| Asian GPS ---> Proceeding ---> 2002---> Trends in GPS data processings |
A comparative study of results from GPS data processing software
Supriya Likhar, Madhav N. Kulkarni*, V.S. Tomar, Praveen Pillai
Department of Civil Engineering, I.I.T. Bombay, MUMBAI, India - 400076
*kulkarni@civil.iitb.ac.in
Supriya Likhar, Madhav N. Kulkarni*, V.S. Tomar, Praveen Pillai
Department of Civil Engineering, I.I.T. Bombay, MUMBAI, India - 400076
*kulkarni@civil.iitb.ac.in
Abstract
The Global Positioning System (GPS) has become an important tool for various applications, including scientific studies to monitor crustal deformations due to earthquakes. Dual-frequency geodetic GPS receivers are used to carry out the GPS observations. It becomes very important to post-process this data using software that gives the desired accuracy for estimating the crustal deformations. A variety of softwares are available in the market today for processing the GPS observations. Vendor-supplied GPS data processing softwares like the GPSurvey, Trimble Geomatics Office (TGO), SKI, etc., and scientific softwares like Bernese, GAMIT are available, that provide different levels of accuracy. This paper deals with the performance of different softwares used for processing short and long baselines. The softwares under analysis are Bernese, GPSurvey and TGO. The main objective of this study is to investigate the effectiveness of these software packages in different configurations. The work is based on the data collected by the GPS team of Indian Institute of Technology, Bombay (IITB), during the GPS observations carried out in Bhuj region of Gujarat, India, in February 2002.
Key words: GPSurvey, TGO, Bernese, Bhuj
Introduction
GPS related studies have received new impetus all over the world, with the full complement of satellites for adequate coverage, availability of comparatively low-cost receivers, sophisticated post processing software, etc. Many applications of GPS demand precise positions, for which extensive data processing is required. There are many GPS data processing software available: both scientific and commercial. In this study, a comparison of the output from the different software is carried out, in order to understand the effectiveness, precision and utility of different kinds of software (Kulkarni, 2000).
The effective monitoring and management of natural calamities requires the most recent terrain information and quick updating of the digital maps. The GPS technique can be used for this purpose, to yield very accurate results in an efficient and economic way. In February 2001, immediately after the 26th January earthquake in the Bhuj region of Gujarat, the GPS team from IIT Bombay carried out GPS observations, to understand the post-earthquake crustal deformation pattern, and to monitor any future crustal dynamics in this region. This GPS field campaign, with a total of 14 stations, including 5 old GT stations, has been repeated in February 2002. This data has been used in the present study.
GPS data
The GPS data, which basically consists of either pseudo-range or phase differences or both, can yield the three-dimensional position of the receiver antenna at the instance of observation (either in Cartesian Coordinates: x, y, z, or latitude, longitude and height), and the precise time. In addition to this important information, other parameters of interest for specific applications, such as atmospheric (tropospheric and ionospheric) delays, clock errors, satellite orbital elements, Earth rotation parameters and model parameters etc. can also be estimated. The choice of the software and technique to be used for such processing would depend primarily on the level of accuracy desired, and other factors such as the hardware available, surveying technique used to collect the data, etc. The GPS data processing software can be broadly classified into two categories - the receiver-specific vendor-supplied software and the scientific processing software. Even though the GPS data collected using different makes of the receivers is in different format, the facility of converting all these to a common internationally accepted ASCII format, called RINEX (Receiver IN dependent data Exchange format) makes it possible to process any kind of data using the scientific software, for achieving higher accuracy (Kulkarni, 1999).
GPS observation should be post-processed in relative mode using suitable software, to get accurate results. For most surveying applications, use of the vendor-supplied GPS data processing software is sufficient to yield the required accuracy. However, for high precision scientific applications, special scientific data processing softwares have been designed. Some of the scientific GPS data processing softwares being used world wide are: the BERNESE, developed by the University of Bern, GAMIT developed by MIT, GEODYN by NASA, GIPSY by JPL, TOPAS, Germany, GPSOBS by ESA, TEXGAP by Univ. Texas, GEPHARD by GFZ, Germany, etc. Of these, BERNESE software is being used by few research institutes/organizations in India, including Survey of India, IIG, IIT Bombay, CMMACS, WIHG, NGRI, etc., and GAMIT is also being used in more recent times.
Softwares used for comparison of results
This paper attempts to make a comparative study of the results from the GPS data processing software: GPSurvey Version 2.35, TGO Version 1.5 and Bernese Version 4.2. For this purpose the Bhuj data collected in February 2002 is considered, and independent processing of this data is done using all the above software. The steps involved in post processing of different softwares are listed below.
The GPS data used for this study is collected in Bhuj region for earthquake studies. The GPS team from IIT Bombay carried out GPS observations during 2001 and 2002, in the earthquake affected area in Bhuj region, Gujarat, in order to understand the post-earthquake crustal deformation pattern and to monitor any future crustal dynamics in this region. The existing geodetic control network in the earthquake affected Bhuj region consists of several geodetic stations at approximately 20-40 km spacing, of the series of the Great Trigonometrical (GT) Triangulation Network of India. These series had been established during the mid-nineteenth century. GPS observations at these stations would yield valuable data about the crustal deformations in the region due to various causes, including the earthquakes of 1919 and 2001. Immediately after the 26th January Bhuj earthquake, the GPS field campaign was carried out in February 2001. A total 14 stations, including 5 old GT stations, which were found intact, and 9 new stations established close to the GT stations found destroyed/disturbed, have been observed. Four 4000SSi Trimble dual frequency geodetic GPS receivers were used for this purpose. The observations were carried out in four campaigns with 48 hours of continuous observations at every station. In February 2002 the team carried out repeat observations of all the stations in the Bhuj region.
Results and Discussions
A detailed comparison of the post-processed results of GPSurvey, Trimble Geomatics Office (TGO) and Bernese has been carried out. The baseline lengths and the respective Root Mean Square (RMS) values of baselines between the stations using the above software are given in Table 1. In Table 2, the Cartesian coordinates of different stations on WGS-84 datum, estimated using these software, are given.
From Table 1 it can be seen that, the baseline lengths processed using GPSurvey and Trimble Geomatics Office (TGO) differ in the order of few mm. The comparison of GPSurvey and Trimble Geomatics Office (TGO) results with Bernese show a difference in the order of few centimeters. The RMS values of baseline length processed using GPSurvey are in the range of 9-38 mm and that of TGO is in the range of 10-45mm. The baseline RMS values obtained from Bernese processing are in the range 1.4-14.7 mm.
Figure 1: Comparison of Baseline Length (in meter, after decimal point), Estimated Using Different Software
Note- Length: Baseline length in meter, after the decimal point
The baseline lengths obtained from all three software differ only after the decimal point, so for comparison purpose, the baseline values after the decimal only are plotted in the graphical representation of variations, shown in Figure 1. It is observed that out of all these estimated baseline lengths, the length of baseline between Roha and Samdhan (35242.569 m) is same, as computed using GPSurvey, TGO and Bernese software.
Conclusions
From the comparison of results obtained by processing the same dual-frequency geodetic GPS data using these three software, it is seen that the baseline lengths processed using GPSurvey and TGO differ in the order of few mm. The comparison of GPSurvey and TGO results with Bernese show a difference in the order of few centimeters. The RMS values of baseline length processed using GPSurvey are in the range of 9-38 mm and that of TGO is in the range of 10-45mm. The baseline RMS values obtained from Bernese software are in the range 1.4-14.7 mm. Thus, based on the RMS values, it can be concluded that the results form the two software: GPSurvey and TGO are in close agreement with each other, however, the Bernese post-processed baseline results are more precise, as compared to GPSurvey and TGO.
References
The Global Positioning System (GPS) has become an important tool for various applications, including scientific studies to monitor crustal deformations due to earthquakes. Dual-frequency geodetic GPS receivers are used to carry out the GPS observations. It becomes very important to post-process this data using software that gives the desired accuracy for estimating the crustal deformations. A variety of softwares are available in the market today for processing the GPS observations. Vendor-supplied GPS data processing softwares like the GPSurvey, Trimble Geomatics Office (TGO), SKI, etc., and scientific softwares like Bernese, GAMIT are available, that provide different levels of accuracy. This paper deals with the performance of different softwares used for processing short and long baselines. The softwares under analysis are Bernese, GPSurvey and TGO. The main objective of this study is to investigate the effectiveness of these software packages in different configurations. The work is based on the data collected by the GPS team of Indian Institute of Technology, Bombay (IITB), during the GPS observations carried out in Bhuj region of Gujarat, India, in February 2002.
Key words: GPSurvey, TGO, Bernese, Bhuj
Introduction
GPS related studies have received new impetus all over the world, with the full complement of satellites for adequate coverage, availability of comparatively low-cost receivers, sophisticated post processing software, etc. Many applications of GPS demand precise positions, for which extensive data processing is required. There are many GPS data processing software available: both scientific and commercial. In this study, a comparison of the output from the different software is carried out, in order to understand the effectiveness, precision and utility of different kinds of software (Kulkarni, 2000).
The effective monitoring and management of natural calamities requires the most recent terrain information and quick updating of the digital maps. The GPS technique can be used for this purpose, to yield very accurate results in an efficient and economic way. In February 2001, immediately after the 26th January earthquake in the Bhuj region of Gujarat, the GPS team from IIT Bombay carried out GPS observations, to understand the post-earthquake crustal deformation pattern, and to monitor any future crustal dynamics in this region. This GPS field campaign, with a total of 14 stations, including 5 old GT stations, has been repeated in February 2002. This data has been used in the present study.
GPS data
The GPS data, which basically consists of either pseudo-range or phase differences or both, can yield the three-dimensional position of the receiver antenna at the instance of observation (either in Cartesian Coordinates: x, y, z, or latitude, longitude and height), and the precise time. In addition to this important information, other parameters of interest for specific applications, such as atmospheric (tropospheric and ionospheric) delays, clock errors, satellite orbital elements, Earth rotation parameters and model parameters etc. can also be estimated. The choice of the software and technique to be used for such processing would depend primarily on the level of accuracy desired, and other factors such as the hardware available, surveying technique used to collect the data, etc. The GPS data processing software can be broadly classified into two categories - the receiver-specific vendor-supplied software and the scientific processing software. Even though the GPS data collected using different makes of the receivers is in different format, the facility of converting all these to a common internationally accepted ASCII format, called RINEX (Receiver IN dependent data Exchange format) makes it possible to process any kind of data using the scientific software, for achieving higher accuracy (Kulkarni, 1999).
GPS observation should be post-processed in relative mode using suitable software, to get accurate results. For most surveying applications, use of the vendor-supplied GPS data processing software is sufficient to yield the required accuracy. However, for high precision scientific applications, special scientific data processing softwares have been designed. Some of the scientific GPS data processing softwares being used world wide are: the BERNESE, developed by the University of Bern, GAMIT developed by MIT, GEODYN by NASA, GIPSY by JPL, TOPAS, Germany, GPSOBS by ESA, TEXGAP by Univ. Texas, GEPHARD by GFZ, Germany, etc. Of these, BERNESE software is being used by few research institutes/organizations in India, including Survey of India, IIG, IIT Bombay, CMMACS, WIHG, NGRI, etc., and GAMIT is also being used in more recent times.
Softwares used for comparison of results
This paper attempts to make a comparative study of the results from the GPS data processing software: GPSurvey Version 2.35, TGO Version 1.5 and Bernese Version 4.2. For this purpose the Bhuj data collected in February 2002 is considered, and independent processing of this data is done using all the above software. The steps involved in post processing of different softwares are listed below.
- GPSurvey Software Version 2.35
- Trimble Geomatics Office (TGO) Software Version 1.5
- Survey and plan view, displays project data in the graphic window Project bar organizes commonly used tasks as shortcuts in the main graphics window View filters allow observation types to be displayed or hidden Import of data collected from GPS receivers and/or conventional total stations DAT checking dialog allows checking and editing of raw GPS data before import Properties window helps to view and edit all related information for points, as well as all observation types Multiple edit dialog is for quick editing of multiple entities Professionally presented output report in HTML format is generated Standard and custom report formats can be prepared (Trimble, User Guide, 2001).
- Bernese Software Version 4.2
- Campaign module: The first task is to setup the campaign and create the basic directories required for the processing.
- Transfer module: Transfer of DAT to RINEX (TRINEXO) and RINEX to Bernese (RXOBV3)
- Orbital module: Two programs are run PRETAB and ORBGEN to get clock file and standard files. To get more precise results IGS precise file has to be downloaded and used.
- Processing module: In this module five programs are run in order to get the final solution.
- Main Features of Bernese Software:
- Meets highest accuracy requirements
- Processing of all principal observables recorded by high precision geodetic receivers (code and phase data on both carriers)
- Five different linear combinations of L1 and L2 may be used
- Processing and combination of data from various receiver types in the same processing step (including establishing and use of receiver type specific antenna phase center variations).
- Single and dual frequency data may be processed in the same estimation step, use of ionosphere models to minimize impact of ionospheric biases on station coordinates)
- Processing of all static GPS applications
- The parameter estimation programs may be used for baseline in campaign and multiple campaign processing. Many different complex solutions (e.g. annual coordinate and ERP solutions) using (e.g. daily) normal equation systems can be produced without reprocessing observations
- RINEX data interface
- Simultaneous solution for a large number of different parameter types (Hugentobler, et.al., 2001)
GPSurvey is vendor-supplied software, specifically suited for hardware supplied by Trimble Navigation Ltd., USA. It has a program called DAT2RIN, which brings hardware dependent DAT files to RINEX format that can be processed with other GPS data processing softwares. The main modules in GPSurvey are:
| GPLOAD | : | Downloading the data from the hardware |
| WAVE | : | Processing of baseline |
| TRIMNET | : | Adjustment of the Network |
| The steps involved in relative mode data processing are: | ||
| PROJECT | : | TO create new or open existing project |
| LOAD | : | To load data from DAT/RINEX file |
| PROCESS | : | Processing baselines, add files to WAVE module |
| WAVE processor | ||
| EDIT | : | Station Position (check base station "fixed control" coordinates) |
| PROCESS | : | Setup (only to alter processing controls) Combinations of baselines to be processed |
| UTILITIES | : | Project report is generated. Coordinate transformations can also be done over different combinations of reference frame (Trimble GPSurvey Manual, 1998). |
The Trimble Geomatics Office software integrates data from a variety of surveying techniques, including GPS (Real Time Kinametic and post processing), conventional/optical instruments (including servo and robotic), levels, and lasers. The Trimble Geomatics Office software is useful for, GPS baseline processing, survey network adjustment, processing of GPS and conventional topographic survey data, quality assurance and quality control of data (QA/QC), road design data import and export, survey data import and export, Digital terrain modeling and contouring, datum transformation and projections, and GIS data capture and data export
The Key features of Trimble Geomatics Office (version 1.5) software are:
Bernese software has been designed by the University of Bern to meet high accuracy required for scientific purposes. The following are the steps involved in the Bernese processing
| CODSPP | : | Its main task is to compute the receiver clock corrections |
| SNGDIF | : | This program creates the single differences and stores then in files (OBS-MAX rategy). |
| MAUPRP | : | The main task of the program is the cycle slip screening |
| GPSEST | : | The main task is Least Square Adjustment The output of this program gives the baseline length and coordinates precisely. |
| ADDNEQ | : | This program combines the normal equations from different sessions and gives the final solution. |
The GPS data used for this study is collected in Bhuj region for earthquake studies. The GPS team from IIT Bombay carried out GPS observations during 2001 and 2002, in the earthquake affected area in Bhuj region, Gujarat, in order to understand the post-earthquake crustal deformation pattern and to monitor any future crustal dynamics in this region. The existing geodetic control network in the earthquake affected Bhuj region consists of several geodetic stations at approximately 20-40 km spacing, of the series of the Great Trigonometrical (GT) Triangulation Network of India. These series had been established during the mid-nineteenth century. GPS observations at these stations would yield valuable data about the crustal deformations in the region due to various causes, including the earthquakes of 1919 and 2001. Immediately after the 26th January Bhuj earthquake, the GPS field campaign was carried out in February 2001. A total 14 stations, including 5 old GT stations, which were found intact, and 9 new stations established close to the GT stations found destroyed/disturbed, have been observed. Four 4000SSi Trimble dual frequency geodetic GPS receivers were used for this purpose. The observations were carried out in four campaigns with 48 hours of continuous observations at every station. In February 2002 the team carried out repeat observations of all the stations in the Bhuj region.
Results and Discussions
A detailed comparison of the post-processed results of GPSurvey, Trimble Geomatics Office (TGO) and Bernese has been carried out. The baseline lengths and the respective Root Mean Square (RMS) values of baselines between the stations using the above software are given in Table 1. In Table 2, the Cartesian coordinates of different stations on WGS-84 datum, estimated using these software, are given.
From Table 1 it can be seen that, the baseline lengths processed using GPSurvey and Trimble Geomatics Office (TGO) differ in the order of few mm. The comparison of GPSurvey and Trimble Geomatics Office (TGO) results with Bernese show a difference in the order of few centimeters. The RMS values of baseline length processed using GPSurvey are in the range of 9-38 mm and that of TGO is in the range of 10-45mm. The baseline RMS values obtained from Bernese processing are in the range 1.4-14.7 mm.
Table 1: Comparison of Baselines (Baseline length (BL) in meter and RMS in mm)
Table 2: Comparison of Cartesian Coordinates of different stations
| FROM STATION | TOSTATION | GPSurvey | TGO | Bernese | |||
| BL | RMS | BL | RMS | BL | RMS | BL | RMS |
| ROHA | NARA | 49674.188 | 12 | 49674.187 | 19 | 49674.1936 | 1.0 |
| ROHA | NETR | 36435.584 | 17 | 36435.591 | 13 | 36435.6206 | 14.7 |
| ROHA | SAMD | 35242.569 | 10 | 35242.569 | 11 | 352342.5685 | 3.8 |
| SAMD | NARA | 68998.375 | 17 | 68998.371 | 45 | 68998.3852 | 1.4 |
| NARA | NETR | 21156.379 | 14 | 21156.377 | 11 | 21156.4003 | 9.2 |
| CHAR | SAMT | 48832.839 | 09 | 48832.938 | 22 | 48832.8192 | 6.6 |
| CHIT | KNDK | 18604.249 | 11 | 18604.250 | 10 | 18604.2403 | 1.9 |
| KAKA | KNDK | 30364.502 | 38 | 30364.540 | 42 | 30364.5027 | 8.0 |
| KNDK | SUKH | 61907.727 | 22 | 61907.738 | 29 | 61907.6797 | 7.8 |
Table 2: Comparison of Cartesian Coordinates of different stations
| STATION ID | Coordinates From GPSurvey Software | Coordinates From BERNESE Software | ||||
| X (m) |
Y (m) |
Z (m) |
X (m) |
Y (m) |
Z (m) |
|
| CHAR | 2007844.873 | 5513392.009 | 2492146.956 | 2007846.4304 | 5513398.1307 | 2492151.2670 |
| JHUR | 2005326.149 | 550459.688 | 2513518.886 | 2005331.5877 | 5504612.5748 | 2513522.0144 |
| SAMT | 2053499.593 | 5496117.749 | 2493522.669 | 2053501.1340 | 5496123.8817 | 2493526.9819 |
| SUKH | 1989107.148 | 5514136.049 | 2505358.236 | 1989114.0313 | 5514143.7692 | 2505366.7085 |
| KANM | 1918613.683 | 5533606.138 | 2517259.405 | 1918613.7317 | 5533603.8435 | 2517257.9335 |
| PIRP | 1909848.765 | 5529197.495 | 2533436.132 | 1909848.8146 | 5529195.2206 | 2533434.6708 |
| KNDK | 1934469.928 | 552537.745 | 2533754.671 | 1934476.8489 | 5520545.4744 | 2533763.1393 |
| CHIT | 1937389.489 | 5527380.848 | 2516702.804 | 1937396.4280 | 5527388.5907 | 2516711.2897 |
| NARA | 2082953.245 | 5462909.224 | 2540783.068 | 2082954.0083 | 5462907.2854 | 2540780.5734 |
| NETR | 2093648.743 | 5467236.695 | 2523049.719 | 2093649.5412 | 5467234.7238 | 2523047.2114 |
| ROHA | 2075837.214 | 5485938.952 | 2497348.999 | 2075837.9844 | 5485937.0459 | 2497346.5102 |
| ASAP | 2005330.031 | 5504606.496 | 2513517.757 | 2025448.0338 | 5505800.4751 | 2495247.0029 |
| KAKA | 1937389.404 | 5527380.749 | 2516702.795 | 1963263.3299 | 5513381.3727 | 2527281.0733 |
| SAMD | 2105589.746 | 5482818.480 | 247817.696 | 2105590.5120 | 5482816.5720 | 2478716.2023 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
Figure 1: Comparison of Baseline Length (in meter, after decimal point), Estimated Using Different Software
Note- Length: Baseline length in meter, after the decimal point
The baseline lengths obtained from all three software differ only after the decimal point, so for comparison purpose, the baseline values after the decimal only are plotted in the graphical representation of variations, shown in Figure 1. It is observed that out of all these estimated baseline lengths, the length of baseline between Roha and Samdhan (35242.569 m) is same, as computed using GPSurvey, TGO and Bernese software.
Conclusions
From the comparison of results obtained by processing the same dual-frequency geodetic GPS data using these three software, it is seen that the baseline lengths processed using GPSurvey and TGO differ in the order of few mm. The comparison of GPSurvey and TGO results with Bernese show a difference in the order of few centimeters. The RMS values of baseline length processed using GPSurvey are in the range of 9-38 mm and that of TGO is in the range of 10-45mm. The baseline RMS values obtained from Bernese software are in the range 1.4-14.7 mm. Thus, based on the RMS values, it can be concluded that the results form the two software: GPSurvey and TGO are in close agreement with each other, however, the Bernese post-processed baseline results are more precise, as compared to GPSurvey and TGO.
References
- Hofmann-Wellenhof B., H. Lichtenegger J. Collins, "Global Positioning System, Theory and Practice", Fifth, revised edition, Springer-Verlag Wien New York, 2001
- Hugentobler, U., S. Schaer, P. Fridez, "Bernese GPS Software Version 4.2 Documentation", Astronomical Institute University of Bern, Switzerland, 2001
- Kulkarni, M. N., "The GPS Data processing for seismic Hazard Assessment", Indian Surveyor, Vol.52, No.1, pp. 28-29, 1999
- Kulkarni, M. N., "Application of GPS and GIS to Disaster Monitoring and Management in the Next Millennium", Geomatics: Conference on Geomatics in Electronic Governance C-DAC, Pune, pp. DM23-DM26, 21-22 January 2000
- Trimble GPSurvey Manual, "Surveying and Mapping Product Training", Trimble Navigation Limited, USA, 1998
- Trimble, User Guide, "Trimble Geomatics Office, User Guide", Trimble Navigation Limited, USA, Vol. 1-2, 2001