POTENTIAL OF PRECISE POINT POSITIONING USING 1 HZ GPS DATA FOR DETECTION OF SEISMIC-RELATED DISPLACEMENTS

Aca Geodyn. Geomaer., Vol. 9, No. 3 (167), 303 313, 2012 POTENTIAL OF PRECISE POINT POSITIONING USING 1 HZ GPS DATA FOR DETECTION OF SEISMIC-RELATED DISPLACEMENTS Ján HEFTY * and Ľubomíra GERHÁTOVÁ Deparmen of Theoreical Geodesy, Faculy of Civil Engineering, Slovak Universiy of Technology, Radlinského 11, 813 68 Braislava, Slovak Republic *Corresponding auhor s e-mail: jan.hefy@suba.sk (Received May 2012, acceped Sepember 2012) ABSTRACT The Precise Poin Posiioning (PPP) analysis echnique adaped for monioring of moderae high-rae coordinae variaions from GPS observaions is applied for measuring acual displacemens relaed o earhquakes. Two approaches are examined: kinemaic PPP which is suiable for larger displacemens and analysis of residuals from quasi-saic PPP which is suiable for deecion of minor displacemens. Resuls from simulaed horizonal movemens of GPS anenna as well as analyses of 1 Hz GPS daa from M 9.0 Sendai earhquake, M 6.3 L Aquila earhquake and M 4.3 Taabanya earhquake are shown. Our experimens proved he abiliy o measure dynamic seismic-relaed shor-erm coordinae variaions a sub-cenimere level wih PPP and o deec boh srong and weak seismic signals using GPS saellies observaions. KEYWORDS: high-rae GPS, GPS seismology, precise poin posiioning, earhquake relaed displacemens 1. INTRODUCTION The GPS saellie posiioning applied for geokinemaic and geodynamic research is usually aimed a monioring of long-erm sie displacemens accompanied wih periodic variaions, predominanly of seasonal naure. For such applicaions he GPS observaions are sampled in 30 s or longer inervals and he sie coordinaes are evaluaed on daily or weekly basis. The high-rae recordings, e.g. wih 1 s sampling (or even wih sampling frequency up o 100 Hz) which are feasible for majoriy of recen GPS receivers, are used predominanly in navigaion or for kinemaic surveying applicaions. However, here exis several geodynamic phenomena like landslides, volcano econics, earhquakes, ec. where he highrae records of observed GPS saellies may provide subsanial informaion. Poenial of GPS for monioring he sie movemens relaed o he seismic phenomena was firsly recognized by Larson (Larson e al., 2003) and furher elaboraed in (Larson e al., 2007; Larson, 2009). Their analyses of 1 Hz GPS daa using he GIPSY sofware (Lichen and Border, 1987) and precise Inernaional GPS Service (IGS) saellie orbis (Beuler and Mueller, 1994) proved he abiliy o deec seismic waves relaed o srong earhquakes wih displacemens a he level of ens cenimeres. Kouba (Kouba, 2003) poined on possibiliy o apply Precise Poin Posiioning (Zumberge e al., 1997) approach using precise IGS orbis and saellie clocks producs (Kouba and Héroux, 2001) for analysis of seismic waves induced by large earhquakes. Applicaion of GPS a high sampling raes for monioring seismic-relaed phenomena Larson ermed GPS seismology (Larson e al., 2009). Laer on, series of sudies analyzing high-rae kinemaic GPS observaions relaed o grea and medium earhquakes have been published (e.g. Avallone e al., 2011; Ji e al., 2004; Wang e al., 2011). The PPP analysis of 1 Hz sampled daa is exhibiing as he mos suiable mehod for monioring he large earhquakes, however he possibiliies of applicaion of PPP for moderae seismic evens was checked as well (e.g. Avallone e al., 2011). The main advanages of PPP are in relaively simple GPS daa processing and unambiguous inerpreaion of measured sie posiion variaions. The objecive of his paper is o furher examine possibiliies of applicaion of PPP for deecion of posiion oscillaions and posseismic offses induced by earhquakes of various inensiies. Our analysis of GPS observaions wih 1 Hz sampling is based on uilizaion of sofware package ABSOLUTE developed a Slovak Universiy of Technology in Braislava for muli-gnss posiioning using un-differenced code and phase observaions (Hefy and Gerháová, 2011). The algorihms of ABSOLUTE were originally designed for saic posiioning when he sie coordinaes are assumed o be sable during he observaion inerval, usually for period of 24 hours or a leas for periods longer han 1 hour. The opion for kinemaic daa processing yielding new posiion for each observed epoch is implemened for purpose of monioring shor-erm coordinae variaions by inclusion of Kalman filering for final parameer esimaion. For deecion of larger displacemens we will apply sandard approach when he final independen 1 Hz coordinae esimaes are used for descripion of saion

304 J. Hefy and Ľ. Gerháová posiion variaions. In case when only smaller displacemens (less han several cenimeres) are expeced, we will analyze carries phases residuals resuling from srongly consrained kinemaic posiioning. 2. APPLIED ANALYSIS METHOD The Precise Poin Posiioning (PPP) processing sraegy is aking advanage of un-differenced GNSS code and phase observaions and uilizaion of precise saellie orbis and saellie clocks informaion. The absolue posiion geocenric sie coordinaes relaed o he Inernaional Terrresial Refernce Sysem (ITRS) are obained wihou necessiy of differencing GNSS observaions and using observed daa from oher erresrial reference sies. The heoreical background for PPP, including mahemaical model and overview of necessary geomerical and physical correcion models are given in (Kouba and Héroux, 2001). The PPP analysis sofware package ABSOLUTE was developed a Slovak Universiy of Technology for processing of saic GNSS observaions. We implemened all he relevan models for correcions and reducions of observed daa which are necessary for cenimere precision of sie coordinaes. In he preprocessing phase are combined dual-frequency carrier phase and code observaions from coninuous observaions no corruped by phase cycle slips. They are used for esimae of iniial non-ineger carrier phase ambiguiies, ionosphere delays and iono-free pseudoranges for each observing epoch. The final esimae of parameers is based on leas-squares adjusmen. Besides he saic sie coordinaes are compued also he real-valued correcions o iniial ambiguiies, roposphere zenih delays for seleced ime inervals (usually 1 or 2 hours) and receiver clock correcions. In (Hefy and Gerháová, 2011) is exended he applicabiliy of ABSOLUTE o GLONASS and GIOVE saellies by inroducing bias parameers necessary for combinaion of observaions from various GNSS. Nex exension of processing opions of ABSOLUTE applied for analyses in his paper is aimed o kinemaic PPP. We inroduced Kalman filer (Srang and Borre, 1997) in he final phase of daa processing for dynamic recursive esimaion of se of unknown parameers sie coordinaes, correcion o iniial ambiguiies and acual receiver clock correcion. The basic idea of applicaion of Kalman filering for esimaion of ime-dependen moderae variaions of posiion is progressive updae of parameers obained form previous adjusmens by gradually addiion of new saellie observaions. Troposphere parameers are no esimaed as we suppose ha he sandard models are sufficien o eliminae roposphere effec in he case when he analysis is focused on shor-erm coordinae changes. The precise saellie orbis in 15 min inervals and he saellie clocks correcions in 30 s inervals are adoped from he IGS Global Daa Cenre (IGS, 2011). The vecor of unknown parameers consiss of hree ypes of quaniies: coordinae incremens X, Y, Z o he iniial sie posiion X 0, Y 0, Z 0, receiver clock offse dt and correcions N j o iniial ambiguiies N j (j = 1, 2, m), where m is number of all ambiguiies of undifferenced phase observaions wihin he ime inerval of processed daa. N T m X Y Z dt N N (1) The vecor of GPS observaions a epoch i (i = 1, 2, n) consiss of p pairs of iono-free combinaions from code P and phase L measuremens. They are obained from simulaneous observaions of p saellies performed a he epoch i. b P L P L P p L p (2) 1 1 2 2 The quaniies P k and L k (k =1, 2, p) in (2) represen he observed code and phase observables afer subracion of he corresponding quaniies compued wih iniial parameer values X 0, Y 0, Z 0 and N 1, N 2, N m. Linearized observaion equaion has hen form b A e (3) wih design marix A and he observaion noise vecor e. As we assume muually independen observaions heir covariance marix Σ b = E[e e T ] is diagonal wih 2 variances P 2, L characerizing he variabiliy of code and phase measuremens. The general concep of Kalman filering is based on updaing of he iniial values of parameers (or parameers resuling from previous processing). The iniial esimae denoed here for ransparency as old will be improved using he new se of observaions, denoed here as b new. The new (updaed) esimae of parameers new is hen obained as linear combinaion of he iniial esimae old and he new observaions b new new old ne 1 K b w A (4) new where K is he Kalman gain marix and A new is design marix relaing new observaions b new wih se of parameers old. The vecor b new - A new old represens mismach beween new measuremens and old esimae (Srang and Borre, 1997). Eq. (4) can be simply modified as new old (5) where he vecor ε denoes he error of predicion of parameers. We assume, ha he elemens in ε are independen, wih known covariance marix Σ = E[ε ε T ], which is in our case diagonal. Applicaion of Kalman filering for dynamic upgrade of parameers is based on series of GPS old 2 T

POTENTIAL OF PRECISE POINT POSITIONING USING 1 HZ GPS DATA FOR. 305 observaion vecors b relaed o epochs i (i = 1, 2, n) and known design marices A. The prediced esimae of (-), (predicion is denoed here wih sign -) is based on informaion from previous epoch - 1 (Srang and Borre, 1997). As he predicion of parameers is considered he final updaed esimae (denoed here wih +) from previous epoch -1 ˆ ˆ (6) 1 ˆ The correced (updaed) esimae observaion b from epoch is according o (4) ˆ ˆ b A ˆ 1 1 using K (7) For correcing he esimae from epoch -1 hree marices are involved, namely he prediced covariance marix P P (8) 1, he gain marix T T A P A 1 K P A (9) b, and he correced covariance marix I K A P P (10) Oupus of he inroduced processing procedure are he ime series of parameer esimaes ˆ. Our analysis of earhquake relaed displacemens is focused on hree elemens of ˆ - he coordinae ime series X, Y, Z which will be subsequenly ransformed o local coordinaes in Norh-Souh, Eas- Wes and heigh denoed as n, e, v. The ime variabiliy of ˆ is defined by seing of diagonal elemens in he covariance marix Σ. In general, he ambiguiies N j are srongly consrained o heir previous esimaes and he receiver clocks dt are loosely consrained ( he acual clock esimaes are independen from hein previous values). In his paper we applied consrains for all ambiguiies N = 0.0001 m and dt = 1.000 m for receiver clock esimaes. Consraining of sie coordinaes will affec heir applicabiliy for sudies of sie displacemens. The loose consraining of X, Y, Z parameers in Σ will cause ha all earhquake relaed displacemens will be included in esimaed coordinae ime series. Such approach we name as Kinemaic PPP. For earhquake relaed displaycemens we applied here X = Y = Z = 1.000 m. Seing srong consrains on X, Y, Z in Σ will cause ha he earhquake relaed displacemens will be visible in residuals of carrier phase observables L k in (2). This mehod we name as Residuals from quasi-saic PPP. The advanage of such approach is in suppression of coordinae drifs due o effec of non-compleely ambiguiy modelling. Applicaion is suiable for deecion of small coordinae variaions relaed o disan earhquakes and o medium - magniude close earhquakes. For earhquake relaed residuals we applied here X = Y = Z = 0.001 m. The residuals from code P and phase L observaions realized a epoch will be evaluaed as L, P b A ˆ r (11) By exracing p elemens relaed o phase observables from he 2p-dimensional vecor L, P r we will obain he vecor of phase residuals r L. This vecor reflecs besides he observaion errors also he discrepancy beween he acual sie posiion and he posiion deermined from quasi-saic PPP (he acual coordinaes are reaed as sable for limied ime inerval). The r L vecor will be used for esimaing he acual coordinae deviaion from he mean posiion by leas-squares soluion of equaion dx L dy B r dz (12) where B is design marix relaing he residuals of phase measuremens o local coordinae shifs. Finally, he ime series of dx, dy, dz which will be ransformed o local coordinaes n, e and v suiable for furher analyses and inerpreaion. Our experience wih processing of 1 Hz sampled GPS daa proved ha analysis of 20 minues inerval of observaions is sufficien for convergence of esimaed ambiguiies a he cenimere level. However, we generally use for all analysed daa ses relaed o earhquakes a leas 60 minues of GPS observaions. For he coninuiy of esimaed coordinae ime series is convenien o resric he final adjusmen o saellies which are coninuously observed during whole analyzed inerval. 3. TESTS OF DATA PROCESSING METHODS ON SIMULATED HORIZONTAL MOVEMENTS For evaluaion of possibiliies and accuracy limis of PPP for deecion of moderae sie displacemens we performed series of experimenal GPS observaions wih 1 Hz daa recording. Besides he saic GPS posiioning processed in kinemaical mode, we performed experimens wih GPS anenna mouned on mechanical device simulaing periodic horizonal movemens. The simulaor consruced a Insiue of Geodesy a Technical Universiy of Brno enables horizonal harmonic moion in sraigh line wih ampliude 5 cm and wih mean frequency 0.82 Hz. I is possible o se he line of moion in arbirary azimuh; he verical movemens of anenna are expeced o be zero.

306 J. Hefy and Ľ. Gerháová Table 1 Resuls from experimenal evaluaion of applied processing sraegies RMS differences beween modelled and observed posiions of GPS Processing sraegy anenna Norh-Souh (mm) Eas-Wes (mm) Up (mm) PPP kinemaic 7.7 5.1 11.8 Residuals from quasi-saic PPP 7.6 4.2 12.5 Figure 1 shows ime series in Norh-Souh (n), Eas-Wes (e) and Up (v) componens esimaed from 20 min inerval of 1 Hz records of GPS observaions in experimen performed on July 6, 2011. The processed inerval of daa sared a 9 h 30 min wih GPS anenna in saic posiion and ends a 9 h 50 min of GPS ime. A 9 h 43 min sared simulaion of harmonic movemens of GPS anenna lasing 60 sec. Second period of movemens aking 60 sec sared a 9 h 46 min. The anenna during he 2 min inerval beween moving periods remained in opposie posiion. The orienaion of he simulaor enabled horizonal movemens wih 3.7 cm in n componen and 3.4 cm in e componen. The plos documen he sensiiviy of GPS processed in PPP kinemaic mode for deecion of horizonal movemens wih 5 cm ampliude. The deailed view of 30 sec inerval of harmonic moion of anenna mouned on simulaor and he GPS esimaed posiion is in Figure 2. Esimaed GPS n and e coordinae consiuens are a cm level consisen wih he modelled posiions of moving anenna. Because of aliasing (frequency of he simulaor is f s = 0.82 Hz, he sampling frequency of GPS is f GPS = 1 Hz) he observed oscillaion is wih frequency f a = 0.18 Hz. I is eviden ha for deecion and reliable descripion of displacemens wih frequencies higher han 0.25 Hz, he GPS sampling frequency f GPS = 1 Hz is no sufficien and f GPS should be increased a leas o 2 Hz or even o 5 Hz or 10 Hz. The poenial of applicaion of high-rae GPS recordings for kinemaic monioring was successfully examined e.g. in (Avallone e al., 2011; Wang e al., 2011). To invesigae he accuracy limis of PPP for monioring dynamic phenomena during shor inervals we analyzed various daa-ses of observaions: i/ he GPS anenna was saic, ii/ wih anenna changing he saic posiion o harmonic movemens, iii/ he saic posiion and harmonic movemens were alernaed. The analyses were resriced o separaed 300 sec inervals wih he long-erm drif eliminaed from GPS daa. Resuls from 10 ses of observaions performed under various siuaions are summarized in Table 1. Boh processing sraegies Kinemaic PPP and Residuals from quasi-saic PPP lead o similar RMS differences beween modelled and observed posiions of GPS anenna. The values in Table 1 demonsrae he poenial of presened processing and analysis echnique for deecion of sudden displacemens in horizonal coordinaes and heigh. I is worh menioning ha he values of RMS differences obained from comparisons of shorer inervals (150 sec or 90 sec) are slighly lower han hose given in Table 1. I is he evidence of some residual sysemaics ha influences GPS esimaed coordinaes in kinemaic mode. 4. ANALYSIS OF 1 HZ GPS DATA RELATED TO GREAT EARTHQUAKES For demonsraion of capabiliy of kinemaic PPP o monior posiion oscillaions relaed o grea earhquakes we used 1 Hz GPS daa from March 11, 2001 during he Sendai earhquake wih epicenre near Eas Coas of Honshu, Japan (38.297N, 142.372E), magniude 9.0, occurred a 05:46:24 UTC (USGS, 2012). Figure 3 shows disribuion of analyzed IGS saions from which are he high-rae 1 Hz sampled GPS daa available. We used he 1 s observaions from he IGS LEO Working Group daabase (IGS, 2011). The MIZU and USUD (IGS permanen saions in Japan wih disances from epicenre 143 km for Mizusawa and 429 km for Usuda) observaions analyzed wih ulra-rapid precise saellie orbis and clocks shorly afer he Sendai earhquake documened enormous posiion shifs (Banville and Langley, 2011). Figures 4 and 5 show horizonal and verical oscillaions of MIZU and USUD resuling from our analysis using he ABSOLUTE sofware in kinemaic mode. The represened displacemens are relaively o he iniial coordinaes a he firs epoch of he ploed inerval. Boh plos unambiguously demonsrae larger han 0.5 m shor-erm displacemens in all hree coordinae consiuens. Eviden also are he posseismic displacemens in horizonal posiion a meer level for sie MIZU and decimere level for Eas-Wes componen for sie USUD. The verical pos-seismic deformaion (20 cm for MIZU and 7 cm for USUD) is indicaed from he ploing; however he insufficien ambiguiy modelling may give rise o similar effec. The oher four analyzed saions are siuaed more han 2000 km from he epicenre. As shown in Figure 6, he IGS sies SHAO (Shesan, China, 2082 km from epicenre) and PETS (Peropavlovsk, Russia, 2063 km from epicenre) are unambiguously indicaing displacemens relaed o he earhquake. The range of SHAO coordinae variaions (15 cm and more in horizonal consiuens and heigh) is significanly larger han he range of observed

POTENTIAL OF PRECISE POINT POSITIONING USING 1 HZ GPS DATA FOR. Fig. 1 307 Time series in Norh-Souh, Eas-Wes and Up coordinae consiuens esimaed from 20 min inerval of 1 Hz records of GPS observaions wih GPS anenna mouned on mechanical device enabling simulaion of periodic horizonal movemens. The forced movemens were acivaed wice for 60 seconds. Fig. 2 Simulaed harmonic horizonal movemens (solid lines) and GPS measured posiion (dos and dashed lines) deail from 30 sec of analysed daa.

308 J. Hefy and Ľ. Gerháová Fig. 3 Disribuion of IGS saions from which he high-rae 1 Hz sampled GPS daa during he Sendai earhquake are available. coordinae displacemens in PETS (abou 10 cm in horizonal componens and 7 cm in up componen) despie he saions are in same disance from epicenre. The evidence of earhquake relaed displacemens in he oher wo IGS sies GUAM (Dededo, Guam, 2727 km from epicenre) and PIMO (Quezon Ciy, Philippines, 3317 km from epicenre) shown in Figure 6 is no proved for all coordinae consiuens. The GUAM Eas-Wes abou 10 cm variaions are well observed, he Norh-Souh ~ 5 cm displacemens are only slighly exceeding he noise of he ime series. The 10 cm displacemen of he up componen is also visible in he ime series; however i canno be excluded ha i is spurious effec relaed o ambiguiy mismodelling. The PIMO coordinae evoluion is relaively sable wih increased variabiliy of Norh-Souh and Eas-Wes componens (~ 6 cm) 700 s afer occurrence of he earhquake. The causaliy of observed heigh variaions wih earhquake canno be confirmed. The resuls of PPP processing of 1 Hz GPS daa wih ABSOLUTE sofware in kinemaic mode wih Kalman filering using IGS orbis in 15 min inervals and saellie clocks in 30 s inervals are capable o monior seismic-relaed displacemens a sies siuae over 2000 km far away from he epicenre of grea earhquake. 5. POSSIBILITIES OF DETECTION DISPLACEMENTS RELATED TO MODERATE- MAGNITUDE AND LIGHT-MAGNITUDE EARTHQUAKES For sudy of he capabiliy of he PPP based analysis of 1 Hz GPS daa for monioring he displacemens induced by moderae earhquakes we will use daa from 5 GPS receivers observing in he period of he Aquila earhquake (Cenral Ialy). The earhquake wih magniude 6.3 and epicenre 42.334N, 13.334E occurred on April 6, 2009, 01:32:39 UTC (USGS, 2012). Saions INGP (Preuro), INGR (Roma) and RSTO (Rosseo degli Abruzzi) are par of he Ialian: Inegraed Naional Ne GPS (Ree Inegraa Naionale GPS, hp://ring.gm.ingv.i/). The oher wo, namely ROIO (Poggio di Roio) and CADO (Fossa) GPS saions were insalled only four days before he main shock (Cheloni e al., 2010). Figure 7 shows he disribuion of analyzed GPS saions. Three sies are close o he epicenre, namely INGP 5.6 km, ROIO 4.5 km, CADO 12.9 km, and he oher wo are more disan: RSTO (66 km) and INGR (87 km). GPS observaions from CADO, ROIO and INGP unambiguously regisered he earhquake induced moions in all coordinae consiuens. The observed shor-erm horizonal displacemens are from 23 cm (Eas-Wes componen of CADO) o 5 cm (Norh-Souh componen of INGP). The significan variaions in Up componen are observed a CADO

POTENTIAL OF PRECISE POINT POSITIONING USING 1 HZ GPS DATA FOR. 309 (17 cm) and ROIO (12 cm). For ROIO and CADO also he pos-seismic drif is well deermined. The ime series of all he hree sies menioned are obained wih kinemaic PPP approach. A ROIO and CADO saions he GPS receivers were se o 10 Hz rae recordings. Their analysis by Avallone e al. (2011) demonsraed he increased sensiiviy o high-rae coordinae variaions and beer performance in monioring horizonal coordinae consiuens. The expeced minor coordinae displacemens of RSTO and INGR are analyzed using he Residuals from quasi-saic PPP approach (Fig. 8). In his case he effec of earhquake canno be posiively confirmed in none of he ploed series, even he increased variabiliy in RSTO is observed. Observabiliy of he effec of ligh-magniude earhquake in GPS 1 Hz records will be examined on he Taabanya earhquake (Hungary, magniude 4.3, epicenre 47.56N, 18.31E) occurred on January 29, 2011, 17:41:37 UTC (USGS, 2012). We processed here he GPS daa from he saion TATA (Taabanya) from he Hungarian naional acive GNSS nework (www.gnssne.hu) which is siuaed 10.0 km from he epicenre. In Figure 9 are ploed ime series of local coordinae consiuens obained from analysis of residuals from quasi-saic PPP. No one of he series indicaes earhquake relaed displacemens. We emphasise ha he variabiliy of he Eas-Wes componen (Fig. 9) sampled wih 1 Hz rae is less han 0.3 cm. Our experience wih PPP analysis of GPS 1 Hz daa observed in various environmens indicaes ha 0.3 cm is limiing value for deecable seismicrelaed phenomena. The qualiy of informaion relaed o earhquake induced displacemens could be increased if he higher sampling rae (e.g. 10 Hz) GPS recordings are available for analyses. Especially, in case of weak earhquake relaed effecs wih shorerm variabiliy (less han 1 s) he analysis of 0.1 s sampled GPS daa is capable o deec displacemens ha remained undeeced wih 1 Hz sampling as documened e. g. in (Avallone e al., 2011). 6. CONCLUSIONS The applied analysis mehod using he PPP approach adoped for processing of 1 Hz sampled GPS daa demonsraed he possibiliies o deec and o monior he earhquake-relaed horizonal and verical displacemens. The PPP sofware package ABSOLUTE developed a Slovak Universiy of Technology in Braislava was exended for analysis of kinemaic GPS observaions by implemenaion of Kalman filering for final parameer esimaion. Two alernaives depending on degree of consraining he adjusmen of coordinae parameers proved he poenial for deecion of irregular posiional variaions a sub-cenimere level relaed o grea and moderae earhquakes. We succeed in reliable recording earhquake relaed displacemens even for sies more han 2000 km away from epicenre of Fig. 9 Horizonal and verical coordinae ime series of GPS saion TATA during he Taabanya M 4.3 earhquake. The ime axis sars a 17:41:00 GPS ime, he verical line indicaes he occurrence of earhquake in GPS ime (17:41:52). Honshu M 9.0 earhquake and more han 80 km away from epicenre of M 6.3 L Aquila earhquake. The presened mehod is efficien for srong earhquakes effecs and for saions close o epicenre. For weaker effecs more dense records (of order of 10 Hz) may improve he delecabiliy of seismic displacemens. Inclusion of GLONASS observaion also could

310 J. Hefy and Ľ. Gerháová increase he accuracy and sensiiviy of PPP based earhquake monioring. ACKNOWLEDGEMENT This work was suppored by he gran No. 1/0569/10 of he Gran Agency of Slovak Republic VEGA. We are graeful o Radovan Machoka for providing he device for simulaion of harmonic horizonal anenna movemens, and o Alessandro Caporali and Robero Devoi for providing he 1 Hz GPS observaions relaed o L Aquila earhquake. REFERENCES Avallone, A., Marzario, M., Cirella, A., Piaanesi, A., Rovelli, A., Di Alessandro, C., D Anasasio, E., D Agosino, N., Giuliani, R. and Maone, M.: 2011, Very high rae (10 Hz) GPS seismology for moderaemagniude earhquakes: The case of he M W 6.3 L Aquila (cenral Ialy) even. Journal of Geophysical Research, 116, B02305, doi: 10.1029/2010JB007834. Banville, S. and Langley, R.: 2011, GPS shows dramaic posiion shifs from Japan earhquake. GPS World, 22, No. 4 (hp://www.gpsworld.com/survey/news/gps-showsdramaic-posiion-shifs-japan-earhquake-11212). Beuler, G., Mueller, I.I. and Neilan, R.E.: 1994, The inernaional GPS Service for Geodynamics (IGS): Developmen and sar of official service on January 1, 1994. Bullein Geodesique, 68, 39 70. Cheloni, D., D Agosino, N., D Anasasio, E., Avallone, A., Manenuo, S., Giuliani, R., Maone, M., Calcaerra, S., Gambino, P., Dominici, D., Radicioni, F. and Fasellini, G.: 2010, Coseismic and iniial pos-seismic slip of he 2009 M 6.3 L Aquila earhquake, Ialy, from GPS measuremens. Geophysical Journal Inernaional 181, 1539 1546. Hefy, J. and Gerháová, Ľ.: 2011, Specific feaures of observaions of various GNSS affecing heir combinaion in he precise poin posiioning. 3rd Inernaional Colloquium Scienific and Fundamenal Aspecs of he Galileo Programme. ESA Proceedings WPP 326. IGS: 2011, Inernaional GNSS Service Global Daa Cenre (hp://igscb.jpl.nasa.gov/). Ji, C., Larson, K.M., Tan, Y., Hudni, K. and Choi, K.: 2004, Slip hisory of he 2003 San Simeon Earhquake consrained by combining 1-Hz GPS, srong moion, and eleseismic daa. Geophysical Research Leers 31, (17), doi: 10.1029/2004GL020448. Kouba, J. and Héroux, P.: 2001, Precise poin posiioning using IGS orbis abd clocks producs. GPS Soluions, 5 (2), 12 28. Kouba, J.: 2003, Measuring Seismic Waves Induced by Large Earhquakes wih GPS. Sudia Geophysica e Geodaeica, 47, 741 775. Larson, K.M., Bodin, P. and Gornberg, J.: 2003, Using 1-Hz GPS daa o measure deformaions caused by he Denali faul earhquake. Science 300, 142, 1421 1424. Larson, K.M., Bilich, A. and Axelrad, P.: 2007, Improving he precision of high-rae GPS. Journal of Geophysical Research 112, B05422. doi: 10.1029/1006JB004367. Larson, K.M.: 2009, GPS Seismology. Journal of Geodesy, 83, 227 233. Lichen, S. and Border, J.: 1987, Sraegies for highprecision global posiioning sysem orbi deerminaion. Journal of Geophysical Research 92, 12751 12762. Srang, G. and Borre, K.: 1997, Linear Algebra, Geodesy and GPS. Wellesley MA, USA, Wellesley-Cambridge Press. 624 pp. USGS: 2012, USGS Earhquake Hazards Program (hp://earhquake.usgs.gov/) Wang, G., Blume, F., Meerens, C., Ibanez, P., Schulze, M.: 2011, Performance of hihg-rae kinemaic GPS during srong shaking: Observaions from shake able ess and he Chile earhquake. Journal of Geodeic Science, 2, 1-16. Zumberge, J.F., Heflin, M.B., Jefferson, D.C., Wakins, M.M. and Webb, F.H.: 1997, Precise poin posiioning for he efficien and robus analysis of GPS daa from large neworks. Journal of Geophysical Research, 102 (B3), 1997, 5005 5017.

J. Hefy and Ľ. Gerháová: POTENTIAL OF PRECISE POINT POSITIONING USING 1 HZ GPS DATA FOR. Fig. 4 Horizonal and verical coordinae shifs of IGS saion MIZU relaed o he Sendai earhquake. The ime axis sars a 5:46:00 GPS ime, he verical line indicaes he occurrence of earhquake in GPS ime (5:46:39). Fig. 6 Fig. 5 Horizonal and verical coordinae shifs of IGS saion USUD relaed o he Sendai earhquake. The ime axis sars a 5:46:00 GPS ime, he verical line indicaes he occurrence of earhquake in GPS ime (5:46:39). Horizonal and verical coordinae shifs of IGS saions PETS, SHAO, GUAM and PIMO relaed o he Sendai earhquake. The ime axis sars a 5:46:00 GPS ime, he verical line indicaes he occurrence of earhquake in GPS ime (5:46:39).

J. Hefy and Ľ. Gerháová: POTENTIAL OF PRECISE POINT POSITIONING USING 1 HZ GPS DATA FOR. Fig. 7 Analyzed sies wih GPS receivers observing wih 1 Hz sampling in he period of he L Aquila earhquake Fig. 8 Horizonal and verical coordinae shifs of GPS saions ROIO, INGP, CADO, RSTO and ING relaed o he L Aquila earhquake. The ime axis sars a 1:32:00 GPS ime, he verical line indicaes he occurrence of earhquake in GPS ime (1:32:54).