Establishment of Absolute ITRF/WGS84 Coordinates in Dubai Mohammed Abdulla Al-Zaffin, Head of Survey Section, Dubai Municipality. Dubal, U.A.E. | Abstract Of The Paper & The Profile of The Speaker | Speaker Index | Paper Title Index | 1.- INTRODUCTION The present mapping system of Dubai is based on a control system which has evolved since the foundation of the United Arab Emirates till today. Extended over an area of approximately 4,000 km2 it is composed mainly of two non least squares adjusted triangulation networks, based on which non connected traverses have been observed and merely computed. In 1990 a study was conducted with the purpose to assess the quality of the above networks. This was conducted in terms of Gross Error Detection based on Baarda's Data Snooping strategy and computation of Internal and External Reliability Parameters. The results of the investigation were such that approximately 15% of the available observations were found to be beyond the stochastical boundary values. Moreover, the arrived Internal Reliability Parameters were such that in certain areas of the network, the influence of a non detected gross error in the observations would have resulted in distortion of the coordinate values in more than 9.00 meters. In view of the above and with the purpose of establishing and developing a Geographical Information System, the Survey Section of Dubai Municipality decided to establish a new control network fulfilling the high accuracy and reliability requirements of the Geographical Information System as envisaged. In September 1990, the Survey Section received its first 2 GPS receivers along with the pre and post - processing software, including a Least Squares Network Adjustment program of GPS observations. Latter in 1992, the Section acquired two additional receivers and updated software. With the above equipment the Survey Section observed the first order 3D GPS Geodetic Network composed of 62 monumented control points with distances between 5 to 10 Km. All GPS measurements between the above mentioned monumented points can be considered only as 3D vector in the space which needs to be interconnected, adjusted and geo-referenced to a chosen ellipsoid. With regard to the chosen ellipsoid, this was, due to obvious reasons, WGS/84. In relation with the adjustment and georreferentiation of the above 3D free vectors, the absence of other WGS/84 points I networks in the area or the existence of precise and reliable ellipsoidal (Latitude and Longitude) data capable to be transformed to WGS/84 values, imposed limitations to the Geodetic project. To overcome this inconvenience, an international campaign was carried out as a joint venture between the Institute of Applied Geodesy of Frankfurt (Germany) and the Survey Section of Dubai Municipality with the aim to fix WGS/84 point coordinates within the Emirate of Dubai. This paper summarizes this unique Geodetic project. 2.- THE CLARK 1880/ UTM NETWORK 2.1.- Description of the Network The network used for the 1983 mapping of Dubai and its subsequent development is based on 4 third order points located in the SE corner of the Emirate. They are part of what is remaining of the old Trucial Coast Countries 3rd. order Geodetic Control System. This was setup by the British Army at the end of the last century, beginning of the present. The development started during the years 70's due to a high demand of control and mapping in a highly development epoch of the Emirate. It is composed of triangles, quadrilaterals and non related / non connected traverses which have been merely computed and non least squares adjusted. For the observations, different types of theodolites and EDM's were used with different accuracy standards. A priory, one can conclude that this network is non uniform, non homogeneous. As the above Trucial Coast points have their coordinates based on the Ellipsoid of Clark 1880, with planimetric coordinates computed in the UTM Zone 40 Cartographic System, the developed network for Dubai is also based on the above reference. Moreover, the computations were done in the UTM/ Zone 40 only with their corresponding reductions. 2.2.- Geometric Problems of the Network As mentioned in the introduction, during 1990 a research was done in order to assess the geometric quality of the network With this purpose Gross Error Detection based on Baarda's Data Snooping Strategy was conducted. Also Internal and External Reliability Parameters for each observation were computed. The associated statistic of Data Snooping is : Following this procedure, computations of the wi statistic were calculated for each observation of the net and compared with the above critical value. As a result, as mentioned in the introduction, approximately 15% (143) of the observations do not passed the above test. This was due mainly to the weak geometry the network has in some areas, mainly in connection with the non - connected traverses. With regard to the reliability parameters, they were computed in terms of Boundary Values (Internal Reliability) and the effect of the boundary values on the adjusted Coordinates (External Reliability). The used statistics were : Table 1 The above table shows that a maximum boundary value of 5'.49 can be expected for the observed directions and 0.785 m for the observed distances. This can produce a maximum distortion on the final coordinates of 9.03 m. The conclusions of this study was obviously that the network is non homogeneous non precise and not reliable to be used as a framework for the fliture GIS of Dubai. 3.- FIRST ORDER GEODETIC GPS NETWORK In view of the above conclusion the Survey Section decided to observe a new Geodetic Network which can flilfill the accuracy requirements of the GIS of Dubai. 3.1.- GIS TOPOGRAPHIC REQUIREMENTS During 1991 all Technical Departments of Dubai Municipality set-up their base mapping accuracy requirements for their different tasks, both in terms of pointing and relative accuracy. After a deep study a decision was taken regarding the mapping scale and projection system to be used. These were as follows : Projection Ellipsoid : WGS84 Cartographic Projection System : Local UTM with central meridian located at 55deg20' with false Easting of 500,000 and scale factor K=1 This projection system is called Dubai Local Transverse Mercator (DLTM) Map Scale-1:1,000 Height Datum : MSL value determined at Port Rashid. Hence the accuracy requirements of the new Geodetic Network should commensurate with the accuracy requirements of the above selected mapping scale and with the new achieved accuracy of digital mapping technologies. This means an absolute positional accuracy of 10 centimeters or better and a relative accuracy of 1/100,000 or better. Only GPS observations can fulfill the above requirements. In this sense Dubai Municipality acquired during -1991 the first GPS Trimble 4000 along the Pre and Post Processing software. During 1993, two more receivers were purchased. 3.2.- DESCRIPTION OF THE NETWORK The first order Geodetic GPS Network is composed of 62 monumented points with distances between points ranging from 5 to 10 Km. The network is more dense in the city area with distances between points from 3.5 to 5.0 Km. In the desert area the distances between points are from 5 up to 20 Km. Each baseline was planned in such a way to have a satellite constellation of at least 5 satellites during the observation time and a PDOP less than 7 Each line was observed during 90 minutes with a sapling rate of 15 seconds and an elevation cut-off angle of 15 degrees. Carrier face observations with Li and L2 frequencies were acquired and post processed using Trimvec Software and recently with GP Survey from Trimble Navigation Ltd. After post-processing accuracies of the order of 1/ 300,000 or better were arrived for fix optimum solution. 4.- THE DUREF-95 PROJECT At this stage of the observations, it should follow by the corresponding Network Adjustment. For this it is necessary to have control points in the selected Geodetic - Cartographic System (i.e., WGS84 - DLTM). Obviously, the above control points of the original network can not be used due to several reasons, namely : 1.- It was shown that it is non homogeneous, non accurate, non reliable 2.- No possible transformation from the previous system Clark 1880 - UTM Zone 40 to WGS84 - DLTM due to the absence of points with coordinates in the latest system. 3. - No rigorous transformation such as the 7 parameters can be applied due to the lack of knowledge of the geoid undulations in the area (i.e. separation geoid - ellipsoid in each network point). 4. - No good results can be expected from any transformation due to the deformations of the Clarkl88O UTM network (See chapter 2 above). 5.- Others. In view of the above the only possibility to fix the GPS network to the ellipsoid WGS84 was via an International Geodetic GPS Campaign. This was done as a joint venture project between Dubai Municipality and the Institute for Applied Geodesy (IfAG) of Frankfurt - Germany. Since the WGS84 datum do not satisfy the modern needs of geodesy, the International Earth Rotation Service (IERS) together with the International Association of Geodesy (IAG) established the International Terrestrial Reference System (ITRS) whose latest accomplishment is the ITRF93 (International Terrestrial Reference Frame 1993). This satisfies the accuracy requirements of the modern Geodesy. This is a geocentric system with an accuracy of +- 5cm and has been defined by the scale and origin of satellite laser ranging solutions and the 'ERS series of Earth Orientation Parameters. Within the accuracy level of the ITRF93, the velocity of the used stations has to be taken into consideration since plate tectonics velocity can be of several cm per year. To this end, the International GPS Service for Geodynamics (IGS) maintains a network of GPS permanent tracking stations which is used for GPS orbit determination. These stations are reference sites for the connection of individual countries main networks to the ITRS. For the case of Dubai, due to technical problems, receiver types deployed or too large distances to the reference stations, for the DUREF95 campaign only the sites at Graz (Austria), Matera (Italy) and Kitab (Uzbekistan) were used. The coordinates at the above sites were fixed at the epoch 1995.1 and the coordinates of the points in Dubai were deduced from them. 4.1.- DATA ACQUISITION AND PROCESSING The data were acquired during January 1995 and are summarized as follows: - GPS Campaign in Dubai from 25 - January to 29 - January 1995 - 6 GPS Stations in Dubai (4 stations observed all 5 days) - 3 IGS Reference Stations, i.e. Graz, Matera and Kitab - Observations with TRIMBLE SSE receivers in Dubai - Observations with ROGUE receivers at the reference points - Session length 24 hours - All Sessions from 15.00 UT to 15.00 UT of the following day - Elevation cut-off angle 10 degrees - Sampling rate 15 seconds (TRIMBLE), and 30 seconds (ROGUE) receivers Four main points were observed within main Dubai Emirate. They were ET145, BPS, ET228 (Dubai Municipality Head Office) and ET225. In addition 2 other stations were observed for a few hours (ET 152 in Hatta - Dubai and FUG M3 in the Emirate of Fujeira) with spares SSE receivers. No meteorological observations were taken. The antennas were located on tripods or pillar plates with tribrach (ET145, BPS) above the markers. The data reductions and computations were carried via the use of the BERNESE software (IfAG) which forms the baselines between pair of stations. The baselines were taken between the IGS permanent reference stations and DM Headquarters, point ET228, for the ITRS connection and from ET228 to the local points. The Table 2 shows the baseline lengths. From Station To Station Baseline length Head Office Graz 4175632.31m Matera 3888599.49 m Kitab 1875417.52 m ET145 36631.66m ET225 18508.70m BP5 32531.13m ET152 96657.22m Fujeira 71944.18m Table 2 The GPS data processing strategy was as follows: - Computation of the complete network using the BERNESE (IfAG) Software - IGS Combined Orbit in ITRF-93 (GPS Weeks 0785 and 0786) - Reference System ITRF-93, epoch 1995.1 - Pre-processing using 24 hours of -15 seconds interval TRIMBLE S SE data - Elevation-dependent phase center variation file - Processing of double-difference measurements based on ionosphere-free phase linear combination L3 - Float/fixed solution, ambiguity fixing for local network only - Standard troposphere model by Saastamoinen - One troposphere parameter estimated every 2 hours for every session - Sigma values for troposphere parameter: 0.50 m absolute, 0.02 m relative - Daily solutions for accuracy evaluation - Final solution including all days with Graz, Matera and Kitab 4.2.- RESULTS WITH REFERENCE TO ITRF93 The baselines to the IGS sites were of float solutions only, i.e. the integer number of wavelengths could not be fixed due to the long distances. The arrive rms of the baselines were all uniform for all days, showing that the data quality was good. The rms are ranging between 0.0021 m and 0.0079 m. The only exemption was the point in Fujeira where the signal-noise is high due to possible radio-frequency disturbance of the near airport. As mentioned above, the final adjustment was done in two parts: 1) Ambiguity fixed solution with Graz, Matera and Kitab as fixed points, only for stations which have been observed during the 5 consecutive days, with the standard troposphere model of Saatamoinen and estimation of -12 troposphere zenith delay parameters per session (one each 2 hours) for all stations 2) A separate adjustment for the two points (Hatta and Fujeira) which were observed for a few hours only. This adjustment were carried out without fixed points but using the a-priori sigma values for the coordinates of the station point BPS (10cm for all three coordinates). a) Fujeira: float solution (without fixing of ambiguities and without tropospheric model due to small observation time) b) Hatta: fixed solution (with fixing of ambiguities and without tropospheric model due to small observation time) The final results with the arrived accuracies are shown in the Table 3. (32K GIF Image File) 4.3.- ACCURACY EVALUATION The following table (Table 4 - 64K GIF Image File) shows a comparison of the arrived accuracy for each observation day. The standard deviations of the 3 components shown in the table displays the repeatability with respect to the IGS global tracking network. Hence they include the remaining orbit errors and other systematic effects of different days. The positional accuracy is about 1 cm, being the height (as expected) around 2 cm. The overall accuracy estimation is the above 1cm plus the uncertainty of ITRF frame which is 5cm. Hence the final accuracy of the ITRF points of Dubai is 6 cm. In order to assess the internal precision and quality of the data, the day by day results were transformed (7 parameters) to the first day results. The residuals of these transformations are shown in the Tables 5 and 6. Table S shows residuals of single days' solutions after 7 parameters transformation to the first day. Table 6 gives the comparison of row results for single days in the local network, i.e. ET228 DM Head Office as a fix point. The results of the tables shows that the systematic effects from day to day are eliminated and the Standard deviations are bellow one centimeter in all three components. This once again shows the high quality of the data and their consistency. Table 6 shows the arrived standard deviations of the local network on which the connections to the IGS stations are not considered and the point ET228 (DM Head Office) is being kept fixed. The accuracy is such that the uncertainties are well bellow -1 cm for all components. Due to the short observation times and the disturbed data at Fujeira, the points in Hatta and Fujeira are less accurate, especially for the height component. Table 5. Comparison of Single days' solutions after 7 parameter Helmer transformation to the first day. (96K GIF Image File) Table 6. Comparison of row results for single days in the local Network. ET228 Head Office fixed. (64K GIF Image File) 5.- CONCLUSIONS AND RECOMMENDATIONS The DUREF-95 campaign determined the geocentric coordinates of points in Dubai Emirate in the ITRF93. The overall uncertainty of the coordinates were estimated to be about 5 centimeters, being the internal precision in the range of the 5 millimeters. The limiting factor of the accuracy was the global distribution of the IGS tracking sites and the large distance to the next IGS site. Nevertheless, the obtained results comply very well with the International standards required for GIS and moreover with the topographical requirements of Dubai Municipality GIS project. The stations of the Dub ai net relative to each other can be taken as error free reference points for the adjustment of the observed GPS Network (i.e. 62 monumented points). This was the way the adjustment was conducted arriving to a standard deviation of 8 centimeters as demanded by the topographical requirements of the GIS Project of Dubai. 6.- REFERENCES 1.- Seeger H., et al., DUREF-95 report. Institut fuer Angewandte Geodaesy. Frankfurt Germany, -18.8.1995 2.- Seeger H., (-1994), EUREF. The New European Reference Datum and its relationship to WGS84. FIG XX Congress, Commission 5. Melbourne, Australia, March 1994. 3.- Baarda W. (1968), Statistical Concepts in Geodesy. Dutch Geodetic Commission 4.- Baarda W. (-1969), Accuracy and Reliabilty of Observations. Duch Geodetic Commission. 5.- Passini, R. (-1985), Error Detection and Reliability in Aerial-triangulation and Geodetic Networks as applied to Project Parana Medio (Argentina). Ph.D. Dissertation. IPI, University of Hannover, Germany. | Abstract Of The Paper & The Profile of The Speaker | Speaker Index | Paper Title Index | CGIS HOME PAGE CONTENTS