HIGH - ACCURACY GPS ANALYSIS ON THE INTERNET
Chris Rocken and James Johnson
GPS Solutions, Inc.
October, 1999
Abstract
We describe a new way to analyze GPS data using the World Wide Web at www.gps-solutions.com. Users can submit static single point or multi-day multi-station network GPS data via their favorite web browser, and receive the highest accuracy results by email just minutes later. Depending on data quality, length of files, and site/network location, the web-based results are typically good to 0.3-1 cm in the horizontal and 1-3 cm in the vertical. This paper describes the automated analysis technique, and presents results from a variety of test applications. The web-based analysis of 88 global stations, operated by the International GPS Service (IGS), agrees to 0.6 cm in the horizontal and 2.4 cm in the vertical with the published coordinates. We further compare the web-based analysis of all Continuously Operating Reference Stations (CORS) to the published coordinates. Results from a scientific expedition to Mount Everest, a Federal Geodetic Control Subcommittee (FGCS) test network, and a continuously monitored site on an offshore oil platform, are also presented. All of these results demonstrate that experts and novices alike can now use this tool to achieve top quality surveys even in the most remote locations for single points or networks with baselines ranging from 0.1-1000 km, without having to master complex GPS analysis software.
Background
Some applications of the GPS require the highest possible accuracy. State of the art GPS data analysis can achieve 2-5 mm horizontal and 1-3 cm vertical accuracy over baselines 1-1000 km. To obtain these kinds of results, multi-hour static data sets of high quality dual-frequency GPS data, highly sophisticated software, and a high level of data processing expertise are required.
Several software packages can be used for high accuracy analysis. For our web – based analysis we use one of the best-known packages, called the Bernese software (Rothacher et al., 1996). The Bernese software is developed at the University of Berne in Switzerland. It is very versatile and can be used for orbit determination, geodetic surveys, time transfer, ionospheric modeling and other applications. The Bernese software reliably achieves the highest accuracy because of its robust data editing and cycle slip correction algorithms, and its sophisticated models of the GPS observations and orbits. While the software has a menu-driven user interface it still requires significant time to master. We developed scripts that can run the Bernese software in a fully automated way and that can be started from our web interface. The user can now use this powerful software without having to learn how to install or operate it.
The automated GPS processing system uses a sophisticated set of scripts and programs to process the data submitted by the user. At the heart of the system is the Bernese Processing Engine (BPE). The BPE was developed in 1994 as a tool to automate the Bernese GPS processing programs. Since the BPE was first developed, it has been deployed at a large number of processing sites around the world including:
Feedback from the over 130 Bernese users at Universities, government research laboratories, geodetic surveys, and private companies contributes to the ever improving software. Another key component for achieving highest quality results is the use of high quality GPS orbits. We use the orbits that are computed by the International GPS Service (IGS). The IGS (Beutler et al., 1996) operates a global network of about 100 dual frequency GPS stations. These stations operate 24 hours / day and their data is transmitted to the IGS data centers. The data centers then provide the data to the 7 IGS data analysis centers. These analysis centers compute several high quality products that are of great utility to the scientific and survey communities. The most important IGS products are (1) high accuracy GPS orbits, (2) station coordinates and velocities, (3) Earth orientation parameters.
Figure 1. Global Distribution of IGS sites and the ITRF97 station velocities (figure courtesy T. Springer, U. of Bern)
The results from all IGS analysis centers are combined into an official IGS solution. The final IGS orbit is provided about 2 weeks after data collection. This is the highest quality GPS orbit available, and it has reached an accuracy of better than 5 cm. An orbit of this quality contributes less than 3 mm in error to a 1000-km baseline. In fact, the IGS orbit product is so good that orbit errors have become nearly negligible even for very long baselines.
For applications that cannot wait 2 weeks for the final orbit, the IGS produces a rapid orbit, which is made available the day after the data have been collected. These orbits are only slightly worse than 5 cm.
For real-time applications the IGS data centers produce a predicted GPS orbit. The quality of this orbit depends on how far into the future it has been predicted, but typically ranges between 10-50 cm. It should be noted that use of the predicted orbit requires orbit quality checks on the side of the user to detect satellite maneuvers that cannot be predicted.
Simultaneously with the orbits the IGS analysis centers also compute the coordinates of the IGS tracking sites. All of these solutions contribute to the definition of the International Terrestrial Reference Frame (ITRF). The International Earth Rotation Service (IERS), hosted by the Observatoire de Paris, is responsible for coordinating the realization of the ITRF (Boucher et al.1998, 1999). This reference frame is generated from geodetic solutions that are submitted to the IERS from global analysis centers using 4 different geodetic techniques: Very Long Baseline Interferometry (VLBI), Satellite Laser Ranging (SLR), GPS, and Doppler Orbitography and Radiopositioning Integrated by Satellite (DORIS). As position and coordinate solutions improve over time a new set of coordinates and velocities for the IGS stations in the ITRF reference frame are released. Approximately once a year the IERS publishes a new ITRF reference frame that is named after the year the data is collected. Usually there is a three year lag between the time data is collected and the date the reference frame is published.
On August 1st, 1999 the IGS data analysis centers switched from the ITRF96 frame to the ITRF97 frame. This change is realized by setting the a priori coordinates of the IGS tracking sites to the ITRF97 values and determining the GPS satellite orbits relative to these coordinates. The main differences between the ITRF96 and ITRF97 systems are a small rotation of the coordinate system at the 1 cm level, and the addition of several new IGS sites.
Users of the IGS orbits work in the ITRF coordinate frame because the GPS orbits are determined relative to ground tracking sites in that well-defined reference frame. Thus high accuracy GPS results, using IGS orbits, first have to be determined in the ITRF system. Various national geodetic survey institutes provide transformations between the ITRF coordinate system and other local or regional datums. The U.S. National Geodetic Survey (NGS) provides the parameters and software to carry out the transformation between ITRF96 (ITRF97) and the North American Datum NAD83. Our web-based analysis uses this transformation and provides coordinates and baselines in ITRF and NAD83. Other transformations can be added in the future.
The Earth orientation parameters computed by the IGS define the orientation of Earth in space. This information is needed for rotations between coordinate systems that are fixed to the Earth (i.e. the ITRF 97) and inertial coordinate systems which are not rotating with the Earth and in which satellite orbits are often described. Orbit determination software such as the Bernese software require Earth orientation parameters that are consistent with the satellite orbits to precisely carry out the required rotations.
High Accuracy Analysis Summary
A user can submit data from a single station or from an entire multi-station multi-day network survey to the GPS Solutions web site for analysis. Data can be submitted in Receiver Independent Exchange (Rinex) or receiver specific format, they can also be zipped or Unix "tar" files. The mechanics of how this submission is done are explained in appendix 1 of this paper. Two key parameters that the user has to provide correctly are the antenna height and the antenna type.
With the goal to position a GPS receiver anywhere in the world to better than 1 cm the horizontal and to better than 3 cm in the vertical the web based analysis system goes through the following main processing steps.
We estimate tropospheric delay parameters every three hours for each station in a network (no tropospheric data such as surface pressure and temperature are required) to obtain accurate and consistent survey results in varying weather conditions. It is important to apply corrections for phase change effects of the GPS antennas, especially when the user antennas are mixed or different from the antennas used at the IGS reference sites. Without applying these corrections errors as large as 10 cm in the vertical can be incurred. We use phase center correction patterns for a large number of antennas that are made available by the NGS at http://www.grdl.noaa.gov/GRD/GPS/Projects/ANTCAL.
Our processing resolves integer cycle ambiguities using a widelane strategy. Ambiguity resolution is generally more successful for longer (> 3 hours) data sets and it is most important for achieving good east-west coordinates.
Web Site Processing Test Results
Extensive tests were conducted with the web site. We tested an IGS data set, a FGCS test data set, a network from a Mt. Everest expedition, all CORS sites, and we estimated daily positions of an offshore oil platform for 100 days. The purpose of these tests was to examine how robust and accurate the web based analysis strategy performs for a variety of regions, applications, and data sets. These results demonstrate the quality of solutions that can now be achieved by any user of the GPS system - experienced or inexperienced who submits high quality field data.
IGS Station coordinates in ITRF97
The first set of tests involves the global IGS stations. The purpose of these tests was to determine the performance of the web-based positioning for data from anywhere in the world. For this purpose we submitted one IGS station at a time as "user" station. The system then automatically retrieves the three nearest IGS stations and positions the "user" site. This was done for all available IGS tracking stations of day 075 1997.
Figure 2: The difference between the solutions obtained from the automated processing and the published ITRF97 coordinates.
If we ignore (3-sigma outlier test) the 6 worst sites, the remaining 88 points scatter with a standard deviation of (lat=0.6, lon=0.7, ht=2.4 cm) about the published ITRF97 coordinates. A part of this error is due to the fact that we computed the solutions for the test data using IGS orbits in the ITRF96 frame.
Some of the sites with high coordinate differences have known problems such as poor data quality due to old receivers. Other sites are relatively new, and their velocities are not yet well defined. Several of the sites that show large differences are located in areas where the distance to the nearest IGS sites is large, as is the case for African and mid oceanic sites. The automated processing avoids using problematic IGS sites as reference points (those where our analysis shows large differences to the published coordinates) unless there are no other reference sites within 1000 km.
Tests with the CORS Network
The web-based data analysis system was tested extensively with all available CORS data from three days (110,111,112) in 1999.
Over 100 CORS sites were positioned and our results were compared to the results published by the National Geodetic Survey (NGS) at "ftp://www.ngs.noaa.gov/cors/ coord/coord_96/". To investigate the effect of the time length of the submitted files we submitted 24, 12, 6, 3, 1-hour long CORS files for each station and each test day.
For example, the comparisons for 106 CORS sites from the 12-hour run are shown in Figure 3. It can be seen that our 12-hour solutions scatter around those published by the NGS with a standard deviation of 7.6 mm in latitude, 9.1 mm in longitude and 18.6 mm in height.
Figure 3. Comparison of automated processing of 3-day 12-hour solutions and the published CORS coordinates.
We found a height bias between our solutions and those published by the NGS of about 11 cm in the vertical for all Ashtech L1-L2 Geodetic antenna sites. No effort was made to estimate this bias precisely. For comparison with NGS we had to lower all our Ashtech solutions by 11 cm. We assume that this bias is due to differences in the antenna phase centers used by us and those used by the NGS. No bias correction was required for any of the other CORS antenna types. For the comparison between our solutions and the NGS solutions we assumed that the antenna heights in the CORS Rinex files were correct and that the published station coordinates were for the antenna reference points.
It should be noted that the differences in Figure 3 are the combined effect of errors in our solution, errors due to us assuming the wrong antenna type for some of the CORS sites, and uncertainties of the published NGS coordinates and antenna heights. We have contacted NGS to obtain the information needed to correctly interpret the antenna naming convention in the CORS Rinex file headers.
For several CORS sites (i.e. ERLA, FMC1, MOR1, YBHB) our results show large (>10 cm) differences compared to the NGS results. Based on the several-mm day-to-day repeatability of our results for almost all sites, we again believe that the reason for this large discrepancy is an error in the comparison antenna heights or the antenna pattern that we used.
Figure 4 This figure shows the lat/lon/hgt repeatability and the accuracy (agreement of our solution with the published CORS solution) for processed data sets of varying file length (24, 12, 6, and 3 hours).
Figure 4 shows the accuracy (comparison with NGS results) and the 3-day repeatability (precision) for the latitude, longitude and height for different file lengths. We excluded 3-sigma outliers from the comparison and all statistics are based on comparison with ~100 CORS sites. For longer files the solutions are better than for shorter files.
Repeatability in the horizontal components is 2-3 mm for the 24- and 12-hour solutions. This increases (especially in longitude) to 5 and then 15 mm for the 6- and 3 hour solutions. The reason for this large increase in longitude scatter is that we are not able to resolve carrier phase ambiguities for shorter data sets because of the long (typically 300 km and often much longer) baselines from the CORS sites to the nearest IGS stations. 1-hour results were worse than 3-hour results by over a factor of 2. Based on these results a user of the on-line service may decide for how long a station should be occupied.
Based on our test results we suggest that at a minimum one reference point of a user network should be occupied for at least 3 hours. For very remote sites like Guam at least 12 hours are recommended. Longer occupation times will ensure more accurate absolute positioning of the entire network. The benefits of station occupation beyond 12 hours are primarily in the vertical and for solutions in North America rather small. Only surveys that demand the highest possible accuracy or are very remote need to collect data for longer than 12 hours.
Note that these recommendations apply to the reference station of a user network and thus the network’s absolute position in ITRF97 or NAD83. Station occupation within the network may be significantly shorter as the analysis software can resolve integer ambiguities over shorter baselines (< 100 km) for shorter data sets.
In summary, agreement between our solutions and those published for the CORS network by the NGS is better than 1 cm in the horizontal and 2 cm in the vertical for occupation times of 12 hours or more. These results can probably be improved when the correct antenna phase corrections for all CORS sites are used.
Expedition to Mount Everest
An interesting opportunity to test the web site was presented by the American Mount Everest Expedition (AMEE) of 1998. This expedition collected several GPS data sets in the area of Mount Everest, including 1 unique observation on top of Everest.
Figure 5 GPS receiver and antenna near the top of the world. (photo by Wally Berg, courtesy of AMEE)
Data from the expedition were submitted to the GPS Solutions web site for analysis. While we cannot show any results for the height of Everest (those will be published elsewhere) we show the scatter of 3-day solutions for the Everest base-camp from the AMEE in Figure 6. Also shown in the right panel of Figure 6 is the map, with the IGS fiducial sites that were used for the three-day analysis. Different fiducial sites had to be used on different days because not all fiducial data were available for all three days.
Figure 6. 3-Day repeatability of the position of the Everest Base Camp at a lofty height of 7862.01 m (left panel) and the position of Base relative to the IGS sites that were used in this analysis (GPS data for this analysis courtesy of AMEE).
These results demonstrate that surveyors, scientists, and explorers can take GPS measurements to some of the most remote location on the planet, and conveniently obtain highly precise positions in the ITRF global reference frame.
FGCS Test Network Analysis
The following test shall demonstrate that the web-based analysis tool is not only useful for accurate positioning in remote places and with long baselines, but that it can also be used for the solution of small multi-day GPS networks, involving baselines from several 100 meters to several 100 km. Static data file lengths for this test ranged from 3 – 7 hours.
For this test we obtained a FGCS test data set collected by Trimble with the 4700 GPS system November 15th through the 20th, 1998 in and around Gaithersburg, Maryland. Trimble processed the original data set with the GPSurvey post processing software and published the results in the report " Federal Geodetic Control Subcommittee Evaluation of the Trimble 4700 GPS System And GPSurvey 2.30 Software November 20, 1998 Gaithersburg, Maryland".
The results presented here are not intended to test the receiver or compare the web-based analysis to the GPSurvey results. Rather we submitted the entire multi-day data set all at once to compare the automated results against the ground truth.
For the analysis of the Trimble FGCS test data set we submitted all three days during which static data had been collected. Each day of these data was solved and then a network for the three days was formed. Our solutions were computed in the ITRF coordinate frame relative to IGS stations GODE 40451M123 and WES2 40440S020. The (somewhat shortened) automatically generated report file from this run is shown as an output example in Appendix 2.
For comparison we took the ground truth from the Trimble report in NAD83 coordinates and performed a 7-parameter Helmert transformation of our solutions onto the ground truth data. The Helmert transformation accounts for differences in coordinate systems. Residuals of this transformation are the coordinate differences between our solution and the ground truth. Comparisons are shown in Table 1.
HELMERT TRANSFORMATION ---------------------- LOCAL GEODETIC DATUM: WGS - 84 RESIDUALS IN LOCAL SYSTEM (NORTH, EAST, UP) --------------------------------------------------------------------- | NUM | NAME | FLG | RESIDUALS IN MILLIMETERS | | --------------------------------------------------------------------- | | | | North East Up | | | 1 | ATHY | M M | 1.1 3.9 18.4 | | | 2 | NBS0 | M M | -6.4 -6.0 -6.3 | | | 3 | NBS1 | M M | 3.5 -8.1 6.2 | | | 4 | NBS3 | M M | 1.9 2.5 -.8 | | | 5 | NBS5 | M M | -1.2 -.8 -7.3 | | | 6 | ORM1 | M M | -4.4 -2.2 -4.3 | | | 7 | ASTW | M M | 1.0 -2.3 -5.0 | | | 8 | GORF | M M | -3.3 5.9 2.5 | | | 9 | MDPT | M M | 1.5 1.7 2.7 | | | 10 | NC25 | M M | -2.7 7.1 7.2 | | | 12 | SCOL | M M | 8.9 -1.8 -13.2 | | | | | | | | --------------------------------------------------------------------- | | RMS / COMPONENT | | 4.2 4.7 8.7 | | ---------------------------------------------------------------------
Table 1 Differences between the automated solution and ground truth for the Trimble FGCS test network.
It can be seen that the agreement for the entire network is 4-5 mm in the horizontal and 8.7 mm in the vertical. It is not clear yet why the results for ATHY and SCOL are significantly worse than the rest of the network.
We also compared all baselines from our solution with the ground truth. Comparison results are shown in Figure 7.
Figure 7 Comparison of north, east, up components of FGCS test baselines. Automated solution vs. ground truth.
The baseline solutions indicate that the outliers in our solutions are for some of the shorter baselines. It should be noted that all our solutions, even for the shortest baselines, process the ionospheric free linear combination and estimate tropospheric corrections. For very short baselines of 2-3 km or less it may be possible to obtain better results when just L1 solutions are computed and no troposphere is estimated. We also used the antenna phase pattern of the Trimble Geodetic L1/L2 antenna (with ground plane) instead of the correct pattern for the Trimble Microcenter antenna.
While there may be ways to yet improve upon the results from the automated processing this example shows that very good results can be achieved for diverse data sets and multi-station multi-day networks.
Permanent Position Monitoring Example
As a final example we show an illustration of how the automated processing system may be used for permanent station monitoring. This is an application for deformation monitoring. We processed 100 days of data from the IGS site Harvest. The Harvest offshore oil platform is located off the coast of California, just to the north of Santa Barbara. The site uses an AOA TurboRogue receiver equipped with a choke ring antenna. It should be noted that the site is located in a high multipath environment. The data was downloaded to the processing server and then submitted to the standard automated processing system. The 100-day solutions are shown in Figure 8.
Figure 8 Station coordinates for the automated solution of Harvest for a test period of 100 days. The RMS of the above traces are 2.98mm North-South (N-S trace) and 3.12mm for East-West (E-W), and 6.17 mm for the height (U-D).
The coordinates in Figure 8 were computed relative to IGS sites in the western U. S. and Canada.
This last example shows the kind of results that a user could obtain from permanently monitoring a structure over time. The N-S and E-W deformation is due to the tectonic motion of the site in the ITRF frame. Longer monitoring time is needed to validate if the platform is slowly subsiding. Monitoring data can be submitted by the user on a daily or weekly basis or they can be fetched electronically at pre-determined times.
Summary
We have introduced a new option to users of the GPS system for high-accuracy analysis of their GPS stations and networks. This tool is very easy to use and can obtain the highest accuracy GPS surveying results. Experts can use it to verify their results or simply to save time. Less experienced users of the GPS can now take on new and challenging projects. Extensive tests with a variety of global, regional, and local data have validated the web-based analysis for many potential uses. Some applications that can make use of this new service are:
This service is relatively new and we are planning to add some additional features in the near future including:
Interested prospective users of this service are encouraged to provide suggestions for changes and new options to make this web-based GPS processing tool more helpful in the future.
References
Beutler, G., I.I. Mueller, R.E. Neilan , The International GPS Service for Geodynamics (IGS): The Story, in International Assoc. of Geodesy Symposia, No. 115, GPS Trends in Precise Terrestrial, Airborne and Spaceborne Applications, pp 3-13, Springer Verlag, ISBN 3-540-60872-6, 1996.
Boucher C., Z.. Altamimi, P. Sillard, Results and analysis of the ITRF96. IERS Tech. Note 24, Observatoire de Paris, 1998.
Boucher C., Z. Altamimi, P. Sillard, The 1997 International Terrestrial Reference Frame (ITRF97). IERS Technical Note 27, Observatoire de Paris, 1999.
Rothacher M., G. Beutler, E. Brockmann, S. Fankhauser, W. Gurtner, J. Johnson, L. Mervart, S. Schaer, T. A. Springer, R. Weber, The Bernese GPS Software Version 4.0, Astronomical Institute, University of Berne, Switzerland, 1996.
Appendix 1
Data submission and processing example
To show how the automated GPS processing system works we will go through a step by step example of how to process data from a single GPS station. For our example we will use data from a GPS site in Platteville, Colorado. This data is available from the IGS data centers, for example: ftp://lox.ucsd.edu/pub/rinex/1999/070/pltc0700.99o.Z
Note that some browsers may rename the file pltc0700.99o.Z to pltc0700_99o.z. This is ok - the processing system will still recognize the file.
The first step is to get the data file onto the local computer. By local, we mean to the user’s computer running the web browser (i.e. Netscape or Explorer). The GPS-Solutions computer, which runs the automated web site, is called the server computer. When the data file is on the local computer, connect to the GPS-Solutions web site and click the link to the "on-line processing" system. Figure A1 shows the login screen. Here the username and password are entered or obtained.
Figure A1: First the user must login to the automated GPS Processing system. New users can create an account on-line using the "Register For Account" option. The new password will be sent via email within minutes.
When logged into the system, users are given the main menu as shown in figure A2. The main menu is fairly limited because the automated processing site minimizes the required input. The five options are: (1) to upload data to the server, (2) show the data files that have been uploaded, (3) delete files that have been uploaded before, (4) begin GPS data processing, and (5) to log off .
Figure A2: The automated web site main menu.
First GPS data must be uploaded to the server. After selecting the "Upload data to server" option screens are presented as shown in Figure A3. Each user is given a private directory on the processing server where data can be uploaded. This allows the user to upload a number of files over a period of time and then process all of these files together. The user can list the files that have been uploaded and delete them when finished. The upload screen is shown in Figure A3. Users can upload up to ten files at a time. Additional upload fields can be created by selecting the "More Upload Fields" button.
Users can either enter the full path to the file on their computer, or they can use the browse button to look for the file. Figure A4 shows the upload dialog box that is used by the UNIX version of Netscape. The exact look and feel of the upload dialog will be slightly different on different platforms, but they all allow the user to browse the local disk and select a file.
Figure A3: The upload menu. Here we can upload as many as ten files at a time.
When the user selects the "Upload Files" button, the data is sent over the Internet to the server. The speed of this download depends the speed of the user’s connection to the Internet as well as the speed of the connection between the user’s Internet service provider (ISP) and the server. The GPS Solutions server can handle downloads as fast as 500 Kbytes per second, but this is usually not achievable because of congestion somewhere between the user’s ISP and the server. If the user is connected to the Internet by phone modem, then the upload speed will be limited by the modem speed. A typical modem can upload data at a rate of about 3 Kbytes per second. To upload a 1.5 Mbyte file at 3 Kbytes per second takes a little over 8 minutes. Most of the web browsers such as Netscape’s Navigator and Microsoft’s Internet Explorer will wait until all data has been uploaded to the server before presenting the next screen. For users with slow Internet connections, it is better to upload one file at a time, else the web browser will stay on the upload data screen for an extended period. For slow connection it is also recommended to compress or zip the data file(s). Unfortunately the current web browsers do not offer any progress meters to show how much data has been uploaded to the server.
Figure A4: The upload dialog box on the UNIX version of Netscape. Here file pltc0700_99o.Z was selected to be uploaded. When the "OK" button is pressed, the upload dialog shows the full path of the selected file (/home/gpsauto/pltc0700_99o.Z).
When the upload is finished, the screen in Figure A5 appears. This screen shows that one data file of approximately 1 Mbyte was received by the server. It also shows that the server recognized that the file was compressed with the UNIX compress program and automatically uncompressed the file. The system can accept data as either uncompressed files, UNIX compressed files (these files have a .Z extension), UNIX GNU-Zip files (.gz extensions), PC ZIP files (.zip extensions) and UNIX tar files (.tar extensions).
Figure A5: The screen that is shown after file(s) have been uploaded.
Figure A6: The screen listing file(s) that have been uploaded to the screen.
The server will first check if the file is compressed and uncompress it if required. Then the server checks for archive files such as PC ZIP files and UNIX tar files. If it is an archive file (i.e. zip or tar), then all files will be extracted from the archive and placed in the user’s personal directory on the server.
Selecting the "Continue" button in Figure A5 returns to the automated processing main menu. Users can continue to upload data to process multiple files at the same time. However, since we only have one file to process in this example, we continue. Before processing the data, one can check the data on the server with the "List data on server" option. Figure A6 shows the listing that is produced. In this example it shows us one file, it is an ASCII Text file, and its size is 3.5 Mbytes. The file is larger than the original 1 Mbyte file that was uploaded because it was uncompressed. If the data file were a raw data file from a receiver and not Rinex, then the data type would show up as "data" rather than "ASCII Text".
To start processing the "Process Data" option is selected and the screen in Figure A7 appears. All the files that have been uploaded to the server will be listed here. The system attempts to read the data files and fills out as many input fields as it can. For each file the user must specify:
Figure A7: The Process Data screen. All uploaded files will be displayed and processed by default. The Use toggle button on the left must be unselected if the file shall not be included in the processing run. The server will try to fill out as many of the input data fields as it can. In the above examples, the values shown were automatically determined by the server.
When the "Continue" button is selected, the server checks that all the data fields have been filled out for the files that are to be processed. If the server detects an error, the user will be informed and asked to enter any missing data.
Figure A8: The screen showing that the file has been submitted for processing.
Once the server determines that it has all the information required for processing, the screen in Figure A8 appears. The job will be placed on a wait queue to be processed. As soon as a processor is available, the user will be sent an email informing them that their job has started and processing will begin. Currently it takes about 10 min to process a single station for a single day.
In the event that there is a problem processing the job, the user will be sent an email informing them that there was a problem that requires operator intervention. Email will also be sent to the operators of the web site to let them know that there is a special case that needs attention. In the case that the GPS data is can be processed, the results will then be provided with a delay of 1-2 days.
Appendix 2
Shortened Example Output (without Network Map)
--------------------------------------------------------------------------------
AUTOMATIC SOLUTION REPORT GENERATED Mon Aug 9 16:49:04 1999 Mountain Time
GPS Solutions, Inc. (www.gps-solutions.com, support@gps-solutions.com)
OUTPUT VERSION 1.0
--------------------------------------------------------------------------------
A total of 14 stations were analyzed for 3 days.
Stations Processed:
ABRV STATION NAME TYPE CONSTRAINT NORTH/EAST/UP METERS
---- ---------------- ---- -------------------------------
ASTW ASTW USER FREE
ATHY ATHY USER FREE
GODE GODE 40451M123 IGS 0.0001 0.0001 0.0001
GORF GORF USER FREE
MDPT MDPT USER FREE
NBS0 NBS0 USER FREE
NBS1 NBS1 USER FREE
NBS3 NBS3 USER FREE
NBS5 NBS5 USER FREE
NC25 NC25 USER FREE
NRC1 NRC1 40114M001 IGS FREE
ORM1 ORM1 USER FREE
SCOL SCOL USER FREE
WES2 WES2 40440S020 IGS 0.0001 0.0001 0.0001
Summary of Rinex files processed. The values below reflect the results
of Rinex Quality Checking. An elevation cutoff of 15 degrees was used
when scanning the data files. The following information is shown for
each Rinex file:
STA4........... 4-Character abbreviation of site.
START OF Rinex. Time of first epoch in the Rinex file.
HOURS.......... Number of hours of data contained in the Rinex file.
SMP............ Sampling rate of data in the Rinex file.
PERC........... Percent of data collected based on a 15 deg elevation mask.
MP1............ Average of multipath RMS values for all satellites in meters.
MP2............ Average of multipath RMS values for all satellites in meters.
O/SLP.......... Total number of observations divided by the number of slips
detected.
ANTENNA........ GPS Antenna Type.
A HGT [M]...... Direct height of the antenna above the marker in meters.
This is measured to the Antenna Reference Point (ARP).
By Session:
Session 3190
STA4 START OF Rinex HOURS SMP PERC MP1 MP2 O/SLP ANTENNA A HGT [M]
---- -------------- ----- --- ---- ---- ---- ----- ----------------- ---------
ATHY 98-11-15 12:56 7.16 15 48% 0.51 0.63 239 TR GEOD L1/L2 GP 2.0000
GODE 98-11-15 00:00 23.99 30 94% 0.37 0.73 674 DORNE MARGOLIN T .0614
NBS0 98-11-15 12:55 7.09 15 42% 0.51 0.55 154 TR GEOD L1/L2 GP 2.0000
NBS1 98-11-15 12:54 7.12 15 51% 0.34 0.39 513 TR GEOD L1/L2 GP 2.0000
NBS3 98-11-15 13:49 6.19 15 43% 0.24 0.25 380 TR GEOD L1/L2 GP 2.0000
NBS5 98-11-15 12:51 7.20 15 48% 0.23 0.23 488 TR GEOD L1/L2 GP 2.0000
NRC1 98-11-15 00:00 23.99 30 100% 0.32 0.61 9648 DORNE MARGOLIN T .0000
ORM1 98-11-15 12:31 7.00 15 47% 0.28 0.31 454 TR GEOD L1/L2 GP 2.0000
WES2 98-11-15 00:00 23.99 30 94% 0.40 0.51 2030 DORNE MARGOLIN T .0000
Session 3200
STA4 START OF Rinex HOURS SMP PERC MP1 MP2 O/SLP ANTENNA A HGT [M]
---- -------------- ----- --- ---- ---- ---- ----- ----------------- ---------
ASTW 98-11-16 12:58 3.08 15 100% 0.23 0.24 5623 TR GEOD L1/L2 GP 1.8910
GODE 98-11-16 00:00 23.99 30 94% 0.37 0.67 524 DORNE MARGOLIN T .0614
GORF 98-11-16 12:55 3.10 15 100% 0.25 0.27 5590 TR GEOD L1/L2 GP 1.5000
MDPT 98-11-16 12:52 3.16 15 100% 0.24 0.28 5721 TR GEOD L1/L2 GP 2.0000
NBS5 98-11-16 12:55 3.16 15 100% 0.22 0.23 5718 TR GEOD L1/L2 GP 2.0000
NC25 98-11-16 12:51 3.19 15 100% 0.29 0.33 5740 TR GEOD L1/L2 GP 1.8000
NRC1 98-11-16 00:00 23.99 30 100% 0.33 0.61 4824 DORNE MARGOLIN T .0000
SCOL 98-11-16 12:45 3.34 15 100% 0.35 0.35 5962 TR GEOD L1/L2 GP 2.0000
WES2 98-11-16 00:00 23.99 30 94% 0.41 0.51 2281 DORNE MARGOLIN T .0000
Session 3210
STA4 START OF Rinex HOURS SMP PERC MP1 MP2 O/SLP ANTENNA A HGT [M]
---- -------------- ----- --- ---- ---- ---- ----- ----------------- ---------
ASTW 98-11-17 12:51 3.19 15 100% 0.23 0.24 5818 TR GEOD L1/L2 GP 1.8910
ATHY 98-11-17 12:22 3.73 15 97% 0.61 0.68 1052 TR GEOD L1/L2 GP 2.0000
GODE 98-11-17 00:00 23.99 30 95% 0.36 0.70 574 DORNE MARGOLIN T .0614
GORF 98-11-17 12:53 3.12 15 100% 0.24 0.27 5679 TR GEOD L1/L2 GP 1.5000
NBS5 98-11-17 12:46 3.25 15 100% 0.22 0.23 5869 TR GEOD L1/L2 GP 2.0000
NC25 98-11-17 12:50 3.21 15 100% 0.29 0.35 5839 TR GEOD L1/L2 GP 1.8000
NRC1 98-11-17 00:00 23.99 30 100% 0.34 0.62 9652 DORNE MARGOLIN T .0000
ORM1 98-11-17 13:33 2.45 15 99% 0.28 0.31 4650 TR GEOD L1/L2 GP 2.0000
WES2 98-11-17 00:00 23.99 30 94% 0.44 0.52 2605 DORNE MARGOLIN T .0000
By Station:
Station: ASTW
STA4 START OF Rinex HOURS SMP PERC MP1 MP2 O/SLP ANTENNA A HGT [M]
---- -------------- ----- --- ---- ---- ---- ----- ----------------- ---------
3200 98-11-16 12:58 3.08 15 100% 0.23 0.24 5623 TR GEOD L1/L2 GP 1.8910
3210 98-11-17 12:51 3.19 15 100% 0.23 0.24 5818 TR GEOD L1/L2 GP 1.8910
Station: ATHY
STA4 START OF Rinex HOURS SMP PERC MP1 MP2 O/SLP ANTENNA A HGT [M]
---- -------------- ----- --- ---- ---- ---- ----- ----------------- ---------
3190 98-11-15 12:56 7.16 15 48% 0.51 0.63 239 TR GEOD L1/L2 GP 2.0000
3210 98-11-17 12:22 3.73 15 97% 0.61 0.68 1052 TR GEOD L1/L2 GP 2.0000
Station: GODE
STA4 START OF Rinex HOURS SMP PERC MP1 MP2 O/SLP ANTENNA A HGT [M]
---- -------------- ----- --- ---- ---- ---- ----- ----------------- ---------
3190 98-11-15 00:00 23.99 30 94% 0.37 0.73 674 DORNE MARGOLIN T .0614
3200 98-11-16 00:00 23.99 30 94% 0.37 0.67 524 DORNE MARGOLIN T .0614
3210 98-11-17 00:00 23.99 30 95% 0.36 0.70 574 DORNE MARGOLIN T .0614
Station: GORF
STA4 START OF Rinex HOURS SMP PERC MP1 MP2 O/SLP ANTENNA A HGT [M]
---- -------------- ----- --- ---- ---- ---- ----- ----------------- ---------
3200 98-11-16 12:55 3.10 15 100% 0.25 0.27 5590 TR GEOD L1/L2 GP 1.5000
3210 98-11-17 12:53 3.12 15 100% 0.24 0.27 5679 TR GEOD L1/L2 GP 1.5000
Station: MDPT
STA4 START OF Rinex HOURS SMP PERC MP1 MP2 O/SLP ANTENNA A HGT [M]
---- -------------- ----- --- ---- ---- ---- ----- ----------------- ---------
3200 98-11-16 12:52 3.16 15 100% 0.24 0.28 5721 TR GEOD L1/L2 GP 2.0000
.
(some of the example output was deleted here)
.
.
GENERAL SOLUTION INFORMATION:
ORBITS USED : IGS FINAL
TOTAL NUMBER OF PARAMETERS : 793
NUMBER OF OBSERVATIONS : 161430
SIGMA OF SINGLE DIFFERENCE OBSERVATIONS: 0.0026
All data was processed on a daily basis. Baselines were formed in order
to produce the highest possible number of observations. All baselines
were then analyzed together to estimate the coordinates and site specific
tropospheric delays. This method of analysis correctly takes into
account the mathematical correlation between all observations and the
resulting coordinates represent a best fit to all possible baselines.
The normal equations for each daily solution were then combined to give
a statistically correct estimation of each coordinate. The sigma of the
single difference observations is in units of meters and is an indication
of how well the observations were modeled. This value should be less
than 0.0050 meters.
COORDINATE SOLUTIONS:
STATION COMPONENT SOLUTIONS NEW-APR RMS 2D ERROR ELLIPSE
---------------- --------- ----------------- ------- ------- ----------------
ASTW X 1097047.9503 0.0049 0.0003 0.0002 135.7000
Y -4897244.1956 -0.0053 0.0011 0.0003
Z 3923114.8004 0.0061 0.0009
HEIGHT 34.4102 0.0087 0.0014
LATITUDE 38 12 7.422913 0.0009 0.0002
LONGITUDE -77 22 24.370295 0.0036 0.0002
ATHY X 1090497.2870 -0.0104 0.0003 0.0002 137.4000
Y -4834731.6795 0.0006 0.0011 0.0003
Z 4001278.5747 -0.0097 0.0009
HEIGHT 102.8921 -0.0083 0.0014
LATITUDE 39 06 11.497896 -0.0057 0.0002
LONGITUDE -77 17 21.696610 -0.0100 0.0002
GODE 40451M123 X 1130773.8418 0.0032 0.0001 0.0001 128.3000
Y -4831253.5652 -0.0003 0.0001 0.0001
Z 3994200.3948 -0.0010 0.0001
HEIGHT 14.4974 0.0002 0.0001
LATITUDE 39 01 18.218311 -0.0015 0.0001
LONGITUDE -76 49 36.584608 0.0030 0.0001
GORF X 1130731.3070 0.0031 0.0003 0.0002 137.7000
Y -4831322.5924 0.0005 0.0010 0.0003
Z 3994136.4765 0.0011 0.0009
HEIGHT 18.9399 0.0009 0.0013
LATITUDE 39 01 15.433663 0.0007 0.0002
LONGITUDE -76 49 38.960182 0.0032 0.0002
MDPT X 1106559.8386 0.0063 0.0004 0.0002 131.1000
Y -4882972.9222 -0.0188 0.0013 0.0003
Z 3938005.0076 0.0129 0.0010
HEIGHT -26.1170 0.0235 0.0017
LATITUDE 38 22 24.224814 -0.0021 0.0003
LONGITUDE -77 13 53.465043 0.0020 0.0003
NBS0 X 1096389.6840 -0.0174 0.0004 0.0002 133.5000
Y -4831092.0258 0.0140 0.0015 0.0003
Z 4004050.9302 0.0018 0.0013
HEIGHT 106.7239 -0.0125 0.0020
LATITUDE 39 08 7.270020 0.0125 0.0003
LONGITUDE -77 12 49.026491 -0.0139 0.0003
.
(some of the example output was deleted here)
.
.
These coordinates are in the ITRF96 reference frame at epoch of the
first day processed. The units for X, Y, Z, new-apriori and formal
errors is meters. The geodetic coordinates are computed using the WGS-84
Ellipsoid. The new-apriori values have significance for the IGS sites --
they indicate the difference between the estimated coordinate and the
initial ITRF coordinate that was used. The new-apriori values are not
critical for other sites since they are just the difference between the
final solution and our first estimate of them. The formal errors tend
to be underestimated. The rule of thumb is to multiply them by a factor
of 10 to get a realistic idea of the coordinate error.
Coordinate Repeatabilities:
Another factor to look at for coordinates is the day to day repeatability.
Below is a table that shows the North, East and Up baseline day to day
repeatabilities for the days that were estimated. The values are in mm
and are the difference between the value for a specific session and the
average value.
STATION RMS SES: 3190 3200 3210
---------------- ---------- --- ------ ------ ------
WES2 40440S020 Nrms 0.6 RES: -0.8 -0.4 -0.2
Erms 1.3 RES: 1.3 1.1 0.8
Urms 0.3 RES: 0.2 0.3 0.2
SCOL Nrms 0.0 RES: -1.5 0.0
Erms 0.0 RES: -1.7 0.0
Urms 0.0 RES: -3.8 0.0
NBS0 Nrms 0.0 RES: 1.9 0.0 0.0
Erms 0.0 RES: 1.5 0.0 0.0
Urms 0.0 RES: -6.0 0.0 0.0
NRC1 40114M001 Nrms 0.9 RES: 0.7 0.5 0.9
Erms 1.0 RES: -1.1 0.9 0.1
Urms 2.1 RES: 0.3 -0.5 -2.9
NBS1 Nrms 0.0 RES: 2.1 0.0 0.0
Erms 0.0 RES: 1.7 0.0 0.0
Urms 0.0 RES: -5.3 0.0 0.0
MDPT Nrms 0.0 RES: -1.6 0.0
Erms 0.0 RES: -1.7 0.0
Urms 0.0 RES: -3.7 0.0
NBS3 Nrms 0.0 RES: 2.0 0.0 0.0
Erms 0.0 RES: 1.6 0.0 0.0
Urms 0.0 RES: -6.1 0.0 0.0
ATHY Nrms 3.4 RES: -1.3 3.2
Erms 1.4 RES: -0.8 1.1
Urms 25.3 RES: -15.9 19.6
NC25 Nrms 1.3 RES: -1.3 -0.3
Erms 1.8 RES: -1.7 -0.5
Urms 7.5 RES: -0.5 7.5
.
(some of the example output was deleted here)
.
===============================================================================
======================== POSITION INFORMATION ==============================
===============================================================================
% NBS5 --------- ITRF96 1998.871233
WIN 1998 319.00000000 51132.0000000 1998 321.99964120 51134.9996412 1
X Y Z
CRD 1096351.1277 -4831475.5438 4003597.9591
CRD rms [mm] 0.3 0.9 0.8
LAT LON HEIGHT (WGS84)
GEO_1 39 7 48.39700 - 77 12 54.12572 104.3397
GEO 39.1301102786 -77.2150349221 104.3397
GEO rms [mm] 1.1 0.5 0.2
CHANGE [mm] 4.7 2.0 0.0
% NBS5 --------- NAD 83 1998.871233
WIN 1998 319.00000000 51132.0000000 1998 321.99964120 51134.9996412 1
X Y Z
CRD 1096351.6870 -4831477.0100 4003598.0970
CRD rms [mm] 0.3 0.9 0.8
LAT LON HEIGHT (WGS84)
GEO_1 39 7 48.36868 - 77 12 54.11652 105.6320
GEO 39.1301024106 -77.2150323662 105.6320
GEO rms [mm] 1.1 0.5 0.2
CHANGE [mm] 4.7 2.6 0.6
.
(some of the example output was deleted here)
.
===============================================================================
======================= NETWORK BASELINE INFORMATION =======================
===============================================================================
%ITRF 96 Baseline from NBS0 to SCOL [SCOL minus NBS0] is 6689.139 m
X [m] x-rms [mm] Y [m] y-rms [mm] Z [m] z-rms [mm]
-5851.005 0.5 1178.695 1.7 3020.100 1.4
%NAD 83 Baseline from NBS0 to SCOL [SCOL minus NBS0] is 6689.139 m
X [m] x-rms [mm] Y [m] y-rms [mm] Z [m] z-rms [mm]
-5851.005 0.5 1178.695 1.7 3020.099 1.4
-------------------------------------------------------------------------------
%ITRF 96 Baseline from NBS1 to NBS0 [NBS0 minus NBS1] is 183.996 m
X [m] x-rms [mm] Y [m] y-rms [mm] Z [m] z-rms [mm]
-43.891 0.4 107.135 1.5 143.004 1.2
%NAD 83 Baseline from NBS1 to NBS0 [NBS0 minus NBS1] is 183.997 m
X [m] x-rms [mm] Y [m] y-rms [mm] Z [m] z-rms [mm]
-43.891 0.4 107.135 1.5 143.005 1.2
-------------------------------------------------------------------------------
.
(some of the example output was deleted here)
.