Towards a GNSS Data Accuracy Standard for Georeferencing


 

By Dr. David Siriba, LS (K), MISK

Department of Geospatial and Space Technology, University of Nairobi, P.O Box 30197, Nairobi

[email protected]

Introduction

Georeferencing is a buzzword and its definition well established. Although the term is generally associated with GIS and remote sensing, the underlying concepts that include reference surfaces, map projections, and coordinate systems – are primarily land surveying. The knowledge and application of these concepts are important for the proper utilization of technology for georeferencing.

The Land Registration Act, 2012 requires that parcel boundaries on cadastral maps be georeferenced and surveyed to such standards to ensure compatibility with other documents required in the land registration act. What is clear in this provision is that, for land registration, it is not just georeferencing for the sake of it, but ensuring that any such georeferenced land parcels are surveyed according to standards – in this case, the cadastral standards. The question then is: do we have standards for georeferencing and how do we carry out georeferencing to achieve those standards?. The purpose of this article is to revisit georeferencing in the context of cadastral surveying concerning boundary systems and accuracy standards.

Georeferencing revisited

Georeferencing has traditionally been used to refer to the process of associating a digital image with locations in physical space in terms of coordinate systems. The term is also used when assigning or determining the spatial locations of geographic objects within a geographic frame of reference. When combining geospatial data from different sources it is necessary to have a common reference system, otherwise, there will be geometric differences between the data layers. The differences could also be due to different data resolutions and accuracies.

To avoid these geometric differences and ensure the compatibility of data from different sources, a common reference system is usually specified. The existence of many reference systems makes it necessary to specify an official reference system – usually by the authority responsible for cadastral surveying. For historical reasons, the existence of different coordinate systems has limited the realization of a consistent nationwide cadastre. To avoid using technology (in georeferencing) to perpetuate the existing status, it is important to define a nationwide geodetic reference frame (in our case KENREF) consistent with Africa and international reference frames, and also specify required positioning accuracy thresholds.

General and Fixed boundaries

A land parcel boundary is one of the key elements of a cadastre. A land parcel boundary, alternatively referred to as cadastral boundary, defines the spatial extent where homogeneous rights, responsibilities, and restrictions exist. Land parcel boundaries are required to be unambiguously defined both on the ground and on the map. The legality and nature of the parcel boundaries depend on whether they have been registered and whether they are accurately defined. If the boundaries are only explicitly marked on the ground using walls, fences and hedges, they are considered general. Otherwise, the boundaries are fixed and legal if they are marked on the ground by permanent monuments, which are surveyed by ground methods with corresponding measurements displayed on the map; and the boundaries are connected (fixed) to an official geodetic network.

In general boundaries, the boundary line between adjoining land parcels is left undetermined or kept vague with the main advantage primarily being less demand on the standard of surveying accuracy; In this way, the land registrar can ignore small changes in the position of the boundary agreed by two parties while guaranteeing the title. In fixed boundaries, the boundary line is accurately surveyed so that any lost boundary corners can be recovered accurately from survey measurements. The main advantage of fixed boundaries is the confidence which the landholders can have as to the precise extent of their properties.

Land parcel boundaries are legal entities and therefore their documentation and official registration are important. The description and documentation of boundaries can be done in various ways with varying degrees of technical complexity and financial costs. The approaches include textual description, graphical representation, or representation by coordinates.

Textual description (also called metes and bounds) uses landmarks as points of reference to describe the route for a land parcel boundary. This approach has been used to describe the administrative boundaries in Kenya(see Districts and provinces Act of 1992).

Graphical description uses graphical survey techniques, for instance, photogrammetry to survey and map land parcel boundaries, without involving any direct geometric measurement of coordinates. The boundaries are plotted either directly in the field (from GNSS tracks) or from aerial/satellite photographs. In Kenya, the Preliminary Index Diagrams (PIDs) and the Registry Index Maps (RIMs) are graphical cadastres that represent general boundaries.

Description of boundaries by coordinates entails a mathematical description of the boundaries using bearings and distances of the individual land parcels determined from computations based on actual land survey. The dimensions of individual parcels are measured and referenced to either local or national coordinate system. Mathematical description ensures high positional accuracy of boundary definition and therefore the required coordinate accuracy can be stipulated precisely. Depending on the survey regulations, numerical cadastres are generally considered to be of the best positional quality and the boundaries are fixed.

In Kenya, some of the maps, in particular, the PID, used for land registration are not georeferenced and the boundaries are general. Although the RIMs can be considered to be georeferenced, the boundaries are however general. Only the survey plans are georeferenced and the boundaries are fixed. Any attempt to georeference parcel boundaries using handheld GNSS receivers in areas based on PIDs will not result in accuracy improvement, because of the inherent inaccuracies in the PIDs. Therefore the method of the survey to adopt becomes an important consideration to meet accuracy standards.

 

Cadastral Accuracy Standards

Traditionally, survey accuracies have been expressed by precision ratios. This method for evaluating survey accuracy is well known and published in many surveying textbooks and manuals. The allowable limits of closure for surveys are derived from the summation of all of the latitudes and departures along the surveyed lines of a closed traverse.

It should be noted that the standards were created regarding the accuracy limitations and technology of the time and do not reflect the way modern field surveys are conducted. The lack of a survey accuracy standard that addresses all types of surveys has created an opportunity for some practitioners to even use handheld GNSS receivers to carry out land surveys even though they could be inadequate to meet the accuracy requirements in particular instances. Accuracy standards are needed mainly to ensure that surveys meet a specified standard of accuracy.

Using a navigation GNSS receiver will result in disproportionately high error propagation in area for relatively small land parcels. Figure 1 illustrates the percentage error against the land parcel area (assuming a parcel with 4 corners) for a handheld GNSS receiver with a nominal point position accuracy of 3 metres (in blue), compared with the cadastral level accuracy (in red). From the figure, the error in area is less than 1% only when the land parcel is more than 100 hectares when using a navigational receiver – a lot of caution should therefore be exercised.

Figure 1: Error propagation in area.

The allowable limits of closure (misclosure) expressed in precision ratios, for example, for a cadastral traverse (i.e., fourth-order traverse) on condition that no misclosure shall exceed 3metres, are specified in table 1 as extracted from the survey manual.

Table; Traverse Standards of accuracy

Farm surveys Traverse between two fixed points Loop traverse
1)   Over level country 1/5,000 1/7,000
2)   Over hilly country 1/4,000 1/6,000
Township surveys 1/6,000 1/8,000

Using a precision ratio is a well-understood principle that during a resurvey the limit of closure or standard in place at the time of the original survey depends on the past survey measurements. The use of precision ratios is not directly applicable when using GNSS equipment, which affords even better measurement accuracies. Instead, a different concept of accuracy is used – that of loop closure.

Loop closure equivalent to precision ratios is a procedure by which the internal consistency of a GNSS network is determined. The procedure entails adding together a series of baseline vector components from more than one GNSS session, forming a loop or closed figure. The closure error is then the ratio of the length of the line representing the combined errors of all the vector’s components to the length of the perimeter of the figure. Any loop closures that only uses baselines derived from a single common GNSS session will yield a closure error of zero, because they are derived from the same simultaneous observations

Both the precision ratios and the loop closure serve only to imply the general magnitude or quality of the relative precision of a closed traverse and GNSS network respectively. The loop closure has minimum redundancy and does not evaluate scale or rotational errors. The least squares analysis fills this gap by adjusting the observations after which a complete statistical analysis can be made from the results. Based on the sizes and distribution errors, various tests can be conducted to determine if a survey meets acceptable tolerances or whether measurements must be repeated. Through least squares analysis precision is depicted in a more comprehensible way using error ellipses for individual points, the relative error of all of the points in a network, or the range of precision within a large network.

Some geospatial positioning accuracy standards exist like the FGDC and ASPRS, only a portion of these standards apply to cadastral surveys. The standards however allow for flexibility by omitting threshold values that data must achieve. The FGDC, in particular, describes two sets of values that must be reported: network accuracy and local accuracy – which are potential for application in cadastral surveying when using GNSS. Network accuracy is a value that represents the coordinates of a control point with respect to the geodetic datum at 95% confidence level. Local accuracy is an average measure (e.g. mean, median, etc.) of the relative accuracies of the coordinates for a point with respect to other adjacent points at the 95% confidence level. Table 2, presents local accuracy standards, for control and cadastral measurements.

Table 2: Local Accuracy Standards

95% Confidence Circle Application
Less than 0.050 (m) Cadastral Project Control
Less than 0.100 (m) Cadastral Measurements

Cadastral Project Control is the network of the GPS stations, tied to the national geodetic datum, which is surveyed to control all subsequent GNSS Cadastral Measurements. The Cadastral Project Control network should be established by either static or fast-static survey methods, while cadastral measurements can be carried out using static, fast-static or even RTK methods. It is therefore important that the standard also includes the different GNSS survey methods with their corresponding achievable accuracies.

Concluding Remarks

As GNSS becomes a common tool in cadastral surveying, the GNSS equipment must be used in a particular way to achieve cadastral accuracy standards. Since the existing cadastral accuracy standards are not adequate for GNSS surveying, a more appropriate approach is to adopt one that specifies the confidence level for each individual coordinate point surveyed. In addition, the specification should also include the accuracy limits of the control points used in the cadastral survey, in relation to the national geodetic network. Different GNSS techniques yield different accuracies – the standard should therefore also include a prescription of the techniques and the accuracy limits. Additional information that the standard should specify include: field survey procedures, control for cadastral surveying, cadastral measurements, data processing and analysis, project documentation and reporting.

Further reading

  • American Society of Photogrammetry and Remote Sensing (2013).ASPRS Accuracy Standards for Digital Geospatial Data. Photogrammetric Engineering & Remote Sensing
  • Craig, B.A. and Wahl J.L. (2003). Cadastral survey accuracy, standards. Surveying and Land Information Science.
  • Districts and Provinces Act, 1992
  • Federal Geographic Data Committee’s (FGDC). Geospatial Positioning Accuracy Standards, July 1997.
  • Michael D. Londe (2002).Standards and Guidelines for Cadastral Surveys Using Global Positioning Methods
  • Survey Manual