Scribd Logo
Upload

Welcome to Scribd!

  • Geomatica Vol 4 No 42000

    Uploaded by

    Cristian Poschina
    5 views8 pages

    Original Title

    GeomaticaVol4No42000

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    5 views8 pages

    Geomatica Vol 4 No 42000

    Uploaded by

    Cristian Poschina

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    of 8
    GE QO Mimo cA A PRACTICAL EVALUATION OF THE GPS RAPID STATIC METHOD MN M102!0 C. Santos, Department of Geadesy and Geomatics Engineering ea University of New Brunswick, Fredericton, New Brunswick Cristiano B. Souza and Sfivio R.C. de Freitas, Departamento de Geomatica Universidade Federal do Parand, Curitiba, PR Brazil The rapid static method is till an atractve method of surveying with GPS. 1 combines high accuracy and short observation periods, bing useful in applications when redundancy of observations is desired, suchas in Tega surveys. Tis paper describes an experiment and its corresponding dasa analysis aimed at evaluating the GPS rapil static method in a production envirownent. The evaluation demonsirates the envelope for the ‘method: occupation times, rang, reliability and accuracy: For this purpose a rapid static survey was carried ‘out over geodetic markers used as “ground tush”. In the experiment, baselines of various lengths were used, fiom 388km 10 75 kn, Dexa were collected using dual-frequency receivers. The data processing generated both ‘single and dua.frequency solutions for thee different lengths of sessions: 5, 10 and 15 minutes. The results were analysed in terms of accuracy and repeatability The analysis suggests that LI-only soltions are reliable ‘only for very short baselines. The scae analysis of results indicates thatthe dual frequency solution is reliable for baselines of up to 15 km There is practically no difference among solusion from the three sessions La méthode statique rapide avec le GPS est encore une technique de levés atrayante. Elle alle une haute précision et de courtes périodes observation; elle est utile dans les applications oi fa redondance des ‘observations est souhattée, notamment dans les levés officiel. Cet article décrt une expérience et Canalyse connere des données, aus fins de Vvaluation de la méthode statique rapide avec le GPS dans un contexte de production, Lévaluation est consacrée &Verveloppe de la méthode : temps aceupation, porée,fabilité et précision. A cente fn un lve statique rapide «été réalisé sur des reperes géodésiques servant de « vérté sol », Pour cette expérience, des bases géodésiques de diverses longueurs ont été utilisées ~ de 3,8 km & 75 km. Les données ont dé collectées& Taide de récepteurs bifréquences. Le traitement des données a produit des solutions monojréquence er bifréquence au cours de mois séances de durées différentes : 5, 10 et 15 minutes Les résultaus ont été analysés pour leur précision et pour leur répétabilité. Lianalyse suggere que les solutions exclusivement L-I ne sont fables que pour les bases géodésiques irs courtes. La méme analyse des résultats Indique que ta solution bifréquence est able pour les bases géodésiques atteignant jusgu'a 15 kom. I n'y a ratiquement pas de difference entre les solutions des trois séances. Introduction ‘The rapid static method is among the most ‘atraetive tebnigues for surveying with GPS. tis an altemative method to conventional surveying trres- ‘cial techniques. Within the surveying methods using GPS, the rapid static method can be placed between the traditional static and the real-time kinematic (RTK), The traditional static method is based on a Jong occupation time offering the highest accuracy and the most reliable results. RTK is the fastest sur- ‘vey method based on single-epoch occupation (at the Timit), but iti restricted to short baselines. Between thems the rapid state method, offering beter accu- racy and reliability than RTKCand faster surveying than the traditional fong occupation static method. ‘The rapid static method also allows forthe postion- ing of points with an accuracy comparable to that of the GPS traditional static method, for shorter base- lines, in a much shorter period of time (less than 15 minutes, boing the eason forthe adjective “rapid”. The last characteristic isa consequence ofthe devel ‘opment of fast ambiguity resolution techniques. "The rapid static method his been widely used in support of many activities. Just few foun elsewhere in the Hiteatue will be mentioned here. Wie and Lin 1995} tried to get te best from the method for height ‘determination. Shepherd et al. [1998] report on the application of the method in ground deformation ‘monitoring. Coe eral. (2000) applied raid static sur veys in restricted but mountainous area in Colorado, using dual-frequeney receivers, for landslide move- ‘ment monitoring. These papers tend to agtee on The fast ambiguity resolution for long baselines is very difficult mostly due to ionospheric effects... GEOMATI formal errors less than or equal to | em in the hod ‘ontal and 1.5 em in the vertical, over basclines 0 Jonger than 15 km. These errs correspond rela tive errors in the onder of several millimetres pls 1 ‘ppm. By “Yormal error we mean here the standard eviaton as estimated during data processing. is imeresting to review typical valves tha are ‘commonly quoted among various GPS roeciver vendors (nip.teww.ashtech com, hup:tivw aoa gps.com, —— hutp:/vwwleica-geosystems.com, Inpttowsarindlecom), Generally speaking, fo dual frequency reeiver elaive eres vary between (0.5 em plus 0.5 ppm) and (1 em plus 1 ppm) for the horizontal component, and berween (0.5 cm pls [ ppm) and (2 em plus I ppm forthe vertical ‘component. For single-frequency receivers the ers range bersen (0.5 em ad em plas 2 pom) forthe | horizontal component. Even though the word | ~accuraey’ is used sometimes, it may be thought that these relative enor refer to foal er (precision), ‘This paper presents an experiment and its coe responding data analysis, which was designed t© evaluate the accuracy of the rapid static method in a ‘reduction environment, There are thre interesting questions related 10 (a) increasing the distance Fperwoen the reforenee station and the unkown points (b) using only L observations as opposed to LI and 12 observation, and (e) using observa tion sessions of watious duration, namely, 5, 10 and 15 minutes Tong. The fast static survey Was carried cout ever geodetic markers located on the Federal. | ‘University of Parand (UFPR) campus. A preliminary evaluation for a small sample was presented in Santos et af. (1998). In this current paper a wally Aitforon evaluation is made, using the published coordinates ofthe geodetic markers as ground rath, The internal precision ig ako assested by means of the shon-erm repeatability of the different solutions. Revisiting the Rapid Static Method ‘The rapid static method has been very well the literature [e.g. Kleushere 1990; va Sickie 1996]. The general idea ‘behind this method is dat fixed reference receiv er remains stationary on a base station of known coordinates. A rover receiver occupies the points of interest fr a shor period of time, typiealy between 5 and 15 minutes. This operational characteristic hasan advantage over the sate meta per se, but ‘comes with restriction in terms of baseline length ‘whieh should not he longer than 10 kilometres ca “The length ofthe session may be a funetion of the number of satellites in view since the method requires a great number of observations fora faster ambiguity resolution. Therefore pre-planning. may be necessary even today withthe full constelaton This certainly will change withthe inclusion of a sew GPS signal [Hutch e al. 2000) ‘The fast stalic method became Feasible de 10 the development of computational algorithms allowing a fast ambiguity resolution, generally Known as “on-the-fly” [Abidin 1992], These tech- rnigues can be divided into. three large groups, namely, the “exttaide lang” [Wibenna 1989), the ambiguity mapping function {Counselman and Gouverich 1981], and the least-squares [el 1990), The “extrawide laning” method is based on eating atifcial observations by linearly combin- ing the respective observations at both LI and L2 frequencies, being the so-called wide-lane and nar= row-lane the 1wo artificial observations: mostly used, These erificial observations possess avant geous characteristics for a fast ambiguity resol tion, such asa larger wavelength for the wide-lane. The ambiguity mapping function and the method based on Teas squares use a search technique rely- ing om statistical criteria In the forme, a statistical tes is performed om a function formed by a combi- nation of observations, on either Lt or L1 and L2, between receivers and satellites, In the later, ast tistical testis operated on tho a posterior’ variance factor resulting fom the adjustment of observa tions [Venice and Krakiwisky 1986). Either the mapping function or the « posterior variance Factor corresponds tothe corect ambiguity in statistical sense. More effort has been made in order 10 increase the efficiency of those methods. suet as the least squares LAMBDA method (Teunissen et ‘al, 1997). The fast ambiguity resolution for long baselines is vey dificult mostly die to ionospher ie elfecs ad is sill matter under investigation (kim and Langley 2000, There is reliability issue related vo ambiguity resolution, which relates to how well the resolution actually occurs and the capability of the data pro- cessing software to identify it. There are four pos sible seenarios: (a) the ambiguities are comect resolved and the sofiware indicates such: (by the ambiguities are resolved, perhaps correctly perhaps rol, and the software indicates they are correctly resolved; (¢) the ambiguities are resolved, perhaps cconectly perhaps not, and the software indicates that there isa possibilty the resolutions are wrong; and, () the software gives up tying Wo resolve the ambiguities (and soles for a Moat solution, tor example), but informs the user what is happening. ‘The reliability problem iy ease (). Experiment Description ‘An experiment was devised to assess the per formance ofthe fast static method in a practical sit uation aimed at matching as best as possible a generic proton envionment. The idea was 10 carry out « rapid static surey on some geodetic nurkers, and 19 compare the computed coordinates with their published values. The geodetic markers tsed were the three ground monuments lacate With: in he test area of UFPR's Space Geodesy Lab. They are known as points RMO1, RMO2 and RMO3. They are reference ations ofthe GPS permanent station ‘PARA that belongs tothe Brazilian GPS Continous Monitoring Network (RBMC).As such, their eo0e dinates are known with ilimre accuracy, ll lat fed 10 the South American Geoventric Reference System (SIRGAS) [/8GE 1997). ‘The control points used as reference stations belong 10 a GPS network maintained by the Electricity Company of the State of Parand {COPEL}, also connected to SIRGAS at the mil Timetre level, It was decided to use some of the points of dhe COPEL network, not only because of the quality oftheir coordinates, bat also because of| their increasing distance with espect to the test area, spanning from 38 km up to 72 km. This meant that the useline would he Of various lengths. Table L shows the station names, code. altitude and distance to theest area. Al stations, including the ones atthe test ate, are located atthe Sern do Mar plateau hhving neaty the sme height. The exception is for station CRI whieh is located close to sea-level ‘Three Ashtech Z-12 receivers were used in the experiment. Tw of them, the ones that helong to the LUFPR, were used for the rapid static survey of the geodetic markers RMO1, RMO2 and RMOB. The tied ‘one, which belongs fo COPEL, was used to occupy the reference stations belonging to the COPEL net ‘work. The survey took place in July 1997, over two ‘consecutive days. Stations UBRB and CRLT were occupied on the frst day; the others on the second ey. Each geodetic marker was ozeuped fora period of 15 minutes, with 5 scons sampling ate this overall scheme, there are alays Ue solaions (or the tree geodetic markers) associated with each one ‘ofthe reference sation ‘The points RMOL, RMO2 and RMO3, and the COPEL points were ted to SIRGAS at an epoch pre- ious to that of the rapid stale survey. In order to ‘make the analysis temporally consistent, the cooed- nates were made compatible with the epoch of the rapid state survey by spying the NNR-NUVELIA plate motion model [Dede eta. 1994]. The use of {his model is recommenced by the HERS [McCarthy GEOMATICA “Tnble 1: COPEL network points used as reference stations, and thelr height and distance to the test aren, NAME, CODE _ HEIGHT (m) Uboraba—_|_UBRB 911.196 ‘Aub REN 928.808 Filarzinho | 985.699 eos KMOs 952.840 ‘Campo Comprido | SCCO 976293 Bateas SBAT gui | ERLE Para 1996) For the Bravlan shield the NNR-NUVELIA ‘made shows good agreement with recent soltions Taso on spice gondetic tchnigues such as the APKIMS0 [Drewes 1993], and also with the recent resulls ofa study carried out atthe Federal University of Parad by Cosa [1999] A diference of 25 em exists between the two epochs and would have been translated into the result if no taken into acco Data Analysis The whole dataset was processed using Ashloch’s Prism soliwne, The idea behind using ‘commercial rar than seieniie software was to simulate a peeduction environment. For the same reason, broadcast ephemerides were used in the data pracessing Te data collected atthe geodetic markers (nthe test area) were divided into 5, 10 and 15-minute ‘segments in such away ato have ifleent solutions computed from diferent sessions fr each one of the _eodic markers. In adition, the data were process ting only L1 abservtions or LI and L2 observ tions. These two types of solutions are hereafter refered as L1-only solutions and V1.2 solutions. The resulting set of coordinates was fist analysed exteraly, by comparing thei respective published coordinates, and hen internal, by means ‘of short-term repeatability. The whole data analysis, which follows, was aimed at answering the three ‘questions raised in the Introduction. Another inter- sting featur isthe comeltion between observation type, baseline length, duration of the observation session andl ambiguity resolution, which becomes vient inthe following analy ‘he first and fundamental aspect to note is how and whether the ambiguities were solved bythe sof ware. The only LI-only solution in which all the ambiguities were resolved i the ane forthe shortest teseine (3.8 km in length) with a 15-minute session In all other Li-only solutions, no ambiguity was suceessfully solved. As far as the L1/L2 solution is 38 69 95 Tht 14 304 cy DISTANCE (km) GeO Mia tc cconcemed, all solutions related to the longest base- Tines (reference stations SBAT and CRLD dd not have any ambiguity solved, For the shortest base- lines (up t© 14.1 kp, all ambiguities were solved, forall sessions (5, 10 and 15 minutes long). The exception is fortwo of the solution related to refer- ence sation SPIL (baseline length: 9:5 km), forthe S and 10-minute observation sessions, in which no ambiguity was solved, The reason for this could not ob © © Distance (kd | gu yl 5 os © Sm, E06 © 10min Eo asin i 02 z Figure 1: Horizontal difference based on the Lt-only solution. & $s 5 8 |e Sain Es 4 24 : 6 | nin 3 a § 4 1smin 5 = 2 oe a Distance (kim) Figure 2: Veta difference based onthe L1-only solution 12 1 08 © Shin. © 10min, 4 4.19mi. of o 7 © © Distance (km) Horizontal difference (m) & Figure 3: Horizontal difference based onthe L1/L2 sol I cA be detected. As far asthe reliability issue related to ambiguity resolution is concemed, te data analysis Indiate that we are in eases (a) and (the software used was capable of corretly indicating whether ambiguities were solved or not. The first analysis was made to verify how close the estimated solutions of markers RMOI, RMO2, and RMO3 were with respoct to their eonesponding published values, regarded as “ground tra”. The Aitferences in latitude, longitude and height were all ‘expressed in length units. After that, the dference in the horizontal component was computed Figures | and 2 show the horizontal and vertical dlfeences, plotted against distance from a reference station, forthe LI-only solution. The three obser- ‘ation sessions are indicated, Each session contains 2 difference for the markers RMOL, RMO2, and MOB. Large and widely spread differences can be seen inthe L1-only solution because the ambiguities could not be property solved. The only exception is for the 3.8-km baseline, fr the 15-minute session, in Which the horizontal difference is below Sem. These Plots stress the importance of ambiguity resolution, "The results indicate that L1-only solutions can only handle ambiguities successfully for baselines shorter ‘than the ones used in the experiment Figure 3 shows the horizontal difference, plot ted against distance from & reference tation, forthe LVL2 solution. Larger differences occur for the longest baselines when no ambiguity was solved. All 10 and 15-minute sessions for the L1/L2 solations forbaselines up 1o 11-1 km are below or atthe S-cm level (withthe exception of two solutions fr station SPIL), as can be better seen in Figure 4, The reli aveuracy, involving only the solutions in which ambiguities were solved (L1/L2 solutions. with baselines up to 14 km, except two of the solutions for station SPIL), was adjusted from Figure 4 and is ‘equal 10 5 ppm. Figure 4 also indicates that there is practically no difference between the results com ing from a S-minute, a W-minute or 4 15-minute session ifthe ambiguities are resolved, {As for the vertical differences for the L1/L2 solution, they are shown in Figure 5. The differ: fences between the thre sessions for the baselines up to 14 km are better seen in Figure 6, being below 0.25 m. The average relative accuracy for the vertical component, involving only the L1/L2 solu- tions in which ambiguity was solved, was ajusted from Figure 6 and is equal 13 ppm. ‘The comparison between Figures 1 and 3, and Figures 2 and 5, clearly indicate the effet of using LL¥-only or L1/L2 solutions, The L1-only solution is widely sprea and with larger differences as oppesed to the L1/L2 solution. The incapacity ofthe L1-only solutions for solving the ambiguity is evident, with GE solitary exception for the solution with a shorter haseline and longest session. The remaining analy- sis involves only the L1/L.2 solutions because ambi puities were resolved in most of them, “The differences shown in Figures 110 6 are a litte larger than typical values (eg those mentioned inthe Introduction, which seem to be mostly relat- ‘ed 10 intemal precision, With this in mind, it was decided 10 Took into the internal precision as wel using the shor-term repeatability as an analysis tool ‘The short-term repeatability is an indicator on the scatter about the mean ofthe solution. The short-term horizontal and vertical repeata- bility 6, was computed by o where is the number of solutions associated with a reference station (in this ease, m equals 3: one solution for RMO1. one solution for RMO2, and one Solution for RMO3)x refers to the quantity under investigation (hee either horizontal or vertical com. ponents) isthe average value of x, and os the

    You might also like