Please use this identifier to cite or link to this item:
AuthorsSerpelloni, E.* 
Devoti, R.* 
Cavaliere, A.* 
TitleAnalysis of High-Rate GPS Data Collected During the L’Aquila Seismic Sequence
Issue Date16-Nov-2009
Keywordshigh-rate gps data
Subject Classification04. Solid Earth::04.03. Geodesy::04.03.06. Measurements and monitoring 
04. Solid Earth::04.03. Geodesy::04.03.07. Satellite geodesy 
04. Solid Earth::04.03. Geodesy::04.03.09. Instruments and techniques 
AbstractFour days before the 6th April M5.8 L’Aquila main-shock, a few GPS receivers recording at 10Hz and 1Hz sampling rates have been set up by INGV in the area affected by the seismic swarm in place by mid-January 2009. These data allowed us to measure for the first time in Italy the dynamic co-seismic displacements with periods ranging from fractions of seconds to several minutes and the full time spectra of the surface co-seismic and early post-seismic deformation with GPS instruments. We use TRACK, the kinematic module of the GAMIT/GLOBK software package, to perform epoch-by-epoch solutions of GPS raw data to obtain 3D time series of surface displacements. TRACK uses floating point LC (L3) observations between pairs of stations and the Mebourne-Wubena Wide Lane combination, with ionospheric constraints, to determine integer ambiguities at each epoch. It requires a fixed station and one, or more, kinematic stations. Usually, the static station is chosen to be far enough from the epicentral area not to be affected by the co-seismic displacements. Since no automatic processing engine exists for TRACK, we built a new shell script, which take full advantage of the Linux CPU-cluster routinely used to analyze 30 seconds GPS data with the GAMIT at INGV-Bologna. The new tool allows to automatically process pairs of stations (i.e., a network) and getting raw time series of several stations simultaneously (depending on the number of cluster nodes available) in a few seconds or minute, depending on the length of the session analyzed. TRACK uses broadcasted, ultra-rapid (containing predictions), rapid and final IGS orbits, thus making quasi-real time processing possible, and actually limited by the access to remote raw high rate GPS data archives. Since that the only two stations recording 10Hz data in the L’Aquila region are located close to the main shock epicenter and no data were available at other sites in Italy, we built a new tool to generate a virtual far field reference station acquiring 10Hz data by interpolating the available 1Hz RINEX data. The interpolated sites permit to properly solve the epoch-by-epoch position of the epicentral sites with the TRACK module. High frequency GPS data are severely affected by multipath noise, which can reach the same magnitude of the co-seismic displacements, and need to be removed consistently. For this reason, we investigate the effect of time and space-wise filters (sidereal and common mode filters) and set up a Matlab tool to perform time and spatial filtering on the raw time series produced by our processing tool. High rate data allow to measure the real static co-seismic offsets, which are not contaminated by early afterslip, which may occur in the next few hours after the earthquake. We analyze 10Hz data from 2 stations (Fig. 1) belonging to the CAGEONET network (Anzidei et al., 2009), and 1Hz data from 75 continuous GPS stations, located in central, southern and northern Italy, for which data are available for the 6th of April. A data quality inspection of the available high rate rinex files has been used to select the reference station, and single baselines solutions have been then resolved. We apply both spatial (common-mode) and temporal (sideral) filters to improve the signal to noise ratio of the observed displacements and estimate the epoch and the static co-seismic offsets. The 3D co-seismic displacement field has been used to invert, using rectangular (Okada, 1985), uniform-slip dislocations embedded in an elastic, homogeneous and isotropic half-space and a constrained, non-linear optimization algorithm (Burgmann et al., 1997), the best fit rectangular dislocation geometry and fault slip distribution, which has been compared with the fault geometry and slip model obtained from the analysis of standard 30 sec 24 hours data.
Appears in Collections:Conference materials

Files in This Item:
File Description SizeFormat 
poster_GNGTS09_hgps.pdfposter file16.75 MBAdobe PDFView/Open
Show full item record

Page view(s)

checked on Apr 23, 2017


checked on Apr 23, 2017

Google ScholarTM