Verical Control Network Of Al Ain Region

Dr Kamal Abdellatif Abdalla
Planning and Survey Sector
Municipality of Al Ain
P. o. box 1339
Al Ain, United Arab Emirates

Abstract:
Horizontal and vertical geodetic control networks of the Eastern region of Abu Dhabi Emirate have been established by a number of different methods, however today, GPS has become the most widely used method. The current height network in the Region is based mainly upon a series of leveling runs carried out by BKS Surveys between 1983 and 1987. WGS84 geoid is determined from GPS measured heights and the available orthometric heights of the region.

An exercise of establishing a local geoid for the Eastern Region of the Abu Dhabi Emirate has been attempted in two steps, first the EGM96 model was compared against the geoid undulations as derived from actual survey data. Secondly the resulting bias was then used to best fit the EGM96 model to the local conditions in an attempt to improve the results, hence transforming the bias corrected EGM96 to a best-fit local geoid. The paper assessed the quality of the vertical control of the Eastern Region of Abu Dhabi Emirate and outlined its accuracy and limitations.

1. Introduction
In order to place individual surveying projects into a larger spatial context, a geodetic control network is necessary. Such a network consists of a number of points spread across the area under consideration, these points are usually in the form of monuments established and placed in the ground, along with a high-accuracy positional value for each point. Traditionally, there have been completely separate networks for horizontal and vertical control, but some and recent networks combine the two on common monuments. By referencing field measurements to such a network, the resulting data and information from multiple local survey activities can be accurately connected. The accuracy of each activity or project is no higher than the control network to which it is referenced. A control network itself is established by highly precise surveying methods followed by a statistical adjustment to reconcile all of the measurements. For each monument the result is a published positional value along with a stated accuracy level.

Actually, geodetic control is typically separated into two components: horizontal (latitude/longitude) and vertical (elevation). This is because latitude/longitude and elevation are based on completely different concepts and measurement methods. Even today when GPS can provide extremely high-accuracy horizontal results, a more traditional method is required to establish vertical control. Geodetic control surveys are usually performed to establish a basic control network (framework) from which supplemental surveying and mapping work are performed. The required accuracy for a control survey depends primarily on its purpose. Factors that affect

accuracy are: type and condition of equipment used, field procedures adopted, and the experience and capabilities of personnel employed. Different sets of detailed standards of accuracy and specifications for geodetic surveys are to be considered to get reliable control points.

Horizontal geodetic control networks can be established by a number of different methods, however GPS has become the most widely used method due to its efficiency and superior results. These networks provide positional information with reference to a mathematical surface called an ellipsoid (horizontal datum) defined to model the size and shape of (all or some part of) the Earth. The high accuracy reference network is usually being established in many countries. It will be based upon GPS positioning and produced highly accurate results that can be the basis for the national datum adjustment. It also provides the surveying community with a network of highly reliable positional coordinates to control their surveys.

Vertical control networks are a series of points on which precise heights, or elevations, have been established. Vertical control stations are typically called Bench Marks (BM) as part of a vertical information network, the benchmark's elevation is known relative to a vertical datum, usually approximating the mean sea level. The vertical accuracy standard as referenced in the 1998 FGCS standard specifies a linear value (plus or minus) within which the true or theoretical location of the point falls 95 percent of the time.

During previous years more than 2000 precisely located, in-ground or monumented reference points were installed to measure heights in Al Ain region. The classical line-of-sight measurements do not provide the real-time accuracy needed for today's positioning technologies and applications. By using GPS pinpoint-positioning accuracies can be provided 24 hours a day. The combination of an improved height system along with GPS, will offer the ability to obtain precise vertical measurements in real time.

In the Eastern Region of Abu Dhabi, because much of the existing vertical control has been lost to destruction or disturbance, and because the network was not dense enough to support use of GPS to derive elevations, Al Ain Town Planning initiated a project for height determination based on gravimetric determination of geoid.

Prior to GPS, horizontal control was most easily established by measuring between hilltops. Similarly, vertical control was (and still is) most easily established by leveling along railroad or highway corridors. Even once GPS became a practical technology, vertical control monument locations were still problematic as control stations since some are near buildings and along the roads covered by trees and others are not easily accessible. Nevertheless, it can be seen in Al Ain that its a great advantage in property surveys to have control values distributed all through the developed areas, because the surveyor saves the time otherwise required to survey ("run") control in from nearby (sometimes several miles or more away) geodetic stations.

There are several phases involved in establishing a geoid surface that can be used to correct GPS derived heights so that they are consistent with the existing orthometric height reference system. Initially, it is necessary to assess the existing control, and confirm its validity and consistency. Only once this has been done can a geoid be determined and then fitted to the existing vertical control points.

2. Existing vertical control.
The current height network in the Eastern Region is based mainly upon a series of leveling runs carried out by BKS Surveys between 1983 and 1987. A total of about, 214 levelling circuits were carried out over a total distance of 2396 km. The accuracy was specified as 10 mm per square root of levelling run in kilometres. The aim was to provide one point for every square kilometre in areas to be covered by 1:500 and 1:1000 mapping, and one point for every 5 km2 in areas to be covered by 1:2000 mapping. Of the 2000 ground markers, 814 were spirit levelled. The rest were heighted by reciprocal vertical angles. Many of these points have subsequently been destroyed during development work, and that only as few as 25% may still exist in some areas. The datum used for this levelling was the Port Rashid datum of Dubai, established by levelling carried forward from Dubai to points BTP219 (Al Faqa) and BTP226 (Schweib). These two points mark the northern extent of the levelling work; the southern extent is at Al Qua. In addition to the levelling and trigonometric heighting established by BKS, two other sources of height information are available in Al Ain Town planning database. The first of these are the points identified by the initial prefix “K”, which relate to a boundary survey carried out along the border with Oman by KLM. The second set of points is that identified by the initial prefix “G”, which were geodetic points established by the Survey Section of Al Ain Town Planning in the 1990s. Concrete monuments were installed in all non-rocky and non-sandy locations. In rocky locations where excavation proved difficult a shortened central tube was used. In sandy locations liable to erosion, the three-metre pipe marker was installed with witness posts. A total of 1712 monuments of all types were installed to provide the vertical control network of the eastern region.

Vertical control analysis showed that there are several discrepancies in the control points used in the region. In the first place, the heights of the KLM points were actually determined by trigonometric heighting which leads to a lower accuracy, as well the points were refereed to Ras Ghantut datum, while that used by BKS to determine the heights of the BKS points referred to Port Rashid mean sea level datum. The comparison between BKS and KLM indicates a difference between datums of approximately 4 m, which indicates that, it is possible, that there is a further confusion between datums. Secondly, there is also a problem with the way in which the “G” points were established. These points were positioned with GPS during the 1990s, using “B” points as control and deriving transformations using the SKI software. The effect of this is to derive heights that are neither orthometric nor ellipsoidal, but what can be termed “linearly corrected orthometric”: that is, they are ellipsoidal heights that have been converted (implicitly, rather than by design) to orthometric using a model of the geoid that is defined as a plane through the values at the control points. This approach is generally valid over short distances (several hundred metres, or perhaps some kilometres), where to a certain extent the geoid can be modelled as an inclined plane. The evidence here, however, is that it has been done over several tens of kilometres: over these distances, the necessary assumptions break down completely. An examination of the form of the EGM-96 geoid shows that errors of around 0.5 m could be introduced in this way, and possibly larger since EGM-96 is a smoothed model of the geoid. In addition, if any “K” points were used as control, then the errors would be even worse than this. It is therefore necessary to ignore both the “K” points and the “G” points in assessing the quality of the vertical control network, this leaves only BKS points to from the vertical control of the Eastern Region.

3. Relations between Orthometric heights and Geoid
The GPS measured heights are measured from the ellipsoid, therefore, they need to be converted into an orthometric height system. The current methods of converting GPS elevations to orthometric elevations (Acharya and Popp, 1994) are:

To incorporate a priori geoid undulation data in three-dimensional adjustment which holding the benchmark elevations fixed for stations with known values determined by spirit leveling. The minimum number of benchmarks should be four, well distributed through out the region.
Determination of orthometric heights from GPS vector baseline data involves performing a 3-d adjustment without using geoid undulation data. In this method the bench marks elevations are held fixed while using zero values for geoid undulations in a 3- dimensional adjustment. This interpolates the geoid undulation values for the rest of the stations in the region. Here also minimums of four known benchmarks are needed and preferably more than four well distributed to achieve valid results (better interpolation of geoid undulation).
The best method is to compute the actual geoid undulation difference details from gravity anomalies for the desired stations where ever centimeter accuracies are derived.

The ultimate aim for the vertical network is to determine a geoid surface across the region, in such a way that GPS observations can be corrected so that they agree with the orthometric height datum. An initial assessment of problems associated with the geoid in Al Ain region were made by Hansa Luftbild (2001); using the observed WGS-84 coordinates with GPS of a selection of points across the region. From the differences between GPS and orthometric height, the initial values for the geoid separation, N (Fig. 1) was determined. This figure was then compared with the value given by the EGM-96 global Earth model, and the results are shown in Table 1. After correcting for an overall bias (which would be expected since the heights use different datums), the residual variations are shown. However, considering the known accuracy of EGM-96, the type of terrain, and the area covered, these variations are far greater than would be expected

The residuals shown in table.1 are of a magnitude that is consistent with an underlying more detailed structure of the geoid than is modelled by EGM-96. The analysis showed that the “B” points are in an orthometric height system and are sufficiently widely spread to form the basis of an on-going vertical control system. The “K” points are few in number and based on the Ras Ghantut local datum; the “G” points are not in an orthometric height system. In adopting the “B” points as the sole basis for the orthometric height datum, however, an assessment will have to be made of the extent to which “K” and “G” points have already been used to control mapping, and appropriate corrections will have to be applied.

Fig. 1 WGS84 Geoid of the Eastern region of Abu Dhabi Emirates

4. The Earth Geo-potential Model EGM96
The EGM96 model is the result of collaboration between the National Imagery and Mapping Agency, the NASA Goddard Space Flight Centre, and the Ohio State University. Major terrestrial gravity acquisitions by NIMA since 1990 include airborne gravity surveys over Greenland and parts of the Arctic and the Antarctic, surveyed by the Naval Research Lab (NRL), and cooperative gravity collection projects, several which were undertaken with the University of Leeds. These collection efforts have improved the data holdings over many of the world's land areas, including Africa, Canada, parts of South America and Africa, Southeast Asia, Eastern Europe, and the former Soviet Union. In addition, there have been major efforts to improve NIMA's existing 30' mean anomaly database through contributions of various countries in Asia.

Presently the best available model of the Earth is the EGM96 covering the Earth gravity field up to degree and order 360. This corresponds to a spatial resolution of up to 55 km and models the geoid within an accuracy of about 40 cm (global average). For a further refinement of the EGM96 in the Eastern Region, surface point information has to be introduced. A point separation of less than 30 km is needed to cover the short wave length part of the harmonic development of the Earth gravity field (degree 361 . . . 10000, 0.5 m . . . 0.01 m geoid undulation, point separation 30 km . . . 1 km (grid wise)).

Fig. 2 shows the part of the EGM96 model that covers the United Arab Emirates. Proceeding along the 24-degree latitude circle from west to east (from Abu Dhabi to Oman), a steep increase of the geoidal height of about 5 m (from -32.5 m to -27.5 m) occurs over a short distance of less than 100 km.

Fig.2: EGM96 Geoid Model for the United Arab Emirates

The exercise of establishing a local geoid for the Eastern Region of the Abu Dhabi Emirate has been attempted in two steps, first the EGM96 model was compared against the geoid undulations as derived from actual survey data. Secondly the resulting bias was then used to best fit the EGM96 model to the local conditions in an attempt to improve the results, hence transforming the bias corrected EGM96 to a best-fit local geoid.

The initial analysis of the first processing step revealed a difference between Al Ain region, orthometric heights and the corresponding EGM96 heights of the magnitude of 1.82m. EGM96 values were then uniformly corrected by this regional default value. However, the mismatch between bias corrected EGM96 values and actual GPS height minus orthometric heights was definitely more than expected.

An agreement within ?40 or ?50 cm could be expected considering the overall quality of the EGM96 model. In this example, the rms difference between the observed and the EGM96 bias corrected heights was ?1.63 m (without bias correction ?2.45 m), which is 3 to 4 times more than expected.

In the northern part (above 24 deg) the residuals are mainly positive and in the southern part large negative residuals of the bias fit are evident. The extreme values were +1.63m (maximum difference) and ?3.63m (minimum difference). In the eastern mean part of the investigation area the station distribution was denser and showed an inhomogeneous behaviour of the residuals with positive and large negative values. The second processing step of the local geoid determination showed a best fit transformation of the bias corrected EGM96 values to the actual values as derived from local observations. Following the transformation the rms difference between the observed and the EGM96 bias/tilt corrected heights was ?1.33m.

The residuals indicate that, similar systematic errors are inherent in adjacent stations, especially in the northern part of the region. This could be due to the use of different height datum points in the local levelling networks.

Al Ain Town Planning has provided a fundamental component of determining a local geoid that of an accurate homogeneous control networks extending throughout the region. The exploitation of this valuable resource to define the relationship between the heights from GPS surveys and the heights above mean sea level is one of the main advantages of such a resource, hence allowing the various user groups of spatial data to maximize their investment in GPS surveying technology (Hansa Luftbild, 2001). The residuals of each station are shown in table.1. The extent of the surveying activities within the region will be dependent on the uncertainties required within the local geoid model. Greater height accuracies will be required within the developed and developing areas and therefore smaller uncertainties will be required within the local geoid within these areas. Outside these areas a less rigorous determination of the model will be required, which can be updated inline with the ever-expansive survey requirements.

Table 1: Comparison between WGS84 and EGM96 geoids in Al Ain region

5. Gravimetric Geoid determination
The initial step in the computation of a geoid surface that will be used to correct GPS observations is to determine a “pure” gravimetric geoid in a global datum. This will then be tailored to the datum of the existing control points, during which process the long-wavelength errors in the levelling will be absorbed. To determine a gravimetric geoid to an accuracy of 5 cm, the ideal requirements (Cross et. Al, 2002) are:

Use of the EGM-96 for long-wavelength effects.
Observations of gravity anomalies on a 5 km grid across the whole of the region where the geoid is required, and in a buffer zone extending 60 km beyond in each direction.
Provision of a digital elevation model to describe the topography across the region where the geoid is required, at a grid density of 100 m.

In practice, some of these requirements may present some problems in this case, in particular because:

The Eastern Region has borders with countries (such as Oman); in such case an arrangement or permission is required for the acquisition of gravity data.
Digital Elevation Models (DEMs) are not consistently available across the region. Resort would have to be made to globally available models, which are generally at a density of 30” grids (around 1 km) or more better models in the future through missions such as the Shuttle Radar Topographic Mapper.

The exact effect of these omissions is almost as complicated computation as that of the geoid itself, and would require access to appropriate data sets. However, as suggested by cross et. Al (2002) an estimate of geoid can be made as:

For areas of the Eastern Region adjacent to areas where gravity cannot be observed, it is likely that there would be “edge effect” errors that may amount to up to 20 cm. The points could control a component of this where GPS and levelling data are available, but there would still be some short- to medium-wavelength effects that were not modelled.
Acting on the fact that DEMs are not available for the Eastern Region, the error effect would be dependent on the topography. For regions that could be described as gently undulating (perhaps +/- 50 to 100 m with respect to the average trend) the effect would not be too serious, perhaps an additional 2 – 3 cm of error. For regions of rough topography (for example the mountainous features where sharp variations of +/- 500 to 1000 m are encountered) the effect is likely to be more pronounced: around 5 – 15 cm over wavelengths of 5 – 15 km.

Once the gravimetric geoid has been computed, it is then necessary to fit this surface to the locally established Port Rashid datum. This was done by determining some geoid separation values using GPS observations at existing control points as shown in table (1). The fitting process suggested by cross et. Al (2002) involves warping the gravimetrically derived geoid onto the orthometric tie points, by a collocation process. It is possible to do this with too much flexibility and thereby lose the inherent quality of the gravimetric geoid; it is preferable to keep certain “stiffness” in the geoid when this is carried out, and thereby retain the short-wavelength quality of the gravimetric geoid whilst absorbing the long-wavelength trend of the original orthometric network.

The eventual result of the fitting procedure will be that, the new observations (from a “GPS+geoid-model” approach) will be broadly in sympathy with existing height data in areas where “B” points were available in the tailoring process. Some discrepancies may be apparent, as some outliers may be identified at the fitting stage. There will also be (potentially more serious) discrepancies where no tie points were available. It is recommended that in these situations the correct definition of the datum be as given by the geoid model in combination with GPS observations. In other words, the geoid model should be regarded as a better approximation to a true orthometric height system; the archive data should be regarded as containing errors. If this approach is not adopted (that is, if the existing height datum based on a mixture of “K”, “G”, and “B” points is retained) then it is likely to lead to long-term problems,

Observation of gravity anomalies across the Eastern Region and – where access permits – across a 60 km wide edge zone, are to be made on a grid at 5 km spacing, although precise adherence to a grid pattern is not essential. Observations should be tied into international gravity networks and be of an accuracy that is compatible with geoid determination. Then compute a gravimetric geoid across the Eastern Region, using: the observed gravity anomalies; a high accuracy global model of the geoid such as EGM-96; and extraction of DEMs from other data sources. Fit the gravimetric geoid to the GPS/orthometric points supplied to derive a GPS correction surface. The fitting procedure is to be the equivalent of a collocation-type approach with a correlation length in the region of 50 km. Final product should be accompanied by an assessment of the accuracies achieved, both in the gravimetric geoid and in the fitting procedure, accompanied some standard quality control data as the details of rejected data and so on. The final product should be in the form of data files for the gravimetric geoid and the GPS correction surface, as well as a software routine that is able to interpolate the values at given locations in an interactive mode. An area where mapping exists was examined, but no tie points have been used to control the geoid surface (for example, where heighting has been carried out with respect to “G” or “K” points). Evaluate the discrepancies between heights now defined by “GPS+correction surface” and archive heights. If these discrepancies are of a magnitude whereby problems are likely to be induced, then a once off datum shift to the archive data is to be carried out, in smoothly varying steps.

6. Conclusions
The paper showed that the basis for the orthometric heights in the Eastern Region should be in consistent with the control points obtained by BKS, which are based on the vertical datum of Port Rashid of Dubai. The mean accuracy of the adjusted BKS vertical network is given as 9mm. Vertical control analysis showed that there are several discrepancies in the control points used in the Eastern region. In the first place, the heights of the points obtained by KLM referred to Ras Ghantut datum were actually determined by trigonometric heighting which leads to a lower accuracy, as compared to the vertical control points obtained by BKS. The comparison between BKS and KLM indicates a difference between datums of approximately 3 m.

As much of the existing vertical control has been lost due to destruction or disturbance, and because the network was not dense enough to support use of GPS to derive elevations, a precise geoid model should be determined. The study recommended that Al Ain Town Planning have to initiate a project for height determination based on gravimetric determination of geoid.

References

Abdalla, K. A (2003). Datum Transformation and Geoid Determination effects in the Quality of Geodetic Control in Developing Countries. Presented in XXIII General Assembly of the International Union of Geodesy and Geophysics, Sapporo, Japan, Nov. 2003.
Bishwa Acharya and Richard Popp (1994). Project Planning for a GPS Geodetic Control Network> Surverying and land Information Systems> Vol. 54, No. 2, pp 69 – 75.
Fashir, H. H., Salih, A. B. and Abdalla, K. A. (1989)."The Transformation Between the Doppler Co-ordinate system and the geodetic Co-ordinate system in Sudan". Published in Australian Journal of Geodesy, Photogrammetry and Surveying, Australia.
Hansa Luftbild (2001). Updating of Topographic and Utility Maps, Al Ain Primary and Secondary Geodetic Network. Adjustment Report, Al Ain Town Planning Department.