Content uploaded by Licia Faenza
Author content
All content in this area was uploaded by Licia Faenza on Nov 23, 2018
Content may be subject to copyright.
ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7238
The ShakeMaps
of the Amatrice, M6, earthquake
LICIA FAENZA*, VALENTINO LAUCIANI, ALBERTO MICHELINI
Istituto Nazionale di Geofisica e Vulcanologia,
Centro Nazionale Terremoti, Italy
* licia.faenza@ingv.it
Abstract
In this paper we describe the performance of the ShakeMap software package (Wald et al., 1999; Worden
and Wald, 2016) obtained from the fully automatic procedure to estimate ground motions, based on
manually revised location and magnitude, during the main event of the Amatrice sequence with special
emphasis to the M6 main shock, that struck central Italy on the 24th August 2016 at 1:36:32 UTC.
Our results show that the ShakeMap procedure we developed in the last years, with real-time data ex-
change among those institutions acquiring strong motion data, produces a reliable and useful description
of the ground motion experienced throughout a large region in and around the epicentral area.
The prompt availability of the rupture fault model, within three hours after the earthquake occurrence,
provided a better description of the level of strong ground motion throughout the affected area. Progressive
addition of station data and manual verification of the data insures improvements in the description of the
experienced ground motions. In particular, comparison between the MCS (Mercalli-Cancani-Sieberg) in-
tensity shakemaps and preliminary field macroseismic reports show overall agreement within the limita-
tions imposed by the station geometry. Finally the overall spatial pattern of the ground motion of the main
shock is consistent with reported rupture directivity toward NW and reduced levels of ground shaking
toward SW probably linked to the peculiar source effects of the earthquake.
I. INTRODUCTION
hakeMap is a software package [Wald et
al. 1999a; Worden et al. 2010, Worden and
Wald, 2016] that can be used to generate
maps of ground shaking for various peak
ground motion (PGM) parameters, including
the peak ground acceleration (PGA), peak
ground velocity (PGV), and spectral accelera-
tion response (PSA) at 0.3 s, 1.0 s and 3.0 s, and
instrumentally derived intensities.
The primarily aim of the implementation of
the ShakeMap code [Michelini et al., 2008] at
the Istituto Nazionale di Geofisica e Vulcanologia
(INGV; National Institute of Geophysics and
Volcanology) is to support the Dipartimento
della Protezione Civile (DPC; Civil Protection
Department) providing a first order assess-
ment of the experienced ground shaking to
better direct the rescue teams and planning the
emergency responses in the first few hours fol-
lowing a damaging earthquakes.
At its core, ShakeMap is a seismologically based
interpolation algorithm that exploits the avail-
able data of the observed ground motions and
the available seismological knowledge to
pro-
duce maps of ground motion at local and re-
gional scales. Of particular importance when cal-
culating
the maps is the availability of
observed
data to accurately reproduce the ground
shaking
experienced, especially in the near source.
Thus, in addition to data that are essential to
derive realistic and accurate results, the fun-
damental ingredients for obtaining accurate
maps are: the ground-motion prediction equa-
tion (GMPE), as a function of distance at dif-
ferent periods, and for different magnitudes;
and realistic descriptions of the amplifications
S
ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7238
that the local site geology induces on the in-
coming seismic wavefield; i.e., the site effects.
In its current version, ShakeMap relies on re-
gional attenuation laws and local site amplifi-
cations based on the S-wave velocities in the
uppermost 30 m (VS30) to generate its PGM
maps [Michelini et al., 2008].
In this report, we start with a chronicle of the
generation of the shakemaps for two main
events of the sequence that struck Central Italy
the 24 August 2016 between the towns of Ama-
trice and Norcia [Scognamiglio et al., this is-
sue; Michele et al., this issue] and we conclude
with a comment on the procedure we adopted.
The main shock caused severe damage in the
small towns in Central Italy, including Ama-
trice and Accumuli and in dozens of villages
located along the river Tronto and almost 300
casualties [Azzaro et al, this issue]. The shak-
ing was felt throughout central Italy.
In the recent past, moderate seismic events
struck this area (Gubbio 1984, Mw 5.6; Colfio-
rito 1997, Mw 6.0; Norcia 1979, Mw 5.9;
L’Aquila 2009, Mw 6.1), all with focal mecha-
nisms consistent within the regional NE–SW
extension of the stress field. The main shock of
the Amatrice sequence occurred along a fault
alignment which extends from Mt. Vettore to
Mt. Gorzano, which is external (to the E) of the
tectonic alignment that develops from Gubbio
to Colfiorito and, to the south, extends to the
area struck by the 2009 L’Aquila sequence.
As of October 6, 2016, shakemaps have been
determined for a total of 74 earthquakes with
M>=3.5.
II. THE AUGUST 24, 2016 ML 6.0
EARTHQUAKE
In this section, we present a concise de-
scription of the evolution of the ShakeMap
de-
termination for the August 24, 2016, Ml 6.0 earth-
quake.
i) The automatic final earthquake location
(origin time, 01:36:32 UTC; latitude, 42.69 N; lon-
gitude, 13.2 E; depth, 4 km) was available within
5 minutes of the origin time (01:41:37 UTC).
ii) The manually revised location became
available 17 minutes (1:53 UTC) after the ori-
gin time, with a similar location (42.71 N, 13.22
E), and depth (4 km).
iii) For the magnitude estimation, the first
automatic determination, which became avail-
able within about 5 min from the origin time,
was ML 6.0 The manual revision, which was
available after 12 min, confirmed the same value.
The first moment magnitude was available 1.5
h later, as Mw 6.0 [Scognamiglio et al., 2009;
Scognamiglio et al., this issue].
iv) The first shakemap based on the auto-
matic location and magnitude became avail-
able at 1:43:16 UTC. This map included only
the first data available and lacks of near epi-
center stations.
v) The first map that included the available
RAN (“Rete Accelerometrica Nazionale” man-
aged by the Department of Civil Protection,
DPC) strong motion data and the more distant
broadband data of the Italian seismic network
(RSN) including the local networks of the Uni-
versities of Genova and Trieste, OGS, AMRA,
among others became available at 02:27 UTC,
~46 minutes after the automatic location and
~33 minutes after the revised location (version
3, figure 1). The spatial coverage of the epicen-
tral area is somewhat denser though the near
finite source area is still only partly covered
(only NRC and RQT station PGM data became
available).
vi) Based on the time domain moment
ten-
sor solution [Scognamiglio et al., this issue], the
scaling laws [Wells and Coppersmith, 1994],
and
the geology, with the analysis of the active tec-
tonic structures in the area and their orienta-
tion, the first maps with the fault
included (ver-
sion 5) were published at 04:39 UTC,
about 3
hours after the earthquake occurrence (figure
2). This map better constrains the shaking in
the epicentral area by taking into account the
fault finiteness. Insertion of the fault is based,
however, on a manual procedure which re-
quires the availability of the manually revised
moment tensor and rapid selection of one of
the rupture planes.
ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7238
Figure 1. The Shakemaps of the main shock using automatic processing for the INGV PGM data and the PGM data
provided by DPC. Left: MCS derived instrumental intensity; Center: PGA; Right: PGV.
Figure 2. The Shakemaps of the main shock as in Figure 1 after inserting the fault. Left: MCS derived instrumental in-
tensity; Center: PGA; Right: PGV.
In the following days, the maps were updated
with more close-in and distant stations and the
Mercalli Intensity instrumentally derived scale
of Wald et al. [1999] was replaced by the MCS
[Mercalli-Cancani-Sieberg; Sieberg, 1930] in-
tensity scale calibrated for Italy (Faenza and
Michelini, 2010, 2011). This second change was
adopted because MCS intensities have been
found more informative to non-expert audi-
ences unfamiliar with instrumental ground
motion parameters. More specifically, in the
INGV ShakeMap implementation [Michelini et
al., 2008], the instrumentally derived intensity
values are derived from the conversion of
PGM into intensity values as proposed by
Wald et al. [1999b]. This regression, however,
is based on the Mercalli Modified scale cali-
brated using intensity and PGM data collected
in California. In Italy, the analysis of historical
seismicity through the use of the macroseismic
intensity data has a long tradition and the
MCS intensity scale has been long adopted. To
attain homogeneity between the instrumen-
tally derived intensity maps and the observed
Italian macroseismic intensities, new regres-
sion
relations between PGM and MCS intensity
ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7238
Figure 3. The Shakemaps of the main shock using the revised data obtained from the engineering strong motion DB
(http://esm.mi.ingv.it). Left: MCS derived instrumental intensity; Center: PGA; Right: PGV.
data were proposed by Faenza and Michelini
[2010, 2011] but never implemented into the
ShakeMap procedure at INGV since it was
sought to maintain consistency with similar
instrumental values elsewhere worldwide.
These new relations for MCS were inserted in
ShakeMap starting with the mainshock of this
sequence to better support local needs in Italy.
The shakemaps are by their nature determined
very rapidly right after an earthquake using
automatic procedures. In the following days,
however, new PGM data became available de-
termined from manually revised waveforms.
At INGV the strong motion data are verified
and archived in the Engineering Strong
Mo-
tion DB (ESM; http://esm.mi.ingv.it) to-
gether
with the associated PGM parameters.
Therefore on September 22, 2016, once the re-
vised data became available we replaced the
PGM readings and re-determined the maps
(figure 3).
Overall, this procedure allows a progressive
improvement in the quality of the shakemaps
as additional data become available and man-
ual intervention is performed.
In figure 4 we have summarised the improve-
ments obtained by calculating the differences
(for PGA) between the final shakemaps shown
in figure 3 and the maps obtained with the
automatic processing (figures 2 and 1). In sum-
mary, we note that i) the addition of the AMT
station data condition strongly the values of PGA
at the southern end of the rupture plane and ii)
including the fault is important to improve
and extend the pattern of ground motion near
and above the fault.
The shakemaps shown in figures 1-3 show the
maps of the main shock as they result at the
end of the steps outlined above. The largest
values of the ground motion occur next or
above the fault plane as resulting from the
largest values of acceleration recorded by the
three closest stations (AMT, RQT and NRC)
that all recorded values around 40%g. This
whole area featured values of PGV on the
horizontal components larger than 20 cm/s
(intensity level VIII). One important feature of
the PGA and PGV maps is that relatively large
accelerations have been recorded from NW to
NE of the earthquake epicenter (see the 7%g
contour line) when compared to those re-
corded SE and especially to the SW. This pat-
tern is likely dependent on the source directiv-
ity observed for the main shock both from the
raw data [INGV-ReLUIS Working Group,
2016] and from the preliminary finite fault in-
version results using strong motion data [Tinti
et al, 2016].
ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7238
Figure 4. Differences between the different PGA shakemaps. Left: final shakemap with revised data and fault map com-
pared with automatically processed data and no fault; center: final shakemap with revised data and fault map compared
with automatically processed data and fault inserted; right: shakemap with fault and automatically processed data com-
pared with the same data but no fault.
III. DISCUSSION
In August 2016, a seismic sequence struck the
Apennine in Central Italy, an area that has a
long history of destructive earthquakes, as
known from historical and macroseismic anal-
ysis [Locati et al, 2015]. In this study, we have
described the determination and the pro-
gressive updating of the shakemaps of the
main shock as additional and more accurate
information became available.
In our experience, the inclusion of observed
data is of fundamental importance for the cal-
culation of shakemaps. Indeed, the quantifica-
tion of the shaking near the epicentre using
only the PGM prediction equations comple-
mented with site-effect corrections is difficult
and prone to macroscopic errors and bias
[Faenza et al. 2011; Lauciani et al, 2012].
Moreover, for larger earthquakes that saturate
the recordings of the velocimeters at and near
the epicenter, the accuracy of the shakemaps
depends also on the prompt availability of
strong-motion data, which, for the Amatrice
main shock, become available shortly after its
occurrence. We have found that the area
stricken by the sequence has in general a dense
enough station coverage to produce reasona-
bly accurate maps of the strong ground shak-
ing. The installation of the temporary stations
in the epicentral area improved, however, the
coverage for the subsequent events.
Comparing figures 1 and 3, it is possible to see
the improvement in the quantification of the
ground shaking with the inclusion of the
source model and new review data (see figure
4 for the differences in terms of PGA). The first
preliminary shakemap (figure 1) remained on-
line for only 3 hours. Figure 2 shows a differ-
ent pattern in the near-source shaking because
of the adoption of the Joyner-Boore distance
measure from the fault location, leading to an
underestimation of the PGM values in the near
source. We note also that this time, we have
not encountered the time delay experienced
previously in the strong motion data exchange
since both the RAN data and the INGV data
were readily available.
Since the intensity scale in our maps adopts the
relations obtained from the regression between
PGM parameters and the MCS intensity values of
Faenza and Michelini (2010, 2011), we have com-
pared the final shakemaps with the preliminary
macroseismic maps available at the time of writ-
ing this work. In figure 5 we show the two maps
represented using the same color scale. We note
ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7238
Figure 5. Comparison between the reported macroseismic intensities and the estimated intensities obtained using
Shakemap represented using the same color palette.. Left: preliminary macroseismic map compiled by the field macro-
seismic teams (from “Rapporto sugli effetti macrosismici del terremoto del 24 Agosto 2016 di Amatrice in scala MCS” a
cura di P. Galli e E. Peronace, Coordinamento del rilievo macrosismico MCS a cura di P. Galli e A. Tertulliani, 2016).
Right: MCS intensity values obtained with the final shakemap updated with the manually revised data.
that the MCS intensity shakemap although
much blurred since it relies on essentially three
main data points in the near fault region (AMT,
NRC, and RQT) can nevertheless provide a
very first information on the ground shaking
in the near fault region. For example, and by
simply determining the population within e.g.
the level VIII MCS, it is possible to obtain very
rapidly an initial estimate of the population
exposed to that intensity level as has been pro-
vided by the USGS that publishes the PAGER
estimates (Earle et al., 2009). For the Amatrice
earthquake we found that by using the MCS
VIII contour as polygonal line within which to
extract the population from the LandScan
popu-
lation DB
(http://web.ornl.gov/sci/landscan/),
it would have
resulted almost immediately that
~10,000 people would have been exposed to
strong ground shaking. Similar estimates can
be done for lower intensity levels.
IV. CONCLUSIONS
Our analysis has shown that for the M6
Ama-
trice earthquake of August 24, 2016, the shake-
maps
produced by INGV
i) became available within a few minutes of
the main shock and they already had an amount
of data that insured a relatively good assess-
ment of the ground shaking experienced in the
Amatrice and nearby villages and towns;
ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7238
ii) inclusion of the finite fault within ~3
hours of the main shock contributed to improve
the accuracy of the maps;
iii) inclusion of additional data as they be-
come progressively available is important to
improve the quality of the maps;
iv) inclusion of thoroughly reviewed data is
equally important to avoid that the maps could
be possibly contaminated by processing errors
always present in automatic procedures;
v) comparison between the MCS intensity
shakemaps and a preliminary map of the mac-
roseismic report compiled by the teams that
have evaluated the macroseismic intensity in
the field indicates a remarkable similarity be-
tween estimated and reported intensities.
vi) the pattern of spatial ground motion ob-
tained is consistent with the preliminary reports
that indicate rupture directivity toward NW
and relatively reduced levels of ground motion
toward SW from the earthquake source.
REFERENCES
[Azzaro et al., 2016] Azzaro R., A. Tertulliani,
F. Bernardini, R. Camassi, S. Del Mese, E. Erco-
lani, L. Graziani, M. Locati, A. Maramai, V.
Pessina, A. Rossi, A. Rovida, P. Albini, L.
Arco-
raci, M. Berardi, C. Bignami, B. Briquela, C.
Ca-
stellano, V. Castelli, S. D’Amico, V. D’Amico,
A.
Fodarella, I. Leschiutta, A. Piscini, M. Sbarra.
The Amatrice 2016 earthquake: macroseismic
survey in the damage area and preliminary
EMS intensity assessment, Annals of Geophys-
ics, 59, Fast Track 5, doi:10.4401/ag-7203.
[Earle et al., 2009] Earle, P.S., Wald, D.J.,
Jaiswal, K.S., Allen, T.I., Marano, K.D., Ho-
tovec, A.J., Hearne, M.G., and Fee, J.M (2009).
Prompt Assessment of Global Earthquakes for
Response (PAGER): A system for rapidly de-
termining the impact of global earthquakes
worldwide. U.S. Geological Survey Open-File
Report 2009-1131.
[Faenza and Michelini, 2010] Faenza L. and A.
Michelini, Regression analysis of MCS inten-
sity and ground motion parameters in Italy
and its application in ShakeMap, Geophys. J.
Int, 180(3), 1138-1152, doi:10.1111/j.1365-
246X.2009.04467.x.
[Faenza et al., 2011] Faenza, L., V. Lauciani and
A. Michelini (2011). Rapid determination of the
shake maps for the L'Aquila main shock: a
critical analysis, B. Geofis. Teor. Appl., 52, 407-
425.
[Faenza and Michelini, 2011] Faenza, L., and A.
Michelini (2011), Regression analysis of MCS
intensity and ground motion spectral accelera-
tions (SAs) in Italy, Geophys. J. Int, 1-16,
doi:10.1111/j.1365-246X.2011.05125.x.
[INGV-ReLUIS Working Group, 2016] INGV-
ReLUIS Working Group (2016), Preliminary
study of Rieti earthquake ground
motion data
V5, 1-87, doi:10.13140/RG.2.2.2
7933.92641/1.
[Available at http://www. reluis. it].
[Lauciani et al., 2012] Lauciani V., L. Faenza
and A. Michelini (2012), SchakeMap during
the Emilia sequence, Annals of Geophysics, 55,
4, 2012; doi: 10.4401/ag-6160.
[Locati et al, 2015] Locati M., Camassi R., Ro-
vida A., Ercolani E., Bernardini F., Castelli V.,
Caracciolo C.H., Tertulliani A., Rossi A., Azza-
ro R., D’Amico S., Conte S., Rocchetti E.
(2016).
DBMI15, the 2015 version of the Italian Macrosei-
smic Database. Istituto Nazionale
di
Geofisica e
Vulcanologia; doi:http://doi.org/10.6092/INGV.
IT-DBMI15.
[Michele et al., 2016] Michele M., Di Stefano
R.,
Chiaraluce L., Cattaneo M., De Gori P.,
Monachesi G., Latorre D., Marzorati S., Valor-
oso L., Ladina C., Chiarabba C., Lauciani V. and
M. Fares. The Amatrice 2016 seismic sequence:
a preliminary look to the mainshock and after-
ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7238
shocks distribution, Annals of Geophysics, 59,
Fast Track 5, doi:10.4401/ag-7277.
[Michelini et al., 2008] Michelini A., L. Faenza,
V. Lauciani and L. Malagnini (2008). Shake-
Maps implementation in Italy, Seismol. Res.
Lett., 79, 688-697.
[Scognamiglio et al., 2009] Scognamiglio, L., E.
Tinti, and A. Michelini (2009), Real-Time De-
termination of Seismic Moment Tensor for the
Italian Region, Bull. Seismol. Soc. Am., 99 (4),
2223-2242.
[Scognamiglio et al., this issue] Scognamiglio
L, E. Tinti, M. Quintiliani, The 2016 Amatrice
seismic sequence: Fast determination of the
time domain moment tensors and finite fault
model analysis of the ML 5.4 aftershock, This
issue.
[Sieberg, 1930] Sieberg A. (1930), Geologie der
Erbbeben, Handbuch der Geophysik, 2, 4, 552-
555.
[Tinti et al, 2016] Tinti, E., L. Scognamiglio, A.
Michelini, and M. Cocco (2016), Slip heteroge-
neity and directivity of the M L6.0, 2016, Ama-
trice earthquake estimated with rapid finite-
fault inversion, Geophys. Res. Lett, 1-8,
doi:10.1002/2016GL071263.
[Wald et al., 1999a] Wald, D.J., Quitoriano, V.,
Heaton, T.H., and Kanamori, H., (1999a), Rela-
tionship between Peak Ground Acceleration,
Peak Ground Velocity, and Modified Mercalli
Intensity in California, Earthquake Spectra, 15
(3), p. 557-564.
[Wald et al., 1999b] Wald, D.J., Quitoriano,
V.,
Heaton, T.H., Kanamori H, Scrivner, C.W.
and
Worden C.B. (1999n). Trinet ‘ShakeMaps’:
rapid generation of peak ground motion and
intensity maps for earthquakes in southern
California, Earthq. Spectra, 15, 537.
[Wells and Coppersmith, 1994]. Wells D.L.,
and K.J. Coppersmith (1994). New empirical
relationships among magnitude, rupture length,
rupture width, rupture area, and surface dis-
placement, B. Seismol. Soc. Am, 84, 974-1002.
[Worden et al., 2010 ] Worden, C.B., Wald, D.J.,
Allen, T.I., Lin, K., Garcia, D. and Cua G. (2010).
A revised ground-motion and intensity inter-
polation scheme for ShakeMap. B. Seismol.
Soc. Am. 100, 3083-3096.
[Worden and Wald, 2016] Worden, C. B., and
D. J. Wald (2016), ShakeMap Manual, version
2.0, 1-113.
