ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 1 

Coseismic displacement waveforms for 

the 2016 August 24 Mw 6.0 Amatrice 

earthquake (central Italy) carried out 

from High-Rate GPS data 
 

A. AVALLONE1,*, D. LATORRE1, E. SERPELLONI1, A. CAVALIERE2, A. HERRERO3, G. 

CECERE1, N. D'AGOSTINO1, C. D'AMBROSIO1, R. DEVOTI1, R. GIULIANI4, M. MATTONE4, S. 

CALCATERRA5, P. GAMBINO5, L. ABRUZZESE1, V. CARDINALE1, A. CASTAGNOZZI1, G. DE 

LUCA1, L. FALCO1, A. MASSUCCI1, A. MEMMOLO1, F. MIGLIARI1, F. MINICHIELLO1, R. 

MOSCHILLO1, L. ZARRILLI1 AND G. SELVAGGI1. 
 

1Istituto Nazionale di Geofisica e Vulcanologia, Centro Nazionale Terremoti 
2Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Bologna 
3Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Roma 1 

4Ufficio Rischio Sismico e Vulcanico, Dipartimento della Protezione Civile, Roma 
5Ufficio Geofisica, Istituto Superiore per la Protezione e la Ricerca Ambientale, Roma 

*antonio.avallone@ingv.it 

Abstract 

Abstract: We used High-Rate sampling Global Positioning System (HRGPS) data from 52 permanent sta-

tions to retrieve the coseismic dynamic displacements related to the 2016 August 24 Mw 6.0 Amatrice 

earthquake. The HRGPS position time series (named hereinafter "GPSgrams") were obtained with two dif-

ferent analysis strategies of the raw GPS measurements (Precise Point Positioning [PPP] and Double-

Difference [DD] positioning approaches using the Gipsy-Oasis II and the TRACK (GAMIT/GLOBK) 

software, respectively). These GPSgrams show RMS accuracies mostly within 0.3 cm and, for each site, an 

agreement within 0.5 cm between the two solutions. By using cross-correlation technique, the GPSgrams 

are also compared to the doubly-integrated strong motion data at sites where the different instrumentations 

are co-located in order to recognize in the GPSgrams the seismic waves movements. The high values (most-

ly greater than 0.6) of the cross-correlation functions between these differently-generated waveforms 

(GPSgrams and the SM displacement time-histories) at the co-located sites confirm the ability of GPS in 

providing reliable waveforms for seismological applications. 
 

I. INTRODUCTION 

O n August 24, 2016, at 01:36:32 (UTC 
time, http://cnt.rm.ingv.it, Marchetti et al., 

2016), a Mw 6.0 earthquake struck the region of 

the Central Apennines (Italy) between the 

towns of Norcia and Amatrice (Figure 1), 

where peak ground accelerations values up to 

0.9g were recorded 

(http://ran.protezionecivile.it/). The main 



ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 2 

shock triggered an aftershock sequence that 

involved a crustal volume extending SE-NW 

for ~30 km and down to ~15 km of depth. The 

main shock and the largest part of aftershock 

events show focal solutions 

(http://cnt.rm.ingv.it/tdmt.html, Scognamiglio 

et al., 2016; http://www.bo.ingv.it/RCMT) 

characterized by almost pure extension on 

NW-SE fault planes, in agreement with geo-

detic measurements of the main event (INGV 

CNT GPS Working Group, 2016; Cheloni et al., 

2016) and with the interseismic SW-NE exten-

sion characterizing this sector of the Apen-

nines (e.g., Galvani et al., 2012; D’Agostino, 

2014). 

The ground displacements associated to the 

August, 24 Mw 6.0 main shock were recorded 

by a number of High-Rate sampling (from 1 to 

0.05 s) continuous GPS (HRGPS) stations (Fig-

ure 1) belonging to several networks devel-

oped for both scientific and surveying purpos-

es. In detail, raw phase data were obtained 

from the following GNSS networks or agen-

cies: RING (INGV RING Working Group, 

2016; http://ring.gm.ingv.it), ISPRA 

(http://www.isprambiente.gov.it), DPC 

(http://www.protezionecivile.gov.it), Regione 

Lazio (http://gnss-regionelazio.dyndns.org), 

Regione Abruzzo 

(http://gnssnet.regione.abruzzo.it), ITALPOS 

(http://it.smartnet-eu.com) and NETGEO 

(http://www.netgeo.it). The main shock was 

also recorded by several strong motion (SM) 

stations belonging to the INGV Rete Sismica 

Nazionale (RSN, INGV Seismological Data 

Centre, 1997) and the DPC Rete Accelerometi-

ca Nazionale (RAN, 

http://ran.protezionecivile.it) accelerometric 

networks.  

This work is within the line of GPS seis-

mology (Larson et al., 2003; Bock et al., 2004, 

Larson et al., 2009), in which the estimation of 

the coseismic dynamic displacements is car-

ried out following different methods (i.e. Pre-

cise Point Positioning [PPP, Zumberge et al., 

1997; Bertiger et al., 2010], Double-Difference 

(DD) Positioning [Herring et al., 2010]), vari-

ometric [Colosimo et al., 2011] and ambiguity 

constrained time-differences [Li et al., 2013] 

approaches). In this work we will describe the 

high-rate GPS (HRGPS) time series (named 

hereinafter "GPSgrams") following the PPP 

and DD approaches by using two different sci-

entific packages (Gipsy/Oasis and 

GAMIT/GLOBK), and compare the results. The 

GPSgrams have been also compared with the 

displacement time histories carried out from 

the double integration of the SM data at some 

stations, where the different instruments are 

approximately co-located (Figure 1). 

II. HRGPS TIME SERIES 

The available HRGPS data were analyzed 

with two different approaches (PPP and DD 

strategies), using two non-commercial, geodet-

ic-quality, software modules for kinematic 

processing: the TRACK module of 

GAMIT/GLOBK developed at the Massachu-

setts Institute of Technology 

(http://wwwgpsg.mit.edu/~simon/gtgk/, Her-

ring et al., 2010), and the GD2P module of 

GIPSY/OASIS II, developed at the Jet Propul-

sion Laboratory (http://gipsy.jpl.nasa.gov, Ber-

tiger et al., 2010).  For the description of the de-

tailed aspects related to the analysis strategies 

we refer to Avallone et al. (2012). For the gen-

eral description of the detailed aspects related 

to the analysis strategies we refer to Avallone 

et al. (2012). However, with respect to this pre-

vious work, during the analysis we applied 

improved available models of the troposphere 

(Bohem et al. 2006) and second order iono-

sphere (Kedar et al., 2003) estimations 

http://smartnet.leica-geosystems.it/


ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 3 

 
Figure 1: Locations of the HRGPS stations (blue diamonds) and of the SM stations (red triangles) operating during the 

Amatrice 2016 main shock. The star represents the location of the main shock, whereas the green dots show the distribu-

tion of the aftershocks occurred within the first two days from the main shock (http://cnt.rm.ingv.it, Marchetti et al., 

2016). The focal mechanism (Scognamiglio et al., 2016; http://cnt.rm.ingv.it/tdmt.html) is normal. The GPSgrams of the 

labeled HRGPS sites are shown in Figure 2A-B, whereas the open squares indicate the couples of HRGPS and SM sites 

used for the comparisons in Figure 3. 

 

(VMF1 grid and IONEX files, respectively). It 

is worth noting that the two software packages 

are based on a completely different approach 

in the reduction of the raw phase data: the 

GD2P module follows the PPP approach, 

where the kinematic station only needs fiduci-

al high-rate satellite orbits and clocks infor-

mation; the TRACK module performs relative 

kinematic positioning and it needs at least one 

reference station out of the epicentral area (i.e., 

assumed fixed in the time interval affected by 

the dynamic displacements in the epicentral 

area), and one (or more) kinematic station. In 

this study, after an analysis of the quality of 

the satellite tracking at different potential ref-

erence (fixed) stations (i.e. sky plots, multipath 

and cycle-slips metrics) and of the resolved pa-

rameters (i.e. ambiguity fixing), the station 

CASS (Cassino, Lat: 41.49; Lon: 13.83, ~140 km 

from the epicenter), belonging to the Regione 

Lazio GNSS network, was chosen as the refer-

ence site for the analysis with TRACK of all the 

stations located within 50 km from the epicen-

ter. The relative double-difference positioning 

performed in the TRACK analysis likely al-

lows to remove, or at least minimize, a com-

mon-mode regional contribution that could be, 

on the contrary, present, as low-frequency con-

tribution, in the PPP solutions. For this reason, 

we applied to the GD2P solutions a spatial fil-

tering commonly used for filtering long-term 

(daily positions) time series (Wdowinski et al., 

1997). In detail, we firstly selected the 

GPSgrams not affected by coseismic dynamic 

displacements in the time interval 0-40s. These 

GPSgrams, corresponding to the 9 sites be-



ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 4 

longing to the Regione Lazio GNSS network 

located outside the epicentral area (more than 

50 km from the epicenter), were then stacked 

and averaged. Finally, we removed this aver-

aged signal from the GPSgrams located in the 

epicentral area.  

By using the SAC software (Goldstein et al., 

2005), distributed by IRIS (Incorporated Re-

search Institutions for Seismology), the North 

and East components of the measured ground 

displacements at each site were rotated with 

respect to the epicenter and presented in the 

Radial-Transverse coordinate system, since the 

radial (R) component mainly contains SV and 

P-wave arrivals, while the transverse (T) com-

ponent presents predominantly SH-wave 

modes. This representation may be not optimal 

in near source regions as in the case of NRCI 

and AMAT stations (Figure 1). Nevertheless, 

in this particular case, we can adopt the RT 

representations because both the stations are 

located along the strike of the M6 event and 

the RT components correspond to the fault-

parallel and fault-normal components, respec-

tively. 

In Figure 2, some examples of comparison 

on both radial and transverse components as 

obtained by the two different solutions are 

shown for the sites labeled in Figure 1, where-

as the comparison between both the solutions 

for all the 52 sites are shown in Figure SM1 of 

the Supplementary Material. A first compari-

son is focused on the two GPS sites located 

closer to the epicenter (AMAT, Amatrice, ~9 

km to the SE, and NRCI, Norcia, ~14 km to the 

NW, Figure 2A). AMAT acquired data at 0.1 s 

sampling rates, thus allowing potentially low 

aliasing artifacts in the observations (Smalley, 

2009; Avallone et al., 2011), whereas NRCI ac-

quired data at 1 s sampling rate. Because of 

their proximity to the epicenter, they experi-

enced the largest coseismic dynamic displace-

ments: AMAT shows larger values of peak-to-

peak displacements on the radial component 

(~16.6 cm) than on the transverse one (~8.2 

cm), whereas NRCI shows generally lower 

peak-to-peak displacements than AMAT, but 

larger on the transverse component (~15.0 cm) 

than on the radial one (~8.2 cm). The compari-

sons between the GD2P and TRACK results 

are also shown in Figure 2B at some sites lo-

cated between ~35 and ~50 km from the epi-

center, where the static deformation due to the 

near field is expected to be negligible (as 

shown in Cheloni et al., 2016) or within the ac-

curacy of the HRGPS time series. The 

GPSgrams shown from the top to the bottom 

of the Figure 2B are sorted by increasing azi-

muth angle and seem to point out the SH arri-

vals on the T component and the surface 

waves arrivals on the R component in the time 

ranges 10-15 s and 15-25 s, respectively.   

Due to power failures occurred a few sec-

onds after the main shock, likely related to the 

S-wave arrivals ground motion, the available 

data and the resulting GPSgrams at some sites 

(LNSS, ASCC, ASCO and CAMR) are sudden-

ly truncated. One of these examples (ASCC) is 

shown in Figure 2B. 

The comparison between GD2P and 

TRACK solutions (in both the Figure 2B and 

the Figure SM1) shows a remarkable agree-

ment in the detection of the higher-frequency 

coseismic dynamic displacements, observing 

comparable peak-to-peak values. To quantify 

the noise level of our GPSgrams, for all the 

sites we calculated the RMS of the position 

time series in a 10-s time window before the 

earthquake origin time. In Figure 2C, the his-

tograms of the RMS distributions of all the 

available sites are shown for both the R and T 

components and for the GD2P and TRACK 



ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 5 

  
Figure 2: A) Comparison between the GD2P and TRACK GPSgrams for the two stations closest to the epicenter 

(AMAT, Amatrice, and NRCI, Norcia) on the Radial and Transverse components; B) Radial (left) and Transverse 

(right) components of the GD2P (red) and TRACK (green) GPSgrams obtained for the sites labeled in Figure 1, between 

35 and 50 km from the epicenter in a time interval of 40 s starting from the main shock origin time (01:36:32 UTC, 

http://cnt.rm.ingv.it). In both the panels, from top to down the sites are sorted by the increasing azimuth degree. Labels 

and numbers on each time series represent the name of the sites, the azimuth degree, the distance from the epicenter and 

the sampling rate, respectively; C) RMS histogram distributions calculated in a pre-seismic 10-s time window for GD2P 

(red) and TRACK (green) on both Radial (top) and Transverse (bottom) components for all the 52 solutions. D) RMS 

histogram distributions of the differences between the GD2P and TRACK solutions in a time interval of 0-40 s on both 

Radial (top) and Transverse (bottom) components for all the 52 solutions. 

 



ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 6 

solutions. Two features are worth noting: a) 

92% to 98% of the GPSgrams reveal values of 

accuracy within 0.3 cm; b) the radial and 

transverse components seem to show compa-

rable accuracies' distribution between GD2P 

and TRACK. Finally, to quantify the consisten-

cy between the GD2P and the TRACK solu-

tions, for each site, we firstly calculated the 

epoch-by-epoch differences between the two 

solutions and then we estimated the RMS dis-

tribution of the so determined residuals. The 

histograms of the residuals' RMS distribution 

(Figure 2D) show that for most of the sites, the 

two solutions agree within a value of 0.5 cm. 

The results obtained in this work show scatter 

values in the solutions reduced by about 50% 

with respect to a similar analysis carried out 

for the Emilia sequence (Avallone et al., 2012). 

III. COMPARISON BETWEEN HRGPS AND 

STRONG MOTION 

In this work we selected 8 examples of SM 

stations approximately co-located with the 

HRGPS sites. For most of the sites (six exam-

ples), the relative distance between the co-

located instrumentations ranges from 0 to 1.1 

km. However, we also included the FOC and 

ASP SM stations whose relative distances with 

GPS instrumentation (CESI and ASCC, respec-

tively) are higher, about 2.4 and 4.6 km, re-

spectively.  

In order to compare HRGPS and accelero-

metric time series, we transformed in dis-

placement the SM recordings by applying a 

double-integration after removing the mean of 

the signal calculated in a 4-s time window be-

fore the P-wave first-arrival. Moreover, since 

SM data were recorded with higher sampling 

rate (0.005 s) than HRGPS data (1 s and 0.1 s, 

Figure 2), we firstly filtered the SM-data at the 

associated GPSgram Nyquist frequency to 

avoid any aliasing effect, and then we deci-

mated each SM displacement waveform at the 

sampling rate of the co-located GPS instru-

ment. To quantify the similarity between the 

SM displacement waveforms and the 

GPSgrams, we performed a normalized cross-

correlation (NCC) between the signals record-

ed in a time window of 10 s after the P-wave 

first arrival. The analysis of source radiation 

recorded at the co-located SM and GPS sta-

tions is shown in Figure 3 in terms of their 

ground-motion displacement in the RT coor-

dinate system, following the same procedure 

used to compare the two GPSgrams solutions. 

As a first approximation, we find a good 

agreement between the processed GPSgrams 

and the doubly-integrated SM data. This 

match can be also deduced from the cross-

correlation function computed for windows of 

signals that include the first coseismic dynamic 

displacements. Due to the minimum pro-

cessing applied to the SM data, we observe a 

strong t2-drift, shortly after the S-wave arrival, 

for stations located in near field (AMT, NRC). 

Such drifts are well known in literature and 

are the effect of small distortions in accelera-

tion, often due to static displacements, that 

appear when waveforms are double integrated 

(e.g., Boore et al., 2002). Other smaller offsets 

are observed at other stations more distant 

from the source, but they are difficult to quan-

tify over this time window. The main differ-

ence between integrated accelerograms and 

GPSgrams in terms of peak of displacement is 

observed for the two stations located near the 

fault (AMT-AMAT and NRC-NRCI). For the R 

component of the AMT station, the integrated 

accelerogram shows only one peak, while the 

GPS shows a comparable (in amplitude peak), 

and a second later peak with amplitude similar 



ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 7 

 
Figure 3: Examples of comparison between GPSgrams (GD2P and TRACK solutions, dark and light green lines, re-

spectively) and the SM displacement time-series (red lines) at 8 co-located sites. Data are presented along the Radial (1st 

column) and the Transverse (3rd column) component. In 2nd and 4th columns are also plotted the normalized autocorre-

lation function (ACC, red line) and the normalized cross-correlation (NCC) functions between GD2P and SM time-

series (dark blue), and between TRACK and SM time-series (light blue) computed for both R and T components. The 

maximum NCC value for each GPS-SM time series is reported in the figure. ACC and NCCs were computed between 

signals recorded in windows of 10 s (gray area), 1-2 seconds after the P-wave first arrival time, picked on the strong mo-

tion waveform in acceleration and indicated with a vertical black bar. Only for signals at GUMA and ASP stations the 

window is smaller (9 seconds) due to the interruption of the SM recording in the first case and of the GPS acquisition in 

the second one. The text in the figures explicit the distance (dist) and the azimuth (azim) with respect to the epicenter, 

the sampling rates, the inter-distance (Id) between the SM and the GPS and, finally, the type of installation of the GPS 

antenna (building or free-field). 



ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 8 

to the first one. GPS and SM displacements are 

similar for the first peak on the T component 

whereas the following portions of the signals 

present significant differences with higher GPS 

amplitudes. These differences cannot be ex-

plained with eventual inaccuracy of the pro-

cessing because these kinds of frequencies are 

very well reproduced on GUMA station for 

instance. Since the GPS antennas at NRCI and 

AMAT sites were installed on the top of build-

ings, possible amplifications related to the re-

sponse of the building to coseismic shaking 

should be more carefully evaluated, consider-

ing both its structure and orientation. Howev-

er, due to the similarity of the first seconds af-

ter the P-first arrival, we find that the NCC 

value for AMAT is still very high on the R 

component (0.84-0.86). Similar characteristics 

are observed for NRCI that shows a high NCC 

value on the T component (0.72-0.73). Here, the 

differences in amplitude are mainly due to the 

down-sampling of the SM data at 1 s, while we 

find that the PGD (Peak Ground Displace-

ment) on the T component is not driven by the 

same peak possibly due to site effects. 

RM33-MTER co-located stations are the 

first that present very coherent signals on both 

the components. The double-integrated SM 

signal and the GD2P and TRAK solutions 

show maximum NCC values of 0.98 and 0.97 

on the R and T components, respectively, indi-

cating that the GPS is able to perfectly repro-

duce the coseismic dynamic displacements. In 

this case, the MTER station has the advantage 

to be installed in free-field, avoiding possible 

problems of amplifications related to the in-

stallation.  

The global pattern of PGD on four co-

located stations ASP-ASCC, GSA-INFN, 

TRN1-TERI and FOC-CESI is compatible with 

the focal mechanism of the M6 event. Indeed, 

these set of stations are located along the nodal 

planes that reduce the amplitude of the SH-

wave, generally identified along the T-

component. The agreement between GPS and 

integrated SM signals is globally good, even 

for stations as ASP-ASCC and FOC-CESI that 

are not perfectly co-located (inter-distance of 

2.4 km and 4.6 km, respectively, Figure 1). This 

is visible on the cross-correlation functions that 

have high maximum values, but are shifted in 

time with respect to the autocorrelation func-

tion computed on the SM signal. 

The amplitudes recorded at the GUMA sta-

tion are compatible with a directivity toward 

the Nord. This is evident by comparing the 

amplitudes recorded at the other stations, in-

cluding RM33 that is closer to the epicentral 

area (Figure 1). At GUMA station, the accel-

erometer recording stopped just after the peak 

over the T-component. After that, a greater 

peak is observed on the R-component of the 

GPSgrams. This signal corresponds to a Ray-

leigh-type surface wave since a similar peak is 

also identified on the Z-component with the 

typical dephasing of π/2. The same kind of 

wave is also observed on the co-located FOC-

CESI stations, demonstrating that the PGD is 

rapidly associated to the surface wave when 

the epicentral distance increases. 

IV. CONCLUSIONS 

The analysis of high-rate coseismic dynam-

ic displacements of the 2016 August 24 Mw 6.0 

Amatrice earthquake were carried out at 52 

continuous GPS stations. The observed ground 

motions in the GPSgrams were consistently 

associated with the propagation of the seismic 

waves. As expected, the HRGPS ability to de-

tect details in the seismic waves increases with 

the increasing of the sampling frequency of the 

acquired raw data. Furthermore, acquiring da-

ta with high sampling frequencies (> 5 Hz) not 



ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 9 

only strongly reduce aliasing artifacts in the 

resulting GPSgrams near the source (Smalley, 

2009; Avallone et al., 2011), but it also allows to 

observe details in the seismic wave propaga-

tion in the far field. For this earthquake, in fact, 

appreciable differences in the observations of 

the surface waves arrivals still exist between 

the 10-Hz-sampling (i.e., GUMA) and the 1-

Hz-sampling (i.e., CESI) GPSgrams even at 

larger distances from the epicenter (35-50 km).  

For the moderate earthquakes occurred in 

Italy in the last 8 years (2009 Mw 6.3 L'Aquila; 

2012 Mw 5.9 and Mw 5.8 Emilia main shocks, 

2016 Mw 6.0 Amatrice) we observe a relatively 

good spatial coverage of the HRGPS stations 

around the epicenter, although the contribute 

of non-geophysical GNSS networks in this pic-

ture is still significant. Further efforts in devel-

oping denser HRGPS networks, with efficient 

coupling of the antenna installations with the 

solid Earth, and operating at sampling fre-

quencies ≥ 10 Hz would provide additional 

constraints to characterize the source process 

and discriminate peculiar site effects (Avallone 

et al., 2014).  

With respect to the case of the 2012 Emilia 

earthquake (Avallone et al., 2012), the ob-

served GPSgrams for different software pack-

ages show a factor-2 lower RMS values (0.3 

cm) and residuals RMS values (0.5 cm) sug-

gesting a very good accuracy and agreement in 

the observation of the coseismic dynamic dis-

placements. In addition, the high values of the 

cross-correlation functions between the 

GPSgrams and the SM displacement time-

histories at the co-located sites confirm the 

ability of GPS in providing reliable waveforms 

for seismological applications. 

The GPSgrams described in this work rep-

resent a potential contribution to further stud-

ies on the earthquake source. Significant im-

provements of (1) the source rupture kinematic 

modeling, in (2) definition of source directivity 

generated by the main shock, or (3) for esti-

mating the magnitude of the event could be 

obtained by the joint use of GPSgrams and 

strong motion waveforms. In fact, in seismolo-

gy and engineering seismology, the estimation 

of the PGD is useful for studies on earthquake 

processes, seismic design and structural moni-

toring (i.e. building, bridges). However, using 

conventional seismological approaches this es-

timation is still challenging. On the other hand, 

positions correspond to basic observations for 

the GPS. In this sense, the GPSgrams could to-

day provide important and decisive contribu-

tions to investigate earthquake radiation pat-

tern and source directivity estimating the PGD 

distribution around the seismic source. 

REFERENCES 

Avallone et al. (2011), Very high rate (10 Hz) 

GPS seismology for moderate‐magnitude 

earthquakes: The case of the Mw 6.3 

L’Aquila (central Italy) event, J. Geophys. 

Res., 116, B02305, doi: 

10.1029/2010JB007834. 

Avallone, A. et al. (2012). High-rate (1 Hz to 20-

Hz) GPS co-seismic dynamic displacements 

carried out during the Emilia 2012 seismic 

sequence, Ann. Geophys., 55 (4); 

doi:10.4401/ag-6162. 

Avallone, A. et al. (2014), Waveguide effects in 

very high rate GPS record of the 6 April 

2009, Mw 6.1 L'Aquila, central Italy earth-

quake, J. Geophys. Res. Solid Earth, 119, 490–

501, doi: 10.1002/2013JB010475. 

Bertiger, W., S. Desai, B. Haines, N. Harvey, A. 

Moore, S. Owen, J. Weiss (2010), Single re-

ceiver phase ambiguity resolution with GPS 

data, J. Geod., 84, 5 (May 2010), pp. 327-337. 

Bock, Y., L. Prawirodirdjo, and T. I. Melbourne 

(2004), Detection of arbitrarily large dy-

http://dx.doi.org/10.1002/2013JB010475


ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 10 

namic ground motions with a dense high-

rate GPS network, Geophys. Res. Lett., 31, 

L06604, doi:10.1029/2003GL019150. 

Boehm, J. Werl, B. & Schuh, H. (2006), Tropo-

sphere mapping functions for GPS and very 

long baseline interferometry from Europe-

an Centre for Medium-Range Weather 

Forecasts operational analysis data. J. Ge-

ophys. Res., 111, B02406, doi: 

10.1029/2005JB003629. 

Boore, D. M., C. D. Stephens and W. B. Joyner 

(2002), Comments on Baseline Correction of 

Digital Strong-Motion Data: Examples from 

the 1999 Hector Mine, California, Earth-

quake, Bull. Seism. Soc. Am., 92, 4, 1543-

1560, doi: 10.1785/0120000926. 

Cheloni, D. et al. (2016), GPS observations of 

coseismic deformation following the 2016, 

August 24, Mw 6 Amatrice earthquake 

(central Italy): data, analysis and prelimi-

nary fault model, Ann. Geophys., 59, FAST 

TRACK 5, doi:10.4401/ag-7269. 

Colosimo, G., M. Crespi, and A. Mazzoni 

(2011), Real-time GPS seismology with a 

stand-alone receiver: A preliminary feasi-

bility demonstration, J. Geophys. Res., 116, 

B11302, doi:10.1029/2010JB007941. 

D'Agostino, N. (2014), Complete seismic re-

lease of tectonic strain and earthquake re-

currence in the Apennines (Italy), Geophys. 

Res. Lett., 41, 1155–1162, doi: 

10.1002/2014GL059230. 

Galvani, A. et al. (2012). The interseismic veloc-

ity field of the central Apennines from a 

dense GPS network. Ann. Geophys., 55, 5, 

2012; doi: 10.4401/ag-5634. 

Goldstein, P., D. Dodge, M. Firpo, Lee Minner 

(2003), "SAC2000: Signal processing and 

analysis tools for seismologists and engi-

neers", Invited contribution to "The IASPEI 

International Handbook of Earthquake and 

Engineering Seismology", Edited by WHK 

Lee, H. Kanamori, P.C. Jennings, and C. 

Kisslinger, Academic Press, London. 

Herring, T., King, R.W. & McClusky, S., (2010), 

GAMIT Reference Manual, Release 10.4, 

Massachussetts Institute of Technology, 

Cambridge, MA. 

INGV-CNT GPS Working Group (2016), Co-

seismic displacements for the August 24, 

2016 Ml6, Amatrice (central Italy) earth-

quake estimated from continuous GPS sta-

tions, doi:10.5281/zenodo.61355. 

INGV RING Working Group (2016), Rete Inte-

grata Nazionale GPS, doi:10.13127/RING. 

INGV Seismological Data Centre (1997), Rete 

Sismica Nazionale (RSN), doi: 

10.13127/SD/X0FXNH7QFY 

Kedar, S., G. A. Hajj, B. D. Wilson, and M. B. 

Heflin (2003), The effect of the second order 

GPS ionospheric correction on receiver po-

sitions, Geophys. Res. Lett., 30, 1829, 

doi:10.1029/2003GL017639, 16. 

Larson, K. M., P. Bodin, and J. Gomberg (2003), 

Using 1-Hz GPS data to measure defor-

mations caused by the Denali fault earth-

quake, Science, 300, 1421–1424, doi: 

10.1126/science.1084531. 

Li, X. et al. (2013), Real-time high-rate co-

seismic displacement from ambiguity-fixed 

precise point positioning: Application to 

earthquake early warning, Geophys. Res. 

Lett., 40, 295–300, doi:10.1002/grl.50138. 

Marchetti et al. (2016). The Italian Seismic Bul-

letin: strategies, revised pickings and loca-

tions of the Amatrice seismic sequence, 

Ann. Geophys., 59, FAST TRACK 5, 

doi:10.4401/ag-7169. 

Scognamiglio, L., E. Tinti and M. Quintiliani 

(2016), The first month of the 2016 central 

Italy seismic sequence: fast determination 

of time domain moment tensors and finite 

fault model analysis of the ML 5.4 after-

http://dx.doi.org/10.1029/2003GL019150
http://dx.doi.org/10.4401%2Fag-7269
http://dx.doi.org/10.1029/2010JB007941
http://dx.doi.org/10.1002/2014GL059230
http://dx.doi.org/10.1002/2014GL059230
http://dx.doi.org/10.1029/2003GL017639
http://dx.doi.org/10.1126/science.1084531
http://dx.doi.org/10.1002/grl.50138


ANNALS OF GEOPHYSICS, 59, FAST TRACK 5, 2016; DOI: 10.4401/ag-7275 

 

 11 

shock, Ann. Geophys., 59, FAST TRACK 5, 

doi: 10.4401/ag-7246. 

Smalley, R. Jr. (2009), High-rate GPS: How 

high do we need to go?, Seism. Res. Lett., 80, 

1054-1061, doi:10.1785/gssrl.80.6.1054. 

Wdowinski, S., Y. et al. (1997), Southern Cali-

fornia permanent GPS geodetic array: Spa-

tial filtering of daily positions for estimat-

ing coseismic and postseismic displace-

ments induced by the 1992 Landers earth-

quake, J. Geophys. Res., 102(B8), 18057–

18070, doi:10.1029/97JB01378. 

Wessel, P., and W. H. F. Smith (1998), New, 

improved version of generic mapping tools 

released, Eos Trans. AGU, 79, 579. 

Zumberge, J. F. et al. (1997), Precise point posi-

tioning for the efficient and robust analysis 

of GPS data from large networks, J. Ge-

ophys. Res., 102(B3), 5005-5018. 

* Acknowledgements: The ASCII files of the HRGPS 
position time series for all the sites used in this work for 

both the GD2P and TRACK solutions are available at the 

following link: 

ftp://gpsfree.gm.ingv.it/amatrice2016/hrgps/. We thank all 

of the agencies, cited in the main text, which have provid-

ed the GPS observations used in this study. Particular 

thanks are addressed to Dr. Domenico Collalti (Regione 

Abruzzo), Ing. Simone Patella (Regione Lazio), Francesco 

Matteuzzi (Leica Geosystems S.p.A.), Dr. Nicola Perfetti 

and Ing. Giuliano Molinelli (Netgeo) for the rapid distribu-

tion of the HRGPS data belonging to their permanent net-

works and all the technical staff involved in the mainte-

nance of these networks. We also thank the staff involved 

in the development and maintenance of the RAN and 

INGV accelerometric networks. Figures have been pre-

pared using GMT software package (Wessel and Smith, 

1998)

 

 

http://dx.doi.org/10.1029/97JB01378