5.2.2 GPS Survey Planning

PROJECT DESIGN

Designing the project layout is one of the most important planning tasks of the GPS surveyor. The final network / project design is usually a compromise between technical requirements and economics, worked out within the framework of explicit recommended practices for GPS surveys (or at the very least, prudent practices that ensure the job gets done to the appropriate standards of accuracy and reliability). The surveyor must take the following factors into account:

Definition of the network: size and shape of the overall network, the number of stations, station spacing, any intervisibility requirements, new and existing (known) stations.

Spacing of the existing (known) stations: for what purpose are they intended? as a quality check, for densification of existing control, for the determination of transformation parameters, etc.

Accuracy requirements (defined by client) and standards (defined by the geodetic control authorities), for both horizontal and vertical surveys.

If the number of stations (known and new) to be surveyed is greater than the number of instruments, the GPS survey will have to be carried out over a number of observation sessions. Hence "project design" and "observation scheduling" are inextricably linked together.

Propagation of the GPS Survey

One of the significant advantages of the GPS survey technique over conventional surveying techniques is that sites may be placed where they are required, irrespective of whether intervisibility between stations is preserved. Generally, the GPS stations would be "clustered" around the project focus, for example a road, dam, powerline corridor, etc. This is in contrast to traditional geodetic control that was generally evenly spaced and the stations located in prominent locations such as at the tops of hills, to ensure that they were visible from afar. In addition, extra survey stations that "carry in" the control from the nearest geodetic control stations to the project area are not usually necessary for GPS work. Hence even spacing of stations and selection of stations on the basis of terrain are no longer important considerations (see Figure 1 below).

Figure 1. Terrain need not influence site selection.

Once the number of GPS stations has been decided upon, and their approximate locations have been determined, other considerations may influence where additional stations may need to be located, or where refinements to the network design could be made:

• Some intervisible stations may need to be included to define starting azimuths for subsequent conventional surveys. This may be best satisfied by the provision of some additional azimuth marks surveyed by GPS, rather than altering the original network design to incorporate station intervisibility. The reference marks may be set up a couple of hundred metres away (and, depending on the terrain, etc., perhaps up to a kilometre or more distant).
  
• There may be several existing stations in the area which could be included in the GPS survey. There are many reasons for occupying already established stations (apart from the obvious one: the client requests it!). Generally the reasons range from simple expediency (save on establishing new monumentation), to necessary ties to previous work, or to ensure datum definition or the calibration of GPS heights (Figure 2 below).
  
• There is generally no need to establish GPS stations to "connect" the main net to surrounding control unless distances are large. Although relative error is a function of interstation distance ("parts per million" is the ratio of baseline vector error to baseline length), it is not a linear relationship. For distances up to 20-30km an "ambiguity-fixed" solution is generally possible (section 8.1.3), and the accuracy of the position solution is of the order of 1 to 5 ppm. When, at longer distances, ambiguity resolution is not possible, the "ambiguity-free" solution is generally weaker by a factor of 2 to 3 compared to the "ambiguity-fixed" solution. It remains at this new (lower) relative accuracy for distances up to 100km. The optimal solution for baselines longer than 100km (if using commercial GPS software) is one based on triple-differenced phase observables.
  
• Certain baselines may be designated as primary ones, for example, the baseline(s) defining the axis of a tunnel or engineering structure. These are critical to the final network and must be at both a higher accuracy and higher reliability. Hence they may need to be measured several times, or special instrumentation and procedures used. Network design may therefore not simply involve points, but baselines (or figures) as well.
  

Figure 2. New, old and datum GPS stations.

Why New AND Known Stations?

A GPS survey typically requires the occupation of new stations and stations whose coordinates (2-D, 3-D, or height) are already known, in either the GPS datum or the local geodetic datum.

Some reasons for surveying both existing (known) and new stations:

Required by the relevant GPS survey standards & specifications.

For the determination of local transformation parameters between the GPS datum and the local geodetic datum.

For quality control (QC) purposes.

In order to determine the geoid-spheroid separation.

In order to connect new GPS points into surrounding geodetic control.

However, a minimum of one known station must be used as the datum station in the GPS survey -- its coordinates must be known in the WGS84 system.

The number and distribution of known stations, and the accuracy with which the coordinates of the known stations are required is strongly dependent on the use to which the known stations will be put. Guidelines can usually be found in the relevant GPS "standards & specifications" (section 10.2.5). For most purposes a minimum of three or four stations around the perimeter of the survey project area is sufficient.

These known stations may be occupied in the course of the GPS survey, or they may be continuously operating "base" or "fiducial" stations. For example, there are many IGS and other continuously operating base stations in many countries around the world which may be incorporated into the survey network if they are in the vicinity of the survey area. In future it is likely that there will be many more of these permanently operated base stations, maintained by government departments, universities, or private companies, to support a range of GPS activities (navigation, surveying and geodesy).

Accuracy Issues

Accuracy of New Stations:

• Required accuracy is defined by the project specifications, but the accuracy classification may be defined by GPS survey standards & specifications.
• Latitude is typically better determined than longitude. Height is worse than both. (The actual accuracy is dependent upon the instrumentation used, observations made and software used for the phase data reduction.)
• Typically, the horizontal coordinates are required in a local geodetic datum, and the vertical results in the orthometric height system. GPS derived 3-D coordinates are nominally in the WGS84 system.
• The accuracy of the GPS derived 3-D coordinates depend on the accuracy of the datum station(s) incorporated into the GPS solution (GPS phase solutions are relative, hence something has to be held fixed in WGS84).
• The final coordinate accuracy (in relation to the local geodetic datum) depends on the precision of the transformation between the GPS derived coordinates and the coordinates of the known stations in the local network.
• However, after transformation, the internal consistency (or precision) of the GPS results is unchanged.
• The final height accuracy depends on the precision of the height transformation: the geoid-spheroid separation, which can be determined in a number of ways (Chapter 11).

Accuracy of Known Stations:

Both GPS derived coordinates and the coordinates of geodetic control are essentially RELATIVE. However, GPS derived coordinate accuracy relative to the local geodetic datum origin depends on:

• the datum station(s),
• the transformation process, and
• any secondary distortion of the GPS network resulting from fitting to the datum defined locally by the known geodetic stations.

The following should therefore be noted:

• Datum station accuracy is generally defined by GPS survey standards & specifications. This influences the accuracy and precision of the GPS results.
• Final transformed GPS coordinate accuracy depends on the quality of the transformation parameters (themselves dependent on the accuracy of the known stations used to derive the transformation parameters).
• If the GPS transformed results are also to be fitted into the surrounding control, the final transformed GPS coordinate accuracy and precision depends on the accuracy and precision of the known stations in the local geodetic system.

Accuracy as a Classification Criteria:

Different geodetic control authorities may have adopted different terminologies for the classes of GPS survey, and have varying numerical accuracy limits used to define the categories, but all distinguish these categories by some relative accuracy measure. For example, in Australia and the U.S., relative error is defined for the various categories of survey by the specification of the maximum allowable "base error" (a) and "line-length error" (b), at the 95% confidence level, for the relative error ellipse (or ellipsoid):

e = a + b.L	(for Australia)	(5.2-1a)
	(for the USA)	(5.2-1b)

where L is the interstation distance in kilometres, the quantities e and a are in millimetres, and b is expressed in "parts per million" (ppm).

There are essentially two classifications for accuracy:

An internal one based on the minimally constrained adjustment of the GPS-only survey.

An external one based on a constrained adjustment, where existing geodetic stations of the GPS network are held fixed to the published values of the terrestrial geodetic datum.

In Australia, the former corresponds to a survey's CLASS, while the latter defines its ORDER (section10.2.2). For details concerning the Australian classification system see ICSM (1994), and for the U.S. see FGCC (1988). Table 1 below is an extract from the Australian Standards and Practices for Control Surveys. Table 2 is the U.S. equivalent. Note the significant difference in definition of base error and line-length error between the Australian and U.S. standards. The Australian standards address all types of surveys, not just GPS.

Table 1. Australian horizontal ( 2-D ) survey accuracy classifications.

CLASS* Minimum geometric accuracy standard (95% confidence level) a -- Base error (mm) b -- Line-length error (ppm)
3A 0.2 2 2A 0.6 8 A 1.5 18 B 3 35 C 6 75 D 10 125 E 20 250

*CLASS is a function of field and reduction procedures.
ORDER is a function of both CLASS and fit to existing control.

Table 2. U.S. three-dimensional GPS survey accuracy classifications.

ORDER-CLASS Minimum geometric accuracy standard (95% confidence level) a -- Base error (mm) b -- Line-length error (ppm)
AA 3 0.01 A 5 0.1 B 8 1 1 10 10 2-I 20 20 2-II 30 50 3-1 50 100

The product of the Network Design Process is generally a project sketch depicting such information as: geographic graticule, symbol for station type (known and new), non-trivial baselines to be measured, repeat baselines, number of independent station occupancies, azimuth reference stations, and any other information that may help in the logistical design. See Figure 2 above.

Network Shape

As in the case of conventional surveys, there is an impact arising from "structural" considerations.

Some networks are superior to others with regard to "strength".

Only independent baselines contribute to network strength.

GPS networks may have different shapes (Figure 3), as well as different "strengths" arising from the number of independent baselines observed over a number of sessions (Figure 4).

Figure 3. Network shapes: "wide" and "narrow"

Figure 4. Network strength:
a function of the number and location of independent baselines.

Back To Chapter 5 Contents / Next Topic / Previous Topic

© Chris Rizos, SNAP-UNSW, 1999