Rapid static GNSS data processing using online services

Transcription

1 J. Geod. Sci. 2014; 4: Research Article Open Access M. Berber*, A. Ustun, and M. Yetkin Rapid static GNSS data processing using online services Abstract: Recently, many organizations have begun providing online GNSS (Global Navigation Satellite System) data processing services. Currently, only one of these organizations i.e., OPUS (On-line Positioning Users Service) provides a rapid static data processing option. In case of static online data processing, the users are required to submit at least two hours of data to get reasonably precise results. To provide processing option with less than two hours of data, NGS (National Geodetic Survey) developed OPUS RS for rapid static data processing so that users may submit as little as 15 minutes of dual frequency GNSS data. In this study, multiple observation sessions are conducted at the same locations to compare OPUS-RS generated coordinates among the different sessions to see whether separate values agree with each other. The results indicate that with OPUS-RS results the differences in horizontal coordinates agree with each other within 3.5 cm and vertical coordinates agree within 7.2 cm. For an independent check, OPUS-RS results are also compared against LGO (Leica Geo Office) produced Static results; this comparison yielded up to 4.5 cm variations among horizontal coordinate differences and variations among vertical coordinate differences are up to 11.4 cm. Keywords: Global Navigation Satellite Systems; horizontal and vertical coordinates; online data processing services DOI /jogs Received April 5, 2014; accepted September 8, Introduction GNSS data processing using online services (see the following section) is a subject of recent interest. Soler et al. *Corresponding Author: M. Berber: Dept. of Civil and Geomatics Eng., California State University, Fresno, CA, USA, muberber@csufresno.edu A. Ustun: Department of Geomatics Engineering, Selcuk University, Konya, Turkey M. Yetkin: Department of Geomatics Engineering, Katip Celebi University, Izmir, Turkey (2006) investigated how the accuracy of derived threedimensional positional coordinates depends on the length of the observing session using OPUS. Ghoddousi-Fard and Dare (2006) compared the performance of online services by submitting different data sets varying in time and location. Liu and Shih (2007) compared five of the online services in terms of success rate, cost, length of waiting period and choice of a specific reference frame. Tsakiri (2008) aimed to assess the four globally available online GPS processing services from the point of view of a user wishing to take advantage of such a service for datum realization. In a three part series Snay et al. (2011a, b and c) discussed the precision obtainable with OPUS. Further, Ebner and Featherstone (2008) investigated whether online services can be used to establish geodetic control networks. Among the above references Snay et al. (2011a, b and c) discusses both static and rapid static positioning. The impetus behind online rapid static positioning is that OPUS system requires at least two hours of dual frequency GNSS carrier phase data to resolve the integer ambiguities. If the ambiguities are correctly resolved, OPUS produces coordinates of the order of a few centimeters. Otherwise, the coordinates might be precise to a few decimeters. To provide processing option with less than two hours of data, NGS developed the OPUS-RS (Snay et al. 2011a) where RS stands for Rapid Static. OPUS-RS was designed to give an option to users that they may submit as little as 15 minutes of dual frequency GNSS data. OPUS-RS processing strategy differs from the one utilized in OPUS-S where S stands for Static. The precision of coordinates produced by OPUS-S depends primarily on duration of the observing session. On the other hand, the precision of coordinates computed with OPUS-RS depends primarily on three variables: (1) the duration of the observation session, (2) the geometry of the CORS/IGS (CORS: Continuously Operating Reference Stations and IGS: International GNSS Service) stations being used for the OPUS- RS solution and (3) the distances from the user s GNSS receiver to various CORS/IGS stations being used for the OPUS-RS solution. OPUS-RS solutions depend on network geometry and the distances from the rover to the CORS/IGS stations because it interpolates/extrapolates the atmospheric refraction values experienced at the CORS/IGS sta M. Berber et al., licensee De Gruyter Open. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.

2 124 M. Berber et al. tions to estimate the corresponding values at the rover (Snay et al. 2011b). Snay et al. (2011c) states that multiple observation sessions at the same location be conducted to compare OPUS- RS generated coordinates among the different sessions to see if separate values agree with each other within their specified error estimates. Following the suggestion, in this study, rapid static measurements are collected at seven NGS points along Road 714 in Martin County, FL, USA and these measurements are processed using OPUS-RS. For comparison purposes, two other online services are also used and this is explained in the following sections. 2 GNSS Data Processing Using Online Services Currently there are seven online GNSS data processing services freely available to users. These are: OPUS (Online Positioning User Service), APPS (Automatic Precise Positioning Service), SCOUT (Scripps Coordinate Update Tool), CSRS-PPP (Canadian Spatial Reference System Precise Point Positioning Service), GAPS (GPS Analysis and Positioning Software), AUSPOS (Geoscience Australia Online GPS Processing Service) and Magic GNSS. In addition, UNB s PPPSC (Precise Point Positioning Software Centre) compares solutions from online PPP applications. Auto- Bernese provided by GPS Solutions requires payment for its services. Snay et al. (2011c) pointed out that multiple observation sessions at the same locations can be conducted to analyze OPUS-RS generated coordinates; however, two other services, namely, CSRS-PPP and GAPS are also used in this study for comparison purposes. The URLs of these services are given in Table 1. To benefit from the online GNSS data processing services, a single GNSS receiver and a single field crew are needed. Thus, these services significantly reduce equipment and personnel costs, pre-planning and logistics compared to conventional approaches. They also lessen the need to learn to use scientific GNSS processing packages (Ebner and Featherstone, 2008). With just one dualfrequency receiver, the observations taken are postprocessed based on differential methods using reference stations or precise point positioning using globally available precise satellite orbit and clock data. The online processing services deliver solutions to users without any cost and with unlimited access. At most the users are required to get an account. The basic requirements that the user needs to take advantage of these different services are almost the same in each case: access to the Internet and a valid address. Users send a RINEX (Receiver Independent Exchange Format) file or compressed RINEX file to the service and within a short period of time the estimated position of the GNSS system used to collect the data is sent back to the user in the form of s or report files (Ghoddousi-Fard and Dare, 2006). Note that OPUS also accepts most of receiver binary format files. In this process, disadvantages are that the quality of the solutions might be affected by the availability of precise orbit and clock data and the results that are returned to the user may be delayed by server processing speed and the internet connection. The online processing services provide the coordinates in a recognized datum. However, it should be mentioned that while producing these coordinates, all services depend on the quality of the data and the length of data span supplied to them by the user (Tsakiri, 2008). Generally, these services produce coordinates in the ITRF (International Terrestrial Reference Frame) or in their national geodetic reference frame. Transformation of coordinates is possible from one reference frame to another either using the published transformation parameters or using available online software from governmental organizations such as NGS. 3 Applications and Results For this project, rapid static survey is conducted by collecting short period of measurements at seven NGS points along Road 714 in Florida using triple frequency GNSS (Leica GX1230) receivers (see Figure 1). To bring the control into the project area, GAMIT (GPS Analysis at MIT) software is used. The remaining data are processed using LGO (Leica Geo Office) software. Figure 1: Seven NGS points along Road 714 in Martin County, FL, USA (image from Google).

3 Rapid static GNSS data processing using online services 125 Table 1: URLs of online GNSS data processing services used in this study. Service OPUS CSRS-PPP GAPS URL To take the observations, with the first session, three receivers are set up on tripods over points AJ8516, AJ8515 and AJ8514. The receiver over AJ8514 is kept running to serve as the base station. And then with the second session points AJ8514, AJ8524 and AJ8512 are occupied and with the third session remaining points AJ8512, AJ8511 and AJ8510 are occupied to complete the survey. To be able to solve for integer ambiguity, around 20 min data are collected at these stations. Since AJ8514 and AJ8512 served as base stations for the following observation sessions, each station has around an hour of data. NGS recommends that users utilize OPUS-RS for observing sessions between 15 minutes and 2 hours and utilize OPUS-S for observing sessions longer than 2 hours. It means that all the data collected at these seven points must be submitted to OPUS- RS. Day and time information for these measurements are given in the appendix. As a matter of fact, along the same line, first, a static survey is run to coordinate these seven sites. Two and a half hours of data are collected at these points yet since AJ8514 and AJ8512 served as base stations for the following observation sessions, each of these stations has around five and a half hours of data. In order to be able to coordinate the points, data from CORS station Okeechobee (OKCB) is downloaded in RINEX format and uploaded to GAMIT. Using this data, point AJ8512 is coordinated (baseline length between OKCB and AJ8512 is approximately 37 km) and then AJ8512 s coordinates are considered fixed and the other points are coordinated using LGO. Since static method yields the most precise results, for comparison purposes, static data files are submitted to OPUS-S and the differences between the LGO produced static results and received OPUS-S results are calculated and these differences ( φ, λ, h) are listed in Table 3. The same data are also submitted to CSRS-PPP and GAPS and the differences are included into Table 3. So, LGO produced results are considered the truth and the differences are tabulated in Table 3 for comparison and portrayed in Figure 3 for easier interpretation. CSRS-PPP gives the option to the users that they can either determine the coordinates in NAD83 (North American Datum of 1983) or other datums. OPUS-S provides the coordinates both NAD83 and IGS08 datums. GAPS provides the coordinates in IGS08 only. Therefore, in this study, the coordinates of the points are determined in IGS08 datum. To produce results in Table 3, for all seven points the same CORS stations i.e., OKCB, PBCH and CCV6 (distances between the project site and with each of the CORS used are given in Table 2) are used - these stations can be seen in Figure 2. The process is done on July 27, Figure 2: CORS Stations in southern Florida (image from Google). If we examine the static solution in Figure 3, the absolute differences in horizontal coordinates are in the order of centimeters. Except a few observations, as expected, ellipsoidal height differences are larger compared to horizontal coordinates differences. GAPS produced ellipsoidal heights are much worse than the OPUS-S and CSRS-PPP results. If we examine the rapid static solution; first, we need to clarify that since CSRS-PPP and GAPS do not have an option for rapid static processing, the same data sub-

4 126 M. Berber et al. Table 2: Distances between the project site and with each of the CORS used. CORS Station Distance (km) CORS Station Distance (km) OKCB 38.8 ZEFR PBCH 43.5 DLND CCV MTNT LAUD FMYR WACH BRTW NAPL MCD FLC Figure 3: Absolute coordinate differences between OPUS-S, OPUS- RS, CSRS-PPP and GAPS results and LGO produced static and rapid static results. mitted to OPUS-RS is submitted to these two services under static processing option and the results are tabulated in Table 3. As a consequence, although the same data are used, as it is indicated in Snay et al (2011a) since the processing strategy is different for OPUS-RS, different results are obtained. As evident from the overall RMS results, horizontal coordinates for OPUS-RS and CSRS-PPP are of the order of centimeters; nonetheless, we cannot say the same thing for GAPS results. Vertical coordinate differences are in the order of decimeters. OPUS-RS and CSRS-PPP are produced very similar results in terms of ellipsoidal heights, yet GAPS produced ellipsoidal heights are the worst among them. All in all, in terms of horizontal coordinates OPUS-RS results are the best. OPUS-RS processes always used nine stations and three of them, namely, OKCB, PBCH, LAUD were common stations; however, remaining six stations were selected from the nearby stations: WACH, RMND, NAPL, FLC5, ZEFR, DLND, MTNT, FMYR, BRTW, MCD5 (see Figure 2). These processes are done on July 28, Table 3: Comparison of OPUS-S, OPUS-RS, CSRS-PPP and GAPS results against LGO produced static and rapid static results. Static Solution Rapid Static Solution OPUS-S (m) OPUS-RS (m) CSRS-PPP (m) GAPS (m) OPUS-S (m) OPUS-RS (m) CSRS-PPP (m) GAPS (m) φ λ h φ λ h φ λ h φ λ h φ λ h φ λ h φ λ h φ λ h AJ AJ AJ AJ AJ AJ AJ RMS

5 Rapid static GNSS data processing using online services 127 If OPUS-RS results are compared against OPUS-S results, as can be seen in Figure 3, in terms of horizontal coordinate differences OPUS-RS results are under 3 cm (both for φ and λ) whereas differences for OPUS-S results are as large as 8.2 cm (see Table 3). Ellipsoidal height differences are almost the same for both OPUS-RS and OPUS- S. It means that in spite of the fact that a shorter span of data is used for OPUS-RS processes, better results are obtained. As mentioned in the introduction, OPUS-RS uses a data processing engine that differs from the one utilized in OPUS-S and these results endorse the power of this strategy. Although rapid static results for CSRS-PPP and GAPS produced comparable or slightly worse results compared to static horizontal coordinate differences, rapid static ellipsoidal height difference results are worse than their counterparts in the static solution. In order to be able to further investigate rapid static GNSS data processing using online services, measurements are taken only at two points i.e., AJ8516 and AJ8524. Again LGO produced rapid static results are considered the truth and the differences are tabulated in Tables 4 and 5 and portrayed in Figures 4 and 5. These processes are done on July 30, Day and time information for these measurements are given in the appendix. Table 4: Comparison of OPUS-RS results against LGO produced Rapid Static results at point AJ8516 (m). 8 min min h 2 min h 49 min Figure 4: OPUS-RS results for point AJ8516. Table 5: Comparison of OPUS-RS results against LGO produced Rapid Static results at point AJ8524 (m). 8 min min h 1 min h 58 min Figure 5: OPUS-RS results for point AJ8524. In fact, 2 min data were also collected at these points and submitted to OPUS-RS; nevertheless, the following message is returned from the server Your input dataset is too short. OPUS-RS will not attempt a solution with less than 7.2 minutes of data. Thus, with the following surveys as little as 8 min and as much as 1 h 58 min of data are collected at these two points because with OPUS-RS there is also a two-hour artificial restriction imposed by NGS. The variations among the horizontal coordinate differences are up to 3.5 cm in size, yet it is hard to discern a pattern. Again, there is no consistent pattern for the height differences and the variations are up to 7.2 cm. On the other hand, it can be stated with confidence that OPUS-RS generated coordinates from multiple observation sessions at the same location agree with each other. Moreover, Snay et al (2011a) stated that using OPUS-RS users may submit as little as 15 min of data; however, in this project, consistent results are obtained even with only 8 minutes of data. The reason for use of different time intervals in Tables 4 and 5 is that first of all we wanted to have as little data as possible to see how OPUS-RS reacts to short span of data. After finding out from the server that as little as 8 minutes of data can be used, 8 minutes of data at both points (AJ8516 and AJ8524) were collected. Generally 15 to 20 min of data is needed to solve the integer ambiguity; that is why we then collected 15 min of data at AJ8524.

6 128 M. Berber et al. Next, we planned to have around 30 min and 1 hour data, since we already had 41 min data at AJ8516, we used the this data in Table 4. And then, we collected, around 1 hour of data at both points. In the end, we also wanted to have close to 2 hour data at both points and this is the reason that we collected 1 h 49 min at AJ8516 and 1 h 58 min data at AJ8524. Although it is well known that horizontal coordinates are determined at a higher accuracy than ellipsoidal heights using GPS, in Figure 4, it appears that ellipsoidal height determinations of two observations are better than horizontal coordinates. This might happen because of changes in satellite geometry, multipath, GPS signal obstructions etc. Additionally, low GDOP values during short observation sessions probably strongly influence the precision of the coordinates computed with OPUS-RS. Finally, to investigate the accuracy of OPUS-RS coordinates over different time intervals, OPUS-RS results are compared against LGO produced Static results to create a form of an independent check. Table 6: Comparison of OPUS-RS results against LGO produced Static results at point AJ8516 (m). 8 min min h 2 min h 49 min Conclusions Online static data processing results are of the order of centimeters for horizontal coordinates and of the order of decimeters for vertical coordinates. GAPS produced ellipsoidal heights are much worse than the OPUS-S and CSRS- PPP results. In terms of horizontal coordinates OPUS-RS produced the best results. Most of ellipsoidal height differences produced by OPUS-RS are less than a decimeter. Despite the fact that shorter period of data are used for OPUS-RS processes, better results are obtained. This should be attributed to the processing strategy used by OPUS-RS. Further tests with rapid static measurements yield a consistent pattern neither with horizontal coordinates and nor with the vertical coordinates. These results might be ascribed to additional variables with OPUS-RS solutions such as satellite geometry, multipath, GPS signal obstructions and errors in the published CORS/IGS coordinates. In addition, low GDOP values during these short observation sessions probably strongly influence the precision of the coordinates computed with OPUS-RS. On the other hand, since the differences in horizontal coordinates agree with each other within 3.5 cm and vertical coordinates agree within 7.2 cm, it can be stated that OPUS-RS generated coordinates agree with each other with multiple observation sessions at the same locations. It is our recommendation that for better understanding of OPUS- RS results, continuous data for several days at study sites should be collected and then split into sessions. Table 7: Comparison of OPUS-RS results against LGO produced Static results at point AJ8524 (m). 8 min min h 1 min h 58 min As can be seen in Tables 6 and 7, this comparison yields up to 4.5 cm variations among horizontal coordinate differences and variations among vertical coordinate differences are up to 11.4 cm. Acknowledgement: We would like to thank the Division of Research at Florida Atlantic University for their support to our project. We also thank Mehmet Ali Cetin for his help with the field work. The third author wishes to acknowledge the support by the Scientific and Technological Research Council of Turkey for his research at Florida Atlantic University. References Ebner, R. and Featherstone, W. E., 2008, How well can online GPS PPP post-processing services be used to establish geodetic survey control networks?, Journal of Applied Geodesy, 2, Ghoddousi-Fard, R. and Dare, P., 2006, Online GPS processing services: An initial study, GPS Solutions, 10, Liu, J-H. and Shih, T-Y., 2007, A performance evaluation of the internet based static GPS computation services, Survey Review, 39, 304, Snay, R., Choi, K., Gerald, M., Schwarz, C., Soler, T. and Weston, N., 2011a, How precise is OPUS? Part I: Experimental results, The

7 Rapid static GNSS data processing using online services 129 American Surveyor, vol. 8, no. 5. Snay, R., Choi, K., Gerald, M., Schwarz, C., Soler, T. and Weston, N., 2011b, How precise is OPUS? Part II: Available precision estimates, The American Surveyor, vol. 8, no. 6. Snay, R., Choi, K., Gerald, M., Schwarz, C., Soler, T. and Weston, N., 2011c, How precise is OPUS? Part III: The rest of the story, The American Surveyor, vol. 8, no. 7. Soler, T., Michalak, P., Weston, N. D., Snay, R. A. and Foote, R. H., 2006, Accuracy of OPUS solutions for 1 to 4h observing sessions, GPS Solutions, 10, Tsakiri, M., 2008, GPS Processing Using Online Services, Journal of Surveying Engineering, vol. 134, no. 4, Table 2A: Day and time for Rapid Static measurements at point AJ8516 Date From To 8 min 01/07/ :49 15:57 41 min 09/04/2011 9:21 10:02 1 h 2 min 01/07/ :58 17:00 1 h 49 min 12/18/ :03 12:52 Appendix Table 1A: Day and time for Rapid Static measurements Date From To AJ /04/2011 9:21 10:02 AJ /04/2011 9:31 10:07 AJ /04/2011 9:42 10:48 AJ /04/ :34 10:49 AJ /04/ :24 11:47 AJ /04/ :18 11:37 AJ /04/ :10 11:31 Table 3A: Day and time for Rapid Static measurements at point AJ8524 Date From To 8 min 01/07/ :26 17:34 15 min 09/04/ :34 10:49 1 h 1 min 01/07/ :35 18:36 1 h 58 min 12/18/ :12 15:10