Engineering and Construction - Research Paper Example

Essay, Engineering and Construction INTRODUCTION During Global Navigation Satellite Systems (GNSS) measurement, there are a number of errors that degrade the positional ground accuracy. To achieve survey grade accuracies for engineering work, these errors need to be mitigated. Differential (D)GNSS methods are usually used to mitigate the majority of these errors, whereby reference stations with known coordinates permit the measurement of relative errors which can be used as correction parameters in real-time or post measurement. This has been the principal surveying technique used to achieve centimetre accuracies for the past decade. An alternative to this differential method is Precise Point Positioning (PPP). PPP achieves centimetre accuracies using only a single receiver by using precise orbital and clock data instead of the standard broadcast navigation message in post processing. A popular way of carrying out post processing is by using online PPP post-processing services. In recent years, several post processing services have been developed by government agencies, universities and survey companies. The aim of this paper was to evaluate the PPP technique for surveying in terms of the x,y and z accuracies achieved from online PPP post-processing services. To date no such study using PPP measurements has been undertaken in Ireland. As GNSS orbit and clock products are by their nature global, PPP is rapidly becoming a technique that has the potential to replace existing range-limited GNSS relative positioning methods (Marreiros, 2012). This paper provides sufficient detail to allow different industries to determine if it is a suitable alternative to current relative positioning techniques. BACKGROUND PPP is a relatively new positioning technique developed in the 1990’s, it has an advantage over traditional differential methods in that it removes the need for the user to establish their own local reference station or to have access to reference stations operated by others (Rizos et al., 2012). PPP is a very good alternative positioning technique that can be used all over the world as the GNSS orbit and clock products used in the PPP solution are by their nature global (ESA, 2011). Additionally PPP can provide more consistency over large areas because its position solutions are referred to a global reference frame whereas DGNSS position solutions are relative to local base stations (Gao, 2006). PPP offers a valuable solution where a control network or Continuously Operating Reference Stations (CORS) is simply unavailable (Van Sickle, 2009). As there is only one receiver being used and no corrections being sent by a base station, pseudo range observations and phase information are combined to fix the integer ambiguities on undifferenced phase measurements. This is what distinguishes PPP from DGNSS. Currently many autonomous and relative GNSS measurement and positioning techniques can be used to achieve varying levels of accuracies. Table 1 outlines the main techniques and the achievable accuracies currently in use. Table 1: Achievable GNSS accuracies Technique Service Code Carrier Accuracy Autonomous ✔ +/- 10m DGNSS ✔ +/- 3 – 5m Augmentation Service (GBAS) & (SBAS) Marine Beacon ✔ +/- 5m EGNOS ✔ +/- 1 – 3m WASS ✔ +/- 1 – 3m OmniSTAR XP ✔ +/- 15cm OmniSTAR HP ✔ +/- 10cm Static ✔ +/ - 0.005m RTK ✔ Distance dependant 1km – 1cm 10km – 2cm NRTK SMARTNet ✔ +/- 0.01 - 0.015m VRSNow ✔ +/- 0.01 - 0.015m TopNET ✔ +/- 0.01 - 0.020m To carry out PPP three things are required: improved orbit corrections, improved clock corrections and improved error modelling as discussed below. International GNSS Service (IGS) PPP requires precise orbit and clock files to compute the precise position of the receiver. Errors in the satellites orbit and clock data can lead to errors in accuracy. The International GNSS Service (IGS) is a voluntary federation of worldwide agencies that uses a global GNSS tracking network of over 300 stations to collect GPS and GLONASS data used to compute precise orbit and clock products. Many national geodetic agencies and GNSS users interested in geodetic positioning have adopted the IGS precise orbits to achieve centimetre level accuracy and ensure long-term reference frame stability (Kouba, 2009). The IGS products consist of a number of different qualities as seen in Table 2, which are generated by 12 IGS Analysis Centres (AC) around the world. These IGS products can be used in real time or within hours, days and weeks of the initial session with the highest quality coordinates achievable using ‘Final’ products, which have a latency of up to 2 weeks. A significant improvement is seen in the accuracy for the IGS products over the broadcast ephemeris (Tekmon, 2013). As seen in Table 2, the broadcast ephemeris orbital accuracy is 100cm while the IGS Ultra-Rapid orbital accuracy is 5cm. This is significant considering both these ephemerises can be obtained in real time. The daily computation of global precise GNSS orbits and clocks by IGS, with centimetre precision, facilitates a direct link within a globally integrated reference frame which is consistent with the current International Terrestrial Reference Frame (ITRF) (Kouba, 2009). IGS products are used in a number of online post processing services which apply different algorithms to compute the receiver’s precise position. Table 2: IGS Product Table (IGS, 2013) Product Parameter Accuracy Latency Broadcast Orbit 100 cm Real Time Clock 5 ns RMS 2.5 ns SDev Ultra Rapid (Predicted) Orbit 5 cm Real Time Clock 3 ns RMS 1.5 ns SDev Ultra Rapid (estimated) Orbit 3 cm 3 – 9 hrs Clock 150 ps RMS 50 ps SDev Rapid Orbit 2.5 cm 17 – 41 hrs Clock 75 ps RMS 25 ps SDev Final Orbit 2.5 cm 12 – 18 days Clock 75 ps RMS 20 ps SDev Online Post Processing Services In recent years several post processing software products implementing a PPP processing strategy have been developed by government agencies, universities, industries and individuals (ESA, 2011). PPP algorithms using undifferenced carrier phase observations have been implemented in different ways as each use unique baseline tools and processing strategies (Silver, 2013). To avail of these free processing tools users upload their RINEX data, observed in a static mode, to the post processing website and their computed position is returned within a few minutes. Table 3 lists a number of online services currently available. Table 3: On-Line Services On-Line PPP Post Processing Services Trimble CenterPoint RTX PPP Limitations PPP is a great addition to GNSS positioning solutions already available but it does have limitations, the most significant disadvantage compared to other techniques is that is has a long convergence time (up to 20 minutes) compared to a convergence time with DGNSS of only several minutes. The long convergence time is required as it is not possible to uniquely isolate the integer nature of the ambiguities, the higher precision of the carrier phase can still be accessed by estimating a random constant bias in place of the ambiguity (Bisnath et al., 2010). As can be seen in Table 4, many of the biases and errors that require corrections are associated with the satellites and geophysical modelling these include clock corrections, antenna phase offsets, ocean loading, etc. Currently, applications with a need for real-time positioning make no use of the PPP technique. It is anticipated that PPP could replace other GNSS techniques, if some of the aforementioned problems could be resolved or reduced (Huber et al., 2010). Table 4: Biases and errors that need to be accounted for in PPP and DGNSS (Rizos et al., 2012) Correction type PPP Differential GNSS Satellite specific errors Precise satellite clock corrections ✔ ✘ Satellite antenna phase centre offset ✔ ✔ Satellite antenna phase centre variations ✔ ✔ Precise satellite orbits ✔ ✔/✘ Group delay differential ✔ (L1 only) ✘ Relativity term ✔ ✘ Satellite antenna phase wind-up error ✔ ✘ Relative specific errors Receiver antenna phase centre offset ✔ ✔ Receiver antenna phase centre variations ✔ ✔ Receiver antenna phase wind-up ✔ ✘ Geophysical models Solid earth tide displacements ✔ ✘ Ocean loading ✔ ✘ Polar tides ✔ ✘ Plate tectonic motion ✔ ✘ Atmospheric modelling Tropospheric delay ✔ ✔ Ionospheric delay ✔ ✘ METHODOLOGY In this study three separate surveys were carried out (Survey A, Survey B and Survey C) as seen in Table 5. These static surveys differ in terms of: measurement period, duration and instrumentation, thus facilitating a comparative analysis of conditions. A test site was established on the roof of Dublin Institute of Technology (DIT) over an Ordnance Survey Ireland (OSi) passive IRENET control point (D147) as seen in Figure 1. Static baseline data was subsequently post processed from multiple CORS stations in the region using Trimble Business Center (TBC). Table 5: Instrument Settings during Field Survey Trimble R10 (Survey A) Date Carried Out 11th February 2014 Survey Time 10:39:45 -13:59:15 Survey Duration 3h 19m 30s Antenna Height 1.568m Datum WGS84 Elevation Cut-Off 10 Degrees Recording intervals 15 Seconds Figure 1: Instrument set up over D147 Online Services Once the data was converted to RINEX format, it was sent to the online post processing services as outlined in Table 6 and discussed below Table 6: Online Services Submission Information (NS- Not Stated) Online Services Supported Devices Data Required Satellite Data Used Registration Needed R10 5800 Trimble RTX ✔ X RINEX/ .TO/ .DAT GPS/GLONASS Yes Trimble RTX: The R10 data could only be used with this service as the 5800 receiver was not a listed device. As Trimble RTX supports .T02 file formats, the unconverted raw data was submitted. The reference frame selected was ITRF2008. Transformations The online post processing services returned their solutions via e-mail in the ITRF08 co-ordinate system at the current epoch. Some of the services, such as Magic GNSS and CSRS-PPP return the ITRF co-ordinates in Geodetic format (Latitude and Longitude) while the remaining services returned the ITRF co-ordinates in both Geodetic and Cartesian (X, Y, Z) formats. It should be noted that OPUS returns IGS08 co-ordinates, but these are taken as being coincident with ITRF08. As Ireland’s projected co-ordinate system is ITM, a co-ordinate transformation was required from ITRF08 to ITM. By using a Helmert 7-parameter transformation with the published parameters, the ITM co-ordinates were computed from the ITRF08 co-ordinates. A number of different steps were taken to convert from ITRF08 (Cartesian) and ITRF08 (Geodetic) to the final ITM co-ordinates as can be seen in Figure 2 and Figure 3, respectively. The first transformation was from ITRF08 (Cartesian) to ITM. The first stage to this was to transform the ITRF08 co-ordinates to ETRF2000. ITRF08 Cartesian coordinates (X1, Y1, Z1) were transformed to ETRF2000 Cartesian coordinates (X2, Y2, Z2) using the 7-parameter Helmert transformation equation illustrated in Equation 1: Eq. 1 Where: X1, Y1, Z1 - ITRF08 Cartesian co-ordinates D - Scale factor between the two systems RX, RY, RZ – X, Y, Z rotation angle in radians between the two systems TX, TY, TZ – Geocentric X, Y, Z translations (velocity) This equation was simplified further by multiplying the matrices: X2 = X1 + TX + D*X1 + RY*Z1 – RZ*Y1 Y2 = Y1 + TY + D*Y1 – RX*Z1 + RZ*X1 Z2 + Z1 + TZ + D*Z1 + RX*Y1 – RY*X1 The parameters of the transformation were acquired from Altamimi’s EUREF Symposium presentation (2011) on the transformation from ITRF to ETRF2000 as can be seen in Table 7. Table 7: ITRF08 - ETRF2000 Parameters (Altamimi, 2011) ITRF Solution TX mm TY mm TZ mm D RX mm RY Mm RZ mm ITRF2008 52.1 49.3 -58.5 1.34 0.891 5.39 -8.712 Rates 0.1 0.1 -1.8 0.08 0.081 0.49 -0.792 All results provided by the online services were in ITRF08 @ epoch 2014. Using these parameters and their rates, the ITRF08 co-ordinates were transformed into ETRF2000 @ epoch 2014. Errors are introduced during the calculation of the transformation due to the nature of the parameters, but general accuracies of the transformation formulas are 1-2 cm (Jivall, 2013). The ETRF2000 co-ordinates computed were in the Cartesian co-ordinate system but needed to be converted into a Geodetic co-ordinate system to be subsequently used in the OSi’s online co-ordinate converter. ETRF2000 Cartesian co-ordinates (X2, Y2, Z2) were converted to ETRF2000 Geodetic co-ordinates ( 2, 2) referenced to the GRS80 ellipsoid using the sequence of standard equations illustrated from Equation 2 – Equation 7: Eq. 2 Eq. 3 Eq. 4 Eq. 5 Eq. 6 Eq. 7 Where: X2, Y2, Z2 – ETRF2000 Coordinates = Latitude (decimals of a degree) λ 2  = Longitude (decimals of a degree) = Prime vertical radius of curvature a = Semi-major axis of the GRS80 ellipsoid (6,378,137.000 m) e2 = Eccentricity squared of the GRS80 ellipsoid (0.006 694 380 022 90) = Ellipsoidal Height As was an unknown in above the equations, to compute and , an iterative process was adopted (Mooney, 2009). When and were computed, was used as an input into the equation and the process repeated. When converting between Cartesian – Geodetic and vice versa the only errors introduced are rounding errors, to ensure these rounding errors did not affect the solution, the Geodetic co-ordinates were calculated to five decimal places. The co-ordinates were then ready to be used in the OSi’s co-ordinate converter. This allows for the ETRF geodetic coordinates to be projected into ITM. The second set of transformations carried out were ITRF08 (Geodetic) to ITM. The only difference between this transformation and ITRF08 (Cartesian) to ITM was the initial step, the Geodetic co-ordinates had to be converted into Cartesian before the transformation could be computed. ITRF08 Geodetic co-ordinates (,) referenced to GRS80 ellipsoid were converted to ITRF08 Cartesian co-ordinates (X2, Y2, Z2) using the following sequence of standard equations (Eq. 8 – Eq. 12): Eq. 8 Eq. 9 Eq. 10 Eq. 11 Eq. 12 Where: = Prime vertical radius of curvature a = Semi-major axis of the GRS80 ellipsoid (6,378,137.000 m) e2 = Eccentricity squared of the GRS80 ellipsoid (0.006 694 380 022 90) = Latitude (decimals of a degree) = Longitude (decimals of a degree) h = Ellipsoidal Height (metres) Once this initial step was complete, the process was the same as the previous transformation. RESULTS & ANALYSIS The results achieved using the online post processing services were remarkably accurate and are listed in Table 8. Figure 4 illustrates the coordinate differences in E, N and H. Table 8: D147 Known vs. Online Services (Survey A) Figure 4: D147 Known vs. Online Services Graph (Survey A) All horizontal positions computed by the PPP online services were within one centimetre of the known D147 coordinates. The vertical positions were no larger than three centimetres; this was considered acceptable as vertical accuracy is generally 2-3 times less accurate. Considering the observation time of Survey A was 3 hours 19 minutes, these results were very good as both AUSPOS and OPUS recommended 6-hour observation to yield centimetre accuracies. The Trimble RTX service states that with 1 hour of data, 2 cm accuracies can be achieved while 24 hours of data can yield 1 cm accuracies. The results of Trimble’s RTX service surpassed centimetre accuracies with no difference in the Easting, a 2mm difference in Northing and a 4mm difference in Height. This is an accuracy level only usually achievable by a static survey and indicates that RTX could be used to establish reliable control. CSRS-PPP, AUSPOS and OPUS could also be considered for establishing 2D control but the computed heights have a difference greater than 5mm. An accuracy of +/-5mm is an industry standard for setting out vertical control (RICS 2010). Figure 5 illustrates the 2D positions of each point plotted in ITM. This shows the horizontal spatial relationship each point has to D147. Again the Magic GNSS computed position is clearly an outlier. Figure 5: Coordinates Spatial Relationship with D147 (Survey A) The results from Survey B are illustrated in Table 8. This survey had an observation time of 1 hour 9 minutes. In this instance the Trimble RTX results were as quoted on the Trimble website: 1 hour of data yielded 2 cm accuracies. AUSPOS and Magic GNSS both resulted in poor Easting positions but highly accurate Northing positions (2mm). From this survey it was also evident that Trimble RTX and CSRS-PPP had a significantly better height accuracy when compared to the other services. It was therefore deduced that more accurate height values resulted from better online algorithms and more appropriate reference stations, especially evident in the case of Trimble RTX. Table 8: D147 Known vs. Online Services (Survey B) From these surveys, it was found that an observation time of 3 hours is sufficient to produce centimetre accuracies. Observation times of less than 3 hours produce accuracies undesirable for setting-out control. Other than Trimble RTX, it was difficult to assess why one service was better than another. Some of the services returned processing reports, which identified the IGS CORS sites used in the solution. AUSPOS for example used 13 IGS CORS to process the solution as seen in Figure 5.1, but did not produce a better solution than OPUS which only used three stations. Trimble RTX uses its own global tracking network with over 100 reference stations with a regional CORS network used to determine the local atmospheric corrections. With Trimble having a fully operational CORS network in Ireland, the RTX service will invariably produce the most reliable results. Figure 6: IGS CORS Stations used in AUSPOS solution As seen in Table 9 the Rapid and Final solution only increased in accuracy by a small margin from the Ultra-Rapid solution. The Rapid and Final solutions were found to be identical. This is unusual as the Rapid products are available within 17 hours of the observation, while the Final products are only available within 12 days of the observation. The IGS also states that the Final products should be utilized to yield the best positional accuracy. Table 9: Difference between International GNSS Service Products CONCLUSIONS The aim of this study was to determine the accuracies achievable from online PPP post processing services. It was concluded that online PPP post processing services for positioning in Ireland yield remarkably accurate results. The horizontal difference between the TBC post processed coordinates from static measurements and online PPP services coordinates for control point D147 were at the millimetre level. This indicates the processes and algorithms used by the services are robust. Trimble RTX tended to produce the most accurate results and proved to be the most reliable in the vertical component. Thus PPP has proven to be a suitable alternative technique to establish control. In addition, the quality of the online PPP post processing services was shown to depend on the observation time span, and a longer observation time was shown to be an effective means of improving the solution to achieve millimeter accuracy. These results present a methodology for using PPP online services in Ireland and are an initial indication of PPP capabilities. This study has only determined PPP accuracies at a single point with a small sample of data and a further study with respect to determining PPP accuracies on several OSi passive control points with different observation times (1hr, 2hr, 3hr, 6hr, 12hr and 24hr) within Ireland is recommended. This would provide a more robust indication of PPP accuracies. This paper is the first to evaluate the performance of PPP online post processing services for Ireland. The effective methodology developed is intended to provide a guideline for future research. ACKNOWLEDGEMENTS This study was undertaken as part of an Undergraduate Honours Degree in Geomatics at the Dublin Institute of Technology (DIT), Dublin Ireland. Additional detail is provided in the final year Dissertation (O’Mahoney, 2014). All hardware and much of the software used during this study were Trimble products used under license of the DIT. The work was carried out under the guidance of my Dissertation supervisor Dr. Audrey Martin. REFERENCES Altamimi, Z 2011, Transformation from ITRF to ETRF2000, presented to EUREF Symposium 2011, Chisinau, Moldova, May 25-28th. Bisnath, S. Collins, P. Heroux, P and Lahaye, F 2010, Undifferenced GPS Ambiguity Resolution using the Decoupled Clock Model and Ambiguity Datum Fixing, PPP-Wizard, viewed Feburary 2014, . ESA 2011, Precise Point Positioning, European Space Agency viewed October 2013, Gao, Y 2006, Precise Point Positioning and its Challenges, Aided-GNSS and Signal Tracking, Inside GNSS, vol. 1, no. 8, pp. 16-18. Huber, K, Heuberger, F, Abart, C,, Karabatic, A, Weber, R & Berglez, P 2010, PPP: Precise point positioning-constraints and opportunities. Paper presented to FIG Congress 2010, Sydney, Australia, April 11-16th. Jivall, J 2013, Simplified transformations from ITRF2008/IGS08 to ETRS89 for maritime applications, Lantmateriet Sweden, viewed April 2014, . Kouba, J 2009, A guide to using International GNSS Service (IGS) products, Natural Resources Canada, Ontario, Canada Marreiros, J.P 2012, Kinematic GNSS Precise Point Positioning - Applications to marine Platforms, Ph.D. thesis, Universidade do Porto. Mooney, K 2009, Co-ordinate Systems (PowerPoint Presentation), Spatial Information Sciences, DIT, Dublin. O’Mahony, K 2014, An Evaluation and analysis of the achievable accuracies from online Point Positioning PPP Post-Processing Services in Ireland, Undergraduate thesis, Dublin Institute of Technology May 2014. RICS 2010, Guidelines for the use of GNSS in land surveying and mapping, 2nd edn, Royal Institute of Chartered Surveyors, Coventry, UK. Rizos, C, Janssen, V, Roberts, C and Grinter, T 2012, Precise Point Positioning: Is the Era of Differential GNSS Positioning Drawing to an End? Paper presented at FIG Working Week 2012. Rome, Italy, May 6-10th. Tekmon Geomatics 2013, Satellite Orbit Errors, viewed Feburary 2014. Silver, M 2013, Seven Alternatives to OPUS GPS Post-Processing During U.S Federal Government Shutdown, GPS World, viewed October 2013. Van Sickle, J 2009 GPS for Land Surveyors, 3rd edn, Taylor & Francis, New York. Read More