Surveying Using Global Navigation Satellite Systems

Transcription

1 Surveying Using Global Navigation Satellite Systems This document has been designed to provide details of GPS technology to enable the practising surveyor to integrate GPS techniques into their surveying operations. Each section of Surveying Using Global Navigation Satellite Systems has been developed, where possible, to be a standalone learning unit. The sections can be read in any order. Each section contains links to other sections. This enables the surveyor to easily access specific pieces of information which may be of interest. Surveyors who are new to GPS technology are directed towards the introductory sections and the specific surveying sections (cadastral, engineering, control and geodetic). The theoretical sections (control stations, receiver specifics etc.) are designed to provide a basis for the recommendations made throughout other sections. This document will support the surveying community in Victoria by enabling the use of GPS technology for surveying purposes. Prepared by RMIT University, Department of Geospatial Science for Surveyor-General Victoria 2003, Minor amendments 2006.

2 Table of Contents SURVEYING USING GLOBAL NAVIGATION SATELLITE SYSTEMS... 1 TABLE OF CONTENTS... 2 GLOBAL NAVIGATION SATELLITE SYSTEM (GNSS)... 5 Introduction... 5 THE GLOBAL POSITIONING SYSTEM (GPS) - AN OVERVIEW... 5 An Overview... 5 History of Satellite Positioning - The Transit System... 6 The Global Positioning System... 6 Basic Positioning Concept... 7 Timing Reference System... 8 Coordinate Reference System... 9 POSITIONING USING THE GPS SATELLITES Introduction Absolute Positioning...11 Relative Positioning Differential GPS Navigation (DGPS) Carrier Phase-Based Relative Positioning Classic Static Baselines Rapid Static Baselines Post-processed Kinematic Baselines Real Time Kinematic Relative Positioning Processing Techniques Linear combinations Processing Data Differential Correction Approach Post-Processing Real Time Differential Correction/Measurement Sources DESIGNING GPS SURVEYS Accuracy Obstructions Length Of Baselines Occupation Time...22 Recording Rate Measurement Redundancy Satellite Geometry Control Requirements Surveying Technique Personnel Management BEST PRACTICE GUIDELINES FOR GPS SURVEYS Introduction Guideline Objectives The ICSM Guidelines Other GPS Considerations for Victoria THE NAVSTAR SATELLITES Introduction Description Types Of Satellite The Constellation Satellite Orbits Calculating Satellite Position - The Ephemeris Signal Structure The Broadcast Signal Signal Access - Positioning Services GPS Modernisation THE CONTROL STATIONS

3 Introduction Control Stations Master Control Station Australian Fiducial Network (AFN) Victorian GPS Network (GPSnet) THE GPS RECEIVER Receiver Components The Antenna and Antenna Pre-Amplifier The Radio Frequency Section Signal Tracking Loops Microprocessor Power Supply Data Storage User Interface OBSERVATION TECHNIQUES Introduction Static Surveying Rapid Static Surveying Stop And Go Kinematic Surveying Continuous Kinematic Surveying Real Time Kinematic (RTK) Surveying Initialisation Techniques COORDINATES AND GPS Introduction The Spheroid (or Ellipsoid) The Geodetic Datum Geocentric Datum of Australia (GDA 1994) Heights and GPS...65 Control Requirements for GPS Surveys LOOP CLOSURES Introduction Checking Baselines Observed In Multiple Sessions Internal Accuracy NETWORK ADJUSTMENT Introduction Minimally Constrained Adjustment Constrained Adjustment Error Ellipses Independent Baselines THE GPS OBSERVABLES Introduction The Pseudorange Observable The Carrier Phase Observable Survey Receiver Measurements Selecting An Appropriate Observable GPS ERROR SOURCES Clock Errors Satellite Clock Error Receiver Clock Error Satellite Orbits Selective Availability Atmospheric Errors...83 Troposphere Ionosphere Multipath Antenna Phase Centre Measurement Uncertainty HOW TO Introduction Measuring the Antenna Height

4 Performing Static Surveys Performing RTK Surveys TROUBLESHOOTING Introduction Problems Transferring Data to a Computer The Processor Generates Float Solutions GUIDELINES FOR CADASTRAL SURVEYING USING GNSS Introduction Validation of Equipment Legislation Validation Methods A Zero Baseline test (All receivers) A High Accuracy GPS Test Network (Static Techniques) A Coordinated RTK/Kinematic Test Site An EDM baseline test (Static and RTK Techniques) Additional Validation Considerations Surveys Determination of Survey Datum Field Survey...99 Connection of Cadastral Surveys to MGA Additional Survey Considerations Measurements and Dimensions Bearings and Lengths Note: Coordinates Heights Classification and Accuracy of Surveys Independent Checks Reporting ENGINEERING SURVEYING Introduction The Height Component Deformation Surveys Construction Surveys CONTROL SURVEYING Introduction Observation Technique Receiver Type GEODETIC SURVEYING Introduction Observation Technique Receiver Type Multiple Occupations MIXING RECEIVER TYPES Introduction Mixing Receivers from the Same Manufacturer The RINEX Format Real Time Considerations INTERPRETING BASELINE SOLUTIONS Introduction Fixed Versus Float Solution The Ratio Value Residual Graphs The Variance Factor Solution Standard Deviation GLOSSARY OF GPS TERMS ACRONYMS USEFUL LINKS

5 Global Navigation Satellite System (GNSS) Introduction The Global Navigation Satellite System (GNSS) is a term used to describe a group of satellite based navigation systems that provide positional information on or near the Earth s surface. At present GNSS consists of the American controlled Global Positioning System (GPS) and the Russian controlled Global Orbiting Navigation Satellite System (GLONASS). Plans are currently underway for a third GNSS to be launched, referred to as GALILEO, whose primary purpose is to provide positional information for European users. It is expected that GALILEO will be fully operational by the year Currently GPS is the most widely utilised GNSS system, finding many applications within the areas of surveying, navigation and recreation. With the break-up of the USSR, GLONASS, although still existing, now operates at a less then nominal status and consequently is less frequently used. Due to similarities in the operation of the different GNSS systems, the survey procedures and techniques described for GPS in the following documents can be considered relevant to other GNSS navigation systems. The Global Positioning System (GPS) - An Overview An Overview History Of Satellite Positioning - The Transit System The Global Positioning System Basic Positioning Concept Timing Reference System Coordinate Reference System An Overview The NAVSTAR (NAVigation Satellite with Timing And Ranging) Global Positioning System (GPS) is a military controlled venture designed for positioning, navigation and timing purposes. Although not designed for surveying, the use of interferometric techniques have enabled surveyors to use the satellite signals to great effect. The first GPS satellite was launched on the 22 nd February, 1978 and became operational on the 29 th March, almost one month later. More than twenty years since the first satellite launch, GPS has become one of the most widely used systems for marine navigation, aviation, vehicle tracking and management, recreational activities and surveying. The use of GPS is seen as an extension, not a replacement, to the surveyors range of equipment which, when combined with a total station, level and even a steel tape, enable the surveyor to use the most efficient positioning tool available for their client. In the modern business environment, it is often difficult to keep up to date with changing technology. In the surveying industry, changes in microchip technology have seen electromagnetic distance measuring equipment (EDM), electronic total stations (ETS), digital levels, computer aided drafting and design (CADD) software, computer based engineering and mapping software, and now GPS, all become a reality in the surveying profession. Keeping abreast of the rapid changes in technology can be a full time job in itself, let alone trying to stay ahead of the competition in the highly competitive surveying industry. Use of the Global Positioning System requires specialised equipment, data collection techniques and data processing algorithms. This document aims to provide a theoretical and practical foundation for surveyors as they try to embrace GPS technology and integrate use of GPS equipment into their daily business operations. 5

6 The Global Positioning System (GPS) - An Overview History of Satellite Positioning - The Transit System The use of satellites for surveying purposes first became a practical reality with the development of the Transit system by the United States Navy. The Transit system used Doppler measurements from seven satellites arranged in polar orbits to determine position and trajectory. The orbits of the seven satellites form a circular birdcage effect with an orbital altitude approximately 1100km above the surface of the Earth. The system was used for geodetic surveying applications in the 1970 s and 1980's. However, due to the limited number of satellites, positioning was performed by observing for long periods. The low altitude of the satellites also meant that satellites were not visible at all times and gaps of 90 minutes between satellite passes had to be contended with. Typically, several satellite passes were required to position marks accurately on the ground. Another limitation of the Transit system was also caused by the low orbital altitude of the satellites. The 1100km orbital altitude resulted in large forces, which are difficult to model, disturbing the satellite orbits. As a result, the accuracy of position estimates was not as high as ideally required for many applications. Regardless, the development and use of the Transit system provided a solid foundation from which to develop the Global Positioning System. The primary limitations of the Transit system that have been rectified in the development of GPS include the ability to now observe 24 hours per day and to coordinate features to a higher accuracy. The former has been achieved by increasing the number of satellites, the latter by placing the GPS satellites in significantly higher orbits than the Transit satellites. The Global Positioning System (GPS) - An Overview The Global Positioning System The Global Positioning System (GPS) is a space based radio-navigation system designed to satisfy the requirements of the United States Department of Defence (DoD). The system consists of satellites and their signals, a series of control stations which monitor and maintain the satellites, GPS receivers which are capable of recording the satellite signals and users who coordinate themselves using observation techniques designed to achieve certain levels of accuracy. Many texts discussing the use of GPS refer to these components as the three systems; space, control and user. In order to use the GPS satellite transmissions for surveying purposes, a number of concessions must be made. First, a receiver capable of precise measurements is required. Such a receiver may cost in excess of tens of thousands of dollars as compared to a lower accuracy receiver, which may cost several hundred dollars. Second, more than one receiver must be used. The use of more than one receiver is termed relative positioning and is mandatory for all surveying applications of GPS technology. Third, specially designed observation techniques must be used. This is to facilitate resolution of an integer bias that exists in the precise portion of the GPS signal. Finally, sophisticated mathematical algorithms are required to convert the satellite measurements to the user position. In addition to these requirements, the system is only useable in locations with a clear, unobstructed view of the sky. This obviously restricts the use of GPS in urban areas and for underground work. There are, however, a number of advantages of surveying using the GPS satellites. The system has been designed to provide continuous satellite coverage, which can be used at all 6

7 times of day (and night). This provides additional flexibility when it comes to designing surveys. The satellites transmit L-band microwave signals, which are not significantly affected by poor weather conditions. As a result, GPS equipment can be used in all types of weather. The system is also global and can be used in any location. Perhaps the greatest advantage of GPS is that the two receivers, required for relative positioning (commonly termed the reference and rover in kinematic applications), can be separated by several tens of kilometres and do not require line of sight intervisibility. This enables surveyors to coordinate marks to survey accuracy over distances which previously may have required several days of traverse measurements. It is this feature that makes GPS so attractive for control survey work. No longer do marks need to placed on top of difficult to access hills, they can now be placed where they are needed. The use of GPS for control survey work is rapidly becoming routine. The use of GPS technology introduces several concepts which are not used in terrestrial surveying applications. However, surveyors with a basic understanding of geodetic positioning will have little trouble in extending that positioning scenario to satellite use. Surveying with GPS is little more than a distance resection problem with the satellites acting as, albeit moving, control points. The Global Positioning System (GPS) - An Overview Basic Positioning Concept The basic positioning concept used by the Global Positioning System is illustrated by a diagram comprising a satellite which orbits the Earth and continuously transmits signals, the Earth with its geocentre defined as the centre of mass, and a user on the Earth with a receiver capable of interpreting the broadcast satellite signal. The position of the user can be represented by the vector from the Earth s geocentre to the receiver on the Earth s surface. This vector is threedimensional and is unknown. The vector from the Earth s geocentre to the satellite defines the three dimensional position of the satellite and is determined using the ephemeris transmitted as part of the satellite signal. The third vector is between the user on the Earth and the satellite. The magnitude of this vector, in other words, the one-dimensional distance from the receiver to the satellite, is measured by the receiver. If one satellite is observed, the user position lies somewhere on a sphere with radius equal to the distance to the satellite. If a second satellite is simultaneously observed, the user position lies on a circle defined by the intersection of two spheres with radii equal to the two measured distances. If a third satellite is introduced, the receiver position can be determined uniquely by the intersection of the three spheres with radii equal to the measured distances to the satellites. The use of three satellites simultaneously facilitates calculation of the three dimensional position of the receiver. Therefore, GPS is more than just a two-dimensional positioning system, height information is also computed. 7

8 The development of the GPS signal structure required the system to be passive in order to protect the position of military users. To facilitate this, the satellite signals are generated by precise atomic clocks aboard the satellites. The user on the Earth utilises a receiver which generates internal signals, however, a less precise quartz crystal clock is used. The distance between the satellite and receiver is measured by aligning the satellite signal and the internally generated signal. The measurement relies on the satellite and receiver clocks being synchronised. A timing error of one microsecond ( seconds) results in a distance error of approximately 300m. Therefore, the measurement of time is a vital component of the GPS system. To eliminate the timing error from the computed receiver position, a fourth satellite is observed. This enables the three position components and the mis-alignment of the satellite and receiver clocks to be determined. All surveying applications of GPS technology require a minimum of four satellites to be simultaneously observed to obtain position estimates to a suitable accuracy. The Global Positioning System (GPS) - An Overview Timing Reference System The maintenance of precise time is a key element in the effective use of the GPS system. The satellites generate signals which are referenced to a specific epoch. The time is kept by atomic clocks, or oscillators, aboard the satellites. These clocks are also used to generate the signals transmitted by the satellites. The receivers used by surveyors on the Earth also house clocks which generate replica versions of the satellite signals for internal comparison purposes. To be able to use the satellite signals effectively, the time component of the measuring process must be regulated to a common time frame. The synchronisation of time in the entire GPS positioning process is paramount, therefore, a brief description of the timing reference system used by GPS is warranted. A complete description of the GPS timing reference can be found at however, the underlying basis of the system is referenced to the second as defined by an atomic time scale. The United States Naval Observatory (USNO) monitors GPS time (GPST) as defined by the oscillations of an atom. The GPS satellite clock correction parameters are developed to correct the atomic clocks on board the satellites to this time frame. Universal Time Coordinated (UTC) is another atomic time scale, which is modified by inserting periodic leap seconds to keep UTC close to Universal Time (UT). UT is governed by the Earth s rotation and the position of the sun (this is wrist watch time). Therefore, the offset between GPS time and UTC will alter by one additional leap second when the periodic adjustments are made to UTC. An adjustment is expected at the end of June, 1997 which will result in GPS and UTC differing by 12 seconds. The time in the GPS system is referenced by the number of seconds in one week and a week number. GPS time was officially initialised at zero hours on January 6 th, At this epoch, the difference between GPS time and UTC was zero. The week number has incremented 8

9 every 604,800 seconds since this time. In the GPS signal specification, the week number is stored as a 10 bit integer value. The maximum value that the week number can have for an integer of this size is 2 10 = 1024 weeks.. This value was reached at the end of the last millennium, at which point the GPS week number was reset to zero (on the evening of 21 August 1999 / morning of 22 August 1999). For surveying applications, the maintenance of GPS time is a function performed by the GPS receiver and processing software. As a result, there is little intervention required by the user with regards to the elements of time. One point of interest regarding the time scales, the GPS satellite clocks are essentially keeping GPS time as defined by the USNO. After the clock correction parameters are applied, all satellites are synchronised to this highly accurate time frame. The GPS receiver uses an inexpensive quartz crystal oscillator, however, the observation of at least four satellites is required to compute the three dimensional user position and the receiver clock synchronisation error. Therefore, once the satellites are tracked and position computed, the receiver clock has effectively been transformed into an atomic clock as it is now synchronised with the satellite clocks. This is the manner in which precise time can be transferred using GPS equipment. The Global Positioning System (GPS) - An Overview Coordinate Reference System There are two reference systems employed by the Global Positioning System: 1. the atomic GPS time reference as maintained by the United States Naval Observatory (USNO), and 2. the coordinate reference system defined by the National Imagery and Mapping Agency (NIMA) who were formally the Defense Mapping Agency (DMA). The coordinate reference system used by the GPS system is the World Geodetic System 1984 (WGS84). This system is a geocentric based coordinate system with the origin of the defining spheroid located at the Earth s centre of mass. The spheroid has a semi-major axis of 6,378,137.0m and an inverse flattening of The semi-minor axis is computed as 6,356, m. These parameters are the same as those used to define the Geodetic Reference System 1980 (GRS80) spheroid. The semi-major axis and flattening define the shape of the WGS84 spheroid. The centre of the spheroid is fixed to the Earth s centre of mass. The direction of the three Cartesian axes need to be defined to constrain the spheroid in space. The Z-axis is defined as passing through the Conventional Terrestrial Pole (CTP) at epoch as defined by the Bureau International de l'heure (BIH). The X-axis is defined as being the intersection of the WGS84 reference meridian plane and the plane of the CTP's equator. The WGS84 reference meridian passes through Greenwich and is specifically defined by the BIH zero meridian at epoch The Y-axis completes a right handed, Earth Centred, Earth Fixed (ECEF) orthogonal coordinate system, measured in the plane of the CTP equator, 90 degrees east of the X-axis. 9

10 The position of the satellite at the instant of measurement is required in order to compute the unknown receiver position. The broadcast ephemeris is a set of orbital parameters that enable the Cartesian coordinates of the satellite to be easily computed. The resultant position of the satellite is referenced to the WGS84 coordinate datum. Therefore, the position estimates derived from GPS measurements are also referenced to the WGS84 datum. In order to obtain coordinates in other systems, such as the Australian Geodetic Datum (AGD), users must transform their GPS coordinates from WGS84 to the new system. To be more compatible with GPS, Australia introduced a new geodetic datum, termed the Geocentric Datum of Australia 1994 (GDA94) in the year The origin of this new datum almost coincides with the origin of WGS84, which means user can directly obtain GDA94 coordinates from their GPS receivers. This eliminates the need to perform transformation computations to integrate GPS observations to the GDA94 coordinate system. More information regarding GDA 94 can be obtained from the Geoscience Australia web site (formally the Australian Surveying and Land Information Group (AUSLIG)). The Intergovernmental Committee on Surveying and Mapping (ICSM) web page also contains detailed information regarding GDA94. The Global Positioning System (GPS) - An Overview Surveying Using Global Navigation Satellite Systems 10

11 Positioning Using The GPS Satellites Introduction Absolute Positioning Relative Positioning Post-Processing Real Time Differential Correction/Measurement Sources Introduction The use of the GPS satellites for positioning applications is based upon the fundamental concept of trilateration. In many terrestrially observed geodetic networks, new points have been coordinated by observing distance measurements to three fixed control stations. Directions to the same stations are also observed in terrestrial surveys (triangulation). Surveying with GPS is similar in concept to trilateration. Distances are observed to satellites, using either the code or carrier, and the receiver coordinates computed using ephemeris parameters. The major difference between the terrestrial implementation and the GPS implementation is that the satellite based solution has control points which are moving in orbit. The distance to the satellite is provided by the pseudorange measurement which is derived from the binary code modulated on the carriers, or from the carrier phase measurement derived from the carrier signal itself. The use of a single receiver, making measurements of the pseudorange, is referred to as absolute positioning. For surveying purposes, the accuracy demanded by cadastral, engineering and geodetic applications requires use of the more accurate carrier phase observable. The carrier phase provides millimetre accuracy ranges, however, they are biased by an unknown number of whole carrier cycles, termed the integer cycle ambiguity. To successfully use the GPS signal for surveying, the integer cycle ambiguity terms need to be determined. There are also a number of errors that affect the GPS measurements. Many of these errors are spatially correlated and can be removed or minimised by observing using multiple receivers. The use of multiple receivers is termed relative or differential positioning and must be used for all surveying applications of GPS technology. One receiver remains stationary and is termed the reference or base receiver. Other receivers may also be stationary (static surveying), or may move to points of interest (kinematic surveying) and are termed rover receivers. The reference/rover, base/rover terminology is used almost universally by all survey equipment manufacturers. Absolute Positioning Absolute positioning involves the use of a single receiver calculating its position in a previously defined coordinate system using the pseudorandom noise code in the GPS signal. Absolute positioning cannot provide the required accuracy for most surveying applications, however it may be suitable for exploration, navigation, small-scale mapping and military applications. With the removal of Selective Availability, a single receiver in absolute positioning mode can now obtain positional accuracies within the range of meters. As part of the Standard Positioning Service (SPS) absolute positioning is available to anyone with a single handheld or geodetic receiver. Relative Positioning Higher accuracies can be achieved if the position of an unknown point relative to the absolute position of a known point can be measured. This can be achieved by measuring the baseline components ΔX, ΔY and ΔZ between these points. This requires the use of more then one receiver and is known as relative positioning. By using two or more GPS receivers tracking the same satellites simultaneously, it is possible to remove many systematic errors and improve the relative position estimates to metre- or millimetre-level. The relative difference in 11

12 coordinates between the receivers can be determined using a number of techniques which include the following: Differential GPS Navigation (DGPS) The use of the pseudorange observable for relative positioning is often referred to as Differential GPS or DGPS. In DGPS mode, one receiver is located on a point with known coordinates (the base or reference receiver) while another receiver is located on uncoordinated points of interest (the rover receiver). At each epoch the base receiver uses the measured pseudorange to compare its calculated absolute position with its known coordinates. This difference forms the source of a correction which is broadcast to the rover receiver to apply to its own determined position. This is the essentially the basis of such commercial real time DGPS services such as OmniStar or LandStar. DGPS can provide positional information to decimetre level accuracy and is frequently used in asset mapping and vehicle location applications. In most applications DGPS is achieved in real-time however post processing is also possible. More information regarding DGPS can be obtained at the following website: Carrier Phase-Based Relative Positioning To achieve the high level of accuracy required for most surveying applications it is necessary to use the carrier phase observable. Relative positioning using the carrier phase observable can provide positional information to centimetre - millimetre level accuracy if the appropriate survey techniques are used. These techniques include the following: Classic Static Baselines The classic static positioning technique utilises multiple non-kinematic receivers occupying both coordinated and uncoordinated points for extended periods of time (e.g. one hour for an engineering control network or 5 days continuous tracking for geodynamic surveys) If these receivers are separated by distances less then 10 km, then the signals they are tracking will be equally effected by spatially correlated errors such as the ionosphere. Mathematically differencing measurements collected by these receivers will remove the effects of these errors and allow very accurate positioning to be obtained. This technique is known as differencing and is why relative positioning is sometimes referred to as differential positioning. Rapid Static Baselines The major disadvantage with classic static GPS positioning is the extended occupation sessions required, which may be uneconomical. The rapid static surveying technique was developed in an attempt to improve the efficiency of the static survey procedure. The length of the occupation period is less than that required for static surveys, but the final accuracy is also less. Occupation times as short as ten minutes are often sufficient to coordinate points. Accuracies in the order of a few centimetres are possible. Post-processed Kinematic Baselines The continuous kinematic survey and the stop and go techniques are almost identical in field procedure. Both techniques require a stationary base station receiver tracking continuously and also a rover receiver being used to determine necessary details and features. Data from both a downloaded at the end of the survey and then post processed to obtain a data set that has been reduced to the base station. In continuous kinematic surveys the rover receiver must track satellites without interruption, allowing the position of the antenna to be estimated (during post processing) at each 12

13 measurement moment, termed epoch. With care and good checking techniques it may be possible to achieve a positional accuracy of a few centimetres. However, this technique is now mostly used for detail surveys over very large open area (such as beaches, paddocks and open undulating country) where the rover is attached to vehicle or other continuously moving platform. The stop and go kinematic surveying technique is extremely effective in coordinating closely spaced features. The features of interest are stationary and are occupied briefly, generally for less than one minute. Satellite tracking is maintained throughout the survey to achieve positional accuracy of a few centimetres for the points of interest. If cycle slips are present, the survey needs to be re-initialised. Real Time Kinematic The real time kinematic (RTK) GPS technique utilises a data link, usually in the form of a radio, to transfer phase corrections acquired at the reference receiver (set up on a survey mark) to the moving rover receiver. The rover receiver then applies these corrections to its measurements, which allows centimetre level positional information to be achieved in real time. This capability enables surveyors to check coordinates in the field, ensure surveys are being performed successfully and facilitate establishment of features at pre-determined locations (setting out). More information regarding the procedure and considerations required when conducting relative positioning is present in the SP1 Publication produced by the Intergovernmental Committee on Surveying and Mapping (ICSM) Positioning Using The GPS Satellites Relative Positioning Processing Techniques Linear combinations When using the relative GPS technique (over short baselines) there exists a high level of correlation among the signals received at independent sites simultaneously tracking the same satellites. By taking advantage of this physical model many of the correlated errors, such as orbit, clock, and atmospheric biases, can be reduced by forming linear combinations of either the code or phase measured ranges. Differencing the ranges in various linear combinations will lead to the computation of precise baselines. The linear combinations used in GPS reductions are: Single Differences: between-epoch between-receiver between satellite Double Differences: receiver-epoch receiver-satellite Triple Differences: receiver-satellite-epoch As the range derived from carrier phase measurements produces more precise values than that of the pseudo-range measurements, the linear combinations of the carrier beat phase observable will be presented. Differences of the original observations allow it to eliminate or reduce some biases. 13

14 The first linear combination can be constructed from the arithmetic difference between the two simultaneous phase measurements from a single satellite s 1 to two independent receivers r 1 and r 2 (at the single epoch t 1 ). This is known as the single-difference observable. s 1 r 1 r 2 This single difference operation has removed the satellite's transmitted phase. The offset and linear drift of the satellite clock has also cancelled. It has also reduced the errors caused by path effects if the receivers are sufficiently close together (<20 km) because the signal paths will be almost the same. Thus, any bias which is equivalent at each receiver has been removed leaving a residual noise derived from any uncorrelated errors. This will include orbit uncertainties, and atmospheric delays. The single difference however is still contaminated by receiver clock errors. The difference between two single difference observations, constructed from simultaneous phase measurements to satellites s 1 and s 2 and receivers r 1 and r 2 is the so-called doubledifference phase observable. This eliminates errors in both receiver and satellite clocks. s 1 s 2 r 1 r2 The double difference phase observable is formed by differencing the carrier phases measured simultaneously by a pair of receivers tracking the same pair of satellites ie by differencing 2 single difference observations. The integer ambiguity is however still present. 14

15 Since the integer ambiguity does not vary with time, the difference between two double differences observed at two epochs t 1 and t 2, will remove the unknown ambiguity except during loss of lock. This is known as the triple difference. t 2 s 1 t 1 t 1 s 2 t 2 r 1 r 2 Because the triple-difference observable is independent of the integer phase ambiguity (ie the phase ambiguity is eliminated) it can be used to detect any cycle slips which contaminate the single and double difference ambiguity parameter. Also, as tropospheric refraction does not vary rapidly with time then triple differencing considerably reduces this effect. Processing Data As there are usually a multiple number of phase measurements in any one static GPS survey, the baseline components derived from the differenced data set are determined by a least squares approach. Theoretically, any one of the differenced observation sets will yield the same solution, provided all correlations are appropriately computed. However, although the single, double, and triple differences progressively eliminate any physical correlations between observed phase data, the stochastic correlation in the least squares solution increases. Differencing is used for removing or minimising biases, unknowns and errors. Whilst some errors can be essentially eliminated by using differenced phase observations, the primary disadvantages are: that there is a trade-off between removing errors by differencing and losing precision by reducing the number of observations i.e. 2 phase observations are used for one single difference, 4 phase observations are used for one double difference and 8 phase observations are used for one triple difference that the noise factor is increased with each difference made the sets of observations become correlated and then these correlations must be modelled as part of the processing of the observations. Theoretically, there should be an exact number of whole wavelengths (ambiguities) between the receiver and the satellite i.e. the ambiguities the software solves for should be integers. Unfortunately, at this stage of the processing, the values for the ambiguities are not integers. 15

16 This solution is referred to as a double-differenced ambiguity free solution because the ambiguities are solved as real values and are not fixed to integer values. It is generally considered that the best baseline solution would be obtained from a double difference solution where the exact number of integer wavelengths was known between receiver and satellite i.e. a double-difference ambiguity fixed solution. The determination of the exact number of integer wavelengths is termed ambiguity resolution. Ambiguity resolution is a critical process for high precision GPS surveys but it requires very careful strategies. Double-differenced ambiguity fixed solutions can be more than ten (10) times more precise than ambiguity free solutions. However, with long observation times (>12 hours) this can be more like a factor of 3 in east-west and less in north-south. However it must be stressed that if the wrong ambiguities are selected, solutions may be worse than ambiguity free or even triple difference solutions. Thus if the integer ambiguity can be correctly estimated, the double difference solution may produce the better solution. For long observation sessions e.g. 24 hours a double difference solution with real valued ambiguities gives similar results. However, if the ambiguity cannot be resolved adequately the triple difference solution may need to be adopted, as it is free of the unknown ambiguity term. The residual noise, which inhibits ambiguity resolution, is a function of the baseline length. This is largely due to the spatial and temporal nature of the atmosphere. Thus longer baselines, which have receivers in differing atmospheric conditions, will usually be computed by the triple difference data set. The chances of ambiguity resolution is increased when: baselines are short e.g. less than 100 km observation sessions are long processing strategies are undertaken using various L 1 and L 2 linear phase combinations Differential Correction Approach The differential correction approach differs from the double difference approach in several respects. In this approach, corrections to the carrier phase measurements are generated at the reference receiver. These carrier phase range corrections are then applied to the equivalent phase measurement observed at the rover receiver. The corrections are computed based on the knowledge of the reference station coordinates and the position of the satellite. This enables a true range to be determined. The differential correction is the difference between this true range and the measured range, with the measured range corrected for the receiver clock error. The implications of the differential correction approach are that the reference station data is processed separately from the rover station data, unlike in the double difference approach where both receiver s measurements are processed simultaneously. In a real time application utilising this approach, the reference station computes corrections and these corrections are transmitted to the rover receiver. The rover receiver then applies the received corrections to the measured carrier phases. The double difference approach, applied in a real time environment, is different to this approach as the reference station measurements are transmitted to the rover receiver, where all measurements are combined 16

17 and processed at the rover. The majority of manufacturers employ the double difference approach, however, several recent products are using the differential correction approach. Both techniques give commensurate results. The differential correction approach requires estimation of the one-way ambiguity between the satellite and receiver. As the measurements are not differenced between receivers, as in the double difference approach, there are n ambiguities where n is the number of satellites. Post-Processing In many surveying applications, the coordinates of the points being collected are not required while in the field. Examples of such surveys include topographic surveys where a digital terrain model (DTM) is to be generated for further analysis and geodetic surveys where angles and distances are adjusted using least squares techniques to provide best estimate coordinate values. Many GPS surveys are performed in a similar manner. Data is acquired at both the reference and rover receivers and stored on internal memory chips, memory cards or external data collectors. The measurements from all sites are then combined on a computer for processing. This method of data acquisition followed by data processing is termed postprocessing. The disadvantage of post-processing is that the success of the survey and the computed coordinates are not available in the field. This is restrictive for many applications such as subdivision set out. There are a number of advantages of post-processing however. All measurements are simultaneously available and can be processed in a variety of ways. Different observable combinations and multiple passes through the data enable the most reliable coordinate values to be determined based on the acquired data set. The alternative to post-processing, real time processing, requires additional equipment in the form of a communications link. This aspect of real time surveying can be restrictive depending on the type of communication link adopted. Post-processing applications do not suffer from communication link restrictions. This translates to less equipment and, hence, less cost. Perhaps the largest drawback of the post-processing approach is the difficulty in assessing the correct amount of data to collect to ensure the integer ambiguities are resolved. Users are advised to adopt a more conservative approach when performing such surveys because insufficient data acquisition usually results in the survey being re-observed. Therefore, postprocessed techniques suffer from a lack of productivity when compared with real time techniques. Surveyors should use their best judgment when selecting an appropriate technique for a particular survey. The Intergovernmental Committee for Surveying and Mapping (ICSM) provide guidelines for determining the required technique when performing a GPS survey (SP1 Publication). In Australia it is possible to post-process GPS observations via the Internet. Geoscience Australia, offer the AUSPOS online GPS Processing Service which provides users with the facility to submit dual frequency, geodetic quality, GPS RINEX data and receive rapid turnaround coordinates in the International Terrestrial Reference Frame (ITRF) and Geocentric Datum of Australia (GDA) This makes it possible to perform GPS surveys in the field using a single rover receiver. 17

18 Positioning Using The GPS Satellites Real Time The alternative to performing surveys which are processed after the data has been collected, post-processed, is to compute the coordinates of all features as they are being collected in the field. This is referred to as real time surveying and may be either relative positioning using pseudorange corrections (DGPS) or phase corrections (RTK). In a post-processed application, measurements are acquired at both reference and rover sites and merged in a computer for processing. In a real time implementation, a communication link, or data link, is used to transfer either raw measurements or differential corrections from the reference receiver to the rover receiver. The rover receiver then uses the transferred information to compute the rover coordinates in the field. The status of the integer ambiguity resolution process is also available, therefore, real time surveys can be performed with optimum efficiency as the minimum amount of data required to resolve ambiguities can be ascertained. The reference station information is transferred to rover receivers via a data link. This data link may be of the form of cellular telephone, communications satellite or high power UHF radio. Each of these forms of communication can be restrictive due to associated cost or licensing requirements and, therefore, are not popular for surveying applications. Manufacturers provide alternative solutions comprising lower power radios, generally, using FM band frequencies. A number of such devices are available. There is one major limitation of many of the data links sold by receiver manufacturers. To avoid complicated licensing in a variety of countries, most manufacturers provide radios which are low power, to the point where they do not require licensing in most countries. The limitation of this as far as the surveyor is concerned is that the radios only operate on line of site. This defeats one of the major advantages of using satellite technology for surveying. In addition, ranges of only a few kilometres can be surveyed using such radio technology. These two problems can partially be alleviated by employing repeater radio sites. A repeater does nothing more than transparently transfer incoming signals. Repeater sites can be placed on hill tops or in strategic locations to provide better radio coverage for real time applications. If higher power radios are available and have been licensed with the relevant authority, they are generally preferable for real time applications. In Victoria, radios must be licensed through the Australian Communications and Media Authority (ACMA). When performing real time surveys, it is not essential to store the measurements acquired by the satellites. The coordinates of the rover receiver have been computed and can be stored with appropriate user entered descriptors. This reduces the amount of data that needs to be stored when performing GPS surveys. Many of the data collectors used to control real time surveys facilitate the representation of the computed rover position as a total station observation of azimuth, slope distance and vertical angle between the reference and rover receivers. This eases the interpretability of real time solutions. If data storage is not a restriction, users are advised to store the measured pseudoranges and carrier phases for later post-processing. Post-processing is more rigorous than real time processing as the measurements can be manipulated in several ways and data can be 18

19 passed through on more than one occasion. The results of the post-processed survey can be used to check the computed coordinates in the field. Positioning Using The GPS Satellites Differential Correction/Measurement Sources The relative positioning approach using observed phase data is a mandatory operational technique that is employed to obtain survey accurate results from GPS measurements. Unfortunately, the role of the reference receiver is simply to mitigate errors affecting the rover receiver. This forces users to purchase receivers which are not productive in the sense that they do not occupy marks of interest. To alleviate this requirement, a reference station network, termed GPSnet, was established in Victoria. At present, GPSnet consists of eighteen operational reference stations. Data is available for post-processing applications in one hour blocks stored at either one second, five second or ten second data intervals. The data can be obtained in the native receiver format or in the receiver independent exchange (RINEX) format via the Internet. The RINEX format enables data collected from different brands of receiver to be combined and processed. At the present, there is no data link facility to enable real time surveys to be performed using the GPSnet stations. This is partly due to the fact that a standard communication format for real time surveying has not been adopted by receiver manufacturers. The Radio Technical Commission for Maritime services (RTCM) have developed a format which supports both raw measurement transfer for double differencing and carrier phase correction transfer for the differential correction approach. Trimble Navigation have also published the message format used by their real time systems. Acceptance of a standard format is not very far away, at which point, broadcast of reference station information from the GPSnet stations can be expected. With the addition of new reference stations to the GPSnet network, Victorian surveyors have ready and affordable access to reference station measurements for a variety of applications. There are also a number of free to air and commercial real-time DGPS correction services available across Victoria which are made available by either radio or communication satellites. These can be very useful for navigation, mapping and lower accuracy surveys. The Marine Radio Beacon DGPS Service operated by the Australian Marine Safety Authority (AMSA), provides free-to-air corrections via the ABC and JJJ radio bands. Although specifically designed for marine navigation, the AMSA DGPS corrections can still be utilised in inland Australia (up to 300 km). Commercial vendors such as OmniStar and LandStar also provide DGPS services via satellite communication links accessible via user subscription. Single GPS receivers fitted with modulation devices capable of receiving the satellite DGPS signal can obtain meter level accuracy using such providers. These services provide corrections suitable for pseudorange processing and therefore may not meet the accuracy requirements specified for surveying applications. Presently there are no free-to-air phase differential GPS services available in Victoria. Land Victoria is however examining the prospect of real time phase corrections being broadcast from GPSnet reference stations. Positioning Using The GPS Satellites Surveying Using Global Navigation Satellite Systems 19

20 Designing GPS Surveys Accuracy Obstructions Length Of Baselines Occupation Time Recording Rate Measurement Redundancy Satellite Geometry Control Requirements Surveying Technique Personnel Management As with most surveying tasks, GPS surveys are more likely to be successful if properly planned and designed. There are a number of issues that need to be considered before performing a GPS survey. This section presents a number of issues that should be considered before surveys are attempted. The most successful GPS surveys are those which suit the application and are designed accordingly. In Australia, decisions regarding these issues can be assisted using the SP1 Publication document produced by the Intergovernmental Committee on Surveying and Mapping. Further information regarding the design of GPS surveys is also available at the following website: Accuracy Perhaps the first question that needs to be answered pertains to whether GPS techniques are capable of achieving the accuracy required by the project. Manufacturer specifications indicate that carrier phase surveys are capable of achieving centimetre to sub centimetre accuracy, plus one or two parts per million of the baseline length. Surveyors must be aware that these specifications usually correspond to the standard deviation of computed baselines. Doubling (and sometimes greater) the specified value often provides a more realistic assessment of the capabilities of GPS equipment. Users should also be aware that errors due to factors such as multipath are not considered in these accuracy values. Another factor to consider is that the vertical GPS component is, generally, not as accurate as the horizontal component. A rough rule of thumb relates the height error to the horizontal error by a factor of 1.5 to 3.0 depending on satellite geometry. The various GPS observation techniques are also capable of various accuracy levels. Static surveying techniques generate the most accurate results, followed by rapid static surveys. Kinematic and Real Time Kinematic (RTK) surveys are not as reliable due to the shorter occupation periods adopted. This short time on site does not facilitate averaging of random measurement and multipath errors. With these issues in mind, surveyors can now estimate whether the specified accuracy can be achieved. Typical examples may include a horizontal accuracy of 0.05m. This can generally be achieved using any of the GPS surveying techniques. Tighter horizontal accuracy requirements of 0.02m can be met by static techniques over short baseline lengths, but will push kinematic techniques to the limit. Vertical accuracy requirements must include geoid estimation considerations. An accuracy of 0.05 can generally be met over short baseline lengths, but users must be careful to survey under ideal satellite geometry conditions and use the best available geoid model. Static surveys employing long occupation times can be used to survey to a high accuracy and may be suitable for monitoring surveys. When in doubt, surveyors should consult their equipment documentation and apply professional experience when deciding on the applicability of GPS to specific projects. Each new GPS receiver model houses smaller circuit boards with more powerful integrated circuits. Improvements in satellite tracking performance, antenna design, multipath rejection and ambiguity resolution time are all areas of developing technology. Therefore, it can be anticipated that the accuracy of GPS technology may yet improve, albeit not greatly, before systems become as commonplace as total station technology. As a result, surveys which may test the capability of equipment today, may become routine in the next few years. Designing GPS Surveys 20

21 Obstructions In order to apply GPS technology to surveying applications, a clear view of the satellites is required. This precludes the use of GPS technology in tunnels, under bridges and in built up areas with tall buildings. In most instances, surveys of features in well established areas may not be completely suitable to GPS technology. In these situations, portions of the survey, such as the control work, can be performed using GPS technology. The remainder of the survey can be completed using a total station. Newer subdivisions in Victoria have underground power and immature trees. These areas are ideal for surveying using GPS technology. Again, professional experience will decide whether sites are suitable. It is important to note that the points of interest must be free of overhead obstructions, not simply the area. Consider the following rural Victorian property. The area is generally free of overhead obstructions, as is common for most Victorian rural areas. However, there are several mature trees along the property boundaries. If a cadastral survey is to be performed on this property, the general inclination may be to assume that GPS techniques are suitable as the property lies in an open area. However, as the fence line lies in the timbered regions, obstructions may become a problem. In general, these type of environments are ideally suited to GPS technology, however, the survey may require the use of a total station to complete the survey in obstructed areas. A static GPS survey may be used to place control points along the fence line in locations with a clear line of sight. The coordinates of both points will have been determined by the static GPS occupations, therefore, the bearing between the two marks can be computed and used to orient the total station. If the length of the fence is excessively long, the total station can be used to place points on line between the two control marks, enabling shorter radiations to be used to pick up bends in the fence. Designing GPS Surveys Length Of Baselines One extremely important factor to consider when designing GPS surveys is the distance between the two receivers, or baseline length. GPS survey accuracy degrades as the separation between the receivers increases. This is due to spatial correlation of errors at both sites not being as high as if the receivers are adjacent to each other. This fact is reflected by the parts per million component of most accuracy specifications. In addition, the time required to successfully resolve the integer ambiguities generally increases as the baseline length increases. This results in surveys which are less accurate and not as efficient to perform. For static surveys, the occupation time is generally quite long in order to ensure ambiguity resolution, as well as, to average random measurement and multipath effects. As a result, baseline length is not as critical a factor. Users must be aware that a limitation exists if single 21

22 frequency receivers are used as the ionospheric error will cause problems over baselines greater than 10-15km. Baseline lengths should be kept below this length, especially, in the peak of the sunspot cycle activity. For rapid static surveys, dual frequency receivers are used. The aim of such surveys is to resolve the ambiguities as quickly as possible. The most efficient rapid static surveys are performed when baseline lengths are less than five kilometres. Baselines longer than this can be observed, however, the time required to resolve the ambiguities may result in a static survey campaign being more efficient. Kinematic surveys (including RTK) are the most sensitive to baseline length as the resolution of the integer ambiguities in an efficient manner is what enables short occupation times to be used. Unsuccessful ambiguity resolution results in surveys which do not meet required accuracy levels. Most manufacturers will recommend that surveys be performed over baseline lengths of less than ten kilometres. Although longer distances than this can be observed, best results are achieved when the reference and rover receiver are separated by less than five kilometres. These guidelines may appear to be restrictive if mis-interpreted. It must be clear that the survey may extend beyond these baseline lengths, it is the reference-rover separation that should stay within these limits. If multiple reference sites are used, surveys can be performed successfully over extremely large areas. Consider the following example illustrating the use of two reference stations. A forty kilometre stretch of pipeline is to be coordinated. The extent of the survey is forty kilometres. If the pipeline is free of overhead obstructions, which is highly probable, kinematic techniques may be the most efficient means of coordinating the pipe location. If it is decided that the maximum distance between the reference and rover stations is to be ten kilometres, two reference stations can be established at points A and B on the diagram. This enables the easterly 20km portion of the pipe to be coordinated using reference station A, and the westerly half to be coordinated from reference station B. The reference receiver locations can be coordinated using static survey techniques before (or after if the survey is post-processed) the kinematic survey is conducted. Intelligent placement and use of reference stations can result in more efficient GPS surveying. Designing GPS Surveys Occupation Time For a static survey, the occupation time per point surveyed is selected to provide sufficient measurements to enable the integer ambiguities to be resolved. Users must be aware that a change in satellite geometry during the occupation period is required to enable the ambiguities to be solved. This is partly due to the ambiguity being a value which is extremely close to the distance between the satellite and receiver at the start of the survey. As the range to the satellites changes, the ranges and ambiguities start to separate. This enables statistical methods to identify the correct number of integer cycles more easily. Therefore, 100 satellite measurement epochs collected at a one second rate are most likely insufficient to resolve the ambiguities, whereas 100 epochs at a fifteen second rate (i.e. 25 minutes) are more likely to be sufficient. This highlights that occupation time, rather than number of measurements, is the key factor in performing static surveys. The occupation time required for static surveying is a function of a number of elements including the baseline length, number of satellites, satellite geometry, atmospheric conditions and multipath conditions. In general, using modern technology surveying receivers,

23 minutes of dual frequency measurements are usually sufficient to resolve the ambiguities over baseline lengths of less than ten kilometres. An additional ten minutes may be sufficient to extend the baselines to 10-20km. Both these estimates presume continuous tracking of at least five satellites. It should be noted that in the presence of obstructions it may be necessary to increase the occupation time in order to achieve clean measurements. Single frequency users are advised to acquire measurements for twice as long, i.e minutes. If there are six, seven or even eight satellites being observed, experienced users who are familiar with the performance of their equipment may wish to observe for shorter periods than this, say ten minutes, particularly if performing rapid static surveys over shorter baselines. For baseline lengths longer than 25-30km, dual frequency receivers should be used and observation times should not be shorter than one hour to ensure successful ambiguity resolution. Kinematic surveys utilise a short initialisation technique to resolve the integer ambiguities. This initialisation procedure may take several forms, however, the end result is successful constraint of the integer terms required during processing. Once the survey is initialised, each point of interest only need be occupied for several epochs. In order to acquire sufficient measurements to detect if a bad epoch has been recorded, surveyors are recommended to acquire at least ten epochs while at rover sites. The recording rate for kinematic surveys is usually higher than that of static surveys, thus, ten epochs of measurement may correspond to less than thirty seconds. To be conservative, occupation periods between 30 and 60 seconds should be used for kinematic occupations. Shorter occupation times can be used for detail surveys where the position of the points of interest are only required for plotting features on a survey plan. The use of shorter occupation times enables up to 100 points per hour to be collected under suitable conditions. Note: The desired occupation times for the specific GPS survey techniques are also outlined in SP1 Publication produced by the ICSM. Further information regarding the length of GPS observation sessions can be obtained in section (GPS Survey Planning) at Designing GPS Surveys Recording Rate The recording rate represents the rate at which satellite measurements are stored. This rate is often termed the data rate or epoch rate. For static surveys, there is little advantage in storing measurements at a high rate. Typically, a recording rate of 10 or 15 seconds is used for static occupation periods of twenty minutes or more. For longer sessions which may involve several hours of measurement, rates of 30 seconds or even one minute are suitable. For static surveying, it is important to assess the amount of work to be performed and weigh the amount of data against the available volume of storage space. For example, it may be feasible to perform four observation sessions of 45 minute duration in one day. In this instance, the user should verify that the rate chosen is such that 180 minutes of data can be stored. It should also be noted that the amount of data storage required will depend on the number of satellites observed and the manufacturer s ability to compress the acquired measurements into efficient data structures. For rapid static surveys, similar considerations apply. The primary difference between a rapid static survey and a static survey is the shortened occupation period. In order to provide the processing algorithm with sufficient measurements to perform statistical operations, a higher data rate is generally used for rapid static surveys. For example, a ten minute occupation should be performed at a data rate of five or ten seconds, rather than 30 or 60 seconds. The data requirements of a kinematic survey are quite different from a static or rapid static survey. Kinematic surveys are designed to be more efficient than static surveys by employing shorter occupation times. In general, a minimum of ten epochs at each rover site is recommended. If the data rate is set to 60 seconds, the performance of the kinematic survey is no different from a rapid static survey. A recording rate of three or five seconds is, 23

24 therefore, usually adopted for stop and go kinematic surveys. This enables the rover receiver to occupy marks for less than one minute, while still providing sufficient epochs to enable gross measurement errors to be identified. By purchasing additional memory chips or data cards, or using a computer with large hard disk to control the receiver higher sampling rates can be used. In all cases, an increased cost is incurred. The alternative is to transfer the acquired measurements to a computer during the day. This requires the use of a computer, generally powered by internal batteries or by a cigarette lighter adapter from a vehicle, to be available in the field. If these options are not feasible, a compromise can be reached by recording at a slower rate, say 10 seconds, and occupying rover sites for closer to two minutes. The final survey type that needs to be considered is the continuous kinematic survey or RTK survey. In these surveys, the position of the receiver while it is motion is of interest. The recording rate needs to be carefully selected to provide points at desired intervals. For example, a recording rate of three seconds will provide one computed position every fifty metres if the host platform is travelling at sixty kilometres per hour. This is increased to almost 100m for a vehicle travelling on an Australian highway. The selection of the data rate must, therefore, be computed based on a desired point spacing and the estimated speed of the host platform. Users may find that surveys need to operate at rates of one second to be effective for the chosen application. If so, it may be necessary to invest in additional data storage to enable surveys of a practical observation period to be performed. One vital point that must be remembered is that the reference receiver must record measurements at the same rate as the rover. The reference receiver may record faster than the rover, as long as the rate is evenly divisible by the rover rate. For example, a rover rate of ten seconds and a reference rate of two or five seconds is satisfactory. A reference rate of three seconds will mean that two of every three rover epochs are ignored. The measurement epochs in GPS receivers are determined by dividing the GPS time by the recording rate. A remainder of zero indicates that the measurement should be stored. This means that surveyors do not need to synchronise receivers as such, as the receiver clock performs this function automatically. Note: The desired sampling rate for specific GPS survey techniques are also outlined in the ICSM SP1 Publication produced by the ICSM. Designing GPS Surveys Measurement Redundancy Professional surveyors build redundancy into their survey procedures as a matter of course. Control work is performed by traversing, computing and distributing measurement errors. Radiations are often checked using a right angle offset from a traverse line. Even a detail survey has a check of sorts as anomalous terrain variations and large bends in fence lines may indicate measurement errors. Each of these techniques provides the surveyor with the ability to detect gross measurement errors. Surveying with GPS is similar, the only difference is that the observation procedure is less prone to user error as almost everything is done automatically by the receiver. The most likely source of human error is coordinating incorrect marks or naming marks incorrectly, and erroneous entry of antenna height details. The corollary to this is that any errors due to the receiver measurement procedure are often difficult to detect. It can be easy to conclude that GPS is an error free "black box" technology - this is definitely not the case! GPS surveys can be designed to contain sufficient measurement redundancy to enable gross errors to be detected. Surveyors should be aware of any requirements that contractors have regarding redundant measurements in GPS surveys. For example, static control surveys may require the occupation of each mark on at least two separate occasions. The ICSM SP1 Publication suggests that for kinematic surveys (including RTK), each point may need to be coordinated from two reference stations. It may also be stipulated that the kinematic 24

25 occupations are to be independent, in other words, the two reference receivers cannot operate at the same time to enable each point to be occupied once. Surveyors can increase the integrity of their results by planning redundant measurements into their survey procedures. The use of loop closures and least squares adjustments can then be used to isolate problematic measurements. One additional method by which checks can be built into surveys is to occupy as many previously coordinated marks as possible. A minimum number of control points must be occupied and integrated into surveys to enable coordinates to be computed relative to the appropriate coordinate system. Integration of additional coordinates serves two purposes, it assists in identifying erroneous GPS baselines, as well as, integrating the survey into the control coordinate system. This also verifies the control coordinates. Designing GPS Surveys Satellite Geometry When observing a terrestrial resection, if the control points that are being observed are closely bunched in one quadrant, measurement errors will not distribute in a manner which provides optimal results. The following diagram illustrates two resection examples using three control points. When the points are well spaced, the measurement errors (shown in red) tend to cancel. The same cannot be said when the points are poorly spaced. The green triangle indicates the region in which the estimated position will fall. When the control points are well spaced, the estimated region encompasses the desired position. When the control points are poorly spaced, the desired position lies outside the green triangle. It should be noted that both configurations are acceptable if no measurement error exists, however, measurement errors are inherent in all survey practice and need to be considered. Surveying with satellites is the same with regards to the spacing of the satellites as a terrestrial resection is with regards to the control points. Satellites which are well spaced will tend to provide better results than constellations which are poorly spaced. The indicator used to describe the instantaneous satellite geometry is termed the Dilution Of Precision (DOP). A high DOP value indicates poor satellite geometry, a low value indicates strong geometry. The DOP value is calculated from the inverse of the normal matrix in a point positioning least squares adjustment. This is the same matrix that is used to compute error ellipses in a least squares adjustment of terrestrial observations. As the point positioning observation scenario is not used in surveying, relative techniques are required, the DOP values do not theoretically relate directly to surveying. A changing DOP may, in fact, result in faster resolution of the ambiguities. This does not, however, mean that the computed coordinates are more accurate. To ensure that the satellites are well spaced when coordinates are computed, surveyors should survey in periods of good satellite geometry. 25

26 The inverse of the normal matrix, from which the DOP value is derived, contains diagonal elements relating to the three position components and receiver clock error. Therefore, different combinations of the diagonal give rise to different DOP values. For example, the geometric dilution of precision (GDOP) indicates the status of the satellite constellation for computing three dimensional position and time. Time dilution of position (TDOP) indicates the suitability of the constellation for computing the receiver clock error. In surveying, there are three DOP values that are of primary interest. The position dilution of precision (PDOP) indicates the suitability for three dimensional positioning. The horizontal dilution of precision (HDOP) reflects the two dimensional suitability of the constellation and the vertical dilution of precision (VDOP) indicates the satellite geometry suitability for height determination. The ICSM SP1 guidelines state that in practice, users should survey when the DOP indicator of interest is below 8.0. It is possible to survey with higher DOP values, however, accuracy may be compromised. In addition, most receivers will cease to record measurements when the PDOP exceeds The dilution of precision is highly dependent on the number of visible satellites. Generally, when six satellites are observed, DOP values remain below 3.0. Another feature of the GPS satellite constellation is that the VDOP value is generally higher than the HDOP. This is due to receivers only observing, for example, 150 degrees of the vertical plane when a fifteen degree elevation mask is used. In contrast, the satellites occupy a full 360 degree horizon with respect to the horizontal plane. The following chart presents the satellite geometry for the Melbourne reference station operated by Land Victoria. An elevation mask of ten degrees has been used to generate the DOP values. It is clear from the chart that the HDOP value is lower than the VDOP value for the entire observation period. The PDOP values generally range from 2.0 to 3.0, highlighting the suitability of the satellite constellation in the Melbourne area for GPS observations. Users should note that the times noted on the chart are only a guide to illustrate the speed in which the DOP changes. As the satellite constellation repeats four minutes earlier each day, the DOP chart will "move" to reflect the changing satellite position. Most manufacturers distribute satellite planning software with their GPS systems. These programs use the almanac broadcast as part of the navigation message to predict the position of the satellites with respect to the user location. Planning programs can usually provide graphs of the number of satellites versus time, satellite geometry (DOP) versus time and combinations of satellite azimuth and elevation versus time. Users should consult their product manuals for specific functions of their planning programs. Designing GPS Surveys 26

27 Control Requirements Almost all surveys require the computed coordinates to be related to an existing set of coordinates. Even a straight forward re-establishment survey will require the survey to be rotated onto the datum used by a previous survey. This may be performed by setting up on a mark occupied in the previous survey and sighting along a direction determined from the previous survey. This provides the bearing datum, or in effect, determines the necessary rotation parameter to apply to determined coordinates. If the total station being used has been calibrated, the distances can be considered correct and the survey can proceed. GPS surveys are a little different from total station surveys. The coordinates generated from GPS measurements are referenced to the WGS84 datum and are presented in terms of Cartesian coordinate differences between the reference and rover receivers. If the desired coordinate system is different from WGS84, as is the case for AGD or GDA coordinates, a transformation needs to be applied. For surveying applications where the coordinates of the local control points need to be considered when integrating new points, a global or regional set of pre-determined transformation parameters is often inadequate for application to new points as the parameters are not sensitive to errors in the local coordinates. Therefore, surveyors must occupy points with known coordinates in order to integrate new points into local coordinate systems. The guidelines for GPS surveying outlined in the ICSM SP1 Publication suggest that all GPS surveys should be connected to state control when possible for the purposes of survey integration, legal traceability and quality assurance. The number of control points required depends on the application. If horizontal coordinates are required, then a minimum of two points are required with known east and north coordinates in the desired coordinate system. This enables a scale factor, rotation and two translation components to be computed. Note that any error in the local coordinates will be difficult to detect, as there is no redundancy in the transformation parameter estimation process. It is beneficial to observe a third control point in such circumstances. If the height of points is also required, sufficient information must be available to compute the geoid-spheroid separation. If the survey extends for less than ten kilometres, geometric geoid modelling techniques can usually be applied to good effect. Geometric techniques require the survey to be connected to three points with known horizontal and height coordinates. An additional point with known height only is sufficient to check the success of the geoid modelling technique. The number and type of control points required to integrate the survey into an existing coordinate system need to be considered when designing the survey. In Victoria, the Survey Marks Enquiry Service (SMES) can be used to find existing marks that may be suitable for use as control points. Designing GPS Surveys Surveying Technique There are a number of survey techniques that may be suitable for use for any particular application. They are the static, rapid static, stop and go kinematic, continuous kinematic and real time kinematic (RTK) survey techniques. Surveyors must decide which technique is most suitable to the application of concern. In most cases, a combination of techniques is desirable. For example, static survey procedures may be used to connect the survey to control points. Kinematic techniques can then be used in the local survey region and a total station used to complete the obstructed portions of the survey. This is illustrated by the following diagram. 27

28 A control network is established using static techniques where the black triangles indicate points with known coordinates and the red triangles indicate points to be coordinated. Three receivers are used in four sessions to generate the baselines forming the network. The baselines are indicated by the blue lines. The stop and go kinematic or RTK survey technique may then be used from station A to locate the features of interest. The same points are reoccupied, however, the reference receiver is moved to station B. This provides an independent check on the kinematic points. Finally, the features adjacent to the line C-D are surveyed using a total station as there are trees, indicated in green, along the C-D boundary. Line points may be placed between C and D to enable short radiations to be used. There is one other critical aspect of the selected surveying technique that must be decided. The question pertains to whether the survey should be post-processed, or processed in real time. If the coordinates of the points of interest are required in the field, then the survey must be processed in real time. If the coordinates are not required in the field, a decision needs to be made based on a number of factors. If the survey is post-processed, the coordinates will be more reliable as the data can be manipulated several times and a variety of ways. If long occupations times are chosen to enable errors to average, the measurements should be postprocessed. The disadvantage of post-processed techniques is that significant amounts of data need to be stored and the success of the survey is not known until the survey is completed. Real time techniques do not suffer from this problem as the surveyor can see the computed coordinates in front of them as the survey is proceeding. The disadvantage of real time techniques is the need for the communication link. The difficulties of the data link need to be weighed against the real time coordinate update to decide whether post-processed or real time surveying is the most applicable to a specific project. Designing GPS Surveys Personnel Management The final issue to consider when planning a GPS survey is that of personnel management. Consider a static survey where four receivers are used in a series of sessions as illustrated. For this example, assume that it has been decided that the sessions are to be 45 minutes in length. Each session is highlighted in a different colour. The baselines are less than ten kilometres, six satellites are available and dual frequency receivers are used. The first session is planned to commence at 9.00am. The project requires four people, all with receivers and sufficient batteries, if each receiver is to be monitored while it is running to prevent theft. To operate efficiently, each person should have a vehicle. If this is the case, it can be assumed that the receiver can be stopped and packed away in the vehicle within 15 minutes. Allowing 45 minutes for travelling to the next point and a further 15 minutes to set up and start surveying, the second session will not commence until 11.00am. Using this procedure as a guide, four, or perhaps five, sessions can be performed in one day. The organisation of each of the receivers is vital to ensure that the survey proceeds smoothly. The importance of this aspect of GPS surveying, particularly static surveying, cannot be underestimated. Additional considerations may apply if only two vehicles are used. This may limit the day to three sessions. In such instances, the surveyor must weigh up whether the rental of an additional vehicle justifies the additional productivity. Radio 28

29 communication between vehicles is a great advantage in the case where one vehicle encounters a number of locked gates, flat tyre etc. For kinematic surveys, the logistical problems of static surveys are not usually felt. More often than not, the reference receiver is set up and one, perhaps two, rover receivers occupy points of interest. The area covered in a kinematic survey is usually much smaller than a static survey, therefore, timing and movement of personnel is not as critical. Consideration must be given to whether the reference receiver is to be monitored during the survey. In a kinematic survey, it is quite possible that the reference receiver will remain stationary for the entire day. Assigning a person to look after this receiver is not an efficient use of personnel. It may be more beneficial to establish reference receivers in secure locations in such circumstances. In summary, there are a number of issues that need to be considered before embarking on GPS surveys. Many of the issues require some thought to ensure appropriate application of GPS technology. Users should take care when designing surveys in order to perform GPS operations with a high rate of success. Designing GPS Surveys Surveying Using Global Navigation Satellite Systems 29

30 Best Practice Guidelines for GPS Surveys Introduction Guideline Objectives The ICSM Guidelines Other GPS Considerations for Victoria Introduction There are a number of different methods by which surveyors can use GPS technology for surveying applications. The static and rapid static techniques are most common, however, with the development of real time operation, kinematic techniques are becoming increasingly popular. The signals received from the GPS satellites are prone to a number of errors, many of which are removed by using relative positioning techniques. However, errors such as those caused by multipath are site dependent and are not removed using the differential approach. This highlights that each GPS survey is different in that the observing conditions are unlikely to be the same from day to day, month to month or year to year. It is, therefore, extremely difficult to develop a set of procedural guidelines which will ensure that required survey accuracy is achieved in all circumstances. A set of best practice guidelines can be developed to provide instruction to surveyors in a manner that will realise satisfactory results in most circumstances. As with any set of recommendations, it is up to the surveyors professional discretion to judge the most appropriate manner in which a survey should be performed. Best practice guidelines provide a solid framework on which to base survey practice. Guideline Objectives The guidelines recommended in this section are based on those developed by the Intergovernmental Committee on Surveying and Mapping (ICSM). The aim of the guidelines for Victoria are: to provide a solid set of observation procedures that will enable inexperienced users to perform GPS surveys successfully to establish guidelines that can be adapted to specific circumstances by not being too prescriptive - this falls in line with the State survey regulations regarding cadastral surveys to provide conservative operating procedures which will be successful for the majority of observation conditions - surveyors should apply professional experience in adapting these guidelines for specific surveys Best Practice Guidelines for GPS Surveys The ICSM Guidelines The Intergovernmental Committee on Surveying and Mapping (ICSM) initially developed a document titled "Best Practice Guidelines - Use Of The Global Positioning System (GPS) For Surveying Applications". The first version of this document was released in May of The second version was released one year later and incorporates changes to enable application of the guidelines in New Zealand. These guidelines now form section 2.6 (Global Positioning System) of the ICSM SP1 Publication document. The development of this document has involved representatives from all states and territories of Australia, in addition to the army, navy and New Zealand representatives. The guidelines are, therefore, reflective of the views 30

31 of surveyors from across the country. The document is available on the Internet from the ICSM home page. The guidelines presented in the ICSM SP1 Publication are suitable for providing the framework for GPS surveying in Victoria. All surveyors planning to perform GPS surveys should read the guidelines before attempting GPS surveys. The document contains a section discussing methods of classifying survey types which will not be discussed as each client's needs will differ according to the project being undertaken. The operational procedures of the document are of more relevance. The main points of the guidelines are: Equipment Validation The ICSM recommend that if required, all equipment should be validated on an appropriate network of control points. Also, the Licensed cadastral must ensure that the equipment and methods used are capable of meeting the accuracy requirements of Regulation 5 (1) of the Surveyors (Cadastral Survey) Regulations This requirement includes the validation of GPS equipment that may be used to carry out a cadastral survey (see the section Guidelines for Cadastral Surveying Using GNSS). For more information the relevant authority should be contacted. The ICSM guidelines also recommend that a zero baseline test be performed and the baseline length checked to be zero. The zero baseline test involves connecting two receivers to the same antenna using an antenna splitter device. The positions obtained from the measurements of the two receivers should agree to the sub-centimetre level. Selection of Observation Technique The observation technique used to perform a GPS survey will vary depending on the type of survey being performed. The section Designing GPS Surveys discusses several issues that need to be considered when planning a GPS survey. The ICSM SP1 Publication contains a table (no. 25) which provides a guide to the user as to what technique should be used in order to achieve the task being undertaken. The surveying industry is comprised of professionals who, using this type of information, are capable of deciding on a surveying technique which is suitable and cost-effective for the survey task at hand. In Victoria, the selection of the surveying technique can be left to the discretion of the surveyor, however, with the following qualifier - if the accuracy required by the survey is at the limits of kinematic survey capability as defined by the manufacturer's specifications, or if the region of the survey is extremely large, static observation techniques should be adopted. In addition, if there is any doubt as to whether kinematic surveying techniques are appropriate, the surveyor should be conservative and choose the static option. Static surveys require the acquisition of measurements using stationary receivers. This provides a higher solution reliability than kinematic techniques as occupation times per point are longer. General Requirements The general requirements for surveys recommended by the ICSM apply to all types of GPS surveys - static, rapid static,kinematic and real time kinematic. It should be noted that the ICSM also provide guidelines for pseudo-kinematic, sometimes called intermittent static, surveying. This technique involves two short static occupations of the same mark, separated in time by a period of approximately one hour. This technique is extremely difficult to manage in practice and is not recommended for use in Victoria. The user should use rapid static procedures in place of the pseudo-kinematic technique. The general requirements include: referring to the manufacturer's documentation for instructions as to the correct use of equipment all ancillary equipment such as tripods and tribrachs should be in good condition 31

32 users should take extreme care when measuring the height of the antenna above the ground mark receivers and baseline reduction software should be of the geodetic type carrier beat phase observations using two more receivers for baseline measurements are being used the point identifier should be recorded at the time of survey satellite geometry as defined by the GDOP should be less than 8 all receivers must observe at least four common satellites the elevation mask in Australia should not be less than 15 degrees when establishing reference stations, marks with high quality coordinates should be adopted when heights are required, marks with high quality height values should be used field observation sheets (as provided) should be used for all static survey occupations it is not necessary to record meteorological readings and standard models should be used instead during data processing measurements for horizontal coordination purposes should form a closed figure and be connected to at least two marks with known coordinates in the desired coordinate system least squares adjustments should be carried out to ascertain whether required accuracy standards have been met all least squares adjustments should be three dimensional in nature where multipath is likely, occupation time should be increased to allow the effect to be averaged away as satellite geometry changes These guidelines provide a basic framework for performing surveys. The ICSM also provide a set of guidelines for each of the observation techniques. The following details are recommended. Static Surveying the minimum observation period for baselines less than ten kilometres should be in excess of 30 minutes the recording rate should be 15 or 30 seconds the satellite geometry should change significantly during the observation session at least four, but preferably as many satellites as possible should be common to all survey sites simultaneously occupied single frequency receivers may be used for short lines for non high precision applications it is essential that the carrier phase ambiguities are constrained for lines less than 15km 32

33 Rapid Static Surveying enough data should be collected to resolve ambiguities. The manufacturer s recommendations should be consulted on relation to the lengths of observation periods, number and geometry of satellites and suitability of single or duel frequency dual frequency receivers are preferred multipath can be a significant source of errors, particularly when short occupational times are used and special attention should be paid to this issue the recording rate may vary between five and fifteen seconds Stop and Go Kinematic Surveying five or more satellites should be observed receivers should be initialised per the manufacturers recommendations each point should be occupied in a different session with different satellite geometry the recording rate should be between one and five seconds each station should be occupied for between five and ten epochs multipath can be a significant sourse of errors, particularly when short occupational times are used and special attention should be paid to this issue single frequency receivers may be used although dual frequency receivers are preferred Real Time Kinematic (RTK) single frequency geodetic quality receivers may be used, although dual frequency capability is preferred the typical range for RTK surveys is up to 15 km, although meeting required accuracies may limit this range to 10km. precision claimed by most manufactures is 10 mm plus 2 ppm or better real time update may vary according to the application ambiguities must be resolved for all occupations multipath can be a significant sourse of errors, particularly when short occupational times are used and special attention should be paid to this issue to allow sufficient change to the satellite constellation being used and improve detection of errors such as multipath, re-occupations should be made more than 45 minutes apart and with independent ambiguity resolution two independent occupations of all new stations from two base stations are a recommended minimum typically both or all base stations should have known three-dimensional coordinates new base stations on very large projects should be surveyed using static or fast static GPS techniques and coordinates should be calculated before commencing RTK 33

34 The guidelines conclude by recommending that all raw measurements be archived, as they may be needed for future verification of coordinates. GPS observation recording sheets for static surveys are also attached. In summary, the ICSM guidelines provide a framework for the performing of GPS surveys. In Victoria, several modifications are recommended. These modifications are presented in the next section, Other GPS Considerations for Victoria. Best Practice Guidelines for GPS Surveys Other GPS Considerations for Victoria GPS surveys in Victoria should follow the basic principles of the ICSM recommendations. However from experience, the following additional practices are recommended to improve the quality and general precision of GPS surveys.. Guidelines for the different observation techniques, static, rapid static, stop and go kinematic, continuous kinematic and real time kinematic, are provided. Surveyors should also follow the prescriptions of the Surveyors Act, Survey Coordination Act and associated regulations and handbooks for performing surveys. These guidelines do not over-ride the existing legislation and are designed to complement the existing recommendations. General Requirements all surveys should be performed using the relative (differential) observation technique the GPS receiver should be able to record the carrier phase and pseudorange from at least six satellites simultaneously receivers may be single or dual frequency the carrier phase integer cycle ambiguities should be resolved and constrained when generating the final baseline solution in all instances, a sufficient number of redundant baselines should be observed to enable erroneous baselines to be detected by loop closure and network adjustment when post processing Static Surveying in Victoria if the baseline length is less than 10km, the L1 fixed solution is optimal, irrespective of whether the receiver is capable of dual frequency. However an ionosphere-free fixed solution may be acceptable after consideration of other baseline indicators (see Selecting An Appropriate Observable) if the baseline length is less greater than 10km, the ionosphere-free fixed solution may be used (dual frequency receivers required) during a static survey, the satellite geometry may change rapidly, however, this is not a requirement as this may be difficult to obtain with the satellite constellation in Victoria. If it is possible, survey planning software should be consulted prior to the field survey being conducted to increase survey efficiency. the PDOP should not exceed 7.0 during the data observation period occupation sessions must include periods where the PDOP is less than 4.0 the elevation mask should not be less than ten degrees - if processing reveals noisy measurements at low elevations, the mask can be raised in the processor 34

35 the length of the observation period should be in excess of than minutes for all baselines - if the baseline length exceeds 10km, longer observation periods will be required if the survey is being performed for geodetic control establishment, the static occupation period should be a minimum of at least one hour the recording rate is not of interest and should be selected with data storage requirements in mind - surveyors should ensure that sufficient epochs are recorded to enable statistical test to be performed by the processor, therefore, rates of greater than 60 seconds should not be selected and commonly, 5, 10,15 or 30 second epoch rates should be used Rapid Static Surveying in Victoria the baseline length should not exceed 10 kilometres dual frequency receivers are recommended as they enable faster resolution of the integer ambiguities the L1 fixed solution should be used (see Selecting An Appropriate Observable) occupation periods should not be less than ten minutes the elevation mask should not be less than 10 degrees at least five satellites must be observed for an occupation period of ten minutes if the occupation period is ten minutes or less, six satellites should be observed the satellite geometry as indicated by the PDOP should be less than 4.0 the epoch rate should be sufficient to enable statistical test to be performed by the processor - as the occupation periods are shorter than static surveys, the epoch rates should be higher, commonly 5 or 10 seconds Stop and Go Kinematic Surveying in Victoria the baseline length should not exceed 10 kilometres dual frequency receivers with OTF are recommended as they enable faster resolution of the integer ambiguities surveys must be initialised using one of the four available initialisation techniques a minimum of five satellite must be observed when occupying stationary marks the PDOP should be less than 4.0 when occupying stationary marks the epoch rate may be selected as either 1, 5, 10, 15 or 30 seconds the elevation mask should not be less than 15 degrees a minimum of ten epochs must be observed per stationary occupation - if the data rate is set to one second, at least twenty epochs should be observed single or dual frequency receivers may be used if the baseline length is less than 10km, the L1 fixed solution is optimal, irrespective of whether the receiver is capable of dual frequency. However an ionosphere-free fixed solution may be acceptable after consideration of other baseline indicators (see Selecting An Appropriate Observable) 35

36 if the baseline length is greater than 10km, the ionosphere-free fixed solution should be used. Continuous Kinematic Surveying in Victoria The recommendations for continuous kinematic surveying are the same as those for stop and go kinematic, except the user should take care to ensure a constant antenna height while the receiver is moving if height determination is required by the survey. Real Time Kinematic (RTK) The typical range of RTK GPS is 15 km, although some manufacturers state that their equipment functions over longer distances (e.g. up to 40 km). If possible the base receiver should be located as close as possible to the rover receiver, to ensure optimal corrections are obtained and radio link is maintained Single frequency geodetic quality receivers should be used, although dual frequency capability is advantageous for ambiguity resolution and mitigation of the effects of the ionosphere if possible, on mountainous terrain where radio communication is difficult, radio repeaters should be used Ambiguities should be resolved for all occupations. In the presence of obstructions such as trees, the number of epochs recorded should be increased to improve the accuracy of the calculated position The reference station (or stations) should be located on survey points whose coordinates are known to a high level of accuracy If possible, permanent survey marks should be occupied to check the accuracy of the coordinates calculated by the RTK receiver during the survey. Best Practice Guidelines for GPS Surveys Surveying Using Global Navigation Satellite Systems 36

37 The NAVSTAR Satellites Introduction Description Types Of Satellite The Constellation Satellite Orbits Calculating Satellite Position - The Ephemeris Signal Structure The Broadcast Signal Signal Access - Positioning Services GPS Modernisation Introduction In many references discussing the use of the Global Positioning System, the "three segments" are discussed. One of these segments, termed the space segment, essentially comprises the satellites and the signals they emit. The other two sections apply to the control segment which monitors and maintains the satellite constellation and the user segment which comprises GPS users, equipment, data collection and data processing techniques. This section aims to present the details of the satellite constellation used to generate the signals that are measured by receivers located on the Earth. The NAVSTAR Satellites Description The NAVSTAR (Navigation Satellite with Timing and Ranging) GPS satellite constellation is designed to facilitate a system which is capable of satisfying the positioning, navigation and timing needs of the United States military. This is an important concept as many of the procedures used to perform GPS surveys may appear inefficient. It is vital to realise that the system is not designed to be a surveying tool, rather a military asset. The specifications of the system require weather independent, twenty-four hour capability, at any location either on, or near, the surface of the Earth. This is accomplished by a constellation of 28 satellites orbiting the Earth. Each satellite transmits a pseudo-random noise (PRN) code message which is modulated on to a carrier (messenger) signal. A user receives these signals and uses them to measure the distance, or range, to at least four satellites simultaneously. The measurements to the four satellites are processed to derive the three position components of the receiver, and the synchronisation offset between the satellite and receiver clocks. The satellite signals are designed to enable this calculation to performed almost instantaneously. In order to provide position estimates to an accuracy suitable for military requirements, the satellite constellation is designed to enable at least four satellites to be continuously available above an elevation mask of ten degrees. The greater the number of satellites simultaneously tracked, the more efficient the GPS surveying process. Fortunately, the satellite constellation is such that six satellites are generally available above a ten degree mask in Victoria. The following satellite availability chart is prepared for the Melbourne reference station operated by Land Victoria. 37

38 For the most reliable and efficient survey results, users are advised to collect measurements when the maximum number of satellites are available. This is especially true for kinematic surveys where interruptions in satellite tracking can greatly affect the success of GPS surveying. The NAVSTAR Satellites Types Of Satellite In the initial implementation of the system, eleven satellites were launched to provide a short observation window to enable the system to be tested, most of which occurred at the Yuma Proving Grounds in Arizona. These initial satellites were termed the Block I satellites, or the prototype satellites. Each Block I satellite weighed approximately 850kg and was launched in one of two orbital planes inclined at 63 degrees to the equator. This inclination has since changed to 55 degrees for the newer satellites. The Block I space vehicles (SV) were launched between 1978 and The design life of these satellites is 4.5 years, however, many of the satellites survived their design lives by more than five years (the third satellite, launched in October of 1978, operated successfully for more than thirteen years!). This bodes well for the survivability of the newer satellites and highlights the exceptional quality of the GPS satellites. In the mid to late 1980 s, the Office of Surveyor General conducted a large number of GPS surveys using these satellites. The measurements acquired in these survey campaigns have been included in Victoria s geodetic framework. The original Block I satellites have since been decommissioned. The second generation of satellites are termed the Block II satellites. This generation of satellite is used to comprise the first fully operational constellation of GPS satellite. A total of twenty-eight Block II satellites have been launched, the first of which was launched in early This number includes both Block IIA and Block IIR satellites. The design life of these satellites has been increased to 7.5 years, however, the components are designed to last for up to ten years. In addition, each Block II satellite weighs approximately double that of the Block I satellites. The Block II satellites operate in much the same manner as the prototype equivalents, however, have an additional capability which is used to deny accuracy to civilian users (see selective availability and anti-spoofing). The launch of a Block II GPS satellite costs approximately $US50 million. This indicates the large investment made by the Department of Defence in the GPS program. It should also be noted that much of the funding 38

39 approved by the US Congress can be attributed to the fact that GPS is a dual use system, providing service to civilians as well as the military. The successor generation to the Block II satellites are designated the Block IIR satellites. The R" is used to denote that the satellites are replenishment satellites for the first constellation. The Block IIR satellites differ from the Block II satellites in that the on board oscillator is changed from either rubidium or caesium, to a more stable hydrogen maser. Apart from this difference, the basic functions of the satellites, with surveying purposes in mind, are essentially the same. In the original implementation of the satellite constellation, the satellites were to be placed in orbit using the space shuttle. However, with the Challenger disaster in 1986, the GPS satellite launches were modified to use Delta II rockets. Up until early 1997, all launches had proceeded smoothly. In January 1997, a malfunction was detected just after launch of a rocket carrying the first Block IIR satellites, forcing the rocket, with a GPS satellite on board, to be destroyed. A contract for the fourth generation of satellites has been issued by the US Department of Defence. There is current discussion regarding the modification of this generation of satellite, termed Block IIF, to incorporate an additional frequency. The motivation for this is to provide civilians with access to two frequencies to enable ionospheric corrections to be applied. The addition of an additional GPS frequency and the development and launch of the Block IIF satellites is part of the current GPS Modernisation project initiated by the U.S. Department of Defence. The Block II GPS satellites have several features which require discussion. The satellites generate signals which are broadcast through a series of L-band frequency antennas. These signals are generated using on board clocks, termed oscillators. The satellites use highly precise atomic standards, either rubidium or caesium. The clocks are powered by nickel cadmium batteries. These batteries are used to power the oscillators when the Earth blocks the energy of the sun from the satellite. During non-eclipse periods, solar panels are used to power the oscillators and charge the batteries. The panels are designed to remain perpendicular to the sun to maximise the efficiency of the sun s energy. The successful operation of the clocks is essential to the functioning of the satellite. If the signals cannot be generated, the satellite is of no use. There are other functions of the satellites, such as the ability to detect a nuclear explosion, however, none are considered paramount to surveying with GPS. The current status of the GPS constellation is available at the following website The NAVSTAR Satellites The Constellation The completed constellation consists of 24 operational satellites and four active spares. These 28 satellites are in almost circular orbits at an altitude of 26,560km above the centre of the Earth. As the radius of the Earth is approximately 6,378km, the altitude of the satellites above the surface of Earth is approximately 20,200km. The signals are nominally travelling at the speed of light, which is defined as 299,792,458.0 metres per second, hence the time taken for the signals to reach the Earth is less than 0.07 of a second. The time taken for a GPS satellite to complete one orbital pass is 12 sidereal hours. This translates to 11hr 58min of solar time. As the 39

40 Earth rotates about its polar axis once every 24 hours, a GPS satellite will be visible from the same user location every two satellite orbital passes, or every 23hr 56min. This highlights that the GPS constellation repeats four minutes earlier each day, or approximately two hours per month. This is useful to know as surveys can be planned for suitable periods of the day and this time frame will not change rapidly from day to day. The satellites move in their orbits due to the attractive force of gravity which "drags" the satellite towards the Earth s centre of mass. The initial launch trajectory of the satellite places the satellite into the designed orbit. Using the orbital radius and period of orbit, a GPS satellite can be calculated to travel at a speed of approximately four kilometres per second. While this may sound fast, in actual fact, a GPS satellite does not change its position quickly relative to a stationary user on the Earth. This has implications regarding the amount of time required to successfully survey using the system (see GPS errors and positioning modes). The GPS satellite constellation of 28 satellites is arranged in six orbital planes, with four satellites located in each plane. The planes are inclined at an angle of 55 degrees to the equator. This arrangement facilitates the continuous visibility of at least four satellites, 24 hours per day, at any location on the Earth. The satellites are placed in one of four, evenly spaced, slots in each plane. The satellite clocks continuously transmit signals, from which the distance to the satellite is measured by the receiver. The receiver position is then computed using these measured ranges. In order to perform this calculation, the position of the satellites at the instant of measurement must be determined. The NAVSTAR Satellites Satellite Orbits The motion of objects in orbit can be described by the three laws of orbital motion developed by the German astronomer, Johannes Kepler. Kepler published his laws in the early 1600 s and they form the basis of the orbital theory used to describe the orbits of the GPS satellites. Kepler s First Law Kepler s first law of planetary motion states that the orbit of an object follows an elliptical path. In addition, the planet which the object is orbiting around is located at one of the foci of the ellipse. In the context of the GPS satellites, this defines the shape of the orbit by the defining parameters of an ellipse; suitably the semi-major axis, a, and the first eccentricity, e. From these two parameters, the semi-minor axis, b, can be determined by a(1-f) where f is the flattening. The square of the eccentricity is equal to the square of the flattening minus twice the flattening, (f*f-2*f). Kepler s Second Law Kepler s second law relates the velocity of the object to the area swept out by the orbital path. The law states that equal areas are swept for the same time period, regardless of the position of the object in its orbit. The ramifications of this are that satellites in highly elliptical orbits will travel at significantly higher speeds as they approach the Earth when compared to the speed when at the extreme (termed the apogee) of the orbit. The GPS satellites are designed to be in orbits with an eccentricity of less than Therefore, the speed of a GPS satellite throughout its orbit does not vary greatly. Kepler s Third Law Kepler s third law of planetary motion relates the orbital period of the object to the semi-major axis. The relationship can be stated as: the square of the orbital period is proportional to the cube of the semi-major axis. The constant of proportionality is the product of the universal gravitational constant and the mass of the Earth. This relationship enables the orbital period to be computed based on a nominal orbital radius. 40

41 Kepler s Orbital Elements The manner in which satellite orbits are commonly described is using the six so-called Keplerian orbital elements. The elements are; the semi-major axis and eccentricity of the elliptical orbit, the orbital inclination, i, the right ascension of the ascending node, Ω, the argument of perigee, ω, and one of the anomalies, most commonly the true anomaly. The right ascension of the ascending node is defined as the angle measured in the equatorial plane between the vernal equinox and the point at which the satellite crosses the equator, moving from below to above (the ascending node). The point on the orbit where the satellite is at its closest to the centre of the Earth is termed the perigee. The angle between this point and the ascending node measured in the plane of the orbit is the argument of perigee. The true anomaly is defined as the angle, measured in the plane of the orbit, between the satellite and the perigee. Knowledge of the true anomaly enables the mean anomaly and eccentric anomaly to be determined. The six orbital elements form the basis of the ephemeris parameters used to define GPS orbits. The NAVSTAR Satellites Calculating Satellite Position - The Ephemeris The GPS satellites broadcast a navigation message which contains a number of elements used during measurement processing, the most important of which is the broadcast ephemeris. The ephemeris is a set of 16 parameters based on the fundamental six Keplerian elements. The ephemeris is, therefore, said to comprise 16 pseudo-keplerian elements. The ephemeris parameters are, essentially, Kepler s six elements with additional correction terms. The parameters are presented in the following table. Symbol Ephemeris Parameter Description Symbol Ephemeris Parameter Description A 1/2 square root of the semi-major axis e eccentricity i 0 inclination at the orbit reference time Ω 0 longitude of the ascending node at the start of the GPS week 41

42 ω argument of perigee M 0 mean anomaly at the reference time Δn correction to the mean motion i-dot rate of change of inclination Ω-dot rate of change of the right ascension of the ascending node C UC, C US amplitude of correction terms to the argument of latitude C RC, C RS amplitude of correction terms to the orbital radius C IC, C IS amplitude of correction terms to the inclination angle t oe ephemeris reference time A series of equations are used to compute the position of the satellite given a time of observation. The result of this calculation is the Cartesian coordinates of the satellite at the observation time. Each satellite has its own set of ephemeris parameters which are used continuously during position estimation. The ephemeris broadcast by the satellites is computed by the GPS control stations and represents a prediction of the satellite position in the future. A post-computed ephemeris, termed the precise ephemeris, can be used for postprocessing applications. The precise ephemeris is an estimate of the satellite orbit based on actual measurements acquired during the observation period. It is, therefore, more reliable and should be used if available. More information regarding the calculation of satellite ephemerides is available at the following website The NAVSTAR Satellites Signal Structure The manner in which positions are determined using the GPS satellites is highly dependent on the signals being transmitted by the satellites. There are a number of design criteria that guided development of the signal structure. As a result, the GPS signal is quite complex, in order to support some of the following features; one-way (passive) positioning, provide accurate range and velocity (Doppler) measurements, broadcast a navigation message, facilitate simultaneous observation of multiple satellites, provide ionospheric delay correction and prove resistant to interference and multipath. The following signal structure has been designed to meet these specifications. The Pseudo-Random Noise (PRN) Codes The GPS system is designed to enable the distance, or range, to a satellite to be determined. This is supported by transmission of two pseudo-random noise codes, the Coarse Acquisition (C/A) code designated for civilian use, and the more Precise (P) code designed for military use. Each code is simply a series of one s and zero s (binary chips), in other words, a binary sequence, which appears to have no pattern and looks random. The PRN codes are generated using a mathematical algorithm and repeat at defined intervals. The C/A-code is a sequence of 1,023 binary chips which is repeated every milli-second. Therefore, the number of chips generated per second is 1,023,000, hence, one chip is approximately one microsecond in length. The chip length can be multiplied by the speed of light to yield the wavelength, in this case, approximately 300m. Each satellite transmits a slightly different C/Acode, thus, the receiver can identify a satellite by the particular C/A-code being transmitted. As the C/A-code sequence repeats itself every millisecond, a GPS receiver can quickly acquire this code and begin making pseudorange measurements. 42

43 The accuracy with which a PRN code can be measured is partly determined by the chip length, or wavelength, of the binary sequence. Most survey receivers can measure the C/Acode to a precision of approximately half a percent of its wavelength, approximately one and a half metres. In order to obtain more precise range measurements, the P-code was developed. The P-code is generated ten times more quickly than the C/A-code, resulting in a chip length which is one tenth that of the C/A-code and theoretically, range measurements to a precision of several decimetres. However, the length of the P-code sequence is significantly longer than that of the C/A-code, 266 days. Therefore, before a receiver can acquire the P- code, it must lock on to the C/A-code so that it can acquire the hand-over word (HOW) from the navigation message and begin generating the correct section of the P-code sequence. Each satellite does not transmit a unique P-code, rather each satellite is assigned a seven day portion of the code which is initialised weekly. Satellites are identified by a PRN number, corresponding to which seven day portion of the P-code is being transmitted by the satellite. The PRN codes have been designed to be orthogonal to each other, in other words, the correlation between any two codes will always produce a low output (see details about the receiver signal tracking loops). This protects the receiver from accidentally locking on to the wrong satellite. In addition, the pseudo-random nature of the codes makes them almost impervious to jamming, either unintentionally or by United States enemy forces. The disadvantage of the binary codes is that they do not have the ability to be transmitted long distances. To alleviate this characteristic, the codes are modulated onto two carrier signals, which transport the binary sequences to users on the Earth. The Carrier Signals The GPS satellites transmit information on two carrier frequencies, termed L1 and L2. Both frequencies are contained in the L-band microwave section of the electromagnetic spectrum with L1 centred at a frequency of MHz and L2 at MHz. The difference in the two carrier frequencies of approximately 350MHz is suitable for eliminating the ionospheric delay affecting both signals. The carrier signals are pure sinusoidal sequences, therefore, one carrier cycle cannot be distinguished from another. This requires specific mathematical algorithms to be used if the carriers are to be exploited for positioning purposes. The wavelength of the carriers is approximately 19cm and 24 cm for the L1 and L2 carriers respectively. Based on the half a percent rule of thumb for measurement accuracy, this implies that the carrier signals can be measured to about one millimetre. Survey receivers are capable of measuring the phase of the carrier signals to this accuracy. This provides a measurement suitable for surveying purposes. However, it should be noted that use of the carrier for positioning is not the design intention of the system. The carriers are present to transport the C/A-code and P-code. This is performed by modulating the codes onto the carrier signals. The technique used to modulate the two PRN codes involves modifying the carrier to accommodate the one s and zero s of the codes. When a zero is to be modulated, the carrier is unmodified. However, when a one is encountered, the phase is mirrored, effectively causing a phase change of 180 degrees. This modulation technique is termed binary bi-phase modulation. The GPS receiver is capable of demodulating the code from the carrier by demodulating the one s and zero s from the pure sinusoidal carrier signal. The Navigation Message The GPS satellites also transmit a low frequency (50Hz) navigation message which contains information that is required to calculate receiver positions. Due to the low rate of the navigation message, the entire message takes 12.5 minutes to transmit. In order to facilitate rapid position estimation, two of the components of the message are repeated every 30 seconds. The first of these is the broadcast ephemeris which comprises a set of 16 pseudo- Keplerian parameters which describe the orbits of the GPS satellites. The ephemeris is required to compute the instantaneous position of the satellites to be used in the position estimation algorithm. The second component is the satellite clock correction coefficients. Each satellite uses atomic clocks which are extremely accurate. However, atomic standards perform more reliably when not interfered with. As a result, rather than correct the satellite clocks to the GPS reference time frame, they are allowed to drift freely. The satellite clock offset at a reference time and drift since that time are reported as a polynomial expression, the coefficients of which are contained in the navigation message. The other pieces of 43

44 information contained in the navigation message include the almanac, satellite health indicators and the hand over word. The almanac is a set of parameters, similar in nature to the broadcast ephemeris, which define the satellite orbits in an approximate manner. The almanac is used by the receiver to determine which satellites are above the local horizon. This enables the receiver to begin generating the C/A-code for a satellite which is, theoretically, visible. Once acquired, other satellites are quickly tracked. Another use of the almanac is in constellation planning software provided with most GPS systems. The planning software uses the almanac to derive satellite positions which can be used to indicate periods of the day which are suitable for performing survey measurements. If a satellite is not transmitting signals correctly, or is unstable in its orbit, the Department of Defense can designate that satellite as unhealthy, indicating that it should be excluded from position calculations. Another more common reason for satellites being set unhealthy is soon after they have moved to a different orbital position. This orbit readjustment is performed by firing booster rockets on the satellites to change its orbit. Once in the designated orbital position, the satellite may take some time to settle into a stable orbit. During this period, the satellite is usually reported as unhealthy. The navigation message contains a satellite health flag for every satellite in the constellation. Use of unhealthy satellites should be avoided. The period of the P-code sequence is quite long, 266 days. In order for a receiver to quickly start generating the P-code and lock onto a satellite, the receiver needs some additional information to know where in the P-code sequence signal generation should commence. This information is provided in the hand-over word (HOW) contained in the navigation message. However, before the navigation message can be accessed, the C/A-code must be tracked, thus, the C/A-code is tracked by all receivers requiring access to the P-code and the HOW used to facilitate P-code tracking. The final broadcast GPS signal is comprised of a combination of the C/A-code, P-code, navigation message and L1 and L2 carriers. More detail regarding the GPS signal structure can be found at the following website: The NAVSTAR Satellites The Broadcast Signal The GPS signal comprises two PRN codes, the C/A-code and the P-code, a navigation message and two carrier signals, L1 and L2. The codes and carriers are generated at a multiple of a fundamental frequency which is defined as 10.23MHz. The P-code is generated at the fundamental frequency, the C/A-code at one tenth of the fundamental frequency. The L1 carrier is centred at MHz, 154 times the fundamental frequency and the L2 carrier, centred at MHz, is generated at 120 times the fundamental frequency. The L1 carrier signal is modulated by a combination of the C/A-code, P-code and navigation message. The L2 carrier is modulated with only the P-code and navigation message. The codes are combined with the navigation message using a binary modulo 2 addition. If both sequences contain a one or a zero, the resultant is zero, otherwise a one is registered. This enables the navigation message to be superimposed separately with the C/A-code and the P- code. In the case of the L2 signal, the combined P-code and navigation message are biphase modulated on the L2 carrier. However, as the L1 carrier is host to both codes, a modification must be made. The P-code with navigation message is modulated onto the L1 carrier in the same manner as it is modulated on the L2 carrier. The C/A-code with navigation message is modulated onto a shifted L1 carrier. This shifted carrier is the L1 signal advanced by 90 degrees. This technique is termed phase quadrature. The two L1 signals are then combined before combination with the L2 signal which is then broadcast. The NAVSTAR Satellites 44

45 Signal Access - Positioning Services The GPS signal is divided into two distinct positioning services. The Standard Positioning Service (SPS) is defined as the C/A-code and navigation message modulated on the L1 carrier signal and is designated for civilian use. The Precise Positioning Service (PPS), designated for military use, comprises the entire broadcast signal. The primary difference between the two positioning services is, therefore, the P-code on both carrier signals. Survey receivers have been able to acquire range and phase measurements on both frequencies for many years. In fact, one of the earliest receivers, the Texas Instruments 4100, accessed both codes and carriers. This receiver was built in the early to mid 1980 s. The primary benefits of accessing both frequencies for surveyors are realised in terms of position accuracy over long baselines as the ionosphere can be modelled, in addition, to rapid identification of the carrier phase integer cycle ambiguity over shorter distances. In order to preserve a military advantage in the use of the GPS satellites, the United States Department of Defence initially implemented two accuracy denial mechanisms to limit civilian positioning performance. The first mechanism, anti-spoofing, is designed to eliminate access to the P-code. The second mechanism, selective availability, reduces the accuracy with which C/A-code positioning can be performed, but was recently turned off. Anti-Spoofing (A-S) The P-code is a binary sequence that has a chip length which enables decimetre accuracy ranges to be measured. In order to restrict access to a signal with such accuracy, the P-code has been modified by addition of a sequence termed the W-code. This combination of the P- code and W-code yields the so-called Y-code which is transmitted in place of the P-code. Only receivers developed specifically for the US military have knowledge of the Y-code signal. However, as the L2 carrier signal and code are of benefit to surveying applications, survey receivers use innovative techniques which enable L2 measurements to be made. The signal tracking loops within the receiver are modified to facilitate this L2 measurement in the presence of the Y-code. The anti-spoofing policy is sometimes termed encryption. Selective Availability (SA) The selective availability policy was introduced to degrade the position accuracy obtained using a single receiver accessing the C/A-code and navigation message. In the mid to late 1980 s, position accuracy of less than 30 metres was routinely achieved with the SPS. The US military considered this accuracy to be sufficiently accurate to cause a threat in times of 45

46 crisis if used by enemy personnel. The SA policy outlines two methods by which the SPS is degraded to a specified two dimensional accuracy of 100 metres with a 95% confidence. The first part of the selective availability implementation involves intentionally degrading the broadcast ephemeris parameters. This results in erroneous satellite orbit computations. The second method of degradation is by dithering the satellite clock to intentionally cause a range measurement error. When combined, both techniques serve to limit the applicability of single receiver GPS positioning (Absolute Positioning). The recent removal of SA has provided a large improvement in the accuracy attainable with a single handheld GPS receiver. This is illustrated in the Figure below, which shows Circular Error Probable (CEP) error before and after the removal of SA. For surveying applications, multiple receivers are used. When used in conjunction with the precise ephemeris, the selective availability issue is not of great concern for post-processed applications. GPS Modernisation In January 1999, the President of the United States of America announced that a GPS modernisation program would commence to accommodate the increasing use of the GPS by the international community. This modernisation scheme consists of the alteration of existing GPS signals and the introduction of a new GPS carrier phase signal and would essentially provide two new signals for civilian use. The first change is the introduction of the C/A code onto the L2 carrier phase signal for general use to commence with the launch of the new Block IIR_M satellites in A third carrier signal, designated L5, will also be introduced, primarily for aeronautical navigation. The L5 signal will be available on board the Block IIF satellites set for launch in Latest information regarding GPS modernisation can be found at The NAVSTAR Satellites Surveying Using Global Navigation Satellite Systems 46

47 The Control Stations Introduction Control Stations Master Control Station Australian Fiducial Network Victorian GPS Network Introduction The signals transmitted by the GPS satellites are generated by atomic clocks on board the satellite. The code, carrier and navigation message are broadcast continually by the satellites as they orbit the Earth. The launching, maintenance, repair and data validity of the satellites is the responsibility of the GPS control stations. The control segment performs the day to day tasks of computing the satellite orbits for the broadcast ephemeris, computing the polynomial correction coefficients for the satellite clocks, monitoring the health of the satellites and facilitating the transfer of this information to the satellites. The other, non-operational, tasks performed by the control segment include moving satellites into different orbital positions and coordinating replacement satellites. From the surveyors perspective, there is no need to have direct contact with the control stations. The control stations are, simply, in place to compute satellite orbits and clock correction parameters and maintain the general health of the system. The Control Stations Control Stations The broadcast ephemeris is computed based on measurements acquired at five tracking stations located at Ascension Island, Colorado Springs, Diego Garcia, Hawaii and Kwajalein. The coordinates of the tracking stations have been well defined using Very Long Baseline Interferometry (VLBI) techniques. Each station houses a dual frequency GPS receiver which is regulated by a cesium atomic clock. P-code pseudorange and Doppler measurements (derived from the carrier signal) are acquired on both the L1 and L2 frequencies. These measurements are acquired every 1.5 seconds and are then corrected for the effects of the ionosphere and troposphere (the troposphere is modelled using meteorological readings). The measurements are smoothed to provide samples at fifteen minute intervals. The samples are then transmitted to the control station located near Colorado Springs for satellite orbit and clock correction estimation. The Control Stations 47

48 Master Control Station The station located near Colorado Springs is termed the Master Control Station (MCS). The MCS is officially known as the Consolidated Satellite Operations Center (CSOC) and is housed at the Falcon Air Force Base. This is the station where all satellite orbit parameters, clock correction parameters and health indicators are generated. The measurements from the other four tracking stations are sent to the CSOC for processing. Only the measurements from the five control stations are used to generate the broadcast ephemeris. Once the elements of the navigation message have been computed, they are transmitted back to the control stations for uploading to the satellites. Precise Ephemeris Calculation A precise ephemeris is also computed using the five control stations, in addition, to a further five stations independently operated by the National Imagery and Mapping Agency (NIMA). There are a number of other GPS satellite tracking networks in operation throughout the world including the International Geodetic Station (IGS) network (formally the Cooperative International GPS Network (CIGNET)) and the Continuously Operating Reference Station (CORS) operated by the United States National Geodetic Survey (NGS). There are several reasons for establishing a permanent tracking network of GPS receivers, one of which is to generate precise ephemerides for use in a specific region. Geoscience Australia has established a network of fifteen tracking stations in and around Australia that forms the Australian Regional Geodetic Network (ARGN). Within the ARGN is the Australian Fiducial Network (AFN) which is comprised of the ten mainland ARGN stations. In Victoria, there are a number of stations established as part of a network termed GPSnet. The primary motivation for the Victorian network is to provide a level of reference station infrastructure for surveying applications. The Control Stations Australian Fiducial Network (AFN) The Australian federal government, through Geoscience Australia, has established a network of ten permanent tracking GPS stations. These sites form a network known as the Australian Fiducial Network (AFN). Each of the stations comprises a dual frequency GPS receiver in a location coordinated using Very Long Baseline Interferometry (VLBI) techniques. 48

49 There are a number of motivating factors for the establishment and maintenance of the AFN. An immediate need is to form the framework of the Geocentric Datum of Australia (GDA). This new datum for Australia was implemented in the year 2000 and represents a large move from the previous Australian Geodetic Datum (AGD). The new datum is geocentric in nature and, for all practical purposes, is the same as the World Geodetic System 1984 datum employed by GPS. This implies that surveyors will no longer have to perform transformation calculations to integrate GPS measurements into the Australian coordinate system. Another purpose of the AFN is to monitor the performance of the satellites in their orbits over Australia. In the future, it can be anticipated that the AFN will operate as an automated integrity monitoring system for surveyors using GPS receivers. This is extremely reassuring as no longer do surveyors have to rely on information from the United States military regarding details of satellite problems. For cadastral surveys, where measurements may commonly be required to be verified in a court of law, the effective monitoring of the satellites by the AFN stations may prove sufficient in proving that the satellites were operating correctly during the period of survey. It is also feasible that future developments by Geoscience Australia may include the calculation of a precise ephemeris for the Australian region. This would enable postprocessed surveys to be performed with satellite orbits which are more suited to Australia as the ephemerides are calculated using data acquired at sites located in Australia. Precise ephemeris information from other tracking networks can already be obtained via the Internet ( The Control Stations Victorian GPS Network (GPSnet) GPSnet is a co-operative Global Positioning System (GPS) Base Station Network which records, distributes and archives GPS satellite correction data for accurate post-processed position determination, 24/7/365 - statewide. Land Victoria, working in cooperation with Industry, has established public access, dual frequency base station infrastructure to support GPS users across the state. This reduces the cost of equipment purchases for surveyors and also facilitates the use of well known reference station coordinates which can be used to propagate coordinates homogeneously to survey marks throughout the state. GPSnet users save time and money by not having to establish and operate base stations. Each of the tracking stations is equipped with a dual frequency GPS receiver, uninterruptible power supply (UPS), controlling computer and bank of modems. The modems are used to transfer measurements in hour long blocks to the SDC for archival. Currently, there are eighteen stations in operation across the state as illustrated in the Figure below. 49

50 Using GPSnet, position correction accuracies are achievable down to centimetre level (depending on equipment and techniques used). Processing with multiple GPSnet Base Stations can also significantly enhance processing accuracy and integrity. Files for post processing are normally available from approximately 10 minutes after the end of each hour. The data collected at each site made available via the Land Channel. If the Australian Fiducial Network is used to generate satellite ephemeris parameters, there is no need for the Victorian stations to do the same. To perform this task using the GPSnet sites would require additional upgrading of the sites, which would incur additional cost, therefore, it is most likely not cost-effective to do so. However, the stations do have the potential to operate as continuous integrity monitors for signals received in Victoria. An automated error detection scheme running in parallel to a system operated by Geoscience Australia would enhance the integrity of GPS derived coordinates used in Victoria. The Control Stations Surveying Using Global Navigation Satellite Systems 50

51 The GPS Receiver Receiver Components The Antenna and Antenna Pre-Amplifier The Radio Frequency Section Signal Tracking Loops Microprocessor Power Supply Data Storage User Interface Receiver Components The basic GPS positioning concept requires the measurement of the range between satellites and the point of interest. This range measurement is performed by a GPS receiver. The receiver is, basically, a radio which is capable of tuning either or both of the two GPS carrier frequencies, L1 centred at MHz and L2 centred at MHz. Measurements of pseudorange and carrier phase provide satellite-receiver range measurements. For surveying, the carrier phase observable is of interest. This section analyses the components of a GPS receiver, specifically, a parallel tracking architecture GPS receiver. The parallel tracking architecture provides better signal tracking performance for surveying applications. The following block diagram presents a simplified overview of a GPS receiver. There are a number of components, the antenna and antenna pre-amplifier, the radio frequency (RF) section, often termed the "front end", the signal tracking loops, the microprocessor, the power supply, the data storage medium and the user interface. The GPS Receiver The Antenna and Antenna Pre-Amplifier A GPS antenna is required to convert the electromagnetic signals transmitted by the satellites into electric signals that can be utilised by the receiver electronics. For surveying applications, the antenna type used almost exclusively by equipment manufacturers is the microstrip patch antenna. The microstrip element resembles a flat, rectangular, copper looking metal strip. Several of these metal patches are often used. The microstrip design is suited to surveying as it is rugged in its construction, easy to manufacture and has a low profile, in other words, tracks well at low satellite elevations. The current trend in GPS system development is to improve positioning performance using intelligent combinations of both carrier frequencies. This requires the GPS antenna to be capable of receiving signals at both the L-band carrier frequencies. Some antennas are designed for use with L1 frequency receivers, therefore, are capable of receiving the L1 band 51

52 only. Users should note that an antenna designed for use with a single frequency receiver is not compatible with a dual frequency receiver. One of the more difficult errors to eliminate from the GPS measurement is the effect of multipath. Multipath is caused by signal reflection on its path from the satellite to the antenna. One method by which multipath can be reduced is by using an antenna ground plane. The microstrip antenna requires a ground plane, albeit a very small one, to operate effectively. For static surveying applications, the ground plane can be extended to a diameter of approximately half a metre. The use of these larger ground planes is recommended for use with all stationary GPS receivers. One very important feature of a GPS antenna is the stability of its phase centre. The mechanical and electrical centre of the antenna, generally, do not coincide. This causes position discrepancies if the antennas are not aligned correctly. In addition, the electrical centre of the antenna may vary with the azimuth and elevation of the satellite. If an assumption is made that antennas manufactured using the same process have similar characteristics, aligning antennas in the same direction can reduce this effect. Users should be aware that problems may exist when different antennas are mixed in the same survey. Fortunately, the phase centre of most antennas currently being manufactured is quite stable and the magnitude of effects are only a few millimetres. Another feature of the GPS antenna that requires consideration is the antenna gain pattern. The gain pattern describes the ability of the receiver to track signals at certain azimuth and elevations. An ideal antenna can track signals low to the horizon, at all azimuths. In general, the tracking performance of microstrip antennas is suitable for surveying The GPS signals are very weak by the time they reach the antenna. As a result, most antennas have a built in pre-amplifier which boosts the level of the signal before it is passed on to the receiver. The amplifier is generally housed in the base of the antenna structure and is powered by a voltage which passes along the coaxial cable that also transfers the signal to the receiver. Most pre-amplifiers can operate successfully at a range of voltage levels, however, users should be aware that most receivers will only send a voltage that is suitable for the antenna which is designed to operate with the receiver. Therefore, it is possible that the voltage output by the receiver is insufficient to power the antenna pre-amplifier, resulting in a system which is unable to track satellites as the receiver never receives signals from the antenna (or the signals are too weak to use). This situation may occur if different makes and model of receiver and antenna are mixed. Users should be aware that such problems may exist and consult their equipment documentation before attempting to mix antennas. In a worst case scenario, it should be noted that sending too much voltage to the antenna may cause damage to the pre-amplifier. This damage is, most likely, not covered by the equipment warranty. The signal from the antenna is transferred to the receiver via a coaxial antenna cable. The signal being transferred from the antenna to the receiver can be distorted if the antenna cable is badly bent, coiled or otherwise damaged. The antenna cable also serves as the mechanism by which power is transferred to the pre-amplifier. Therefore, the role of the antenna cable is as important as the antenna and receiver. Users should take care of antenna cables by minimising the coiling and wear on the cable. It is recommended that antenna cables be replaced periodically, especially, if the surveys being performed expose the equipment to harsh treatment. Antenna cables generally come in standard lengths ranging from three metres to thirty metres. In instances where a GPS antenna is to be mounted permanently for use as a reference receiver, antenna cable lengths which are greater than the standard lengths may be required. In this situation, an in-line amplifier can be purchased which provides additional amplification of the signal on its way to the receiver. 52

53 The GPS Receiver The Radio Frequency Section The radio frequency (RF) section of the receiver is the first section of the receiver that performs operations on the signal after its amplification by the antenna pre-amplifier. This section of the receiver is often termed the "front end" of the receiver. The purpose of the RF section is to take the incoming signal and reduce its frequency to a more manageable frequency, termed an intermediate frequency (IF). This reduction in frequency is generally performed in several stages by mixing the incoming signal with a pure sinusoidal signal. The sinusoidal signal is generated by the receiver clock, termed the local oscillator. Most receivers use inexpensive quartz crystal sources to generate internal signals. As the quartz crystal receiver clock is less stable than the atomic clocks used to generate the transmissions aboard the satellite, corrections need to be applied to remove this clock error from position solutions. The newly created intermediate frequency signal contains the same modulated information as the original signal, however, is shifted in frequency. The new frequency is simply the difference between the original frequency and the frequency of the sinusoidal signal generated by the receiver clock. This is often termed the carrier beat frequency. This beat frequency signal, with all its modulation, is then passed to the signal tracking portion of the receiver. The GPS Receiver Signal Tracking Loops The antenna of the GPS receiver simultaneously receives signals from a number of satellites (the greater the number of satellites, the more reliable and efficient the positioning process). These signals are amplified in the antenna, then reduced to an intermediate frequency by the front end of the receiver. The next stage of the range measurement process is to distinguish between the different satellite signals that are simultaneously being received. This distinction is performed by the signal tracking portion of the receiver. The receiver isolates the signals using a number of channels. For surveying applications, a set of dedicated channels are employed. Each channel is responsible for isolating one satellite only. Such a receiver uses a number of these dedicated channels, all operating simultaneously, or in parallel, thus the term parallel tracking architecture. Each satellite transmits a unique C/A-code (or portion of the P-code) which identifies it from any other satellite. The signal tracking channel measures the pseudorange derived from the code by demodulating it from the carrier signal using a sequence of tracking loops. The tracking loop enables the receiver to "lock" on to a particular satellite transmission. Generally, two types of loops are used in conjunction with each other to measure the range and phase. A delay-lock loop is used to align the binary pseudo-random noise (PRN) code sequence arriving from the satellite with a locally generated replica. The replica is generated by the local oscillator. The local replica and incoming signal are cross-correlated to produce a large "number" if the satellite sequences are aligned. If a low number is generated, the signals are shifted in an attempt to obtain a high correlation. Once a high correlation is achieved, the satellites is said to be "locked". It should be noted that the satellite code sequences are designed in a manner to ensure that two different satellite sequences always produce low correlation. This prevents an incorrect satellite being tracked. When the satellite is locked, the amount of shifting of the replica signal required to achieve high correlation gives rise to a time delay which, when multiplied by the speed of light, yields the pseudorange measurement. Once the pseudorange measurement has been made, the PRN code can be removed from the signal by mixing it with the locally generated replica. The remaining signal is then passed to a second loop, generally, a Costas loop. The Costas loop is a variant of the phase-lock loop, specifically designed for bi-phase modulated signals. This loop is capable of aligning the 53

54 carrier signals and measuring the difference in phase between the beat signal and the locally generated replica. However, due to the nature of a pure sinusoidal signal, the carrier measurement is ambiguous at the whole carrier cycle level (when expressed as a range measurement). This carrier cycle ambiguity must be identified during data processing to achieve survey accurate results using the GPS signals. The combination of the delay-lock and Costas loops to extract the information from the incoming satellite signals is termed code-correlation and provides the most precise measurements obtainable. It should be noted that this technique requires knowledge of the PRN code for the delay-lock loop to be implemented. Currently, the GPS signal comprises one freely available code signal, the C/A-code. The C/A-code is only transmitted on the L1 carrier frequency (Note: the C/A code will also be available on the L2 carrier phase signal with the launch of the Block IIR_M satellites). This enables the code-correlation technique to be used to measure the L1 signal. Unfortunately at present, the C/A-code is not transmitted on the L2 frequency, rather the classified Y-code is broadcast. This precludes the use of the code-correlation technique to recover the L2 signal. The use of both carrier signals is beneficial to surveying applications. As a result, a number of other techniques have been developed to recover the L2 signal. These techniques include signal squaring, Z-tracking, cross-correlation, and code-aided squaring. Each of the techniques is capable of providing L2 carrier information, however, with some penalty when compared to the code-correlation technique. One feature of note is the current discussion within the GPS industry regarding an additional civilian frequency. Should a second civilian code be available on a frequency other than L1, code-correlation techniques will be able to provide more precise phase measurements than currently obtainable on the L2 carrier. The GPS Receiver Microprocessor Many of the functions of the GPS receiver are digital, rather than analogue, operations. The control of operations such as initially acquiring satellites, tracking the code and carrier, interpreting the broadcast navigation message and computing satellite receiver coordinates, is performed by a microprocessor. The microprocessor runs a program which is stored on a memory chip within the receiver. When the receiver firmware is upgraded to a new version, it is this program which is being updated. The power of the microprocessor defines a number of related characteristics including; computing ability (whether a simple positioning solution or a real time carrier phase survey is supported), speed of signal acquisition and calculation, receiver size, and power requirements. The control of the user interface and communication ports are also controlled by this microprocessor chip. In general, the more powerful the processor, the greater the functionality that can be supported by the receiver. The GPS Receiver Power Supply For surveying applications, receiver portability is a key design aspect. Currently available GPS receivers are generally supplied with a number of batteries. The power supply of the GPS receiver is designed to accept a DC power source, thus the convenient use of batteries. Unfortunately, the power consumption of modern survey receivers is such that a number of small batteries are required to 54

55 operate equipment for a full working day. Alternatively, larger batteries can be used for stationary receivers which do not have to be transported frequently. When purchasing a receiver, an AC to DC power converter is generally provided which enables mains power to be used to power the receiver. This is especially useful for permanent reference stations and for providing a power supply for transferring data from the receiver to a computer. Many manufacturers also offer a vehicle cigarette lighter adaptor which enables a car battery to be used to power the receiver. In accordance with the use of a total station, data collector or any other electronic device used for surveying purposes, users should ensure that batteries are in good working order and are suitably charged. The manufacturer s recommendations for battery maintenance should be adhered to as several battery technologies, such as lead acid and nickel cadmium, may be used. The GPS Receiver Data Storage For almost all surveying applications, some information is required to be stored. This information may comprise pseudorange and carrier measurements for later processing, station identifiers and antenna height details, or position estimates determined by the receiver microprocessor. The nature of the device used to store such information varies greatly between receivers. Examples of data storage devices currently used include custom memory cards, PCMCIA compatible memory cards, internal random access memory chips, externally connected data collectors, and externally connected computers. Many of the storage mediums are compatible with other survey equipment, such as total stations and data collectors. Some devices are more robust and environmentally rugged than others. In summary, each device has its strengths and weaknesses and users should assess the benefits of each type of storage medium for their own implementation. The GPS Receiver User Interface The mechanism by which the user interacts with the GPS receiver is via the user interface. GPS systems are designed for different purposes, thus, the nature of the user interface varies greatly across products. Some receivers employ built in keyboards and liquid crystal displays (LCD). These keyboards are generally limited due to size constraints not permitting a full alphanumeric keyboard. Other receivers modify the display to support keys which perform more than one function. For a complete keyboard interface, users must generally move to a data collector controlled receiver. Data collectors are generally quite rugged, however, have limited storage compatibility and often have limited screen displays. Alternatively, a pen based computer may be able to be connected to control receivers, 55

56 providing large screen real estate, hard drive storage and full keyboard. Again, the type of user interface adopted will depend on the needs of the user. The GPS Receiver Surveying Using Global Navigation Satellite Systems Observation Techniques Introduction Static Surveying Rapid Static Surveying Stop And Go Kinematic Surveying Continuous Kinematic Surveying Real Time Kinematic (RTK) Surveying Initialisation Techniques Introduction There are a number of different operational techniques that can be used to collect satellite measurements from the GPS satellites for surveying purposes. Each technique provides, essentially, the same result; a three-dimensional coordinate difference, or vector, between the reference and rover receivers. This vector is referenced to the World Geodetic System 1984 (WGS84) coordinate datum. The reason for the number of observation techniques is to facilitate more efficient surveying practice under a variety of conditions. Similar observation technique variations are routinely employed by surveyors using terrestrial equipment. For example, a control survey over a small area is generally accomplished by traversing the perimeter of the region of interest. By closing the traverse, a check is provided on the acquired measurements. For a detail survey of the same region, the traverse stations may be used as the framework for a larger number of radiations. The radiation observations do not have the redundancy of the traverse observations. For more precise control work over longer distances, a network of angles and distances may be observed. This provides a high level of redundancy and is a more reliable position determination technique. Similar techniques are employed when performing GPS surveys. Users should be well aware of the limitations of each technique so they can ensure appropriate application of the various GPS surveying techniques. For surveying applications, all surveys are performed using relative positioning techniques. The use of the reference/rover station nomenclature is used throughout the description of each observation technique. The carrier phase observable is utilised in all instances. In addition, it is assumed that the carrier phase ambiguities are identified and constrained to generate the adopted vector solution. All satellite measurements must be observed simultaneously by all receivers. Observation Techniques Static Surveying The static observation procedure is the most commonly used GPS observation technique due to its reliability and ease of data collection. The results generated from static observations are the most robust of the GPS positioning solutions due to the increased length in observation period. All control surveys over reference/rover separations of several kilometres are performed using static surveying techniques. The static procedure requires satellite measurements to be acquired simultaneously at multiple sites by stationary receivers. Static surveys are performed by setting receivers on stable platforms, usually a tripod or survey pillar, and leaving them to record measurements at predetermined intervals for a period of time. Observations are usually collected at a rate of one epoch every 5, 10, 15 or 30 56

57 seconds. The rate of collection is not of prime importance in static work and measurements should be acquired at rates which are related to the amount of available data storage space. More importantly, data needs to be collected for a sufficient time period to enable the integer cycle ambiguities to be determined. In addition, the effects of multipath and random measurement error can be reduced by observing for longer periods. There are no hard rules for determining how long data should be collected for. The time required to achieve a suitable accuracy is a function of the number of visible satellites, baseline length, multipath conditions, atmospheric conditions and satellite geometry. Using past experience and knowledge, the surveyor may choose an observation period that he or she feels is sufficient to resolve ambiguities and obtain an accurate position. To make full use of the acquired measurements, static surveys should be post-processed. This requires the storage of the observables to be merged in the processing software at a later time. This implies that results are not available in the field, as well as, implying that the time period required to obtain a required accuracy is a calculated guess. Experience under similar survey conditions generally defines the observation period, however, for accurate results, observation periods of less than 30 minutes should not be used for lines greater than five kilometres in length. Observation periods will also need to be longer if single frequency receivers are used as wide-lane and narrow-lane combinations will not be able to be formed. It should be noted that the longer the observation session, the more accurate the calculated position would be. If post-processing reveals that results are unsatisfactory, the baseline will need to be observed again. It is, therefore, wise to use caution when estimating observation periods as an additional five or ten minutes per point may be sufficient to prevent further observation. The main drawback of static procedures is the lack of productivity. If, for example, observation periods of 45 minutes per point are adopted, it may only be possible to collect five or six points per day depending on the time required to move between marks. To increase efficiency, multiple receivers can be used simultaneously. Many receiver manufacturers sell receivers in groups of three for this reason. Each receiver remains stationary at the same time to enable three points to be occupied. Each observation period is termed a session. Once each session is completed, the receivers move and begin acquiring measurements simultaneously at three other stations. For each three receiver session, three baselines can be generated. Technically, only two of the baselines are independent as the third uses measurements already used by the other two lines. However, it is recommended that all baselines are processed and adjusted using a least squares estimation procedure. The statistical output for each baseline will reflect the correlated nature of the third baseline. This use of sessions with multiple receivers enables surveys to be formed to generate networks of baselines. The greater the number of receivers, the greater the productivity. The number of baselines per receiver combination can be computed by summing the number of receivers minus one, down to one. Three receivers yield 2+1=3 baselines, four receivers yield 3+2+1=6 baselines, five receivers yield =10 baselines etc. It should be noted that increased receiver numbers generally require increased personnel and organisation. The following diagram shows how three receivers can be used to collect six stations using three sessions. The nature of the three sessions provides a network of connected baselines. 57

58 Observation Techniques Rapid Static Surveying The rapid static surveying technique was developed in an attempt to improve the efficiency of the static survey procedure. Users should note that the observation procedures for rapid static surveying are the same as those for static surveying. The only difference is the length of the occupation period which is less than that required for static surveys. This reduced occupation length is facilitated by mathematical improvements in processing software which enable the integer ambiguities to be determined using less observations. In order to perform surveys with maximum efficiency, a dual frequency receiver which is capable of pseudorange and carrier phase measurements on both carrier frequencies is required. This enables the wide-lane and narrow-lane phase combinations to be used to aid in the estimation of the ambiguities. The use of the pseudorange measurements (after they have been smoothed) also assists in the rapid determination of the ambiguities, therefore, receivers capable of code measurements in the presence of anti-spoofing are desirable. As the occupation period of rapid static surveys is shorter than that of static surveys, it can often be extremely difficult to manage the movement of receivers in sessions. Therefore, rover receivers generally occupy points as efficiently as possible. Occupation times as short as ten minutes are often sufficient to resolve the integer ambiguities over short baselines when at least six satellites are tracked. This results in a series of radiation type vectors from the reference station. To provide some redundancy, a second reference station may be used. This provides two vectors to each point. It should be noted that this method of data collection is unsuitable for detecting erroneous vectors if the rover station is the source of error. An example of such an error is incorrect entry of the antenna height. A second occupation of each station, ideally using a different reference receiver, is preferable, however, this reduces survey efficiency. Surveyors must use their professional discretion in the manner in which rapid static surveys are conducted. The rapid static survey procedure is most efficient when dual frequency receivers are used and baselines are kept below five kilometres. If the points of interest are free of overhead obstructions and six or more sufficient satellites are observed, surveys can be performed in a matter of minutes. In this environment, the time required to set up the antenna on a tripod makes are significant contribution to the time spent at each point. To improve efficiency, a bipod arrangement may be used with the receiver contained in a backpack. Observation Techniques Stop And Go Kinematic Surveying The kinematic survey procedure was developed in the mid 1980 s as an attempt to improve the productivity of surveying with GPS receivers. In applications where a large number of points spaced less than a few kilometres apart need to be coordinated, static procedures are inefficient and are, generally, not cost effective. The kinematic survey technique is ideally suited to such applications where points are closely spaced and easily accessed. 58

59 Once the carrier phase integer cycle ambiguities have been determined, they do not change value if continuous tracking of the satellites is maintained, i.e. there are no cycle slips. The kinematic survey procedure is based on this characteristic of carrier phase positioning. A short initialisation procedure is performed with the primary aim of determining the integer ambiguity values. Once this initialisation is completed, the rover receiver occupies points of interest for a short period, generally, less than one minute. As the ambiguities do not change when satellites are continuously tracked, centimetre level accuracy can be obtained with a brief stationary occupation. Once a point has been occupied, the rover moves to the next point of interest where it acquires another minute of data. During the period in which the receiver is being transported between the two sites, satellite tracking must be maintained. This technique of continually moving, then stopping briefly, is termed stop and go kinematic surveying. The stop and go kinematic surveying procedure is performed to coordinate the position of stationary marks. The receiver must still continuously track satellites while in motion to preserve the integer ambiguity estimates. However, the position of the receiver while moving is not of interest. If cycle slips are experienced while moving, the integer ambiguity term for the interrupted satellite must be calculated again. If at least four satellites are still being tracked, this can be performed automatically by the processing software without user intervention. This is possible as the receiver can calculate its position using the remaining satellites. This position is then held fixed (constrained) and the unknown integer ambiguity estimated. This technique is termed the known baseline initialisation and is performed frequently when cycle slips are present. If the cycle slips cause the number of satellites to drop below four, the survey must be re-initialised to determine the integer ambiguities as their values will have changed. There are four ambiguity initialisation techniques which can be used at any time throughout a kinematic survey. Some of the techniques are more reliable than others and more strongly recommended. Observation Techniques Continuous Kinematic Surveying The stop and go kinematic surveying technique is extremely effective in coordinating closely spaced features. The features of interest are stationary and are occupied briefly, generally for less than one minute. The integer cycle ambiguities are initially estimated using one of four initialisation techniques. Satellite tracking is maintained throughout the survey to facilitate centimetre level position accuracy for the points of interest. If cycle slips are present, the survey needs to be re-initialised. Importantly, the position of the receiver while in motion is not of interest in stop and go surveys. The continuous kinematic survey technique is identical to the stop and go procedure, except the position of the receiver while it is in motion is now of interest. As long as the satellites are tracked without interruption, the position of the antenna can be estimated at each measurement moment, termed epoch. The epoch rate must be set carefully to ensure that position estimates are computed at a desirable frequency. An example of where a continuous kinematic survey may be practical is in the coordination of a train track. The antenna can be placed on the train and 59

60 driven to digitise the track. Surveyors should build redundancy into kinematic surveys by occupying marks on multiple occasions, or in the continuous kinematic case, re-traverse the same route. One feature of continuous kinematic surveys that must be considered is whether the height of the receiver is of interest while the receiver is in motion. If this is the case, then the height of the antenna above the ground must be kept uniform. This can be accomplished using a range of devices, many of which are best developed by the surveyor for a specific use. One example of such a device, which mounts on the tow bar of a vehicle, is illustrated. The device is such that a bicycle wheel rides along behind the vehicle with the antenna mounted on a pole above the wheel. The device is engineered to keep the antenna at a constant height by enabling the bicycle wheel to be at a different height to the vehicle wheels. Observation Techniques Real Time Kinematic (RTK) Surveying The satellite carrier phase measurements collected at both reference and rover receivers can be stored using a number of different media, then combined in a computer for post-processing. The restriction of this approach is that the results of the survey are not known until after the survey has been completed. Real time processing techniques utilise a data link, usually in the form of a radio, to transfer corrections acquired at the reference receiver (set up on a survey mark) to the rover receiver. The microprocessor in the rover receiver then combines the reference and rover information and computes the rover coordinates as the survey is being performed, i.e. in real time. The corrections broadcast are the difference between the known coordinates of the survey mark and the calculated coordinates of the survey mark by the reference receiver. This capability enables surveyors to check coordinates in the field, ensure surveys are being performed successfully and facilitate establishment of features at pre-determined locations (setting out). If the GPS equipment facilitates all four initialisation methods, including on the fly, then surveyors can simply occupy marks and wait until the receiver display indicates that the ambiguities have been resolved by the on the fly method. The antenna swap essentially becomes obsolete as occupation of previous marks (known baseline), static survey and the on the fly technique can be used in combination to perform surveys more efficiently. The major advantage of this field procedure, termed real time kinematic, is that users are aware of the status of the survey as it is being performed. Therefore, surveys can be performed to maximum efficiency as the minimum amount of data required to resolve the integer ambiguities can be determined. A large amount of literature is available with regard to the observation techniques referred to in the above paragraphs. The links section of this document contains a list of literature which contains details descriptions of these techniques. Observation Techniques 60

61 Initialisation Techniques Before a feature of interest can be coordinated to an accuracy suitable for surveying applications, the carrier phase integer cycle ambiguities must be determined. In a kinematic survey, this ambiguity resolution process is termed initialisation. The primary purpose of the initialisation procedure is to identify the ambiguity values. There are four techniques that can be used to perform kinematic survey initialisation, the most frequently used being on the fly resolution. Other less frequently used techniques include static survey, known baseline and antenna swap.. On The Fly On the fly resolution, computes the integer ambiguities while the receiver is motion. To perform this efficiently, a minimum of five satellites are required, however, six or seven satellites are preferred. In addition, on the fly techniques should not be attempted with single frequency receivers as the process is extremely inefficient. The advantage of this technique for surveyors is that previous marks do not need to be located as frequently for initialisation purposes. An example of the flexibility of the on the fly initialisation procedure is where a surveyor collects points, then passes under a bridge. By moving under the bridge, satellite tracking is interrupted and the survey must be re-initialised. A static survey could be performed on the other side of the bridge, however, this may be time consuming if the distance to the reference receiver exceeds several kilometres. With the on the fly technique, the surveyor can continue moving to the next feature of interest. While the surveyor is moving, the satellites will be re-acquired and the on the fly resolution scheme will automatically begin to resolve the ambiguities. In general, the ambiguities are safely resolved in less than five minutes. In most instances, two minutes of tracking six or seven satellites is sufficient. Once initialised, the survey can proceed as normal. Users should note that if there are only four satellites being tracked, the on the fly technique cannot operate. In addition, if five satellites are observed, or the satellite geometry is poor, initialisation times may exceed ten or fifteen minutes. Before a kinematic survey is performed using on the fly techniques, users should consult their equipment documentation for specific details of the supported functionality. Static Survey The static survey initialisation is identical to a static survey performed to coordinate points of interest. In order to coordinate features using static techniques, the observation period is designed to facilitate the identification of the integer ambiguities. Therefore, performing a static survey results in both the coordinates of the mark, as well as, the ambiguities. To perform a static initialisation efficiently, the reference/rover separation should be kept short to minimise the static survey observation period. If a cycle slip occurs which requires reinitialisation, a new point can be occupied and that point considered to be a static occupation. Once sufficient measurements are considered to have been observed to complete the static survey, kinematic occupations can proceed. Some manufacturer implementations have certain restrictions when initialising using the static method. Users should refer to their equipment documentation for specific details. Known Baseline The known baseline initialisation technique requires knowledge of a previously determined GPS vector. This position vector can be derived from a previous static or kinematic survey. In addition, any point occupied in the current survey can be used to initialise the survey. The known baseline procedure is based upon constraining the position vector and only estimating the unknown ambiguities. Theoretically, one measurement epoch is sufficient to perform this estimation, however, one minute of observation is generally recommended. Users should be aware that the points occupied after such an initialisation procedure are dependent on the success of the initialisation. As a result, it is best to favour a conservative approach when reinitialising surveys. Occupation periods used to initialise surveys should reflect this conservatism. 61

62 If cycle slips occur while the receiver is in motion, however, the number of satellites tracked remains four or greater, the receiver position can be computed using the available satellites. This position is then constrained to enable the unknown satellite ambiguities to be determined. This occurs transparently to the surveyor and does not require modification of the field procedure. In instances where cycle slips cause the number of satellites tracked to drop below four, the known baseline technique can only be used by occupying a previously observed mark. Antenna Swap The antenna swap procedure is a technique which is used to initialise the integer ambiguities at the beginning of a kinematic survey. The main limitation of the antenna swap procedure is that the reference and rover receivers must be within ten metres of each other. In most implementations, two antenna swaps are actually performed and this approach is recommended. The procedure is performed by placing the reference and rover receivers over well defined marks and simultaneously collecting measurements for approximately one minute. The two antennas are then swapped, such that the reference receiver is now located over the rover mark, and vice versa. A further minute of observations are then collected. These two steps are sufficient to resolve the integer ambiguities, however, a further swap is strongly recommended. To complete the antenna swap procedure in this manner, the reference and rover antennas return to their original locations for a further minute of observation. The rover receiver can then proceed to points of interest. It is vital that continuous tracking of the satellites occur during the antenna swap procedure. If a survey is to be performed by one person, the antenna swap procedure may be difficult to perform. The use of a third tripod can be used to accomplish the swap procedure. As mentioned initially, the biggest weakness of the antenna swap procedure is the need to remain adjacent to the reference receiver. This limits the applicability of the procedure in most surveying applications. In addition, the mechanism of the swap procedure is awkward as the reference receiver must be moved. Therefore, a simpler approach is to place a well defined mark within ten metres of the reference receiver and perform a short static survey. As little as ten minutes of measurement is generally sufficient to resolve the ambiguities, even using a single frequency receiver. A known baseline initialisation can then be performed on the same marks to initiate the survey. Observation Techniques Surveying Using Global Navigation Satellite Systems 62

63 Coordinates and GPS Introduction The Spheroid The Geodetic Datum Geocentric Datum of Australia (GDA 94) Heights and GPS Control Requirements for GPS Surveys Introduction The Global Positioning System (GPS) is a satellite based system which is designed for global navigation and positioning. This implies the use of a coordinate reference surface which is global in nature (World Geodetic System 1984). Coordinates determined from GPS measurements are referenced to this global reference surface. In order to correctly use coordinates derived from GPS measurements, an understanding of the coordinate system used by the GPS control segment is required. Coordinates and GPS The Spheroid (or Ellipsoid) The shape of the Earth approximates a sphere which has been "squashed" at the poles. This phenomenon is due to the gravitational field of the Earth and the centrifugal force caused by the Earth's rotation about the axis of the poles. The effect of centrifugal force is a maximum at the equator. The total force of gravity is determined by the gravitational force (directed towards the centre of the Earth) minus the centrifugal force. At the poles, there is no centrifugal force, thus the force of gravity is greater than at the equator. This causes the squashing of the Earth at the poles. A figure which resembles a sphere squashed at the poles is more correctly represented by an ellipsoid (often termed a spheroid). An ellipsoid is an ellipse which has been rotated about its minor axis to form a three dimensional solid of revolution. This surface forms the basis for all geodetic coordinate systems used throughout the world. The coordinates derived from GPS measurements are also directly referenced to a spheroidal surface. The shape of the spheroid can be defined by two parameters, the semi-major axis and the semi-minor axis. A sphere is represented by a spheroid with equal semi-major and semiminor axes. Commonly, the semi-major axis and flattening are used to define the spheroid shape. The flattening is a quantity which is computed from the difference between the semimajor and semi-minor axes, divided by the semi-major axis. It is a dimensionless quantity which provides an indication of the amount of "squashing" of the spheroid. A sphere has a flattening of zero, a plane has a flattening of one. There are a number of spheroids used throughout the world. Some are designed to be used over a specific area of interest, such as Australia, and are termed regional spheroids. The Australian National Spheroid (ANS), which is used as a basis for the Australian Geodetic Datum (AGD), is such a spheroid. Other spheroids are designed for use across the entire 63

64 planet and are termed global spheroids. The GPS system adopts the Geodetic Reference System 1980 (GRS80) spheroid, which is global in nature. Coordinates and GPS The Geodetic Datum A reference surface which can be used as a basis for geodetic coordinates is referred to as a geodetic datum. A datum is comprised of a number of elements - a spheroid which has a defined size and shape, a location, or origin, in three dimensional space, and an orientation of each of its axes. The definition of these elements fixes the datum is space and enables users to reference points on the Earth to the defined coordinate reference system. The datum used by the GPS system is the World Geodetic System 1984 (WGS84). This system is a geocentric based coordinate system with the origin of the defining spheroid located at the Earth s centre of mass. Until recently, the Australian Geodetic Datum 1966 (AGD66) was the most commonly used coordinate reference system for many projects in Victoria. The AGD66 is a regional datum which is designed to best fit the surface of the Earth over the Australian continent. The Australian National Spheroid is used and has a different size and shape to the GRS80 spheroid. The AGD66 is located by the coordinates of the Johnston Geodetic Station. The coordinates of the Johnston Station have been derived from astronomical observations. The orientation of the AGD66 is defined by the BIH zero meridian and the Earth's mean axis of rotation at epoch This definition is different from that of the WGS84, therefore, transformations must be performed to convert GPS coordinates to the AGD66 coordinate system. Geocentric Datum of Australia (GDA 1994) As the use of GPS for surveying, navigation and recreation in Australian increased, it was decided that a more compatible mapping datum with the GPS Datum (WGS84) was required. In the year 2000, Australia moved from the regional Australian Geodetic Datum to a geocentric datum termed the Geocentric Datum of Australia 1994 (GDA94). This new datum is compatible with the WGS84 datum and users in Australia do not have to perform transformation computations and can obtain GDA94 coordinates directly from GPS observations. Also, this means that GDA coordinates will be more compatible with other worldwide geocentric datums. Geoscience Australia have recently published the technical specifications of GDA 94 available at Surveyors should be aware of the effect of changing from a regional datum to a geocentric datum in Australia. In the instance of moving from AGD to GDA, there will be an approximately 200m shift in the north-east direction between the two coordinate systems. There are a number of documents available on the internet explaining the effects of the change from AGD to GDA. The ICSM web page on GDA 94 contains a number of documents explaining the effects of the implementation of GDA94, including a GDA94 technical manual. Coordinates and GPS 64

65 Heights and GPS The coordinates derived from GPS measurements are three dimensional coordinates which are commonly presented in terms of Cartesian coordinates (X,Y,Z) or, equivalently, ellipsoidal coordinates (latitude, longitude, height). The height component, as well as the horizontal component, is referenced to the WGS84 spheroid. Users must be aware that GPS heights are spheroidal in nature and are not directly compatible with mean sea level (MSL) heights. Therefore, the height values associated with GPS coordinates may indicate that water flows uphill! The sea level surface of the Earth can be approximated by an equipotential surface. An equipotential surface is one where the gravitational potential is equal at all points on its surface. Therefore, the direction and magnitude of gravity is an essential component in the definition of an equipotential surface. There are an infinite number of these equipotential surfaces, effectively forming an onion skin around the Earth. One of these surfaces, termed the geoid, is of particular interest. The geoid is defined as the equipotential surface that the Earth's oceans would cover if all land mass was removed. The geoid provides a surface which is sufficiently close to the mean sea level surface, thus can be used as a reference surface for height values. In Australia, the Australian Height Datum (AHD) is the practical realisation of such a height reference surface. Tide gauge readings on the Australian coastline and spirit level observations have been used to define a datum surface which can be considered, for all practical purposes, mean sea level. This National datum was provided by a national adjustment incorporating levelling measurements and tide gauge data in As the direction of gravity changes across the surface of the Earth, due to mass differences, this datum surface, unlike the spheroid, is not uniform. Heights above the geoid are often referred to as orthometric heights. The difference between the geoid (AHD) and the spheroid at a particular point on the Earth is termed the geoid undulation or geoid-spheroid separation. Surveyors must apply the geoid undulation to GPS derived height values to obtain heights which are related to mean sea level. There are a number of methods by which the geoid undulation can be computed. The methods can, however, be grouped into two general categories, gravimetric techniques and geometric techniques. As the name suggests, gravimetric geoid estimation techniques are based on global and regional measurements of gravity. A gravimeter is used to measure the force of gravity at discrete points on the surface of Earth. The measurements are then manipulated to provide estimates of the geoid undulation. This process is quite complex and is not feasibly performed by the surveyor. In addition, surveyors are not in a position to take gravity measurements every time they wish to perform a GPS survey. Fortunately, the Australian government has realised the importance of determination of an accurate geoid model for Australia. Geoscience Australia has made available a geoid model of Australia based on global and Australia wide gravity measurements. AUSGEOID98 is a 2 by 2 database of geoid-spheroid separation values which can be accessed via the Geoscience Australia website. The surveyor can simply access the web site and provide the position of the points of interest to obtain geoid undulation estimates. Geoscience Australia also provide free software which can be used to interpolate geoid-spheroid separation values at particular locations. 65

66 The accuracy of such geoid models is generally of the order of a few centimetres, however, the accuracy of the models is continually improving as new gravity measurements are incorporated into the estimation process. For large survey regions, generally those greater than 10km by 10km, gravimetric techniques are the most appropriate means of estimating the geoid undulation. For smaller survey regions, geometric techniques can be applied to estimate the geoid in the area of interest. The most commonly used geometric technique is the modelling of the geoid to a linear plane. The assumption of the technique is that a linear relationship exists between the spheroidal surface and the geoidal surface. The plane is defined by three parameters, a constant value plus slopes in both east and north directions. In general, this technique is suitable for survey regions less than ten kilometres in extent. Many of the commercially available GPS software packages and data collectors enable this calculation to be performed. The surveyor must provide estimates of the geoid undulation at three points which are well spaced throughout the survey area. The geoid undulation is determined by occupying points with known orthometric height during the GPS survey. The software then uses this information to estimate the parameters of the linear plane and applies the transformation to all remaining points. Coordinates and GPS Control Requirements for GPS Surveys The surveyor must occupy a number of control points if the coordinates derived from GPS occupations are to be integrated into an existing coordinate framework. This is no different from a terrestrial survey where a point with known coordinates is occupied and a reference direction sighted. Alternatively, two points with known coordinates may be occupied, enabling a rotation, scale factor and east and north translations to be determined. In a level survey, an existing bench mark is occupied to enable the measured staff readings to yield heights of points above a known reference datum. Control points and benchmarks can be located using the Survey Mark Enquiry Service (SMES). When performing GPS surveys and assuming that the coordinate datum of interest is AGD66, rather than WGS84 or GDA94, then certain requirements must be adhered to if horizontal coordinates are to be determined in the desired coordinate datum. For horizontal integration, a minimum of two points with known east and north in the desired coordinate system must be occupied as part of the GPS survey. Users should note that these points do not necessarily need be occupied by a reference receiver, although setting up reference receivers on known points is a common practice (see ICSM SP1 Publication). Occupation of two points enables the determination of four parameters, a scale factor, rotation and translations in east and north. The occupation of only two points with known coordinates is, however, not recommended as standard practice. A third point should be occupied to enable the surveyor to check that the control points are homogeneous and that gross errors have not been made. The Victorian Geodetic Network has been built up over many years since the National adjustment of Survey instrumentation and adjustment techniques have been steadily improving since that time with the current usage of GPS (when used properly) giving the most accurate results. It may be that GPS can deliver a higher precision of measurement than those used to define the control points originally, however, to meld the survey at hand to existing values, a professional judgement may need to be made as to best fit. In such instances, an additional point should be occupied to provide a check on the homogeneity of the control points. Needless to say, if more than three control points are available, they should be occupied. Most GPS processing packages facilitate least squares transformation parameter estimation so increasing the number of control points does not cause a computational problem. The example highlighted in the diagram shows three control points with known horizontal coordinates denoted by the blue triangle symbol. There are four points with coordinates which need to be determined, denoted by the red circle symbol. The rapid static surveying technique is selected as two receivers are available for use. One receiver is setup as a reference at 66

67 control point A. The four unknown points and the third control point are occupied by the rover receiver. The reference receiver is then moved to control point B. The remaining five points are then re-occupied by the rover. This provides two independently observed vectors to each of the unknown points and the third control point. The coordinates of A and B (or any combination of the three control points) can be used to determine transformation parameters enabling the GPS vectors to be integrated into the desired coordinate datum. Alternatively, all three control points can be constrained and least squares techniques used to estimate transformation parameters. As a matter of note, the survey could be improved by occupying both control points A and B and including the vector between them. For vertical integration, the control point requirements will depend upon whether the geoid is being modelled using gravimetric or geometric techniques. If gravimetric techniques are being used, one point with a known elevation is required. This is due to the fact that all precise GPS positioning uses the relative positioning method. Therefore, the application of the GPS baseline vectors needs one fixed point to begin propagating height differences. Again, it is strongly recommended that several points with known height values be occupied to enable checks on the geoid model to be performed. In many instances, there are a number of benchmarks in the near vicinity of the survey area as Victoria has an excellent geodetic control framework. These benchmarks should be occupied where practicable. Surveyors should ensure the height values of these benchmarks have been established previously to a sufficient level of accuracy to be used as a control point. Surveyors should not consider this to be a major increase in the amount of work that needs to be performed as a single GPS vector over several kilometres can be easily observed. Surveyors will, naturally, be reminded of the increased time required to run a flight levels between the survey region and bench marks. GPS techniques alleviate this work load, however, generally cannot achieve the same accuracy standards as spirit levelling observations. If vertical integration is to be performed using the linear plane geometric technique, then a minimum of three control points are required. It must be noted that the horizontal coordinates of these control points are also required to enable the slope parameters to be estimated. The horizontal coordinates can, however, be referenced to any datum. This enables the WGS84 coordinates derived from the GPS measurements to be used to provide the horizontal reference for the linear plane. The points should be well spaced throughout the survey region. A minimum of three points are required, however, additional points should be occupied to enable the validity of the linear plane to be tested in the survey region. In real time kinematic surveying, horizontal and vertical transformations need to be performed in the field. Surveyors should ensure that the controller for their real time kinematic system has the capability to perform such calculations. The terminology used by receiver manufacturers for such functionality includes "transformation", "calibration", and "GPS coordinate system". Users should consult their product documentation for the specific details and limitations of their equipment before attempting these integration computations in the field. Coordinates and GPS Surveying Using Global Navigation Satellite Systems 67

68 Loop Closures Introduction Checking Baselines Observed In Multiple Sessions Internal Accuracy Introduction The GPS surveying techniques are capable of generating centimetre accuracy results if the carrier phase ambiguities are correctly identified and constrained during data processing. The results are generally presented as Cartesian coordinate differences, referenced to the World Geodetic System 1984 coordinate datum. These coordinate differences, or vectors, represent the three dimensional coordinate difference between the reference and rover receiver. In addition to Cartesian coordinates, the vectors can be presented in terms of east, north and height differences. This is commonly performed using a local horizon plane projection. Regardless of the manner in which the vectors are presented, closures of connecting baselines can aid in the detection of erroneous measurements. In the same manner in which a traverse misclose is computed, the three dimensional misclose of GPS vectors can also be determined. GPS surveys are not performed to generate traverse measurement equivalents, therefore, surveyors use manually selected baselines to form loops of baselines. The closures can be performed using a calculator, however, some GPS surveying systems provide loop closure utilities with the data processing software. Intelligent use of loop closures can enable erroneous baselines to be identified. Loop Closures Checking Baselines Observed In Multiple Sessions In order for a loop closure to be performed, GPS baselines are required from more than one observation session. If only one session is used, the baselines are correlated and loop closures will tend to always indicate excellent results. This is due to the correlation between the baselines rather than the quality of the baselines. When multiple sessions are observed, a number of strategies for detecting poor quality vectors can be adopted. Consider the following example where several redundant baselines have been observed. One strategy which may be adopted is to check each triangle while trying to isolate any triangle which reveals poor results. If each triangle is closed, it is likely that a bad baseline will affect more than one triangle. This technique results in often checking correlated baselines from the same session. It is also likely, however, that a session which was too short to enable the ambiguities to be correctly resolved will highlight two low quality baselines. Comparing all triangles will enable such instances to be detected if sufficient baselines are observed. In the example provided, if baseline X is erroneous, it can be anticipated that triangles 1 and 2 will highlight a poor closure. By performing a closure around the four sided perimeter of triangles 1 and 2, the poor baseline can be highlighted. In addition, several of the points have been occupied on more than one occasion. Performing loop closures will aid in detecting whether antenna height errors are present in the data set. In order for the processing software to be able to check for poor baselines it is important that the baselines are measured more then once (i.e. over multiple sessions) to obtain independent baselines. For example in the scenario described in the previous paragraph, the 68

69 baseline X could be measured in the first and second sessions (by holding this baseline fixed), to enable the processing software to do a comparison. Loop Closures Internal Accuracy In much the same way a traverse observed with a total station can be given an accuracy based on the computed misclose, GPS baselines can also be treated similarly. The misclose is computed in all three coordinate components and expressed as a ratio to the total distance of the loop. More commonly, however, this internal accuracy can be expressed in parts per million of the total baseline length. This will easily enable an assessment to be made regarding the loop closure performance in comparison to the manufacturer defined specifications which are generally presented in parts per million. An example of a baseline closure is given by the following screen capture. A total of 6 baselines with a combined length of more than eleven kilometres have been selected. The loops close to within a few centimetres, resulting in a part per million error of just over one and a half millimetres per kilometre. Loop closures should be performed on all networks before least squares network adjustments are carried out. This will simplify the network adjustment process as bad GPS baselines should have been detected and eliminated from the least squares adjustment. Loop Closures Surveying Using Global Navigation Satellite Systems 69

70 Network Adjustment Introduction Loop Closure Minimally Constrained Adjustment Constrained Adjustment Error Ellipses Independent Baselines Introduction When performing networks of GPS baselines, a least squares adjustment of the generated baselines is often performed once processing is complete. These networks may comprise static and kinematic baselines, however, static baselines are generally observed. The network adjustment procedure has several functions in the GPS surveying process. The adjustment provides a single set of coordinates based on all the measurements acquired, as well as, providing a mechanism by which baselines which have not been resolved to sufficient accuracy can be detected. A series of loop closures should be performed before the network adjustment procedure to limit the number of erroneous baselines entering the adjustment process. A further feature of the network adjustment stage is that transformation parameters relating the GPS vectors to a local coordinate system can be estimated as part of the adjustment. The adjustment process can be done in several ways. The following sections highlight the major elements of the adjustment process. Network Adjustment Minimally Constrained Adjustment Once the processed Cartesian vectors have been loaded into the adjustment module, an adjustment should be performed where no coordinates are constrained. The adjustment should be performed using the WGS84 datum. In actual fact, the processor does constrain one point internally to enable this adjustment to be solved. This solution provides a mechanism by which GPS baselines which are not sufficiently accurate can be detected. Once the minimally constrained adjustment has been performed, the surveyor should analyse the baseline residuals and statistical outputs (which will differ between adjustment programs) and ascertain whether any baselines should be removed from subsequent adjustments. This process relies on the baseline network being observed in such a manner to ensure that redundant baselines exist. It is the redundant baselines that enable erroneous baselines to be detected. Network Adjustment Constrained Adjustment Once the minimally constrained adjustment has been performed and all unsatisfactory baseline solutions removed, a constrained adjustment can be performed. The constrained adjustment is performed to compute transformation parameters, if required, and yield coordinates of all unknown points in the desired coordinate system. The surveyor must ensure that sufficient points with known coordinates are occupied as part of the survey. The user should analyse the statistical output of the processor to ascertain the quality of the adjustment. Large residuals at this stage, after the minimally constrained adjustment has been performed, will indicate that the control points are non-homogeneous. It is, therefore, important that additional control points are occupied to ensure that such errors can be detected. 70

71 Network Adjustment Error Ellipses The standard deviations of the estimated coordinates are derived from the inverse of the normal matrix generated during formulation of the least squares process. Error ellipses for each point can be computed from the elements of this matrix. The ellipse presents a one standard deviation confidence region in which the most probable solution based on the measurements will fall. Surveyors should base the quality of the adjustment process on the magnitude of these ellipses. Many contracts will specify the magnitude of error ellipses for both the minimally constrained and fully constrained adjustments as a method of prescribing required accuracy levels. The product documentation for the adjustment program will further indicate the manner in which the ellipse values are generated. Network Adjustment Independent Baselines For the least squares adjustment process to be successful, the surveyor must ensure that independent baselines have been observed. If more than one session is used to build the baseline network, then independent baselines will exist. In instances where one session is observed and all baselines adjusted, the measurement residuals will all be extremely small. This is due to the correlation that exists between the baselines solutions as they are derived from common data sets. This is not a problem as long as the surveyor is aware of the occurrence and does not assume that the baselines are of as high an accuracy as implied from the network adjustment results. The inclusion of independent baselines is an important component of GPS survey design and leads to a strong network configuration. Network Adjustment More information on network adjustment can be obtained at the following website: The processing software documentation should also be consulted when processing and adjusting GPS network observations. Surveying Using Global Navigation Satellite Systems 71

72 The GPS Observables Introduction The Pseudorange Observable The Carrier Phase Observable Survey Receiver Measurements Selecting An Appropriate Observable Introduction The term observable is often used to define a measurable quantity. In the context of the Global Positioning System, there are two primary observable types that are derived from the GPS signals. The GPS signal structure is such that two carrier signals transmit modulated binary code sequences. This provides two distinct measurements, one of the code, the other of the carrier. Measurement of the binary code gives rise to the pseudorange observable, whereas measurement of the carrier gives rise to the carrier phase observable. The GPS Observables The Pseudorange Observable A GPS receiver aligns the incoming binary code sequence with a replica generated within the receiver. This is performed by the receiver tracking loops. When the code signal is demodulated from the carrier, a binary sequence of one s and zero s remains. The receiver attempts to align this sequence with an identical signal generated by the receiver clock. As the satellite signal arrives approximately 0.07 of a second after it was transmitted due the orbits of the GPS satellites, the received signal and internal replica will not align in time. The magnitude of the shift required to align the two signals represents, in effect, the time required for the signal to leave the satellite and arrive at the receiver. This shift is a measure of the time delay between the incoming and replica signals. The time delay measured by aligning the binary code sequence can be multiplied by the speed of propagation to yield a distance. The speed of propagation is nominally the speed of light, which yields the pseudorange observable when multiplied by the measured time delay. The pseudo- prefix is included to represent a clock error present in the measured time delay. 72

73 One of the design criteria of the GPS system is to support an unlimited number of users in manner which does not require the user to communicate with the satellite (and hence potentially reveal position). This is facilitated by GPS being a one-way ranging system. The satellite signal is generated using the atomic oscillators on board the satellite and is measured by a receiver which generates replica signals using its own internal oscillator. As the GPS receiver uses an inexpensive quartz crystal source as its clock, there is a natural mis-alignment between the satellite and receiver clocks. As the pseudorange is derived based on a measured time delay, the time delay will inherently contain this clock synchronisation error. Therefore, in order to effectively use the pseudorange observable for positioning, the clock error needs to be removed. The GPS Observables The Carrier Phase Observable After the binary code has been demodulated from the carrier, the pure sinusoidal carrier signal remains. The tracking loop in the receiver is used to align the received carrier signal with a replica generated within the receiver, in a similar manner to that of the delay-lock loop and the code. The incoming sine wave can attempt to be aligned with the internal replica, however, as the carrier signal is sinusoidal in nature, the receiver can only align to the nearest whole cycle. The amount by which the internal replica is shifted to align with the received signal represents the carrier phase observable. Due to the ambiguous nature of the carrier, the carrier phase observable has a maximum amplitude of one carrier cycle. The carrier phase observable is also affected by the same receiver clock error which must be removed before position estimates can be determined. The distance between the satellite and receiver can be considered to comprise a whole number of carrier cycles plus the measured carrier phase which represents the missing portion of a cycle. The whole number of cycles is termed the carrier phase integer cycle ambiguity and must be resolved mathematically if the carrier phase is to be exploited for surveying applications. As a satellite moves along its orbital path, the change in frequency of the signal as it moves either towards, or away, from the satellite is estimated by the receiver tracking loop. This is essentially the Doppler measurement used as the principle observable of the Transit system. If the satellite is continuously tracked, the receiver is capable of determining the change in range, or effectively, the whole number of cycles that have been received since the satellite was initially tracked. Therefore, if satellite tracking is uninterrupted, the integer cycle ambiguity will not change. A corollary to this is that the integer cycle 73

74 ambiguity represents the number of whole cycles between satellite and receiver at the commencement of satellite tracking. If satellite tracking is interrupted, a cycle slip is said to have occurred. In static surveying applications, cycle slips can, generally, be easily repaired as the stations are not moving. In such an environment, cycle slips have the effect of increasing the time required to mathematically resolve the integer cycle ambiguities. It must be noted that the determination of the ambiguities is the primary defining factor in the occupation time required to perform static surveys to a suitable accuracy. In a kinematic environment, cycle slips have the effect of causing the survey to be initialised. This initialisation procedure aims to determine the integer cycle ambiguities and may be performed in number of ways. Again, the surveyor feels the effects of cycle slips in the form of decreased efficiency. It is vital to understand that the integer cycle ambiguity terms must be determined in order to use GPS to accuracy levels required for surveying applications. This is due to the ability of the receiver to measure the carrier phase observable to millimetre accuracy. The codes cannot be measured to a similar accuracy, therefore, are only used in a supporting role during position estimation. The GPS Observables Survey Receiver Measurements A GPS receiver designed for surveying purposes is generally available in one of two forms, single frequency or dual frequency. A single frequency receiver provides access to the C/Acode and navigation message modulated on the L1 carrier. The observables from such a receiver are, therefore, the C/A-code pseudorange (accurate to approximately 1.5m) and the L1 carrier phase (accurate to approximately 1mm). A dual frequency receiver measures the L1 observables, however, also measures the signal on the L2 carrier. Due to the anti-spoofing policy, the delay-lock and Costas loop combination cannot be used for the L2 signal as the details of the Y-code are classified. Receivers designed for surveying applications are capable of providing pseudorange and carrier phase measurements on both frequencies employing modified tracking loop architectures. It should be noted that the measurement accuracy attainable using such techniques is not as high as the code-correlation technique, however, most receivers can measure the L2 carrier phase to an accuracy of a few millimetres. The use of measurements from the second frequency enables longer baselines to be observed as the ionospheric delay error can be estimated. In addition, the time required to resolve the integer cycle ambiguities on shorter lines can be reduced. This is achieved by forming combinations of the two GPS carrier phase measurements. Single frequency users cannot form such combinations and, for this reason, generally cannot operate as efficiently as dual frequency users, however, this is not true in all cases. Single frequency receivers are still valuable and capable of efficient surveying, however, the application of use must be more carefully chosen. In order to use either of the two receiver types, the integer cycle ambiguities must be resolved. The most common observables used to accomplish this are the L1 carrier phase, the L2 carrier phase and three artificially derived observables; the wide-lane, narrow-lane and ionosphere-free observables. The last three observables are not actually measured by the receiver, rather, are formed by linearly combining the L1 and L2 carrier phase measurements. Most GPS processing packages facilitate processing of these observables, therefore, it is important to understand the ramifications of using either of these measurements. The following sections present a mathematical outline of the observables. In each case, subscripts and superscripts used to denote receiver and satellite have been neglected for clarity. In addition, the majority of error terms have been ignored for the purpose of presenting an overview of the measurements. 74

75 The L1 Carrier Phase Observable The carrier phase observable is the difference in phase between the carrier signal emitted by the GPS satellites and the carrier signal replicated within the GPS receivers. This quantity can be measured to within one half a percent of the carrier signal wavelength and is used to obtain very high precision GPS positioning. The carrier phase measurements derived using code-correlation techniques on the L1 carrier signal is presented below in terms of carrier cycles. f c L1 φ L1 =.ρ + NL1 where, A f L1 φ L1 is the L1 carrier phase measurement in cycles, f L1 is the frequency of the L1 carrier ( MHz), c is the speed of light in a vacuum (299,792,458.0 metres/second), ρ is the range between the satellite and receiver (metres), N L1 is the L1 integer cycle ambiguity (cycles), and A is the ionospheric error term based on total electron content. The wavelength of the L1 carrier is given by the coefficient of the range term, frequency divided by the speed of light. This yields a wavelength of approximately 0.19m. The L2 Carrier Phase Observable The L2 carrier phase observable can be presented in a similar manner to the L1 observable. f c L2 φ L2 =.ρ + NL2 where, A f L2 φ L2 is the L2 carrier phase measurement in cycles, f L2 is the frequency of the L2 carrier ( MHz), and N L2 is the L2 integer cycle ambiguity. The wavelength of the L2 carrier signal is approximately 0.24m. The Wide-Lane Carrier Phase Observable The wide lane carrier phase observable is formed by subtracting the L2 phase from the L1 phase. This leads to a longer wavelength observable (approximately 0.86m) that can be used to aid in the identification of the integer cycle ambiguities. φ WL =φl1 φl2 f f 1 1 where, ( ) L1 L2 =.ρ + NL1 NL2 A. c fl1 fl2 75

76 φ WL is the wide lane carrier phase measurement in cycles. The Narrow-Lane Carrier Phase Observable The narrow lane carrier phase observable is formed by adding both the L1 and L2 carrier signals. A shorter wavelength measurement (approximately 0.11m) is formed, however, the noise (measurement uncertainty) of the narrow-lane observable is significantly less than that on the wide-lane observable. During ambiguity resolution, the narrow lane wavelength is effectively doubled due to the "even/odd" relationship between the wide and narrow-lane integer ambiguities. If the wide-lane ambiguity is even/odd, then the narrow-lane ambiguity is even/odd, effectively doubling the narrow lane wavelength once the wide lane integers are determined. The observable is formed in the following manner. φ NL =φ L1 +φl2 f + f 1 1 where, ( ) L1 L2 =.ρ + NL1 + NL2 A. + c fl1 fl2 φ NL is the narrow lane carrier phase measurement in cycles. The Ionosphere-Free Carrier Phase Observable The ionosphere-free carrier phase observable is formed in a manner designed to eliminate the ionospheric carrier phase advance terms present in the other four observables. The observable is formed by scaling the L1 and L2 carrier signals relative to each other. Two coefficients, α and β are applied to the L1 and L2 carrier phase observables in the following manner. φ =αφ +βφ IF L1 L1 2 2 fl1 f L2 f L2 =.ρ + N L1.NL2 c.f L1 fl1 The selection of the two coefficients can be chosen arbitrarily, however, certain combinations of values have advantages with regards to noise minimisation. For simplicity, if the first coefficient, α, is set to a value of 1.0, the second coefficient, β, can be set to (-f L2 /f L1 ) to remove the ionosphere term, yielding an ionosphere free phase observable with a wavelength of approximately 0.48m. One popular choice of values for α and β is 5 and -4 respectively. This combination does not completely remove the ionospheric influence, however, retains the integer nature of the ionosphere-free ambiguity. The wavelength of this measurement combination is approximately 0.10m. The measurement uncertainty of this combination is also desirable as it less than most other combinations with similar properties. The use of the ionosphere-free observable is extremely valuable when the separation between the reference and rover receivers exceeds 30km. At baseline lengths greater than this, the ionospheric effects at both receivers begin to differ, causing errors in position estimates. The use of the ionosphere-free observable eliminates this error source from such solutions. Users should, however, be cautious about using the ionosphere-free observable of short baselines as there is an increase in noise when compared to the L1 carrier phase. A mathematical description of the above observables can be found in many GPS textbooks and documents. Alternatively a number of sources of information also exist on the Internet which include: The GPS Observables 76

77 Selecting An Appropriate Observable With the possible choice of five observables which may be used generate position estimates, a question arises as to which observable to use. Before this can be answered, the motivation for the generation of the wide-lane, narrow-lane and ionosphere-free observables must be considered. The wide-lane is generated because it has a long wavelength of 0.86m, more than four times the wavelength of the L1 observable. This reduces the number of possible integer cycle ambiguity values by a factor of greater than four. In addition, any residual error will appear as a smaller percentage of the wavelength of the wide lane phase when compared to the L1 phase. This eases the process of statistically selecting the appropriate integer ambiguity value. The measurement uncertainty, or noise, of the wide lane observable is significantly higher than that of the L1 carrier phase observable. The narrow-lane observable has a short wavelength, almost half of the L1 wavelength, however, is designed to have a low noise component. When used in combination with the wide-lane observable, the narrow-lane wavelength is effectively doubled to approximately 0.21m. The combination of the wide-lane and narrow-lane integer ambiguity values enable the L1 and L2 ambiguities to be isolated. It is for this reason that the combination of the widelane and narrow-lane solutions is popular. The ionosphere-free observable is designed to reduce the effect of the ionosphere on position estimates. Depending on the manner in which the observable is formed, the ionosphere-free observable may still have integer ambiguity values. In the case where the combined ambiguity is not an integer, identification of the L1 and L2 ambiguities using the wide-lane and narrow-lane observables facilitates the ionosphere-free solution with identified ambiguity values. Based on this summary, it is evident that the wide lane and narrow lane solutions are intended as an aid to the integer ambiguity resolution process, while the ionosphere-free observable is intended for providing solutions free of ionospheric delay. The recommendation of a particular observable can be made based on the noise component of each of the observables. The noise of a measurement is the term often used to refer to the standard deviation of the measurement. It is a reflection of the ability of the instrument to measure the desired quantity. The noise of a GPS observable is directly related to the wavelength of the signal. Signals with short wavelengths are measured more precisely than those with longer wavelengths. It is extremely important to understand that the wide-lane, narrow-lane and ionosphere-free observables are not measured by the receiver, rather they are artificially generated by addition and multiplication. Therefore, the noise of these three observables must be computed using the law of propagation of variances, as the observables comprise the measured L1 and L2 phase quantities. The noise values for the observables can be related to the noise of the L1 carrier to provide a relative comparison. The noise of the carrier phase measurement can conservatively be estimated as being 2mm (1% of its wavelength). Note that most receivers can measure the carrier more precisely than this, using the code-correlation technique. If the P-code was available to civilians, a similar phase accuracy could be expected for the L2 carrier. However, with the Y-code being transmitted, the result of using other signal tracking techniques is a noise value closer to three times that of the L1 carrier phase. The following table illustrates both measurement scenarios. 77

78 Phase Observable Wavelength (cm) Noise (P-code) Noise (P-code) Noise (Y-code) Noise (Y-code) (multiplier) (multiplier) (multiplier) (mm) L1 Phase 19 cm x mm x mm L2 Phase 24 cm x mm x mm Wide-Lane Phase Narrow-Lane Phase Ionosphere Free Phase 86 cm x mm x mm 11 cm x mm x mm 48 cm x mm x mm The results presented in the table indicate that the L1 observable provides the best noise results in the presence of anti-spoofing. Therefore, for baseline lengths that are sufficiently short such that ionospheric errors are not considered significant, the L1 carrier phase measurement should be used to generate the final position solution. For longer baselines where the ionospheric delay error is considered significant, the ionosphere-free observable should be used. This observable should not be used on short lines, however, as the measurement noise is 6.5 times greater than that of the L1 signal. Potential of a Second Civilian Frequency Due to the new US DoD GPS modernisation policy, civilian users will have access to code and carrier information on two frequencies. In this instance, code-correlation tracking techniques will be able to be used by receivers to acquire range and phase measurements. This would result in noise values reflected by the P-code column in the above table. In such an event, the selection of an appropriate observable for generating final position solutions should be reviewed. The GPS Observables Surveying Using Global Navigation Satellite Systems 78

79 GPS Error Sources Clock Errors Satellite Clock Error Receiver Clock Error Satellite Orbits Selective Availability Atmospheric Errors Troposphere Ionosphere Multipath Antenna Phase Centre Measurement Uncertainty There are a number of errors that affect the position estimates derived from GPS measurements. The errors can generally be divided into three distinct groups; satellite dependent errors, propagation dependent errors, and receiver dependent errors. Satellite Dependent Errors: Satellite Clock Error Satellite Orbit Error Selective Availability Propagation Dependent Errors: Ionosphere Troposphere Multipath Receiver Dependent Errors: Receiver Clock Error Antenna Phase Centre Variation Measurement Uncertainty The majority of the errors listed are "spatially correlated", in other words, two receivers at adjacent locations experience similar errors. This characteristic is used to eliminate many of the errors by surveying with two receivers simultaneously. The position error at both sites is assumed to be the same, therefore, the difference in coordinates between the two receivers should be significantly more accurate than the absolute position of each receiver. This technique is termed relative positioning and is used in all surveying applications of GPS technology. GPS Error Sources Clock Errors The maintenance and measurement of precise time is a vital element in the use of the Global Positioning System. The signals transmitted by the satellites and received by user equipment are light waves, therefore, nominally travel at the speed of light (299,792,458.0 metres per second). In order to estimate the position of a receiver on the Earth, any errors in the clocks aboard the satellites, or in the receiver itself, 79

80 must be accounted for. An error of one microsecond ( seconds) causes a range error between the satellite and receiver of approximately 300 metres. To further highlight the importance of maintaining and measuring precise time, consider that the signal transmitted by the satellite takes approximately seconds to travel to the receiver. It is vital that clock errors are either accurately modelled or eliminated from the positioning model. There are two clock errors present in the GPS system; satellite clock error and receiver clock error. GPS Error Sources Satellite Clock Error The satellite clock error is caused by the inability of the satellite oscillator (clock) to maintain the GPS system reference time frame. The satellites use high stability atomic clocks, typically caesium or rubidium, which result in satellite clock errors which are significantly smaller than receiver equivalents, as receivers use inexpensive crystal oscillators. The magnitude of satellite clock error is of the order of seconds. Rather than physically correct the satellite clocks to the satellite reference system time frame, the oscillator is allowed to drift and a correction broadcast as part of a navigation message. This is due to atomic oscillators behaving in a more stable manner if not constantly adjusted. Therefore, the satellite clock error can be modelled using the broadcast satellite clock correction. This broadcast correction is in the form of a polynomial expression with clock offset, drift and rate of drift parameters. However, the satellite clock error is the same for two receivers simultaneously observing the same satellite (relative positioning). This property is used to remove the satellite clock error from the measurements by differencing the measurements between two observing receivers. GPS Error Sources Receiver Clock Error The oscillator used to generate the satellite signal replica within the receiver is commonly an inexpensive quartz crystal source. The primary reason for this is to keep receiver equipment costs to an affordable level. Therefore, there is an error due to the inability of the receiver clock to maintain the GPS system reference time frame, in much the same way that the satellite clocks are unable to precisely maintain GPS time. The stability of the receiver clock error is worse than the satellite equivalent due to the nature of the oscillator (quartz crystal versus caesium and rubidium). Therefore, the receiver clock error is generally larger than the satellite clock error. A GPS receiver has one oscillator which generates all internal signals required to track the GPS constellation. Thus, if the receiver clock is in error, the error will affect measurements to all satellites being tracked by the same amount, in other words, the receiver clock error is identical for all satellites observed simultaneously. To determine the three dimensional position of a point of interest, three unbiased satellites measurements are required. To account for the receiver clock error, an additional (fourth) satellite is observed. Therefore, a minimum of four satellites are required to solve for the three position components and to account for the receiver clock error. GPS Error Sources 80

81 Satellite Orbits The GPS satellites move in almost circular orbits at a nominal altitude of approximately 20,200 kilometres above the surface of the Earth. At this altitude, the force of gravity causes the satellite to move at approximately four kilometres per second. The orbits of the satellites can be described using the six parameters developed by the German astronomer, Johannes Kepler, in the seventeenth century. These parameters are; the semi-major axis, eccentricity, inclination, right ascension of the ascending node, argument of perigee and one of either the true, mean or eccentric anomaly. The navigation message broadcast by the GPS satellites contains an ephemeris (set of 16 orbital parameters) which are based on these six Keplerian elements. There are additional elements to account for slight perturbations to the nominal satellite orbit. However, there are other influences which affect the motion of the satellites which are difficult to account for. These influences include particles of the Earth s atmosphere which cause the satellite to slow down. This effect is not noticeable on the GPS satellites due to the high orbital radius as defined by the semi-major axis parameter. Another effect is that of solar radiation pressure. Photons which make up sunlight impact on the satellite eventually causing a change in its orbital path. The satellites orbit around the Earth, however, also suffer from gravitational attraction effects with the sun and moon. These small forces cause the orbit to be disturbed. Other effects such as those caused by the magnetic field of the Earth and changes to the Earth s gravitational field caused by tides also contribute to disturbing the orbits of the GPS satellites. As a result, the ephemeris parameters will not describe the true orbit due to these small influences. The broadcast ephemeris is computed at the Master Control Station based on measurements acquired at the tracking stations. The parameters for each satellite are then uploaded to the satellites for use in the future, generally, for a period of a few hours before a new set of parameters is broadcast. As a result, the broadcast ephemeris is a prediction of the where the satellites are going to be in the future. This predicted ephemeris is required for real time applications where position estimates are computed in the field. Alternatively, for post processed applications, the precise ephemeris can be used. The precise ephemeris is an ephemeris which is computed based on the position of satellites at some point in time in the past. It is, therefore, more accurate than the broadcast ephemeris as actual satellite positions are used to develop the parameters. For precise surveying applications, use of the precise ephemeris is recommended. However, the use of two receivers ( relative positioning) is successful in eliminating much of the satellite orbit error, in particular, over reference-rover station separations of less than ten kilometres. The precise ephemeris can be downloaded from a number of different websites which include: In the future, it can be anticipated that the Australian Fiducial Network (AFN) will be used to generate ephemerides for use in Australia. This would be advantageous as the stations comprising the AFN will enable satellite orbits which best fit the Australian region to be computed. GPS Error Sources Selective Availability The Global Positioning System was devised by the United States Department of Defence (DoD) to satisfy military positioning, navigation and timing requirements. In the design of the system, military users access the P-code to measure ranges to satellites and compute positions using a single receiver to an accuracy of 2-5 metres. Civilian users access the C/Acode, which is less precise, and obtain position to a lesser accuracy than United States military personnel. In the late 1980 s, civilian users were routinely able to obtain point position 81

82 accuracy of approximately metres. The DoD considered this level of accuracy to be too close to that attainable by military personnel and a threat to military advantage in a time of conflict. To counter this, the selective availability policy was introduced. Selective Availability (SA) was the mechanism which intentionally degraded the performance of the system for civilians to a point positioning accuracy of 100 metres (specifically horizontal position accuracy at a 95% confidence level). This had a significant impact on the use of the system in many navigation applications and was the largest single source of error for absolute positioning. US presidential policy removed selective availability in May 2000 and this has again allowed users of a single GPS receiver accuracy of metres. This effect is evident in the following figure, illustrating horizontal and vertical circular error probable before and after the removal of selective availability. The method by which selective availability was implemented in two ways. First, the satellite orbit parameters contained in the broadcast ephemeris were intentionally degraded to limit positioning accuracy. Secondly, the satellite clock was dithered to cause an additional error which was felt as a range error. For surveying applications, relative positioning techniques are used. Selective Availability was a spatially correlated error, therefore, the impact of SA on surveying applications was not significant. In particular, when the precise ephemeris was used, the orbit error caused by selective availability was eliminated. For real time applications, the orbit error due to SA could be removed, however, short station separations were successful in reducing SA errors. It should be noticed that the US DoD has the prerogative to re-introduce Selective Availability if it deems it necessary for US security. 82

83 GPS Error Sources Atmospheric Errors The Earth s atmosphere has retarding effects on satellite signals which pass through on the way to the receiver. The GPS receiver relies on time synchronisation with the satellite and receiver clock, followed by application of the signal speed, which is the speed of light. However, the speed of light (299,792,458.0 metres per second) is computed based on propagation through a vacuum. Propagation through the atmosphere causes changes in the satellite signal speed. For the two GPS frequencies, there are two portions of the atmosphere which are considered to affect signal propagation. The ionosphere is loosely defined as the portion of the atmosphere between 50 kilometres and 1000 kilometres above the surface of the Earth. The troposphere is defined as the portion of the atmosphere below an altitude of 50 kilometres. In order to control the magnitude of atmospheric errors, an elevation mask is adopted during data collection and processing. The elevation mask is an angle measured positive above the horizon below which satellite signals are ignored. For surveying applications, a mask of between 10 and 15 degrees should always be adopted. GPS Error Sources Troposphere The troposphere is loosely defined as the region of the Earth s atmosphere below an altitude of 50 kilometres. This portion of the atmosphere causes a delay on code and carrier measurements which may reach approximately 2.5 metres at the zenith and almost 30 metres at the horizon. The tropospheric delay varies with temperature, pressure and humidity and the height of the receiver. The error can be broken into two distinct portions, the dry component and the wet component. The dry component closely obeys the ideal gas law relating the surface meteorological readings and contributes almost 95% of the total tropospheric error. Accurate surface measurements enable this component to be modelled to an accuracy of almost 98% of the dry component. The wet component cannot be measured as easily and is the cause of most of the tropospheric error remaining in the GPS measurements. The reason for this is the difficulty in estimating the water vapour content of the atmosphere. An instrument called a Water Vapour Radiometer (WVR) can be used to model the wet component to an accuracy of less than two centimetres, however, these instruments are valued at approximately $100,000. Surface measurements enable the wet component to be estimated to an accuracy of approximately five centimetres. For surveying applications, the relative mode of positioning is always used. The troposphere is highly correlated for reference-rover receiver separations of less than 10 kilometres, especially when the receivers are at the same altitude, therefore, relative positioning techniques are effective in reducing tropospheric effects. It should be noted that tropospheric errors may become problematic when the reference and rover stations are at different altitudes. An example where problems may occur is in the use of GPS for positioning an aerial camera in a photogrammetric application. If the reference station is located at the airport and the aircraft is several kilometres higher and travelling through dense cloud, the spatial characteristics of the troposphere will be different at both sites. 83

84 The majority of GPS processing programs enable a number of different atmospheric models to be used to estimate the tropospheric delay error. Most models will generally provide suitable results, however, the Hopfield model is perhaps the model most commonly used. Other models include the Black, Saastamoinen and Modified Hopfield. GPS Error Sources Ionosphere The ionosphere is the portion of the atmosphere between an altitude of 50 kilometres and 1000 kilometres in which free thermal electrons are present. The number of free electrons is defined by the Total Electron Content (TEC). As the GPS signal passes through the ionosphere, the code is retarded in much the same way as the tropospheric delay, however, the carrier phase is advanced. The magnitude of the advance on the carrier is the same as the magnitude of the delay on the code. This advance can attempt to be modelled using the broadcast ionospheric model transmitted in the navigation message. In addition, ionospheric errors are spatially correlated over reference-rover receiver separations of less than 10 kilometres, hence, relative positioning techniques eliminate much of the error. The Total Electron Content varies with a number a factors including the time of day, location, season and also the period of the 11 year sunspot cycle. In peak periods of sunspot activity, the ionosphere may become a problematic source for surveying applications. However, one characteristic of the ionosphere is that signals are affected proportionally to the signal frequency. As the GPS signal comprises two frequencies spaced by approximately 350 MHz, a combination of the L1 and L2 carrier phases can be generated which removes the ionospheric error. This ionosphere free observable suffers from an increase in random measurement error, however, should be used in periods of high ionospheric activity. In order to generate the ionosphere free observable, a dual frequency receiver is required. If single frequency receivers are used, especially in peak sunspot activity, receiver separations should be kept as short as possible. GPS Error Sources Multipath The basic principle of positioning using the GPS satellites involves the simultaneous measurement of distances, or ranges, between at least four satellites and the receiver. The position of the satellites is obtained from the broadcast or precise ephemeris and the receiver position is determined using trilateration. Multipath is the phenomenon where a signal arrives at the receiving antenna via an indirect path. Objects such as large metal roofs and buildings can cause the incoming signal to be reflected before reaching the antenna. This has the effect of increasing the measured distance between the receiver and satellites and, hence, causes erroneous position estimates. The characteristic of multipath that makes it difficult to remove is that it is a site dependent error. Therefore, relative positioning techniques are ineffective in removing its effects. 84

85 However, there are several observation and equipment related techniques that can be used to reduce the impact of multipath. Antenna Ground Plane The majority of antennas used by GPS surveying equipment manufacturers employ the microstrip patch element. These antennas can make use of a ground plane to help reduce the impact of multipath. The ground plane is, in general, a large metal disk which can be attached to the main antenna body. Some antennas have built in ground planes. Using a ground plane is successful in reducing some of the effects of multipath and should be used at all stationary receivers if available. An alternative to the standard ground plane configuration illustrated is the choke ring antenna. The choke ring antenna ground plane is larger and has circular rings which are designed to provide multipath rejection at the GPS frequencies. Choke ring antennas are not as easily obtained as standard ground planes and the results obtained from them have not been sufficiently conclusive to warrant their additional expense for the majority of applications. This is not to say that a choke ring should not be used if available, rather that the purchase of a choke ring ground plane may be an unwarranted expense. Careful Site Selection In many survey applications, in particular control surveys, the position of points of interest are yet to be defined. For example, a control point is required in a general region. In such instances, users should try to locate features in locations which are free from overhead obstructions, in particular, building and large reflective objects. In Victoria, the GPS satellite constellation is such that satellites do not appear low in the southern sky. Therefore, if obstructions are present and cannot be avoided, marks should be placed to the north of such features. Increase Observation Sessions When the observing receiver is stationary (static surveying), the multipath effects on a particular satellite change slowly due to the altitude of the satellites. Therefore, any multipath effects will remain present for observation periods of several minutes. The effects of multipath can be averaged if longer observation periods are adopted. For geodetic surveying applications, observation periods of greater than two hours should be adopted. For general control surveys, shorter periods can be used, however, if multipath is anticipated to be a problem, users should exercise caution and observe for longer periods than otherwise recommended. Recent Developments in Receiver Technology The reduction of multipath effects is one area of research that is currently advancing rapidly. In the last few years, several equipment manufacturers have developed techniques to detect and correct multipath effects within the receiver itself. This has in effect, hidden, the complexity of multipath rejection from the user. At this stage, there have been tremendous improvements shown in measurement of the codes, however, further research is required to conclusively ascertain effects on the carrier phase measurements used in surveying. Any new receiver purchases should be made with this feature in mind. GPS Error Sources 85

86 Antenna Phase Centre In terrestrial surveying applications using total stations, a reflective prism is placed vertically over a point of interest and distances measured from the total station to the prism. If the physical centre of the prism does not coincide with the reflecting surface, a correction termed the prism constant is applied to the measured distance. A GPS antenna has similar characteristics. The mechanical, or physical, centre is placed vertically over the feature of interest. The electrical centre is the point on the antenna at which the satellite signal is received. There will be a difference between the mechanical and electrical centres of all antennas. This is due to a number of factors, such as the difficulty in aligning both centres in the manufacturing process and the characteristics of the antenna element. In addition, the electrical centre, often termed the phase centre, of the antenna varies with the azimuth and elevation of the satellite. The antenna phase centre variation may approach several millimetres under certain conditions depending on the antenna type. This error can attempt to be minimised by aligning both the reference and rover antennas. Many antennas have a north point marked on the ground plane. If both antennas are aligned in the same direction and have been manufactured using the same process, antenna phase centre errors will be minimised during relative positioning operation. In kinematic surveying applications, this cannot be easily achieved, however, for static survey work, should become a routine procedure in the setting up of both reference and rover receivers. GPS Error Sources Measurement Uncertainty Any measurement made using an electronic measuring device (and non electrical device for that matter) is subject to a random measurement error (or noise). This error reflects the ability of the device to measure the sought quantity. For example, an atomic clock can measure time more accurately than a quartz crystal clock. Therefore, the random measurement error of the atomic clock is smaller than that of the crystal clock. The error is considered random as it averages to a mean of zero as more measurements are made. For GPS surveying applications, the measurement of interest is the carrier phase measured on the L1 and/or L2 frequency. For almost all receivers manufactured since 1990, the carrier phase random measurement error is a few millimetres, generally less than two. Receivers developed more recently are capable of carrier phase measurement to an accuracy of less than one millimetre under ideal conditions, i.e. satellite overhead, limited atmospheric activity, no obstructions etc. This is evidenced by the results generated by receivers in a zero baseline test. In the majority of static surveying applications using dual frequency receivers, linear combinations of the L1 and L2 frequencies are generated. The three combinations of particular interest are the wide-lane, narrow-lane and ionosphere-free combinations. The law 86

87 of propagation of variances can be used to determine the noise on these derived observables. Users should be particularly careful when adopting wide-lane and ionospherefree solutions as the random measurement error is greater than the other three observables. For short reference-rover station separations (less than 10km), the L1 observable should be used to generate the final solution. An exception to this rule may apply in periods of peak sunspot activity where the ionosphere-free observable may be preferable. GPS Error Sources Surveying Using Global Navigation Satellite Systems 87

88 How To Introduction Measuring the Antenna Height Performing Static Surveys Performing RTK Surveys Designing GPS Surveys Introduction This section aims to highlight some of the more common queries regarding the use of GPS technology. There are a number of sections in this web document that contain information relating to various aspects of GPS surveying. This section provides procedural steps which access relevant sections of the Surveying Using Global Navigation Satellite Systems document. How To Measuring the Antenna Height It is important that the height of the GPS antenna is measured when performing surveys, even if the height of the points being surveyed is not of interest. Take, for example, a static survey where a network of baselines is being observed. Each station should be re-occupied with a different antenna height to provide an independent occupation of the mark. The antenna height must be correctly measured so the baseline can be reduced to ground mark to ground mark. This will enable loop closures and network adjustments to be performed successfully. Unlike a terrestrial survey reflector, the phase centre of the GPS antenna cannot be accessed easily. As a result, there are a number of methods by which antenna heights can be measured. Each GPS system will have a slightly different method, however, three techniques are generally adopted. Direct Measurement If the antenna is physically quite small and a ground plane is not used, the height to a predefined mark can be directly measured. As there is no ground plane and the height of the antenna is generally set to be greater than 1.5m, any error caused by the measurement not being perfectly vertical is considered negligible. There is also a device which looks likes a rectangular ruler which contains two right angles to measure correctly around the body of the antenna. This device is ideal for use when the antenna is not physically large. Slope Distance Measurement When the antenna has a ground plane attached, the height to the edge of the ground plane is measured. By knowledge of the distance from the phase centre to the edge of the ground plane, the correct vertical height can be computed. This technique relies on the correct selection of the antenna type so the radius of the ground plane can be used. Incorrect selection will lead to errors in the antenna height which will propagate into baseline solutions. Partial Measurement Some receiver manufacturers provide reference marks on their equipment which are not directly associated with the antenna phase centre. A known offset exists between the reference mark and the antenna phase centre. This is common for receivers where the antenna is built in to the receiver housing. 88

89 How To Performing Static Surveys There are a number of steps involved in performing a static survey. The more common tasks are: design the survey decide on whether GPS technology is appropriate for the accuracy required design appropriate networks, number of receivers, receiver types etc. acquire the measurements process the baselines perform loop closures to detect poor baseline solutions adjust the network transform the coordinates to the desired system How To Performing RTK Surveys There are a number of steps involved in performing a static survey. The more common tasks include: design the survey decide on whether RTK GPS technology is appropriate for the accuracy required. search/locate control marks design appropriate networks, number of receivers, receiver types etc. set up base station and radio link on control point obtain communications link and real time position perform survey if required move base station and continue survey 89

90 Note: these steps are generic in nature and will alter depending on the nature of the GPS survey. A variety of literature exists on how to conduct different types of GPS surveys and should be consulted for a detailed description. The readers are directed to the following web sites for more information on how to conduct a GPS survey: Surveying Using Global Navigation Satellite Systems 90

91 Troubleshooting Introduction Problems Transferring Data to a Computer The Processor Generates Float Solutions Introduction This section aims to provide solutions to some of the more common problems encountered during GPS surveying operations. The problems discussed are not specific to any particular make or model of GPS system, rather, are more generic problems which occur during normal GPS operation. Troubleshooting Problems Transferring Data to a Computer One common problem which is encountered using a variety of GPS products is errors occurring during the transfer of measurements stored on internal memory chips to a personal computer. A communications error may be displayed and the transfer of data aborted before the complete file(s) has been transferred to the computer. The typical messages displayed relate to communications overflow or communications over-run. This error may also occur on different computers, not only with different receivers and survey controllers. For example, the same communication parameters may be used on two different computers, however one computer receives a file successfully while the other aborts on a file of similar size. The solution to this problem is to ensure that the communication parameters are set correctly as described in the equipment documentation. If all seems in order, the user should reduce the baud rate to slow down data transfer. This gives the computer more time to access the incoming data and store it on the internal hard disk. As GPS software improves and communication modules are developed more carefully, this problem should not occur as frequently. Commonly, data transfer rates of 9600, and baud operate successfully. Higher rates are more prone to communication errors, lower rates generally take too much time to transfer. If these higher rates do not transfer data successfully, the user should lower the baud rate to 1200, 2400 or Troubleshooting The Processor Generates Float Solutions If insufficient data has been acquired to successfully resolve the carrier phase ambiguities, a float solution is generated. In the float solution, the ambiguities are not constrained to integers, rather are left to "float" as real numbers. Most commonly, the floating ambiguities will not be close to integers in this instance. The most precise results are obtained when the ambiguities are constrained to integers, therefore, a float solution generally implies that the required accuracy will not be met. In most cases, the baseline will need to be re-observed, however, there are some processing modifications that may be used to alleviate this problem. This discussion assumes that the survey has been performed using the static or rapid static observation technique. Kinematic techniques are generally much harder to resolve and are best re-observed. The surveyor should closely analyse the output provided by their data processing program. In all float solution cases, the ratio of the sum of the squares of the residuals for the potential solutions will be a low number close to one. This indicates that there is no clear solution to the 91

92 integer ambiguities. The meaning and nature in which the ratio is computed will depend on the processing package and users should refer to their documentation. If the ambiguities are displayed as numeric values, they will be real numbers which do not approach integers. In addition, the measurement residuals may be large. In the instance that every ambiguity does not appear to be an integer, the baseline is probably best re-observed. If, however, there is one satellite which is not close to an integer, but the other values are quite close, the data should be re-processed after eliminating this satellite from processing. Similarly, analysing the residual graphs may reveal one or two noisy satellites which can be eliminated from processing to try to generate a fixed ambiguity solution. Another modification worth trying is raising the elevation mask. If data has been observed at ten degrees and residual graphs reveal that measurements are particularly noisy when satellites are low to the horizon, the mask can be raised to, say, fifteen degrees and the processor run again. This may alleviate the problem and generate fixed ambiguities. In general, when problems during processing occur that cannot be resolved using the above suggestions, the baseline must be re-observed. However, by analysing the output provided from the processing program and looking for satellite measurements which may be causing problems, some baselines may be able to be processed to a satisfactory accuracy. Troubleshooting Surveying Using Global Navigation Satellite Systems 92

93 Guidelines for Cadastral Surveying Using GNSS Introduction Validation of Equipment Survey Measurements and Dimensions Classification and Accuracy of Surveys Independent Checks Reporting Designing GPS Surveys Introduction The Surveyor-General of Victoria is responsible for setting and monitoring cadastral survey standards and practices in Victoria under the Surveying Act The Surveying Act 2004 also sets out the role and functions of the Surveyors Registration Board of Victoria which is primarily concerned with the regulation of the training and Registration of Surveyors in the State. For cadastral surveys in Victoria, the Surveying (Cadastral Surveys) Regulations 2005 and Survey Co-ordination Regulations 2004 contain much of the specific detail to which surveyors must conform. The Survey Co-ordination Regulations 2004 makes reference to the Intergovernmental Committee for Surveying and Mapping Standards and Practices for Control Surveys (Special Publication 1), known as SP1. This section assumes that the surveyor has a solid understanding of the concepts and specifics of these regulations and SP1. The use of Global Navigation Satellite Systems (GNSS) (predominantly GPS carrier phase-based positioning) has been adopted and used by the surveying profession. Traditionally, GPS has been used for high precision geodetic survey, engineering and topographical surveys (via post processing and real time techniques). Use by practitioners for cadastral surveys has been limited by equipment costs, unfamiliar operational procedures and the current accuracy requirements of State legislation. However, licensed surveyors are expected to be capable of deciding: whether GPS can be effectively used to achieve the cadastral standards required by Victorian legislation, and the appropriate techniques to achieve this required accuracy. These guidelines outline recommended procedures in the relation to the use of GPS for cadastral surveying in accordance with the Surveying (Cadastral Surveys) Regulations It should be noted that the use of GPS might not be suitable to some areas such as built up areas where satellite visibility is poor. It is also necessary for cadastral practitioners to be well trained and educated in the use of carrier phase-based GPS techniques, the testing and certification of equipment plus the appropriate field/office procedures associated with it. However, the use of GNSS does not differ from conventional surveying techniques in that quality assurance processes must be utilised on a routine basis. This is essential to ensure that satisfactory accuracy specifications can be, and are being, met. In reality, GPS is just another surveying tool and as such may be used in conjunction with traditional methods to provide sufficient information to fix boundaries, marks and occupations. Therefore, no matter what instrumentation is employed, it is the responsibility of the professional surveyor to know and understand the following to obtain the accuracy required of the survey: the limitations of the equipment to be used the observational procedures 93

94 the processing techniques geodetic and map projection reductions for the MGA, suitable practices to ensure measurement redundancies and basic statistical analysis Guidelines for Cadastral Surveying using GNSS Validation of Equipment Legislation It is the prerogative of the Licensed Surveyor to determine the equipment to be used in carrying out a cadastral survey. The Surveying (Cadastral Surveys) Regulations 2005 require that licensed surveyors are required to maintain and compare survey equipment used for cadastral surveys. Survey equipment must be capable of achieving the levels of precision set out in Part B of the ICSM Standards and Practices for Control Surveys (Special Publication 1), referred to as 'SP1'. These requirements are in addition to those specified in the National Measurement Act 1960 and National Measurement Regulations Licensed Surveyors are to retain records of all calibration and standardisation, records may be inspected on request by the Surveyor-General. Thus, it is the surveyor s duty of care, under State legislation, to ensure that the equipment and methods used are capable of meeting the accuracy requirements. Also, from time to time the Surveyor General circulates practice directives that provide instructions for Licensed Surveyors in accordance with the Regulations and inform them of changes to the requirements of the Surveyors Board of Victoria. These directives include issues such as the frequency of equipment validation. Surveyor-General Victoria provides survey instrument calibration services, refer to for the calibration of EDMs, Tapes and Bands, and Staves. However, there are no guidelines for any other equipment, such as theodolites (optical or digital), gyro theodolites or GPS, which may be used for cadastral surveys. The basic vector used in cadastral surveying is the bearing and the horizontal ground distance of a line. It is only the distance that legislation currently attempts to calibrate. A differential GPS determination of the same line is a 3D vector which can then be reduced, like any other surveying method, into the same bearing and horizontal distance. There is still much uncertainty about the use of GPS and legal traceability based on the National Measurement Act A sub committee of the ICSM has been working to address this issue for many years and it is expected that it would recommend specific procedures for GPS validation. SP1 includes the Best Practice Guidelines for the use of GPS for survey applications. Fortunately, the Regulations state that surveyors need only ensure that the process and basis of comparisons with a standard is adequate for the legislated accuracies. A pragmatic approach to equipment verification, e.g. GPS, can therefore be adopted. The Surveyor- General of Victoria has left this procedure to the professional discretion of the surveyor. Therefore, philosophically, any verification/testing process can be devised so long as it can be established that the measurements: have been compared to the appropriate standards, and can be made within the standards of accuracy as specified in the regulations. GPS manufacturers quote specifications for their receivers and processing software, which have been developed from extensive programmes of research and development. The Federal 94

95 Geodetic Control Sub-committee in the USA (FGCS) test all GPS surveying receivers (single and dual frequency instruments) released onto the market. Manufacturers can test and verify quoted accuracy specifications for their products. However, it is the responsibility of the professional surveyor to validate all equipment purchased and used in practice. Validation Methods The GPS validation process should test: equipment measurement techniques together with processing, and transformation and heighting methodologies. Successful validation will also demonstrate the competence of the surveyor in using GNSS technology to achieve the required accuracy. The surveyor should retain the results of validation to comply with Regulations 5 and 14 (2c) where necessary. Ideally, GPS validation would consist of a combination of various methodologies i.e. zero baseline test, a coordinated network, an RTK test site and a coordinated EDM calibration baseline. The combinations would be dependent on the GPS available to a surveyor e.g. static technique only, RTK or a combination. The frequency of validation would be consistent with the current Survey Practice Directive from the Surveyor General. However, it would be advisable for the surveyor to undertake validation as often as deemed necessary to satisfy professional due care, best practice and competence. A Zero Baseline test (All receivers) This can be carried out to check the correct operation of a pair of GPS receivers, associated antennas/cabling, and data processing software. As the name implies, a zero baseline test involves connecting two GPS receivers to the same antenna via an antenna splitter (as recommended by the manufacturer). The computed baseline should be theoretically equal to zero and any variation will represent a vector of receiver errors (usually results should give sub millimetre results). This is a very simple and inexpensive process which: verifies the precision of the GPS receiver measurements, proves that the receiver is operating correctly, and also validates the data processing software Note: A zero baseline test does not examine satellite ephemeris, time or atmospheric errors. However, making measurements and processing data over known baselines or a network of coordinated points can achieve this. A High Accuracy GPS Test Network (Static Techniques) This can be undertaken to ensure that the operation of GPS receivers, associated antennas/ cabling, and data processing software, give high accuracy baseline/coordinate results. Satellite ephemeris errors, clock biases and atmospheric effects must be removed or minimised during baseline processing. Network validation allows GPS equipment to be tested under realistic field conditions which includes the dynamic nature of the satellite constellation and the atmosphere. Mission planning (finding what time of day gives acceptable GDOP for observations) is as essential for GPS validation as it is for real survey applications. The test network should: consist of extremely stable ground marks with almost perfect sky visibility be of a very high precision e.g. first or second order have stations which are ALL coordinated in both the local geodetic system 95

96 (AGD/GDA φ, λ, h or x, y, z) and plane projection (MGA E, N and AHD elevations) have a variety of baseline lengths and directions consist of points with varying elevations to check for the correct modelling of the atmosphere as well as geoid determination to obtain AHD values Example of GPS Test Network Network validation is suitable predominantly for static/ rapid static surveys because the baselines are generally longer than for kinematic surveys. However, RTK GPS equipment/firmware can be checked on a network to validate the equipment and the procedures used to obtain acceptable final results. After observing the network, the surveyor can process the data to produce a network of vectors. These vectors can then be reviewed, adjusted and analysed following conventional methods: a) Independent vectors can be used to determine loop closures and precision b) By holding the values of one of these sites fixed, coordinates for all other sites are derived using the GPS observations initially using a minimally constrained least squares adjustment. The ensuing statistics can be reviewed and assessed. Any flagged outliers can be noted and examined. c) A subsequent adjustment can then be undertaken holding multiple stations fixed to calculate 3D final coordinates of all stations. This ensures that the adjustment has attempted to solve the transformation parameters of the local area. Final coordinates and baseline vectors can be compared to the known values. d) AHD elevations may be required and can determined using two methods. During the final least squares adjustment process using a geoid model (such as AUSGEOID98) within the software package, OR Manually, after the adjustment, by calculating height differences from known benchmark AHD values and comparing with corresponding ellipsoidal heights differences. A geoid model for the local area can be interpolated from this data for all other points. 96

97 A Coordinated RTK/Kinematic Test Site Kinematic surveys are generally restricted to baselines of less than 10km and involve occupying points for a short period of time e.g. less than one minute. A test site, designed for techniques such as RTK (and possibly as a simplistic rapid static check), can be established that is an array of points to be coordinated from a fixed base station. At the end of the observation session final coordinates can be compared to a set of known values (E, N, MGA and AHD elevations). The inclusion of obstructions such as trees could be planned into the array to test the accuracy of the re-initialisation processes of the OTF hardware and firmware. The surveyor should re-observe the array under different satellite configurations to ascertain possible precision under varying conditions i.e. for horizontal coordinates and AHD determination. Example of RTK Test Array An EDM baseline test (Static and RTK Techniques) A pair of GPS receivers (plus ancillary equipment) can be tested over the various pillars of a validated EDM baseline. Measurements would involve setting up one receiver on the start pillar and simultaneous observation would be made to the other one on each pillar along the baseline. EDM calibration baselines have been established throughout Victoria to service the requirements of the surveyors under the Surveying (Cadastral Survey) Regulations Because these baselines are certified annually as subsidiary standards of length, surveyors can then make a comparison of known lengths with EDM measurements. Similarly GPSderived distances can be compared to the standard measurements. Example of a Combined EDM/GPS Coordinated Baseline 97

98 GPS can be used to measure the 3D geodetic vector of a baseline e.g. Δx, Δy, Δz. This can then be reduced to ground distance for comparison purposes. Traditionally, EDM baselines are used to validate the distance component of a measurement and not the vector as a whole. EDM baselines are rarely longer than one kilometre (i.e. well short of the operating range of GPS) and therefore only comparatively short distances can be checked. Finally, if the reduced GPS measurements can verify the known distances between the markers on the pillars of the EDM baseline, it can be considered that: the equipment is in good working order, competency has been proved for the observations technique and reduction processes undertaken, and the GPS receivers are capable of delivering baseline solutions that are within specification. This method of GPS validation would be useful for post processed and real time techniques e.g. static GPS and RTK respectively. It is important that the surveyor is well trained in GPS methodology and has a full understanding of the achievable accuracy of each technique so that baseline comparisons are realistic. One useful addition to an EDM calibration baseline for GPS validation would be for the end points to be coordinated in MGA to a high accuracy i.e. via connection to first/second order marks in a surrounding observed network. Therefore, GPS validation could include a coordinated network followed by a baseline comparison test. This combination would verify final 3D vectors and GPS derived distances. Additional Validation Considerations All GPS equipment, software and procedures should be tested before general usage. Unlike EDM equipment, GPS receivers cannot be calibrated for scale because the definition of scale is inherent in the satellites and orbit data. However, antennae should be checked for centring errors. These should not generally be significant if geodetic quality equipment is used for cadastral surveys. Antenna offsets may also be present when mixing different antenna types. Measuring a line of a few metres with GPS and comparing the results with a direct EDM or taped measurement can easily test this. Tribrachs should be regularly checked and if necessary adjusted to minimise plumbing and levelling errors. The validation of survey equipment is an attempt to ensure the quality of measurements. However, in line with good survey practice, it is recommended that multiple field checks be used throughout a GPS cadastral survey. If any significant modifications or upgrades are made to the GPS receiver or the postprocessing software, then the validation must be repeated. To avoid additional fieldwork for every software upgrade, re-process the original validation raw data with the new version and check for any changes in the results. Another advantage of the validation process is that it allows the surveyor to train and evaluate the competency of staff employed on GPS surveys. This is important for total quality management. Guidelines for Cadastral Surveying using GNSS Surveys Determination of Survey Datum A licensed surveyor making a cadastral survey must adopt and verify a datum with a previous cadastral survey/plan and if practical bring the datum onto the MGA Traditionally a surveyor connects to various existing survey marks, checks their reliability and then 98

99 establishes the survey orientation and scale. GPS observations can also be made directly between appropriate existing survey marks to set up the datum in the same way. Therefore the surveyor undertakes the same procedure of comparisons whatever means of measurement has been adopted. Where a GPS base station is used outside the area of the survey, appropriate existing surveys marks in the area of the survey still need to be connected. Note that a transformation of the GPS data to the local coordinate system of the origin marks may be required. The transformed data must then be used to prove the datum of the survey in terms of the Surveying (Cadastral Surveys) Regulations 2005 e.g. by calculating the GPS vectors between the origin marks and comparing with the bearings and distances between datum marks. Field Survey GPS is only another instrumentation option for the practising surveyor. It provides the ability to operate over greater distances than with conventional equipment. Often base stations outside the area of the survey can be employed. All GPS surveys must be undertaken in accordance with accepted good survey practice such as: 1. GPS observation procedures should be designed to detect and eliminate: ambiguity initialisation errors the effects of multipath interference from electrical interference such as substations, microwave or other spurious radio signals poor satellite geometry due to satellite configuration and/or sky coverage obstructions 2. Observation networks and reduction procedures should be designed to ensure measurements are independent e.g. a multi-baseline static GPS survey observed for only one session provides some dependent baselines which may create uncertainty with the results. 3. Permanent Marks (PMs), Primary Cadastral Marks (PCMs) and reference marks (RMs) placed and measured using GPS should be intervisible, where possible, for ease of subsequent use by all suitable surveying techniques. 4. GPS observations for boundary definition are to be checked by independent observations from another base station. The checks may be made using any suitable instrumentation. The only exceptions to this would be ties to natural boundaries using techniques such as kinematic or RTK i.e. similar to traditional observations. 5. GPS observations from an independent base station can be used to connect survey or boundary marks to PMs, PCMs and RMs. The reference vector connection can be calculated from the independent GPS observations. Such observations must be independently checked to ensure compliance with the Surveying (Cadastral Survey) Regulations Sufficient observations are made to fix boundaries, marks and occupations using traditional best practice concepts for cadastral surveys i.e. not technology dependent. 7. Any boundaries marked using GPS techniques must conform to the accuracy standards of Regulation 7 of the Surveying (Cadastral Surveys) Regulations

100 Connection of Cadastral Surveys to MGA The Surveyor-General of Victoria publishes Practice Directives from time to time to provide surveyors specific practice instructions and interpretation of regulations. Practice Directives are published on The Surveyor-General s requirements for appropriate cadastral surveys to be connected to MGA94 came into effect on 1st July The Surveying (Cadastral Surveying) Regulations 2005 require that a licensed surveyor making a cadastral survey must adopt and verify a datum in accordance with a previous cadastral survey or plan and bring the datum onto the Map Grid of Australia 1994 (MGA94) in a manner specified in Regulation 14(2) of the Survey Co-ordination Regulations Coordinate values for marks contained in SMES with values specified as 3rd order or above result directly from a network adjustment and provide a more homogenous system than previously available with AMG66. Therefore, co-ordinate information is to be presented in terms of MGA94 in cadastral surveys commenced after 1 July 2005, where co-ordinate information is required in support of documentation to be lodged with either Titles Registration Services or the Surveyor-General. Surveys commenced before 1 July 2005 that are current and connected to AMG66 will be regarded as complying with the requirements of the relevant legislation and directives of the Surveyor-General. Generally, cadastral surveyors are requested to connect to coordinated marks. However, if these marks are not within 3 set-ups (traditional surveys), then the Surveyor General will arrange the provision of coordinated marks within the vicinity of the survey. Refer to Surveyor- General Practice Directives and Regulation 11 of the Surveying (Cadastral Surveys) Regulations Victorian surveyors have two possible GNSS services available to support the connection requirements of the Surveying (Cadastral Survey) Regulations 2005 and the Survey Coordination Act These are: 1. GPSnet A permanent GPS Base Station Network which records, distributes and archives GPS satellite correction data for post-processed accurate position determination in Victoria. Land Victoria, working in cooperation with Industry, has established public access, dual frequency base station infrastructure to support GPS users across the state. The surveyor can download the RINEX data from the base stations records and then differentially post-process with their own single/dual frequency receiver data to achieve accurate and reliable GDA/MGA positions. 2. AUSPOS a free online GPS Processing Service operated by Geoscience Australia which: provides users with the facility to submit dual frequency, geodetic quality, GPS RINEX data observed in a 'static' mode, to the GPS processing system and then receive rapid turn-around GDA and ITRF coordinates, and takes advantage of both the IGS product range and the IGS GPS network and works with data collected anywhere on Earth Note: Both these techniques allow precise GDA/MGA coordinates to be determined for GPS stations in the survey area. GPS receivers can be placed directly onto PCMs or RMs without the need for multiple traverse set-ups to a coordinated mark. This is very useful for the coordination of new marks or for checking purposes. 100

101 Additional Survey Considerations 1. GPS is another measurement technique and as such should only be used if it is the most efficient and cost effective method of survey available 2. Use the receiver/antenna configurations recommended by the manufacturer but avoid mixing different types of receivers in a survey. 3. GPS allows the surveyor to place new stations exactly where required without the intervisibility requirements of traditional surveys. 4. Mission planning is the first phase of managing any static or real time GPS survey. This is necessary to define significant aspects of the survey so that it can be performed effectively and efficiently under foreseeable conditions. Commercial planning software is available and by using the latest satellite almanac the surveyor can: visualise and predict satellite availability via graphs and tables, simulate field conditions with respect to satellite selection, time zones, site visibility obstructions, and elevation masks, and determine the best time of day for observation sessions, given necessary constraints on PDOP and sky view obstacles. 5. In order to minimise post-processing errors and biases, calculation of baselines should start from a mark based on known geodetic coordinates i.e. in a datum such GDA. The accuracy of these values should be better than 20 metres both horizontally and vertically. This is because an uncertainty of start position of 20 metres adds a systematic 1 ppm error into baseline results. 6. Take time to plan the baselines to be measured. Where possible the following concepts on GPS baselines should be considered. 101

102 Connect GPS baselines to build up a network which increases the redundancies in the survey Combining Static and Kinematic GPS Surveys Measure between adjacent sites keeping baselines short i.e. baseline length affects accuracy Similarly keep GPS loops small GPS traversing is acceptable between coordinated marks or can be checked via loop closure Unlike conventional surveys, the shape of a GPS network is not significant in the final accuracy. 7. It is vital that all GPS antennae heights are meticulously measured and checked. GPS baselines are required as ground-to-ground vectors. However, GPS vectors are observed from antenna phase centre to antenna phase centre in the field. Baseline processing software reduces these vectors based on antenna phase centre information supplied. Any errors in the measured antennae heights will affect the final reduced baselines i.e. affect both horizontal and vertical components of the vector. 8. Processed baselines can be used as input into network adjustment software but it is important to have the appropriate statistical input. GPS baselines that have been observed simultaneously in the same session are correlated (or linearly dependent). The misclose of any loop within the session would theoretically be zero. As a result there is no independent check or redundant observations for that session and any dependent vector is referred to as a trivial baseline. For example, if stations A, B, C are observed in the same GPS session, then the baselines are correlated and vector A-C is a dependent or trivial baseline. 102

103 N.B. For n simultaneous GPS stations, there are n-1 independent baselines Independent checks, using GPS data, must come from additional observing sessions i.e. observations recorded at a different time. When using multiple receivers it is good survey practice to link sessions by re-observing some common baselines i.e. through pivot stations that are common to two or more sessions. These independent baseline sets (primarily Static techniques) can then be used to build up a network of control points. The established network can be checked for a series of independent loop closures, and then subsequently adjusted without an artificial sense of redundancy. Baselines from different sessions can be added together to form closure checks 103

104 Linking GPS Session with a Common Baseline 9. In certain environments, the GPS antenna may receive multiple signals which have been reflected off nearby objects and surfaces e.g. large water surfaces, buildings and vehicles. Urban environments are the most likely to have multipathing problems. As a result of multipath, baseline vectors are altered and as such the final position of the receiver is in error. Multipath errors are not constant but change rapidly over time due to the dynamic nature of satellite geometry. Therefore, these errors are particularly hard to detect and eliminate. The surveyor can minimise multipath effects by: making GPS observations from stations that are totally clear of objects and surfaces that may introduce signal reflection, or taking a second measurement after a suitable time period has elapsed (after about 30 minutes apart) plus making independent check from another station. 10. All high precision surveying applications require differential carrier phase observations. The processing software/firmware must be capable of determining the integer ambiguities for post processed static surveys (this may be only possible for short lines), prost processed kinematic and real time positioning techniques short line static and real time methods such as RTK. However, sometimes ambiguity initialisation can be incorrect even though the recommended techniques and statistical tests are followed. It is vital that the surveyor adopts procedures (whether real time or post processed techniques) to ensure the correct resolution of ambiguities. These procedures could include: Re-initialisation of real time GPS receivers e.g. OTF initialisation of the RTK rover - the receiver is re-started or by turning the antenna upside down so that lock is lost to all satellites. After re-initialisation, some marks could be checked for a second time. Re-occupation of the same base station at a later time the ambiguities resolution is random and points can then be checked. If the re-occupation is after a suitable time interval then this also provides a check on multipath errors. Occupying and observing from a second base station at later time this provides the most reliable check on ambiguity resolution as well as an independent check on base station coordinates and multipath errors. 104

105 11. Baseline processing provides the surveyor with a series of vectors plus an overall quality of the GPS measurements, but good surveying practice should include a network adjustment of all observations. The subsequent least squares adjustment will provide final results and an analysis of the consistency of the observed baselines within the network. Even RTK observations may be added to a network if the necessary field files are also logged (as recommended by the manufacturer) and then imported into the software for subsequent recomputation and adjustment. Guidelines for Cadastral Surveying using GNSS 105

106 Measurements and Dimensions Bearings and Lengths In line with the Surveying (Cadastral Surveys) Regulations 2005, GPS vectors and boundary lines determined via GPS are to be supplied as bearings, and horizontal distances (either ground distances on a plan or plane distance on MGA). Using two GPS receivers, the relative position or baseline between two station marks is determined i.e. a cartesian vector ΔX,ΔY, ΔZ in a geodetic datum such as WGS84. Post processing usually occurs in the WGS84 datum and then various transformations can be implemented to bring the vector into the required datum of GDA as a cartesian vector (or if necessary a geodetic azimuth and ellipsoidal distance). This is usually done within the proprietary GPS software. Once this has been done the geodetic vector can be projected onto the UTM to produce the MGA vector (ΔE,ΔN) and then into final coordinates (E, N plus Zone) if start values are known. Note: 1. The GPS vector (the final bearing and distance) between two points determined from simultaneous GPS observations at those points is regarded as the measured dimension. 2. GPS observations on a plan shall be shown as the two dimensional polar (horizontal) vector between survey marks, e.g., as a bearing and reduced horizontal ground distance. Conventional GPS Coordinates Where coordinates derived from GPS observations are being shown, they shall be provided as MGA coordinates (ie. E, N plus Zone), and not as geocentric cartesian coordinates (e.g., X, Y, Z) or geographic coordinates (e.g., φ, λ, h) in GDA. Heights Where heights are to be shown on the plan, GPS ellipsoidal heights (h) in GDA, must be transformed to the Australian Height Datum (AHD). This requires knowledge of the geoidellipsoid separation (N) for that particular geodetic datum. The N value can be determined by observations onto Benchmarks (ie. AHD ellipsoidal height comparison) or by adopting a geoid model such as AUSGEOID

107 Guidelines for Cadastral Surveying using GNSS Classification and Accuracy of Surveys The standards of accuracy for GPS data must comply with the Surveying (Cadastral Surveys) Regulations Accuracy applies to traditional cadastral survey techniques as well as the indirect measurements from various GPS methods such as static, kinematic and RTK. It is always the responsibility of the licensed surveyor to use the appropriate instrumentation and procedures to achieve the accuracy of the Regulations. It should be noted that the bearing and distance of a measured line, is a vector that can be determined from conventional surveying or by GNSS. These vectors can then be manipulated in the usual way e.g. traversing, radiations, resections, intersections, and networks. GPS has certain advantages for the cadastral surveyor: Flexibility in designing surveys ie. GPS can be used at all times of day (and night) and not significantly affected by poor weather conditions. The system is also global and can be used in any location. The two receivers, required for differential operation do not require line of sight intervisibility. This enables surveyors to coordinate marks to survey accuracy over distances which previously may have required several days of traverse measurements. It is this feature that makes GPS so attractive for survey work. Marks do not need to be placed for traditional traversing (ie. line of sight such as on top of hills), but can be placed directly where they are needed. The system requires only a clear, unobstructed view of the sky above a selected elevation. Obviously this restricts the use of GPS in urban areas and densely forested areas. Differential GPS methods allow a high degree of precision to be obtained over distances from metres to thousands of kilometres. Thus the notion of a "traverse closure" is not always appropriate to GPS surveys as traversing is not necessarily the most efficient GPS observation procedure. GPS loop closures are applicable if each line is from an independent observation session. Therefore, the surveyor must design procedures to analyse results similar to those already employed by the profession e.g. statistics, least squares. Guidelines for Cadastral Surveying using GNSS Independent Checks The need to perform independent checks on measurements is specified in the Surveying (Cadastral Surveys) Regulations This requirement applies, particularly, to the measurements used to locate and determine survey boundaries. Licensed surveyors are familiar with best survey practices required for checking conventional surveys. Independent checks on GPS surveys should be treated in a similar way to any traditional survey. The reliability of observations can be safeguarded by way of additional or redundant observations (such as traversing, radiations, intersections, distance and offset and distance measurement between radiations). It should be noted that differential GPS measurements between two receivers gives a vector for that observing session. That vector can be re-determined independently by observations 107

108 made at a different time (at least 30 minutes after the first observation) to enable satellite geometry to change and thus ensure that any multipath errors will be detected. Unfortunately, because this vector is not connected to the whole survey then multiple observations of the same vector cannot be accepted as an independent check for cadastral surveys i.e. not good survey practice. However additional observations are always useful to increase the redundancies in the survey. Checks may be made by GPS and/or traditional methods. A few examples include: GPS traversing by using two receivers simultaneously a single vector (bearing and distance) is obtained between stations. By an observation sequence of leap frogging receivers, a consecutive series of vectors analogous to a traverse is obtained. Each traverse line is then independent and conventional loop or traverse closures can be adopted. GPS network incorporating important marks into network observations involving numerous sessions of multiple receivers that are being moved to ensure sufficient redundancies. A minimally constrained least squares adjustment is then carried out and results can be analysed for precision and possible outliers detected. Observations from two or more base stations - using a GPS base station and fixing each mark via a rover receiver and then checked by an additional measurement from at least a second different base station (or alternatively multiple base stations to increase redundancy). This method will check for correct ambiguity initialisation plus multipath errors and also provide an independent check on the base station coordinates. Traditional terrestrial measurements - any three marks placed by GPS may be checked by terrestrial measurements i.e. the three inter distances or two distances and the included angle. Also, conventional survey techniques need to be used when GPS observations are impractical due to vegetation and buildings. Distances Only Angle and Distances Guidelines for Cadastral Surveying using GNSS 108

109 Reporting Under the cadastral regulations, a licensed surveyor must prepare a detailed survey report when lodging an abstract of field records with the Surveyor-General or the Registrar of Titles. However, the regulations stipulate that if a cadastral survey has been performed by methods other than a direct determination of directions and distances, then such methods must be described. Therefore, the use of GPS for a cadastral survey would necessitate an additional section to the survey report required. The additional section would provide the following information to show that suitable GPS observations and reduction procedures were employed for the cadastral survey: 1. List of Equipment Used - the type and model of equipment used. Also, information on any base station service that has been used. 2. The date and validation methods of GPS equipment validation 3. Description of the GPS methods employed - a description of the methods could include: the method of survey used e.g., static, rapid static, stop and go, kinematic, or real time kinematic (RTK) the expected precision from the method of survey used. This may be provided by manufacturers, software providers, other survey literature or the surveyor s experience description of any specific parameters programmed into the receiver or used in processing that would be likely to affect the result of the survey, e.g. use of tropospheric models for static observations, an indication of observation, session times and ephemeredes used in the pots processing i.e. broadcast or precise the mode of operation e.g., single/dual frequency observations, carrier phase, differential pseudorange, or carrier phase smooth DGPS a tabulation of the observations used from any base stations a description of the GPS reduction techniques used including the software used 4. Assessment of GPS data quality the following would help prove the appropriateness of the methodology used for the survey. the repeatability of observations e.g. the maximum difference or standard deviation of repeated observations on each line a comparison of GPS observations with underlying work such as comparisons with traditionally determined vectors summary of independent checks to verify quality assessment e.g. loop closures or network analysis Guidelines for Cadastral Surveying using GNSS Surveying Using Global Navigation Satellite Systems 109

110 Engineering Surveying Introduction The Height Component Deformation Surveys Construction Surveys Introduction The types of surveys that may be classed as engineering surveys are broad in scope. Control surveys for road construction, cross-section surveys for design purposes, detail surveys for volume calculation and monitoring surveys can all be considered related to engineering. Therefore, each of the GPS observation techniques may be applicable depending on the survey being performed. This section focuses on some of the issues which surveyors should consider when performing certain engineering surveys. The ICSM SP1 Publication should be consulted for specific details regarding performing of GPS surveys. Engineering Surveying The Height Component The height component is often of prime interest in surveys for engineering purposes. Surveyors must be aware that the GPS height component is weaker than the horizontal component. In general, the accuracy of the GPS height component is a restrictive factor and this renders GPS unsuitable for many height related surveying applications. The height component is approximately one and half times weaker than the horizontal component due to satellite geometry restrictions. The coordinates derived from GPS measurements are referenced to the WGS84 datum. The height component is, therefore, referenced to a mathematical surface, rather than a physical surface such as mean sea level. The geoid undulation must be computed if GPS heights are to be referenced to mean sea level, or the Australian Height Datum (AHD). Errors in the estimation of the geoid undulation must also be added to the GPS spheroidal height error to arrive at a total error in the height component. N.B. Geoscience Australia provides an online service which allows users to calculate the geoid undulation at any location in and around Australia. This service can be located at the following website: Engineering Surveying Deformation Surveys A deformation survey is performed during excavation operations in order to check that the construction region is not moving due to earthwork operations. Another example of a deformation survey occurs in the routine monitoring of dam walls. In these type of surveys, marks in the region of interest are regularly surveyed and related to other marks which are not in the same vicinity and are considered stable. If movement in the marks in the survey region is detected, then action can be taken to maintain the safety of the workers in the survey area. In general, the accuracy associated with monitoring surveys is extremely high. For example, water was released from the Hume weir in 1996 when movement of approximately five millimetres was detected. As a result, the accuracy attainable from GPS systems may be unsuitable for many deformation monitoring applications. However, it should be noted that GPS techniques may be suitable as long baselines can be measured to a precision which is high when compared to terrestrial survey techniques where errors may accumulate if a large 110

111 number of setups are required (in a levelling run for example). In fact, GPS techniques are in extensive use throughout the world for crustal motion detection. It is left to the surveyor to make a professional judgement as to the suitability of GPS techniques for these types of surveys. If GPS is indeed considered suitable for a deformation survey, the static occupation technique should be adopted. The guidelines for geodetic surveying should be followed. If the highest precision is required, then twenty four hour data collection sessions should be observed. This will ensure that the receiver receives measurements from satellites in every possible position with respect to the specific receiver location. By observing for this length of time, many of the errors that restrict GPS for high precision applications will most likely have averaged and not effect the final computed baseline. Engineering Surveying Construction Surveys Surveyors are frequently asked to undertake construction surveying work, which primarily includes the setout of features such as buildings and roads as well as infrastructure such as pipelines. Historically, this work has been undertaken using more conventional survey equipment, such as the total station. However, the advantages of static and RTK GPS have been widely recognised for this work. The rapid nature of RTK GPS in particular makes this work efficient and cost reductive. If GPS is indeed considered suitable for a construction survey, then the Real Time Kinematic (RTK) technique could be adopted. Furthermore, the ICSM SP1 Publication should be followed. Surveying Using Global Navigation Satellite Systems 111

112 Control Surveying Introduction Observation Technique Receiver Type Introduction A control survey can be considered a survey which is designed to provide marks which will form the framework for subsequent surveys. A geodetic survey can be considered a high level of control survey with special accuracy requirements. In the context of this document, a control survey may be a survey with an accuracy requirement which may be attainable using either static or rapid static observation procedures. As the nature of the survey may vary greatly from application to application, the recommendations for performing control surveys are not as stringent as that of geodetic surveys. Examples of such applications may be the establishment of points for photogrammetric control purposes, or for future detail surveys. Surveyors should use their professional discretion in deciding whether to use these recommendations, or those of geodetic surveying, for their required application. Further, surveyors should use the ICSM SP1 Publication as a guideline for selecting the appropriate GPS survey technique. Control Surveying Observation Technique When performing control surveys, both the static and rapid static surveying techniques may be adopted. The recommendations in the ICSM SP1 publication should be adhered to when performing control surveys. If the rapid static technique is used, sufficient baselines to provide a network adjustment with redundancy must be observed. A bipod arrangement can be used in place of a tripod, however, is not recommended as standard practice. The occupation period will depend on the baseline length and receiver used, however, the minimum occupation period should be twenty minutes. Several marks with well defined coordinates should be occupied as part of the survey, especially if the GPSnet reference stations are not used. The surveyor should adopt the recommendations for geodetic surveys if in doubt as to the suitability of these recommendations to a specific application. Control Surveying Receiver Type The use of single and dual frequency receivers is suitable for control surveys. If single frequency receivers are used, occupation times will need to be increased and shorter baselines must be observed. The ICSM SP1 Publication provides an indication of the observing conditions and occupation periods for static and rapid static surveys. The use of an antenna ground plane is strongly recommended, however, may not be necessary if the accuracy requirements of the survey are lenient. It may be more practical to use a bipod arrangement without a ground plane when performing rapid static surveys. Again, the surveyors professional judgement should be used when deciding whether a ground plane is necessary. Control Surveying Surveying Using Global Navigation Satellite Systems 112

113 Geodetic Surveying Introduction Observation Technique Receiver Type Multiple Occupations Control Requirements for GPS Surveys Designing GPS Surveys Introduction Geodetic surveys can be defined as those performed to coordinate survey marks for the purposes of establishing a reliable and accurate control framework. Geodetic surveys are considered distinct from control surveys in the respect that the marks being coordinated are assumed to be stable, well built structures. In Victoria, an excellent geodetic framework has been established incorporating many years of observations using both terrestrial and GPS techniques. The recommendations made in this section are designed to meet the standards required to maintain the high quality of the Victorian network. In accordance with the theme of this entire web site, the surveyor should use their professional judgement in making final decisions regarding the performing of GPS surveys. It must be recognised that, in most instances, the accuracy requirements of geodetic surveys are the highest of all survey types. In general, GPS technology can easily meet the required tolerances, however, the observation technique must be conservatively designed to ensure that these high standards can be met in a reliable and cost-effective manner. Geodetic Surveying Observation Technique In all instances, when performing geodetic surveys, the static observation technique should be used. The length of the baselines being observed is not of importance if dual frequency receivers are used. If a single frequency receiver is used, baselines should be kept below 20km and ideally less then 10km. Each station should be occupied in sessions of at least one hour duration. A minimum of two hours occupation is suggested if practicable and observation periods of longer than this are encouraged. The GPSnet reference stations should be incorporated as part of the observation network where applicable. If control points are to be placed, they should be established in locations with a clear view of the sky. If obstructions are unavoidable, the mark should be placed such that the obstructions are to the south of the mark. Surveyors should refer to the ICSM SP1 Publication for further details regarding performing of static surveys. Geodetic Surveying Receiver Type The receiver used may be capable of single or dual frequency measurements. If a single frequency receiver is used, baseline lengths should be kept below 20km and, ideally, less than 10km. The platform for the antenna should be stable. Observation pillars are desirable, however, are not generally feasible. A tripod is most commonly used and the surveyor should ensure that the tripod is solid in construction and is in good condition. Similarly, the tribrach should be in good adjustment. Care must be taken when setting the antenna 113

114 over the control point. If antennas from different manufacturers are used they should be oriented to the same direction, commonly north. To minimise the effects of multipath, a ground plane should be used with the antenna in all circumstances. The receiver should be shielded from the sun in an attempt to keep it at a reasonably constant temperature for the duration of the observation session. Geodetic Surveying Multiple Occupations The data acquired at the control stations must be processed using the surveyor's GPS data processing package. The ICSM SP1 Publication indicates the solution type that should be adopted. Each point should be occupied on more than one occasion, with a different antenna height used for each occupation. This will enable loop closures to be carried out to detect the presence of baselines which have not been accurately determined. Once the loop closure process has been completed and erroneous baselines removed, a minimally constrained least squares adjustment should be performed to determine the quality of the baselines. The surveyor should refer to the Intergovernmental Committee on Surveying and Mapping (ICSM) SP1 Publication for details of assessing the quality of adjusted baselines. Geodetic Surveying Surveying Using Global Navigation Satellite Systems 114

115 Mixing Receiver Types Introduction Mixing Receivers from the Same Manufacturer The RINEX Format Real Time Considerations Introduction With the need to observe GPS measurements using more than one receiver, the so-called relative positioning method, there is a possibility that the receivers at each end of baselines may be of different models or makes. Unlike a total station and reflector combination where one end of the line makes all the measurements, GPS receivers at both ends acquire satellite information. This raises a number of issues regarding the compatibility of the measurements acquired. This section discusses some of the issues that must be resolved when mixing receivers. Mixing Receiver Types Mixing Receivers from the Same Manufacturer In general, there are very few problems associated with mixing GPS receivers designed for surveying made by the same manufacturer. There are only really two issues that should be considered. First, the surveyor should note that the receiver capability of the "weakest" receiver defines the capability of the pair of receivers. This mainly applies to the frequency tracking capability of the receivers. If one of the receivers is single frequency, then a second dual frequency receiver effectively performs as a single frequency receiver when combined with the first receiver. Users should be aware of this limitation and design surveys accordingly. The second issue of note when mixing receivers manufactured by the same vendor concerns the antenna height. The manner in which the antenna height is measured will depend on the antenna type. When mixing receivers, it is vital to ensure that antenna height and types are correctly measured and, importantly, documented. In many instances, the person processing the data may not have been present during the recording of the data. The surveyor should ensure that appropriate station summaries and data collection specifics are provided to enable the processor operator to correctly interpret the acquired information. Mixing Receiver Types The RINEX Format The Receiver Independent Exchange (RINEX) format is an ASCII data format that has been designed to facilitate post-processing of GPS data collected using receivers developed by different manufacturers. Conversion programs are provided with GPS processing software to enable proprietary data format files to be converted to RINEX, and to facilitate processing of RINEX files within the processing program. The RINEX format specifies three files, an observation file which contains the measured pseudoranges and carrier phases, an ephemeris file which contains the satellite orbit parameters, and an optional meteorological file which contains station surface meteorological readings. In general, the compatibility between different processing packages and RINEX files is quite sound. There are, however, instances where RINEX data collected from different receivers does not process successfully. Many of these problems are caused by the user misinterpreting the fields required to be entered when generating RINEX files. Some of the 115

116 commercially available RINEX conversion routines are difficult to interpret as they offer a large number of options for conversion. These options include correction for errors such as receiver clock offsets. Surveyors should be aware of these settings and consult their technical documentation for specific details if in doubt. In most cases, it can be anticipated that static data processing will proceed smoothly. Kinematic data files process less reliably, however, this is generally due to the processor not being able to resolve the integer ambiguities on the fly. If known baseline occupations are performed frequently throughout the observation period, kinematic processing is also quite reliable. Mixing Receiver Types Real Time Considerations The RINEX format has been developed to facilitate post-processing of GPS data collected by receivers manufactured by different companies. At this time, there is no commonly used format for transmitting differential data across a data link for real time processing. Each manufacturer uses a proprietary format, therefore, mixing receiver types in real time for surveying purposes cannot be performed at this stage. The Radio Technical Commission for Maritime service (RTCM) have, through sub-committee 104, published a data communications format which contains sufficient messages to enable real time surveying to take place. Survey receiver manufacturers have not adopted this format as readily as the navigation industry which uses RTCM messages as a standard for differential pseudorange positioning (DGPS). Trimble Navigation have recently published their proprietary format. This is a first step in reaching manufacturer agreement on a format which will be universally accepted for real time surveying use. Therefore, surveyors can expect that mixing survey receivers in real time will be a practical reality in the near future. Mixing Receiver Types Surveying Using Global Navigation Satellite Systems 116

117 Interpreting Baseline Solutions Introduction Fixed Versus Float Solution The Ratio Value Residual Graphs The Variance Factor Solution Standard Deviation Loop Closures Network Adjustment Introduction The solutions derived from GPS carrier phase measurements are generally presented as Cartesian coordinate differences between the reference and rover receivers. Some commercially available GPS post-processing packages provide spheroidal coordinates of the rover station based on the user provided reference station coordinates. One of the challenges for the GPS surveyor is interpreting the output of processing packages to evaluate whether or not a pre-defined accuracy level has been met. In general, the surveyor must analyse the solution and ascertain whether the integer cycle ambiguities have been resolved correctly. This task does require some experience, however, there are several indicators provided by most manufacturers that can aid in this process. Surveyors should note that the most reliable way of evaluating whether baseline solutions are satisfactory is to observe redundant vectors and perform loop closures and network adjustments. Interpreting Baseline Solutions Fixed Versus Float Solution There are two types of carrier phase solutions that are generated as a matter of course during GPS data processing. A floating ambiguity solution is initially generated where the integer ambiguities are allowed to float as real numbers. The solution ends up providing the coordinate vector and a set of real number ambiguities.. A second solution is then generated where the ambiguities are constrained to integer values and only the coordinate vector estimated. This ambiguity constrained solution is termed a fixed ambiguity solution. It is this fixed solution that is the most accurate and is required for surveying using the GPS satellites. In reality, the processor in effect generates several fixed solutions and selects the best one. This selection is discussed in the ratio value section. If the correct ambiguity values have been used during the fixed solution estimation process, the coordinate vector is generally accurate to better than two centimetres. If incorrect values are used, however, the solution may be worse than the float solution. There are a number of reasons why an incorrect set of ambiguities may have been selected, including large multipath effects, and differential ionospheric effects over longer lines. It is essential that the surveyor ensure that the correct ambiguities were identified by the processor. If the correct ambiguities cannot be reliably detected, the baselines must be re-observed. It is, therefore, wise to observe more data than is anticipated to be necessary to ensure that re-observation is a rare occurrence. Additional observation enables the satellite geometry to change sufficiently and for random measurement and multipath errors to average, improving the probability of successful ambiguity resolution. Interpreting Baseline Solutions 117

118 The Ratio Value For any particular baseline, the GPS data processor will generate a number of solutions using different combinations of integer ambiguity values. Once the solution has been estimated, the measurement residuals are computed for each carrier phase measurement. The residual is the small difference between the measured double difference and equivalent value computed from the estimated coordinates. These residuals, which may be positive or negative, are then squared and added together to yield, the sum of the squares of the residuals. In a least squares estimation process, this value is minimised by the final solution estimate. If the measurements "fit" the computed solution extremely well, a small sum of squares value is obtained. This indicates good quality measurements as well as confirmation that the mathematical model is correctly chosen. In GPS processing, a small sum of squares value may indicate low multipath, low measurement noise and, importantly, correct identification of the integer ambiguities. As the processor generates a number of solutions using different ambiguity combinations, a number of these sum of squares values can be listed. A ratio value can then be generated by dividing the second smallest sum of squares value by the smallest value. This ratio, by definition, can never be less than one. If the correct ambiguity solution has been selected and the measurements are of a high quality, the denominator of the ratio will be a much smaller value than the numerator, yielding a high ratio. This indicates successful identification of the integer ambiguities and, most likely, a solution accuracy that is suitable for most surveying applications. If the best and second best sum of squares values are close to each other, a ratio approaching one is generated. This indicates that the processor has had difficulty and the selection of the ambiguity values is questionable. If the ratio is less than two, the float solution is often adopted as the best available solution. The surveyor must check ratio values, if provided, for suitable values. One word of caution regarding the ratio indicator value, a high value does not always guarantee that the correct ambiguities have been chosen. This is especially true if the occupation period is short. A multipath effect which does not change throughout the observation session may incorrectly influence the ambiguity resolution process. This effect, may or may not, influence the ratio value. Surveyors should be extremely cautious when analysing solutions generated from short observation sessions. Interpreting Baseline Solutions Residual Graphs It is becoming more common in post-processing software packages to provide more information to the user regarding the quality of the generated solutions. With the graphical capabilities of modern computer operating systems, manufacturers are presenting numerical information in easily interpreted graphical formats. Some of the newer versions of processing programs display the measurement residuals in graphical form. Surveyors should analyse these graphs for small gaps in the data (cycle slips), checking that they are not frequent. Data which is prone to cycle slips can be unreliable. The residuals should also be quite small, generally less than a few millimetres for an L1 carrier phase measurement. Larger values can be expected for the other phase observables. If the solution has been resolved with the correct integer values, the measurement residuals should all be small. 118

119 Interpreting Baseline Solutions The Variance Factor The variance factor, sometimes called the reference variance, is a parameter that results from the least squares adjustment process. It is a dimensionless number which indicates whether the initial standard deviation of the measurements is realistic as indicated by the standard deviation of the solution. If the factor is one, then the variance assigned to the measurements agrees with the estimated solution. This factor is very sensitive to the variance values used, therefore, it is not generally practical to arrive at values close to one. In fact, data sets of 24 hour duration, processed using the precise ephemeris, may still reveal variance factors of greater than ten. This would initially indicate that the solution is unreliable as the measurement standard deviation is generally quite well known for most receivers. The effects of multipath, for example, are not reflected in the measurement variance, thus the possibility of such large variance factor values. Surveyors should check for excessively large, or small, values as this may indicate an incorrect choice of ambiguities. In general, however, the variance factor derived from GPS processing packages cannot be treated with great confidence. Interpreting Baseline Solutions Solution Standard Deviation The Cartesian coordinate vector solutions generated from processing packages are presented in terms of differences in the X, Y and Z coordinate axes. Each of the components will have an associated standard deviation. This standard deviation value is derived from the normal matrix used in the least squares formulation. Unfortunately, the standard deviation values are, more often than not, optimistic estimates which are generally not of any practical use. The reason for this is due to the processor dealing with a large number of measurements, which statistically indicate an averaging of all random effects. The many errors affecting GPS measurements are generally not accounted for in this process. It is, therefore, not uncommon for solutions to have standard deviations of less than one millimetre, even for baselines greater than 20km! As a result, the standard deviation of the solutions is not recommended as a measure of the baseline accuracy. On the other hand, if the standard deviation values are high (greater than a few centimetres), the surveyor should be wary of the solution vector. 119