam-
mometer, directly samples its seismic moment. ples the spectral plateau of an M w 4 earthquake of this region
In fact, equation (4) shows that the peak value on the (the same vertical-dashed line is plotted also in Fig. 2a). Be-
Wood–Anderson measurement is proportional to the rms cause the dominant frequencies of events larger than M w 4 do
average of the recorded time history. Because of the limited not sample their spectral plateaus, we conclude that M w ∼ 4
bandwidth of the filter made of the Wood–Anderson seis- represents, for our data set (and for most cases), the upper
mometer coupled with the anelastically attenuating crust limit of validity of equation (10).
(Fig. 2b), if the event is small enough that its corner frequency
is significantly higher than the Wood–Anderson’s natural fre- Moment Magnitude Calibration and M w –M L Scaling
quency (f c ≫ fn ), the recorded peak value is thus propor-
tional to the seismic moment of the event, and we can write Despite the fact that the quantification of M w for small
earthquakes is of great importance in seismology, its deter-
EQ-TARGET;temp:intralink-;df9;55;481 M LSMALL
log10 M0   C: 9 mination is not yet a routine procedure due to technical dif-
ficulties. The classical determination of M w using moment
From equations (3) and (9), we can write tensor solutions relies upon the point-source hypothesis and
2 the availability of low-frequency data (i.e., frequencies that
M wSMALL
M LSMALL  C′ ;
EQ-TARGET;temp:intralink-;df10;55;437 10 are much lower than the event’s corner frequency). As such,
3
Mw may be measured only above a minimum threshold, in
which is the relationship to be used for small earthquakes in which the recorded ground motions at long periods are rou-
Mw –M L regressions. tinely stronger than the background noise. This minimum
For both equations (9) and (10), M LSMALL are from events magnitude is typically about Mw 3 for local events (Herr-
for which corner frequencies are larger than the Wood– mann et al., 2011; Malagnini et al., 2012).
Anderson natural frequency, and C and C′ are constants that The moment magnitude of an earthquake is generally
represent the crustal attenuation coupled with the gain of the computed by reproducing low-frequency seismograms, but
Wood–Anderson seismograph. due to the difficulty in obtaining realistic high-frequency
In fact, the ground displacement filtered through a Green’s functions for complex media, it turns out that the
Wood–Anderson instrument (see Uhrhammer and Collins, latter is virtually an impossible task to perform on the small
1990, for a comprehensive analysis of its instrument response) earthquakes. Given the mentioned difficulties, the only via-
may be thought of in terms of a high-pass-filtered version of ble option for the calculation of accurate M w for small earth-
the true ground displacement, above the natural frequency of quakes is the spectral correction.
the short-period instrument. Because of the coupling between In order to develop a technique for spectral correction of
the natural low-pass filter of crustal anelastic attenuation and seismic waves, we separate the contributions of source, site,
the relatively high natural frequency of the Wood–Anderson and crustal attenuation (Ⓔ Figs. S1 and S2) and define a spec-
seismograph, the complete filtering chain (instrument plus tral model that accurately reproduces the latter at regional dis-
Earth, see Fig. 2b) is a relatively narrow band-pass filter that tances (Akinci et al., 2001; Malagnini et al., 2007, 2008; Ford
samples the seismic spectra well below the events’ corner et al., 2008). This is the first study to calibrate a tool for the
frequencies of the small earthquakes. RVT (in the form of calculation of M w by spectral correction for the seismically
equation 4) shows that the peak values observed on the Wood– active region of the Upper Tiber Valley.
Anderson time histories are proportional to the seismic mo- Following the technique recently defined by Malagnini
ments of the corresponding small earthquakes. For larger and Dreger (2016), we compute the seismic moments (and
events, the 1:1 scaling between M w and ML is the result of magnitudes) from the rms averages of the low-frequency spectral
the interference between the events’ corner frequencies and plateaus (Fourier amplitudes) of the 100 largest events (arbitrary
the band-pass filter operated by the coupled responses of the choice). On the peak amplitudes, we compute all moments by
Wood–Anderson filter and the Earth. taking the ratios of the rms-averaged peak amplitudes in a
We now need to quantify the upper threshold of validity suitable frequency range. By looping through all events and
of equation (10). Figure 2c shows the theoretical displace- all the 100 largest events, we obtain the averages and stan-
2406 Short Note

(a) (b)

(c)

Figure 2. (a) Theoretical moment-rate spectra of earthquakes for different magnitude values (black solid lines) and their magnitude-
dependent stress-drop values, based on the Brune (1970, 1971) earthquake spectral model. The light-green band represents a relatively
narrow band-pass filter that samples the seismic spectra below the events’ corner frequencies. The vertical-dashed line marks the dominant
frequency of the Wood–Anderson-filtered displacement spectrum at Mw 4.0 (∼1:3 Hz). (b) The red line shows the Wood–Anderson
transfer function, whereas blue and green lines represent the attenuated Wood–Anderson filters (instrument plus Earth) at 20 and
40 km, respectively. (c) Blue and green lines are the Wood–Anderson-filtered displacement spectra for different magnitudes at
20 and 40 km, respectively. The vertical-dashed line marks the dominant frequency of the Wood–Anderson-filtered displacement spectrum
at Mw 4.0 (∼1:3 Hz).

dard deviations for the seismic moments of all earthquakes in 2

M w
M L  1:15: 11
EQ-TARGET;temp:intralink-;df11;313;111

the data set. Based on these values, Figure 3 shows Mw ver- 3

sus ML for our data set. The dashed black line, obtained by All information about the earthquakes of our data set (includ-
trial-and-error tests, represents the relationship ing event IDs, hypocentral coordinates, origin times, local
Short Note 2407

Some figures were made using the Generic Mapping Tools

4 v. 4.2.1 (www.soest.hawaii.edu/gmt, last accessed December
2015; Wessel and Smith, 1998).

3
Acknowledgments
We would like to thank Raffaele Di Stefano and Luisa Valoroso for
2 providing The AltotiBerina near fault Observatory earthquake catalog. Elisa
Tinti and Marco Cattaneo are acknowledged for their important suggestions
MW

and helpful discussions that improved this work. The authors are also grate-
ful to Francesca Bianco for reviewing the article.
1
This work was supported by Progetto PREMIALE 2011: Studio multi-
disciplinare della fase di preparazione di un terremoto (cap. 3,1.01.01, cod.
team 0551.020) and by Progetto Terremoti 2016: Reconciling Differences
0 Between Spectral- and Recurrence-Interval-Based Earthquake Source Char-
acterization, and Scaling.
Mw = (2/3) ML+ 1.15

−1
References
Aki, K., and P. G. Richards (2002). Quantitative Seismology, Second Ed.,
−1 0 1 2 3 4
University Science Books, San Francisco, California, 704 pp., ISBN:
ML
0-935702-96-2.
Akinci, A., L. Malagnini, R. B. Herrmann, N. A. Pino, L. Scognamiglio, and
Figure 3. Comparison of Mw with ML for events in the Alto H. Eyidogan (2001). High-frequency ground motion in the Erzincan
Tiberina region. The dashed black line is derived from the relation- region, Turkey: Inferences from small earthquakes, Bull. Seismol. Soc.
ship shown in the figure. Am. 91, 1446–1455.
Barchi, M., R. Minelli, and G. Pialli (1998). The CROP 03 profile: A syn-
thesis of results on deep structures of the northern Apennines, Mem.
magnitude, moment magnitudes, seismic moments, and their Soc. Geol. Ital. 52, 383–400.
standard deviations) is provided in Ⓔ Table S1. Bethmann, F., N. Deichmann, and P. M. Mai (2011). Scaling relations of
local magnitude versus moment magnitude for sequences of similar
earthquakes in Switzerland, Bull. Seismol. Soc. Am. 101, no. 2,
Conclusions 515–534, doi: 10.1785/0120100179.
Boncio, P., F. Brozzetti, and G. Lavecchia (2000). Architecture and seismo-
We use RVT to show that the M L of a small earthquake is tectonics of a regional low-angle normal fault zone in central Italy,
proportional to the logarithm of its seismic moment. Conse- Tectonics 19, 1038–1055.
quently, we define a general relationship (equation 10) that is Brune, J. N. (1970). Tectonic stress and the spectra of seismic shear waves
from earthquakes, J. Geophys. Res. 75, 4997–5009.
globally valid up to M w 4, in which the constant C′ only de-
Brune, J. N. (1971). Correction, J. Geophys. Res. 76, 5002.
pends on the characteristics of the crustal attenuation of the Cartwright, D. E., and M. S. Longuet-Higgins (1956). The statistical distri-
study region that needs to be locally calibrated. We test our find- bution of the maxima of a random function, Proc. Math. Phys. Sci.
ings on a high-quality data set of 30,000 selected waveforms 237, 212–232.
from 1191 well-relocated earthquakes in the 0 ≤ M L ≤ 3:8 Chiaraluce, L., A. Amato, S. Carannante, V. Castelli, M. Cattaneo,
M. Cocco, C. Collettini, E. D'Alema, R. Di Stefano, D. Latorre, et al.
range. For our studied region (ATF area in the northern Apen-
(2014). The Alto Tiberina near fault Observatory (northern Apennines,
nines, Italy), the obtained free parameter of equation (10) is Italy), Ann. Geophys. 57, no. 3, doi: 10.4401/ag-6426.
C′
1:15 (see equation 11). Based on our general relation- Chiaraluce, L., C. Chiarabba, C. Collettini, D. Piccinini, and M. Cocco
ship the M L values in catalogs of past earthquakes may be (2007). Architecture and mechanics of an active low-angle normal fault:
correctly converted to Mw , allowing a more effective use of Alto Tiberina fault, northern Apennines, Italy, J. Geophys. Res. 112,
no. B10310, doi: 10.1029/2007JB005015.
the events magnitudes, for studies on earthquakes source
Chiaraluce, L., C. Collettini, M. Cattaneo, and G. Monachesi (2014).
physics or hazard assessment. The shallow boreholes at The Altotiberina near fault Observatory
(TABOO; northern Apennines of Italy), Sci. Dril. 2, 1–5, doi:
10.5194/sd-17-31-2014.
Data and Resources De Luca, G., M. Cattaneo, G. Monachesi, and A. Amato (2009). Seismicity
in central and northern Apennines integrating the Italian national and
Seismograms and the earthquake catalog used in this regional networks, Tectonophysics 476, 121–135.
study were collected and generated in the framework of Deichmann, N. (2006). Local magnitude, a moment revisited, Bull. Seismol.
The AltotiBerina near fault Observatory project (http:// Soc. Am. 96, no. 4, 1267–1277.
taboo.rm.ingv.it, last accessed March 2016). The raw wave- Ford, S., D. S. Dreger, K. Mayeda, L. Malagnini, and W. S. Phillips (2008).
Regional attenuation in northern California: A comparison of five 1-D
forms are available at the European Integrated Data Archive
methods, Bull. Seismol. Soc. Am. 98, no. 4, 2033–2046, doi: 10.1785/
repository at http://145.23.252.222/eida/webdc3/index.html 0120070218.
(last accessed April 2016), whereas the products are listed Hanks, T. C., and D. M. Boore (1984). Moment-magnitude relations in
in Ⓔ Table S1 (available in the electronic supplement). theory and practice, J. Geophys. Res. 89, no. B7, 6229–6235.
 2408 Short Note

Hanks, T. C., and H. Kanamori (1979). A moment magnitude scale, J. Geo- sections in the northern Apennines (Italy), Tectonics 30, TC6002, doi:
phys. Res. 84, 2348–2350. 10.1029/2011TC002890.
Herrmann, R. B., L. Malagnini, and I. Munafò (2011). Regional moment Petrosino, S., L. De Siena, and E. Del Pezzo (2008). Recalibration of the
tensors of the 2009 L’Aquila earthquake sequence, Bull. Seismol. magnitude scales at Campi Flegrei, Italy, on the basis of measured path
Soc. Am. 101, 975–993, doi: 10.1785/0120100184. and site and transfer functions, Bull. Seismol. Soc. Am. 98, no. 4, 1964–
Jost, M. L., T. Büßelberg, Ö. Jost, and H. P. Harjes (1998). Source parameters 1974, doi: 10.1785/0120070131.
of injection-induced microearthquakes at 9 km depth at the KTB deep Richter, C. F. (1935). An instrument earthquake magnitude scale, Bull. Seis-
drilling site, Germany, Bull. Seismol. Soc. Am. 88, no. 3, 815–832. mol. Soc. Am. 25, 1–32.
Kanamori, H. (1977). The energy release in great earthquakes, J. Geophys. Uhrhammer, R. A., and E. R. Collins (1990). Synthesis of Wood–Anderson
Res. 82, 2981–2987. seismograms from broadband digital records, Bull. Seismol. Soc. Am.
Malagnini, L., and D. Dreger (2016). Generalized free-surface effect and 80, 702–716.
random vibration theory: A new tool for computing moment magni- Valoroso, L., L. Chiaraluce, R. Di Stefano, and TABOO Working Group
tudes of small earthquakes using borehole data, Geophys. J. Int. 206, (2014). Seismic activity along a low-angle normal fault: The case study
no. 1, 103–113, doi: 10.1093/gji/ggw113. of the Alto Tiberina fault (northern Apennines, Italy), San Francisco,
Malagnini, L., A. Akinci, K. Mayeda, I. Munafò, R. B. Herrmann, and A. American Geophysical Union (AGU) Fall Meeting, Oral presentation,
Mercuri (2011). Characterization of earthquake-induced ground mo- San Francisco, California, 15–19 December 2014.
tion from the L’Aquila seismic sequence of 2009, Italy, Geophys. J. Wessel, P., and W. H. F. Smith (1998). New, improved version of generic
Int. 184, 325–337, doi: 10.1111/j.1365-246X.2010.04837.x. mapping tools released, Earth Space Sci News 79, doi: 10.1029/
Malagnini, L., R. B. Herrmann, I. Munafò, M. Buttinelli, M. Anselmi, A. 98EO00426.
Akinci, and E. Boschi (2012). The 2012 Ferrara seismic sequence:
Regional crustal structure, earthquake sources, and seismic hazard,
Geophys. Res. Lett. 39, no. 19, doi: 10.1029/2012GL053214.
Malagnini, L., K. Mayeda, R. Uhrhammer, A. Akinci, and R. B. Herrmann Istituto Nazionale di Geofisica e Vulcanologia
(2007). A regional ground motion excitation/attenuation model for the via di Vigna Murata 605
00143 Rome, Italy
San Francisco region, Bull. Seismol. Soc. Am. 97, no. 3, 843–862, doi:
irene.munafo@ingv.it
10.1785/0120060101.
luca.malagnini@ingv.it
Malagnini, L., L. Scognamiglio, A. Mercuri, A. Akinci, and K. Mayeda lauro.chiaraluce@ingv.it
(2008). Strong evidence for non-similar earthquake source scaling in cen-
tral Italy, Geophys. Res. Lett. 35, L17303, doi: 10.1029/2008GL034310.
Mirabella, F., F. Brozzetti, A. Lupattelli, and M. R. Barchi (2011). Tectonic Manuscript received 29 April 2016;
evolution of a low angle extensional fault system from restored cross Published Online 6 September 2016

View publication stats