INVESTIGATING ACCURACY ENHANCEMENT OF GLOBAL NAVIGATION SATELLITE SYSTEM. Federation. Nigeria.

Transcription

1 INVESTIGATING ACCURACY ENHANCEMENT OF GLOBAL NAVIGATION SATELLITE SYSTEM F. Zarzura,1, R. Ehigiatr Irughe 2, B. Mazurv,3 1 Department f Higher Gedesy, Siberian State Gedesy Academy, Nvsibirsk Russia Federatin 2 GeSyatems and Envirnmental Engineering, 14 2 nd East Circular Rad, Benin City Nigeria. 3 Department f Higher Gedesy, Siberian State Gedesy Academy, Nvsibirsk Russia Federatin ABSTRACT: Fast and accurate relative psitining fr baselines is pssible using dualfrequency Glbal Psitining System (GPS) receivers. In spite f the fact that DGPS technique increased the GNSS accuracy by reducing the errrs, it can't remve the rbital errr, insphere errr and trpsphere errr. This paper utlines the use f accurate relative psitining fr prcessing GPS data (advantages-and disadvantages) and cmpares the results with relative psitining f the same pint using Mecca permanent GPS bservatin netwrk. The results and analysis f integrated system are presented. In cnclusin, using precise ephemeris frm Internatinal GPS Services ( IGS ) Netwrk with Klbucher insphere mdel with Hpfield r Saastiminen trpsphere mdel imprve the accuracy f DGPS measurements. KEYWORDS: DGPS, RTK, Klbucher mdel, Hpfield, Saastiminen and IGS. INTRODUCTION Our research is fcused n imprving the accuracy f differential GPS and Real Time Kinematic (RTK) bservatins using wide area GPS systems. GPS bservatins cntain bth systematic and randm errrs; Differential GPS (DGPS) and Real Time Kinematic (RTK) is an bservatin technique that can be used t remve r reduce the insphere effects arising in rdinary GPS [Alves 24].In rder t btain precise crdinates fr a pint frm GPS data, a number f nuisance parameters first need t be remved frm the data. These may be classified as satellite errrs, atmspheric errrs, and receiver errrs. Satellite errrs include errrs in the reprted satellite crdinates and satellite clcks, atmspheric errrs include signal delays due t the trpsphere and insphere while receiver errrs include receiver clck errrs. Let us cnsider fr a mment hw each f these errrs might be remved r mitigated. Trpsphere errrs are largely remved by either applying a mdel which attempts t mathematically simulate the signal delay as in mst cmmercial sftware r by estimating the signal trpsphere delay alng with the receiver crdinates (as in mst research sftware). Insphere errrs are remved by bserving bth GPS frequencies (L1 and L2) and cmbining the tw bservatins t derive an insphere-free bservatin. Errrs in satellite 1

2 psitins can be reduced by using precise satellite rbits available frm the IGS and any remaining errr (except multipath) largely cancels ver shrt distances. That leaves satellite and receiver clck errrs as the dminant errrs t be death with and this is where relative psitining cmes t the fre [ Kba, 23]. Precise Pint Psitining The vast majrity f cmmercially available sftware utilizes the principles f relative psitining. Hwever, in the late 199s, the Jet Prpulsin Labratry (NASA) pineered a new technique that did nt require differencing t btain precise psitin. The labeled it Precise Pint Psitining (PPP) and implemented it in their, GIPSY/OASIS II GPS prcessing sftware [Rulstn, et al 2]. The largest difference between relative prcessing and PPP is hw satellite and receiver clck errrs are handled. Instead f between-receiver differencing t remve the satellite clck errrs, PPP uses highly precise satellite clck estimates. These satellite clck estimates are derived frm a slutin using data frm a glbally distributed netwrk f GPS receivers. Instead f between satellite differencing t remve receiver clck errr, PPP estimates theses as part f the least squares slutin fr the crdinates. Cnsequently, precise abslute crdinates fr a single receiver at an unknwn lcatin may be btained withut the need f a secnd receiver at a knwn lcatin. A nte f cautin at this pint is necessary. It may be pssible t get PPP cnfused with anther frm a pint psitining that many GPS users will be familiar with i.e, Single Pint Psitining (SPP). SPP is different frm PPP in tw ways. Firstly SPP des nt use precise satellite clck values and secndly, nly the pseud rage bservatins are used. PPP uses bth the pseud range and mre precise carrier phase bservatins [Witchayangkn, 22]. The difference between these methds in terms f crdinate accuracy is larger; SPP prducers crdinate accurate at the 1-3 m level while PPP can prduce crdinates accurate at the.1 m level with 24hurs bservatins. Cnsequently, PPP allws crdinate determinatin with a precisin that is cmparable t relative prcessing. Since n base statin is required in PPP, a further questin is:" what datum are the crdinates in?" Fr PPP, the datum is hidden in the satellite crdinates-the satellite reference frame (datum) will be the unknwn grund site reference frame. This means that t btain crdinates in a different reference frame the user needs t perfrm a usually straight frward crdinate transfrmatin [Zhng-yi, etal. 22]. OBSERVATIONS METHODOLOGY This prcedure invlves fur bservables fr each f the visible satellites in each epch. The tw pseud range and carrier phase bservables can be linearly cmbined, thus reducing the effects f the insphere refractin. The use f a trpsphere mdel, tgether with parameterizatin techniques, can reduce the trpsphere refractin effects. The IGS ephemerids supply satellite crdinates and clck errrs, with accuracy in the rder f 5cm and.3 ns, respectively, and are essential in PPP. Hwever, variatins due t gephysical phenmena shuld be remved using apprpriate mdels. These crrectins include Plarmtin (Kba et al 2). Accrding t Salam et al. (22), it is pssible t btain precisin f a few millimeters and a few centimeters in the hrizntal and vertical cmpnents, respectively. Such levels f accuracy can be btained fr static pint psitin, using a perid f 24 hurs f data. Once 2

3 the crdinates fr all statins are daily estimated, a slutin fr a specific epch can be btained. As there is n crrelatin between the crdinates f different statins, such a slutin may be btained independently fr each statin [Shen, 22]. Observatin Sites and used Instruments The lcatin fr the prpsed GPS netwrk is shwn in Figure (1). A pilt netwrk has been established ver the Hlly Mecca. The system invlves permanently running GPS reference statins, at spacing up t 3 km, then feeding GPS data t a central prcessing cmputer. Five LEICA GPS SR53 dual frequency receivers cllected the GPS data n 12 th February 26, where pint (G182) was used as reference pint. At first, the static bservatins with rate in legal tw secnds are perfrmed. Fur receiver f the same LEICA type is setup at the ther pints fr mre than 24 hurs. The Reference Statins are designed t supprt highprecisin psitining ver a wide area. Figure (1): The shape f Mecca netwrk The weighted average psitin f pints btained frm the slutin f pints in cde_ phase slutin with Hpfield trpsphere mdel, Klbucher Insphere mdel, precise rbit and mask angle 15. This value is t be adpted as the psitin t be used as a reference t test the accuracy and precisin in all subsequent investigatins. Table (1): Tw reference slutins, namely Cde slutin and Cde-Phase slutin pint Cde slutin Cde_ phase slutin (adpted value) Nrth(m Ht.(m East (m) Nrth(m) Ht.(m) East(m) ) ) G-79 G-97 G-164 G-182 referen ce M The difference E( m) N( m) H( m)

4 Observatins Analysis Leica Gestatinary Office prgramme ( LGO ) is used fr data analysis. The sftware is particularly well suited fr the rapid prcessing f small-size single and dual frequency surveys, permanent netwrk prcessing, ambiguity reslutin n lng baselines, insphere and trpsphere mdeling, clck estimatin and time transfer, cmbinatin f different receiver types, simulatin studies, rbit determinatin and estimatin f Earth rtatin parameters and the generatin f s called free netwrk slutins. The results and analysis f bservatins will be intrduced int three steps as fllwing: Orbital errrs T study the effect f satellite psitin n the slutin, the prcessing f Mecca netwrk Cde-Phase bservatins was dne twice. Every run utilized the same prcessing parameters except that the first run used the bradcast ephemeris, and the secnd run used the precise ephemeris as prduced by Internatinal GPS Service IGS. The results is presented in table (2) belw. Frm table (2), the hrizntal Psitin crdinates varies in a wide range frm.13 mm t.71 m The range fr height varies frm.83 mm t 1.94 m As a clsing remark fr this sectin, ne can easy detect the cntributin f precise ephemeris in imprving the slutin against the bradcast slutin. Table (2): The differences between the default cde- phase slutin and the cde-phase slutin by replacing the rbit mdel t precise mdel at Mecca netwrk. Pint E(mm) N(mm) H(mm) G G G M Insphere errrs T study the effect f insphere errr n the slutin, the prcessing f Mecca netwrk Cde-Phase bservatins was dne several times in this bservatin using klbucher insphere mdel with the adpted values [Witchayangkn, 22]. Every run utilized the same prcessing parameters except that the first run utilized an insphere mdel frm the fllwing mdels: Cmputed Mdel Standard Glbal/Reginal Table (3): The differences between the default slutin and the slutin by replacing the insphere mdel t cmputed mdel at Mecca netwrk. pint E(mm) N(mm) H(mm) G-79.1 G G-164 M

5 Table (4): The differences between the default slutin and the slutin by replacing the insphere mdel t standard mdel at Mecca netwrk pint E(mm) N(mm) H(mm) G G G M Table (5): The differences between the default slutin and the slutin by replacing the insphere mdel t glbal mdel at Mecca netwrk pint E(mm) N(mm) H(mm) G-79 G G-164 M As it is indicated in tables (3, 4and 5), the crdinates vary in a clear range frm sub millimeter with respect t all types f Insphere mdels. Trpsphere errrs T study the effect f trpphere errr n the slutin, a prcess f Mecca Cde-Phase bservatins was carried ut several times. Every run utilized the same prcessing parameters except that the first run utilized a trpsphere mdel frm the fllwing mdels: Simplified Hpfiled Mdel Saastiminen Mdel Essen & Frme Mdel N Trpsphere Mdel The differences between the resulted crdinates fr each trpsphere mdel used and the riginal values are depicted in tables (6 ) t (9). Table (6): The differences between the default cde- phase slutin and the cde-phase slutin by replacing the trpsphere mdel with simplified Hpfield at Mecca netwrk. pint E(mm) N(mm) H(mm) G G G M As it is shwn tables in (6), the differences between the Simplified Hpfield trpsphere mdel values and the cmputed values ranging between -.1mm and -.2mm in east 5

6 cmpnent,.5mm t 1. mm in nrth cmpnent and ranging between 9.6 t 26.3 mm in height cmpnent. By changing the trpsphere mdel t Saastaminen mdel the difference t phase slutin is shwn in table (7). Table (7): The differences between the default cde- phase slutin and the cde-phase slutin by replacing the trpsphere mdel t Saastaminen mdel at Mecca netwrk. pint E(mm) N(mm) H(mm) G G G M The differences between the Saastiminen trpsphere mdel values and the cmputed values ranging between.3mm and 1. mm in height cmpnent and.1mm differences in east and nrth cmpnent. By changing the trpsphere mdel t Essen and Frme mdel the difference t phase slutin is shwn in table (8). Table (8): The differences between the default cde- phase slutin and the cde-phase slutin by replacing the trpsphere mdel t Essen and Frme mdel at Mecca netwrk pint E(mm) N(mm) H(mm) G G G M The differences between the Essen & Frme trpsphere mdel values and the cmputed values ranging between.1mm and.2mm in east cmpnent, -.2 mm t -.6 mm in nrth cmpnent and ranging between -.7 t mm in height cmpnent. By changing the trpsphere mdel t N Trpsphere mdel the difference t cde-phase slutin is shwn in table (9). Table (9): The differences between the default cde- phase slutin and the cdephase slutin by replacing the trpsphere mdel t n trpsphere mdel at Mecca netwrk. pint E(mm) N(mm) H(mm) G G G

7 British Jurnal f Earth Sciences Research M The differences between the cmputed values and the values, which n trpsphere mdel used, ranging between -13.6mm and 16.7mm in east cmpnent and -5.6 mm t 25.9 mm n nrth cmpnent and in height cmpnent ranging between t m Finally, it is revealing that trpsphere mdels have higher effect n the height cmpnent than n the east and nrth cmpnents. The difference between the trpsphere mdels was very small but it higher if n trpsphere mdel used, s the use f any trpsphere mdel is better than when n mdel used. Kinematic cde & phase slutin. In this mde, there are tw prcessing categries based n the number f the references statins. The first prcessing categry is perfrmed based n using ne single reference statin while the secnd categry is based n using three reference statins. The results f bth categries are utlined and analyzed in the fllwing sectin. By changing the trpsphere mdel t simplified Hpfield, as an example fr the cmmn trpsphere mdel, the difference in phase - cde slutin is shwn in figure (2). Figure (2) shws the differences between the Simplified Hpfield trpsphere mdel and the cmputed values. The difference in east cmpnent varies in ranges between -.8mm and 1.4 mm, the mean differences equal.12 mm and the standard deviatin.47 m The difference in nrth cmpnent varies in ranges between -1.4mm and 1.6 mm, the mean differences equal.3 mm and the standard deviatin.6 m The difference in height cmpnent varies in ranges between 11.4 mm and 32.2 mm, the mean differences equal 18.77mm and the standard deviatin 4.4 m single referance height errr Minimum.114 Maximum.322 Mean.1877 Standard deviatin nrth errr Minimum Maximum Mean Standard deviatin east errr Minimum -.8 Maximum.14 Mean Standard deviatin GPS Time(sec) simplified hpfield phase Figure (2): The differences between the default cde- phase slutin and the cde-phase slutin by replacing the trpsphere mdel t Simplified Hpfield at Mecca netwrk By changing the trpsphere mdel t n trpsphere mdel the difference in cde-phase slutin is shwn in figure (3). Figure (3) shws the differences between the n trpsphere mdel used and the cmputed values. The difference in east cmpnent varies in ranges between 13.3mm and 37.5 mm, the mean differences equal mm and the standard deviatin 6.44 m The difference in nrth cmpnent varies in ranges between -2.2mm and 7

8 British Jurnal f Earth Sciences Research 25 mm, the mean differences equal 9.66mm and the standard deviatin 5.51 m The difference in height cmpnent varies in ranges between mm and -189 mm, the mean differences equal mm and the standard deviatin 21.8 m single referance height errr Minimum Maximum Mean Standard deviatin Minimum -.22 Maximum.25 Mean nrth errr Standard deviatin east errr Minimum.133 Maximum.375 Mean.2147 Standard deviatin GPS Time (sec) n trpsphere phase Figure (3): The differences between the default cde-phase slutin and the cde-phase slutin by replacing the trpsphere mdel t n trpsphere mdel at Mecca netwrk By perfrming the secnd run using Cde-Phase nly, with multi references, we use the main default parameters with cde-phase bservatins and Hpfield mdel fr trpsphere. In the fllwing prcesses the trpsphere mdel is replaced nly, with fixing all ther prcessing parameters. By changing the trpsphere mdel t simplified Hpfield the difference t cdephase slutin is shwn in figure (4) multy-referances height errr Minimum Maximum.825 Mean Standard deviatin nrth errr Minimum Maximum.265 Mean Standard deviatin east errr Minimum -.23 Maximum.25 Mean Standard deviatin GPS Time(sec) simplified hpfield phase Figure (4): The differences between the default cde- phase slutin and the cde-phase slutin by replacing the trpsphere mdel t Simplified Hpfield at Mecca netwrk fr multi references CONCLUSIONS In this research 5 statins f Hlly Mecca were prcessed with Leica Gestatinary Office sftware (LGO) and cmpare the result with thse btain in ITRF2. The result shws the general agreement. Slutin is better than 4mm fr daily slutins, and the repeatability is abut 2mm, 35mm, 45mm fr N,E,H cmpnents. The difference between the crdinates and thse btain in relative mde in ITRF 2 are due t the ambiguity reslutin and cmbinatin f slutin in sftware. The difference between baseline cmputed and relative mde is better than 3 m The use f precice ephemeris rather than bradcast ephemeris, Klbucher insphere mdel, andhpfield r saastiminen trpsphere mdel wuld give an apperciable imprvement fr all baselines. Als, The trpsphere mdels have the same effect n the all bservatin teqniques, the Hpfield mdel give the same results with the Saastiminen mdel as additin f midel result values between the Simplified Hpfield mdel and Essen & Frme mdel. 8

9 REFERENCES Alves P. (24) Develpment f tw nvel carrier phase-based methds fr multiple Reference statin psitining, PhD, The University f Calgary, Canada. Hfmann et al (1997), "GPS:Thery and Practice". 4th revised ed., Springer Wien New Yrk 389p.. Kuba, J.(23) "A Guide t using internatinal GPS Services (IGS) Prducts".GSD,Ottawa, Ontari Canada, February. Kuba, J. and Herux, P.(2), "GPS Precise Pint Psitining) using IGS Orbit Prducts", GSD,NRCan. Rulstn, A. Talbt N.and Zhang K.(2), " An Evaluatin f Varius GPS Satellite Ephemeris, "Prceedings f ION GPS 2,Salt Lake City, UT. Salam, M.A.Ga, Y. and Shen, X. (22), "Analyzing the Perfrmance Characteristics f a Precise Pint Psitining System", Prceedings f ION GPS,Prtland Shen,X.(22), "Imprving ambiguity Cnvergence in Carrier Phase-based Precise Pint Psitining" MSc. Theses, Department f Gemetrics Engineering, University f Calgary. Witchayangkn, B.(22), "Elements f GPS Precise Pint Psitining", M.Sc.Thesis, the University f Maine December Zhng-yi,C,Cheng. H; Xia-Gng,H.(22), "Slutin f Reginal GPS Netwrk Using Precise Pint Psitining with Undifferenced Data", Chinese Astrnmy and Astrphysics 26,p