Publication by Gartrell Group Principal Bryce Gartrell, 1996

Drawing Lines in the Sand: Representing Property Data Qualitatively in a GIS

by Bryce Gartrell and Tim Hodson
University of Maine
Department of Spatial Information Science and Engineering
1996

1. Abstract

Due to shifting economic and political factors, projects involving initial registration of real property have greatly expanded in recent years. The great expense and time involved in conducting what are often unnecessarily precise boundary surveys, and a lack in sensitivity to customary modes of defining land rights are among the chronic hindrances to land titling projects as they are currently undertaken. This paper investigates the possibility of electronically capturing the locations of occupied parcels and occupier or neighborhood testimony of use limits.

The use of hand-held Global Positioning System (GPS) instruments and sketch books will be proposed as an enhancement to the adjudication procedure typically used in most current mass land titling projects. Methods for evidencing agreements between neighboring landowners, incorporating pre-existing paper evidence of boundary locations and ownership, and recording the findings of adjudicatory teams will also be explored. Means for effectively merging lower accuracy parcel boundary locations with higher accuracy surveys will be explored and the usefulness of "rough" surveys as alternative means of representing boundaries that may later be updated will also be suggested. A mock-out of the technical component of the model is used to explore and refine the concepts. The conclusions describe the problems encountered and suggests solutions.

2. Clarification of Terminology

It is necessary to define what is meant by `qualitative' versus `descriptive' information. The word 'qualitative' is used extensively throughout this paper. The meaning is intended to describe any value which is not based on a formal unit of measure. In terms of spatial dimension - a qualitative value may be based on comparison. A length, for example, is twice or three times as long as another, it can be 'long', 'short' [Downs, Stea `77], or defined in terms of `close' and `far' [Gahegan `95] "[Q]ualitative models deal with magnitudes, which can sometimes be seen as abstractions from the quantitative details" [Egenhofer, Mark `95]. `Qualitativeness' alludes to descriptive methods and is distinct from qualitative reasoning [Ibid ]. This `qualitativeness' is an important component of this model and to avoid confusion will be referred to as descriptive data.

3. Introduction

Cost/benefit analyses of land titling effectiveness have often shown unfavorable ratios in developing countries. This is particularly evident in cases where land values are relatively low. The spatial data gathering component of titling procedures has been identified as one of the most costly, as it requires considerable time and money to be initiated and executed. This is one of the primary reasons why alternative surveying and mapping methods are needed to help assure the success of land titling projects. [Holstein `96]

The goal of a cadastral information system implementors in a territory where land registration is taking place for the first time is to establish what parcels exist, by whom they are owned, what their approximate value is, and thereby, to support the certification or titling procedures. This paper will investigate the potential of alternative surveying and data capture techniques. Specifically, the use of GPS and GIS technologies in combination with hand-drawn sketches will be examined as a means of collecting and representing quantitative, qualitative, and descriptive information about land parcels.

The techniques discussed in this paper are intended to meet the basic requirements of a fiscal cadastre, in the form of a GIS, which is at least the preliminary goal of most initial registration projects, and which serve as a base upon which more extensive, higher accuracy cadastres may be developed as demands require and resources allow. The qualitative and descriptive methods are not seen as a replacement for precise quantitative methods, but rather as an alternative means to represent property boundaries when time, monetary and social constraints preclude efficient implementation of conventional methods.

4. Background and Concepts

Jackson [1995] suggests the high value of participative rural appraisal, recognizing that "rural communities are seen to be adroit at expressing local knowledge in abstract graphical forms." The map is seen to empower local people when they have created it themselves, since it enforces their independence of conventional western techniques. These community-based maps often lack geometrical correctness, but this is compensated for by their accurate topological correctness, which assures, for instance, addressability (correctness of street/parcel addresses). [Ibid] The technique proposed in this paper allows the owners of the land parcels to essentially communicate both their boundary positions and the forms of their boundaries to the adjudicator who, in turn, uses a combination of sketched and written documentation to represent the parcel data in a qualitative and descriptive fashion. There are no restrictions envisioned in allowing local residents themselves to provide sketch information. A community-generated map, such as that outlined by Jackson, is intended to reinforce local participation and endorsement of an officially recognized form of tenure. The middlepoint idea, referred to by [Jackson '95] as "fuzzy tenure", uses a centrally identified point within each parcel to which a right of unmeasured extent is attached. We propose that this "fuzzy tenure," when integrated into a GIS with low precision GPS locations, will allow for the topologically correct community map representations to be associated with the absolute locations of GPS points. Although it should be noted that Jackson's "fuzzy tenure" is somewhat looser-fitting than the techniques we are advancing, the argument presented in this paper is that this "looseness" can be allowed to co-exist with the closer fitting quantitative GPS values. This flexibility is seen to be facilitated by the relatively new and previously unavailable higher precision hand-held GPS receivers and nascent GIS technologies.

5. Range of Applicability

The hand-held GPS/sketchbook adjudication method is subject to the same limitations that affect any attempt at GPS data capture. Its practicability will be largely determined by such factors as the presence or absence of overhead and nearby foliage and structures, ionospheric interference, the availability of satellites, and the accuracy requirements of a given application. Furthermore, its diminished locational accuracy will render it unsuitable to circumstances where a precision of greater than five meters is requisite. Use of a differential solution is proposed, and was used in the mock-up adjudication process described below, making use of the University of Maine base station. This method has been used extensively; in the case study by [Scmidt `95], a maximum error of 4.35 meters was reported, with the average error being approximately 1.3 meters. The base stations in the Schmidt study were approximately 35 miles distant, as compared with fewer than three miles in this mock-up adjudication. A full investigation of the errors in this survey are not conducted in this paper since the capture of qualitative and descriptive data is of greater concern.

The proposed method is particularly applicable where these performance criteria and local cultural factors will allow for a substitution of highly accurate quantitative locational data with both qualitative and descriptive testimonial data combined with lower accuracy positional information. This survey strategy is designed with the expectation that the data which it provides can and will be incorporated into the national geodetic grid datum. Furthermore, it is intended to be useful in the event that changing property interests, valuations, or uses will require higher accuracy survey and mapping work in the future.

6. Context

As the successes of land registration have been given considerable press in recent years, and titling projects have increased, there have been a significant number of similar conclusions reported by titling practitioners. Although the survey methods discussed in this paper deal most obviously with technical aspects of initial titling projects, they respond to a number of these conclusions pertaining not only to survey strategies but also to institutional concerns, and issues of finance, law, and policy. A brief overview of these matters may help define the contribution which hand-held GPS/sketchbook adjudication makes in the context of initial titling of property.

6.1 While land titling has, in many cases, aided in the delivery of such benefits as increased access to credit for small landholders and increased productivity of land, it has not been universally successful. One of the biggest challenges in land titling, and one which has actually been the undoing of some efforts, is the need to accommodate customary systems of tenure into updated or replacement systems. Western countries, which have been the primary exporters of registration system designs, typically record individual property rights with the primary intention of assisting a free market economy by creating an atmosphere suitable for efficient land transactions. It follows that most land registration systems which have been recently modernized or implemented have been based on western concepts of land tenure where individual rights to property are stressed and land is foremost a commodity and an agent for economic growth. These systems are in many cases inadequate to the circumstances encountered in countries where large quantities of land are held under customary tenure and where communally held land rights are common. If land registration systems are to promote the development of a market economy, and avoid the risk of abandonment by title holders through assuring the protection of their traditions, then they must be designed to accommodate customary land tenure data [Rakai and Williamson `96].

6.2 The importance of decentralizing land records and land administration is also being emphasized with growing ardor. Many existing land administration systems suffer from the over-centralization of responsibilities which prevent land-related ministries from effectively and efficiently performing their duties and thus increase widely shared feelings of contempt on the part of the rural poor, the supposed beneficiaries of titling. If land records system implementors are to avoid the unfortunate consequences of losing out in competitions with informal systems or of reinforcing the power of the "haves" at the expense of the "have-nots," [Barnes `96 ] they must provide greater transparency in land administration procedures, greater access to land data, and solicit public involvement in process of deciding how land is used, valued, regulated, and divided among groups and individuals in a society.

6.3 Strong public policy, built through public education, debate, consensus building, and other democratic processes is often lacking in places where land administration projects are undertaken. While laws should be used to enforce policy, antiquated legal requirements are often used to interpret policies; for instance, in many countries in Latin America, policies are often inferred from agrarian reform laws dating back to the 1950's and 1960's, though these laws are neither relevant to the actual state of affairs nor to the present agenda of most governments [Barnes '96 ].

6.4 Cost is, of course, another matter of great concern in land titling projects. Especially in cases where initial titling is being sought, the question of whether the expense of introducing a formalized land records system will be disproportionate to its likely benefits looms large. Surveying and mapping procedures typically consume a majority of project budgets, so the need for cost-efficiency is particularly acute here. Although aerial photo-mapping techniques and pioneering efforts at using GPS equipment to gather boundary location and dimensions show considerable promise, there is still ample room for improvement. One widely recognized hindrance to surveying and mapping cost reduction is an unnecessary insistence on high precision work even in circumstances where much lower levels (between 1 and 5 meters) would satisfy titling requirements. The following survey/adjudication method responds directly to all of the above concerns.

7. Description of Technique

As with any geographical data we are interested in the what and the where of the spatial object under investigation. Qualitative and descriptive data may be documented from physical monumentation at each end of a mutually agreed boundary (fiat boundaries), or from existing physical features (bona fide boundaries). [Smith `95] Defining a land parcel in terms of a `spatial object' is conceptually consistent with GIS design, and is useful terminology for conceptualizing the attachment of other objects to it, such as a legal right, a sketch, an owner, and so on.

Although the proposed hand-held GPS/sketch book adjudication procedure emphasizes content-related parcel data over precise boundary locations, it proceeds from locational information. Hand-held GPS units (the specific models must be determined by the accuracy needs and budgetary constraints of a given project) are used to establish the coordinate positions. The number of positions used will be determined by the shape and size of the parcels concerned; a single, centrally located GPS position will be used in the higher density areas where parcels are small relative to the 5-meter precision of the GPS position.

Where parcels are large, locations will be established at each of the monumented points, being closer to the conventional surveying concept of locating the physical extents of the land parcel. In cases where parcels are irregularly shaped, or are long and thin, GPS locations can be used to provide a generalization of parcel shape. GPS positions will serve to locate not so much the individual monuments of a parcel's boundaries but to position and orient the parcel as a whole, a "spatial object." Of course, in many cases, it will be convenient to observe positions on the monuments themselves.

It is envisioned, as a general rule of thumb, that where boundary lengths are of an order less than three times the maximum expected precision, an approximate central GPS position should be used. For example, a parcel with lengths of less than fifteen meters could be located with a single GPS position of five meter precision.

Because quantitative data is being minimized in the interests of cost and possible conflicts with locally-determined methods for boundary representation, the sketch book data assumes an important role in defining parcels. Accordingly, certain simple but standardized procedures should be used. The following sequence of steps are offered as one possible model:

• Before sketching or interviewing begins, the practitioners are accompanied by a community leader and perform a reconnaissance survey to investigate the area, familiarizing themselves with the general lay of the land and the apparent organization of settlements. They should take particular note of the relative sizes of parcels.

• Land valuation is a significant component of a modernized cadastre, and is typically based on such factors as parcel area and productivity.

A qualitative estimate from the adjudicator or the village representative could give a comparative areal value. For instance, a parcel might be described as being a medium, large, or small property relative to others in the region. This could then be slightly refined. For example, on a scale of one to five, a medium-sized parcel might be given a magnitude of three [Gahegan '95]. The assignment of these areal magnitudes will be restricted to instances in which all the vertices of a parcel boundary are not located.

• Local landmarks, especially those that are widely visible should also be identified. If practical, these can be used as points for base-stations, and may also serve as references for compass readings or general directional references for orientation, especially if parcels are only defined by single GPS locations.

• Adjoining neighbor's indicate their boundary lines, which might be defined by natural features, extant monuments, or might be placed according to mutual agreement. In circumstances where customary, informal tenure has been the norm, irregular linear features, such as fences, rivers, and trenches will often have been used to divide property interests. These types of bona fide boundaries are likely to have intuitive value in informal tenure systems as opposed to the rigidly defined boundaries supplied by conventional surveys and should be accommodated.

• After interviews with neighboring landowners regarding mutual boundaries, which will, in many cases, include strolling bounds, identifying important features, landmarks, and monumentation, a rough polygon representation of the parcel is sketched out. GPS position points are then marked and labeled on the sketch, and significant and identifying features occurring in and around the parcel are drawn. These might include rocks, trees, gardens, houses, monuments, fences, waterbodies, drainage, topographic features, vegetation, pathways, roads, et cetera.

• Each GPS point is given a unique identifying number and description that will link it to the parcel and all associated data. The parcel is given an identification number as well, and, in some cases, a further number can be used to order and place it within a given cluster of parcels being adjudicated. Clusters can be similarly identified.

• Recorded positions will be merged with sketch book data. If the sketch book is digitally interfaced with a GPS unit this may be done in the field. Otherwise it will become part of the post-processing operations; paper sketches will be digitally scanned and imported with positional data into a GIS environment. This latter method is used in our mock-adjudication.

• Depending on the nature of the land administration system that parcel data will join, information about use limits and agreements, ownership, use/productivity of land, soil/vegetation type etc. can be added to the same sketch page, or recorded in separate forms and cross-referenced to the sketch and GPS points.

• If there are pre-existing parcel documents, on paper or in some other format, these may also be joined by reference to the new data.

• The signatures and testimony of neighboring landowners may be added to the sketch page and accompanying documents. If disagreements exist or arise from the process of boundary determination, disputants will be referred to the tribunal function of the adjudication process. Typically, locally reinforced bodies will offer final rulings on contested boundaries. When conflicts over property persist beyond adjudication procedures, they may be noted alongside sketches or on accompanying documents or extracted through query functions in a GIS.

8. Mock-out Adjudication

A study area was chosen near the Universty of Maine. Three "parcels" were defined, a parcel of approximately 5000 sq. metres was used as the 'small' parcel and two larger parcels were established. Based on assumed input from residents, corner markers and boundary features were captured in sketch format on paper, and later scanned into a digital format. At each corner monument of the larger parcels and at the approximate center of the small parcel, a GPS location was captured, using a hand-held Trimble Navigation unit. These GPS rover files recorded at each point were assigned names which were clearly notated on each sketch describing the monument, except in the case of the smaller parcel, where a central point was used. In this instance the name was attached to a second level sketch.

A representation of the ArcView interface shows how the polygons, boundary identifiers, and GPS points, described below will appear.

At this implementation stage the concept of a two-tiered approach for sketches was found to be the most useful. The first level sketches dealt with each parcel as separate objects and was used to represent qualitative and semi-quantitative components. The approximate locations of the GPS points were also documented, with reference to the locations of defined monuments. Please use the following links to see the first level sketches: Sketcha, Sketchb, Sketchd.

Also, see what these look like in ArcView.

The second level sketches allowed a greater levEl of detail, with descriptive information attached to each monument, giving location of the GPS point, and explicit details about physical objects of a permanent nature around the monument, within a radius of approximately 20 paces. For examples of these sketches please make use of the link from each of the rover files defining the GPS locations:

r042518a, r042518b, r042519a, r042519b, r042519c, r042519e, r042519f.

Using Trimble's Pathfinder Office software, the rover files were downloaded and differentially corrected to obtain the final latitude and longitude positions. The base-station data was obtained from the Department of Forest Management at the University of Maine where the base - station is located.

The coordinate points were exported in DXF format, imported into ArcView 3.0, and converted into a shape file. The polygons were established in ArcView by visually linking the coordinate positions with the 'draw polygon' feature. Separate themes were created for the GPS points, the polygons, and the boundary identifiers. Using the paced distances around the smaller parcel, an approximate polygon was drawn to circumscribe the rough central GPS location. The shared boundary of the adjoining parcel established through low precision GPS locations, served as a good reference base to work from. Of course, in a real-world application where the 'central' GPS location is used consecutively for many adjoining parcels, this may be somewhat more difficult, though the general layout would still be well defined.

Hot-links were then established for each of the polygon vertices to cooincide with the relevant corner monument information. In order for the data about boundaries between owners to be established, the 'Neighbors' theme was created. A circular symbol was used at the approximate mid-point along each shared boundary line. On each first level sketch all adjoining owners are noted and the signatures represent their attestation to the boundaries as depicted and described. The instrument shown in these sketches is a basic form and could be more complex depending on requirements.

Tables for each of the three themes were created; the polygon theme table gives information about parcel ownership,rough area, the number of points used for the parcel location, and names of adjoining owners.

The GPS point theme table gives information about the GPS point, such as Northings and Eastings, the properties to which points are anchored, rover file name, and a short description of location.

The table for the 'neighbors' theme has fields for the GPS locations defining each end of the line,a line theme, the names of the owners sharing the line, and the status of the line, being either "DISPUTED" or "UNDISPUTED." Please refer to the ArcView window image.

9. Problems Encountered / Possible Improvements

There was limited memory in the GPS unit; we were unable to log data for periods greater than approximately one hour.

Only after the data had been collected was it found that additional information was needed to adequately perform the office implementation. Firstly, in order to streamline the process of polygon creation using the GPS point positions, the first level sketches should also be annotated to contain the rover-file names. This was not a problem in this particular exercise, however for a large number of parcels in a high density area, this would be useful. Secondly, the first-level sketches needed to have a little more information about the actual boundary line. It is not entirely clear, in some instances, where the boundary line is located. For example, the road was used as the boundary line, however the GPS location on the rock, is 3 paces from the road edge. Whether this is significant or not depends on the user and the required precision. the additional physical evidence gathered at this location was the granite rock 15 paces from the original. This was intended as a 'pointer' marker as described in the second level sketch.

10. General Conclusions

The proposed representation of parcels would be neither sufficient nor equitable in cases where land values are at a premium. However, in cases where rural, low-valued lands are being initially titled and the primary administrative interest is simply to add them to the register, the information provided by this method will be adequate.

The GPS/sketch book adjudication method described in this paper makes significant contributions to the solution of contextual problems surrounding land titling efforts. Its demonstrated flexibility in recording parcel data, landowner testimony, and the nuances of divided interests in land is a marked improvement over the rigidities of many contemporary adjudication, demarcation, and survey procedures which fail to accommodate customary understandings of land rights. The emphasis this method places on community participation helps lead toward a more decentralized, transparent land registration system that is more responsive to local needs and practices. The organization it requires on the part of communities and landowners also promotes education about land rights and titling benefits, and welcomes expanded input in the development of policy. This method also greatly reduces cost. In circumstances where GPS/sketch book adjudication is employed, surveys can be accomplished with less advance planning and less time spent in the field. The instrumentation is less expensive than the total stations and high accuracy GPS equipment deployed in most conventional surveys and it is also easier to operate so the need for professional, highly-compensated surveyors is reduced. Perhaps most importantly, this method allows for the concentration of what are typically several separate procedures into a single, more efficient operation. While most current titling efforts divide adjudication, boundary demarcation, surveying, and recording operations, this method allows for their simultaneous accomplishment. Problems such as boundary movement, which often emerge due to a time lag between completion of one operation and the initialization of another, are more easily avoided in the case of GPS/sketch book adjudication. And finally, while the development of a bare-bones cadastre that the GPS/sketch book adjudication method implies might be suspicious in that it would seem to require later reconstruction, this method provides base data that can easily be combined with and enhanced by any higher accuracy information that may follow it in time.

11. Future Work

A refinement of the sketching techniques used is necessary. Though it is envisioned to allow as much flexibility as possible, the method will require a set of adjudication rules, such as the amount of data recorded in order to define parcel boundaries.

An investigation could be conducted for the integration of digital images of aerial photographs as base-maps within the electronic sketch-book. These uncorrected photo-images would be less expensive since there would be no photogrammetric post processing, and would also serve as a good ground-reference system. The photography could occur after the adjudication process, and where possible monument positions could be pre-marked during adjudication. This might be the next step towards a more precise system.

The extension of the same semi-qualitative system into a more accurate quantitative `numerical cadastre' is another topic which could be explored in greater detail.

12. References

Barnes, Grenville. 1994. LIS in Latin America. FIG vol.3.

Downs, Roger M; Stea, David. 1977. Maps in Minds, Reflections on Cognitive Mapping.

Egenhofer, Max J; Mark, David M. 1995. Naive Geography (Proceedings of COSIT95, eds M Egenhofer, A Frank, Springer Verlag)

Gahegan, Mark. 1995. Proximity Operators for Qualitative Spatial Reasoning (Proceedings of COSIT95, eds M Egenhofer, A Frank, Springer Verlag)

Holstein, Lynn. 1996. Towards Best Practice from World Bank Experience in Land Titling and Registration (International Conference on Land Tenure and Administration - 1996, Orlando, Florida)

Jackson, Johnathan. 1993. Maps and the Poor. Surveying World, Vol3 (2).

Jones, Andrew C; Abidin, Hasanuddin Z. 1996. An Investigation into the Potential Application of Low Cost GPS Receivers to Cadastral Surveing in Indonesia (International Conference on Land Tenure and Administration - 1996, Orlando, Florida)

McKenna, John R. 1996. Modernization of Cadastres and Land Registries in Latin America: The Experiences of Paraguay, Brazil and) Nicaragua (International Conference on Land Tenure and Administration - 1996, Orlando, Florida)

Schmidt, Joel J. 1995. Evaluation of Hand-Held GPS for Registration of Scanned Cadastral Maps (GIS/LIS Proceedings - 1995)

Smith, Barry. 1995. On Drawing Lines on a Map (Proceedings of COSIT95, eds M Egenhofer, A Frank, Springer Verlag)

Gartrell Group Home Page