ESTABLISHING A DIGITAL CADASTRE IN ISRAEL

Kiril Fradkin and Yerahmiel Doytsher
Department of Civil Engineering (Geodesy)
Technion – Israel Institute of Technology
Haifa 32000, Israel

The paper describes a new method for generating a spatially accurate, legally supportive and operationally efficient cadastral database of the urban cadastral reality. The definition and compilation of an accurate cadastral database (achieving a standard deviation of less than 0.1 meter) is based on a new analytical reconstruction of cadastral boundaries rather than on the conventional field reconstruction process. The new method is based on GPS control points and traverse networks for providing the framework; the old field books for defining the links between the various original ground features; and a geometrical and cadastral adjustment process as a conceptual basis. A pilot project that was carried out in order to examine and evaluate the new method is described.

L’article decrie une nouvelle methode pour genere une base des donnes cadastrale, precise, legale et efficiente. La definition et la compilation d’une base de donnes cadastrales (en obtenant une erreur de moins de 10 cm.) sont basees sur une reconstruction analytique des frontieres cadastrales plus que la reconstitution des frontieres sur le terrain. La nouvelle methode est fondee sur des points de control mesures par GPS et des reseaux polygonaux et les liaisons entre differents elements de terrain qui sont definies dans les carnets de terrain. La conception est basee sur adjustment geometrique et cadastrale. Un project experimental a ete realise pour verifier et evaluer la nouvelle methode.

Dieser Artikel beschreibt eine neue Methode fur die Herstellung einer raumlich genauen, juristisch annehmbaren, und praktisch funktioniernden Kataster Databasis von der staetischen Kataster Wirklichkeit. Die Definierung und Zusammenstellung von einer genauen Kataster Databasis (bei Ereichung einer Standard Deviation von weniger als 0.1 meter) basiert sich mehr auf einen neuen analytischen Wiederaufbau von katastralen Grenzen, als auf den herkoemlichen Wiederaufbau Prozess durch Vermessungen. Die neue Methode basiert sich auf GPS Kontrol-Punkte, und ein polygones Arbeitsnetz fuer die Darstellung des Rahmen; und die alten Fachbuecher fuer die Definierung der Zusamenhaenge zwischen den verschiedenen urspruenglichen Grundelementen; und ein geometrischer und Katastraler Adjustierungsprozess als eine conceptuale Basis. Ein Pilot Project, das ausgefuehrt wurde, um die neue Methode zu pruefen, wird beschrieben.

Cadastral mapping started in Israel in 1926, based on the Torrens method - Registration of Titles (Dale, 1976), the most advanced and innovative system at the time. This system defines the cadastral blocks and parcels based on official surveying and mapping that is carried out (in Israel) by the state and is linked to the national coordinate network. Measurement results are recorded in field books and used to determine the boundaries of the block and the parcels, as well as other features (buildings, fences, electric poles, etc.) on a field plan sheet. Maps of the blocks are prepared based on the field sheet, consisting of all parcels in the block and all included features. These maps contain neither the measured data nor any dimensions whatsoever of the parcel boundaries. The current cadastre in Israel can therefore be defined as a measurement base cadastre of a graphical nature.

Most measurements for cadastral mapping were performed by using the chain surveying method. This method necessitates prior marking and measurement of control points in the field (a traverse network), and linking them to the national control network (the trig points). Until the 1980s, the cadastral blocks in Israel were based on a control network of very low accuracy. The calculation of the control network (traverses) was based on separate adjustment of each traverse and not on a rigorous adjustment as a uniform network. The adjustment method used was the Bowditch method (first handling the angular misclosure, then following with the linear misclosures) (Berthon, 1972). Most blocks were plotted manually based on chain surveying measurement, without calculating coordinates

The quality of the cadastral maps improved with the development of modern measuring instruments and introduction of the polar method in cadastral measurements (theodolite and electrooptical distance measuring). Concurrently the new block maps were being drawn by plotters based on calculated coordinates (Adler, 1974). Another improvement in quality and accuracy of cadastral maps took place in the course of the 1980s and 1990s following the increased accuracy of the Israeli control network.

Accurate, accessible and updated cadastral information constitutes the basis for planning and implementation of a variety of real estate related operations in many areas. In the present form of the graphic cadastre in Israel, the existing cadastral information does not fulfill these needs. Most information is currently kept on paper (field books, field sheets), which does not permit its computerized management. Inherent contradictions and inaccuracies in the existing material cause difficulties in its use and delays in tracking and measuring the changes, and in updating the cadastral map. It is important to note that according to the current surveying regulations in Israel, restoring of boundaries must be performed according to the original measurements from the field books and not according to calculated coordinates. The final decision regarding the location of the boundary is made on the basis of adjustment between the measurements and the actual situation in the field.

The solution for the problems of the current graphic cadastre is to establish an analytical cadastre in which the location of each entity is determined by the state plane coordinate system, coordinates that are sufficiently accurate for legal validation. The data of the analytical cadastre will thus constitute a spatial information system (LIS) defining the statutory land division. Obtaining the digital mapping data is the bottleneck in the process of establishing the analytical cadastre. Several basic sources exist for obtaining such data: field measurements of land boundaries; digitizing existing maps; and processing the existing surveying data.

Resurveying all land boundaries could constitute a radical solution for the problem. To this end it would be necessary to restore the boundaries according to the existing information in the field books and sheets, and measure these in relation to a new control network. Restoring the boundaries in itself constitutes a tremendous effort, whose estimated cost is extremely high. The cost of measuring the restored boundaries further on in the process is estimated to be also very high.

A possible alternative for restoring the boundaries could be the surveying of the existing boundaries. This solution would in practice cancel and expunge the current statutory validity of the land boundaries and would in fact institute a new real estate order. This solution is impractical in countries with a legal real estate order, and moreover, it has no economic justification (Perelmuter, 1993).

Another solution would use the original measurements and would be based on recalculating the field books. Calculating the books as a stand alone process for establishing an analytical cadastre is unsuitable due to several reasons (Perelmuter, 1993):

In addition to all these factors, from the economic standpoint, a conservative estimation of the work involved in calculating more than 20,000 existing measurement field book amounts to hundreds of man-years.

Converting graphic data on existing maps to digital data would of course be the least expensive and the fastest procedure of obtaining such data. There are about 15,000 cadastral block maps in Israel, prepared in different periods, employing different methods of measurement, calculation and drawing, as well as on different scales. Simply digitizing the existing maps will not contribute to obtaining a database that would be adequate for the analytical cadastre, both from the accuracy aspect and from the judicial validity aspect. This is due to the heterogeneity in the quality of the maps, and the accuracy limitations inherent in the digitization process itself.

As a result of the need to improve the quality of the data digitized from the cadastral maps, in recent years the option of integrating external information with digital data derived from maps was being considered. Reconstruction of geometric and cadastral conditions is generally performed on each map using adjustment techniques (Doytsher, 1980; Morgenstern, 1989; Perelmuter, 1993). Two types of external information can be distinguished: "cadastral" information and "geometrical" information. Cadastral information is the registered area of a parcel, the length of a front, right of way, etc. The geometric conditions define the shape of the entities described in the map (closed polygons, straight lines, circular arcs, parabolic curves, etc.), as well as the existing spatial relationships between the various entities (parallelity of lines, perpendicularity of lines, etc.).

In order to examine the degree of improvement in the accuracy of coordinates obtained from the cadastral maps by employing external geometric and cadastral conditions, experiments were carried out on several representative maps. To this end, maps were selected from urban areas, which are the critical areas from the standpoint of accuracy of determining boundary points and which are also important from the economic, engineering and planning aspects. The selected cadastral maps have the following characteristics:

The cadastral maps were scanned (at a resolution of 400dpi) and the coordinates of the boundary points of the cadastral parcels were digitized manually from the scanned maps. The accuracy of the coordinates of the digitized points was estimated at 0.04 m. (about 0.5 pixel at the scanning resolution).

The common method for transforming coordinates picked up from scanned maps into the state plane coordinate grid, is affine transformation. Affine transformation is based on a minimum of three control points, when in practice more than the required minimum is used to avoid gross errors, as well as to enable statistical analysis of the transformation quality. Cadastral maps in Israel generally contain many tens of control points (usually traverse points). To examine the relation between the number of the transformation points and the precision of the transformed boundary cadastral points (not the residuals of the points used for the transformation), several transformations of the same maps were performed. The transformations were based on more points than the minimum requirement and were performed each time with a different equally distributed number. Since the measurements in the field books served both for plotting the maps and for calculating the coordinates, smaller differences between the calculated and transformed coordinates are supposed to indicate better and more accurate transformation. Values reflecting the results of these comparisons are presented in Table 1. It can be seen that there are no significant differences between the transformation results (from the coordinate differences aspect) even though the parameters of the various transformations differ considerably from one another (translation, for example). The explanation for this is based on the local nature of these differences. The differences do not indicate any significant trend whatsoever in their direction or size. The reason for the local nature of the differences becomes clear when they are plotted together with the appropriate measurement lines (see Figure 1). An obvious correlation exists between the differences of points that are linked to a measurement line, and not between different measurement lines. It is therefore recommended to prefer to use a limited number of control points as long as they are equally distributed on the cadastral map.

Figure 1. Differences between calculated coordinates and coordinates digitized from the map.

On the background of the parcels (right); on the background of the measurement lines (left).

MSE X	Mean X	MSE Y	Mean Y	Control points
0.20	-0.04	0.18	0.02	3
0.22	-0.02	0.19	-0.04	9
0.21	0.01	0.19	0.02	20
0.21	0.02	0.18	-0.02	31

Table 1. Transformation accuracy as a function of the number of control points (in meters)

The process of digitizing the contents of a map point by point cannot take into consideration the fact that these points describe geometric entities. Ignoring the existing spatial relationships in the map during the digitizing and transformation process leads to mutual displacements of the points. The irregular nature of these changes combined with the drawing inaccuracies of the map itself, leads to distortions of the geometric shapes of the entities that were picked up, compared with the real entities. Correction of these distortions may both reconstruct the correct geometric shapes of the entities, as well as improve the accuracy of the coordinates of the points that define them. Reconstruction of the geometric conditions was performed by adjustment, forcing the coordinates of points whenever it is needed, to be placed on straight lines. It turned out that the change in coordinates as a result of the adjustment did not exceed a few centimeters (less than 0.05 meter), thus not significantly affecting the differences between the calculated and transformed coordinates

Another attempt to improve the transformation results was made by integrating the cadastral constraints - lengths of fronts of the parcels that appear on the field sheets. The results of this procedure (MSE of 0.18 m. and 0.20 m. in the X and Y directions) indicate that the use of cadastral (and geometric) conditions did not reduce the existing differences between the transformed and calculated coordinates. This can be explained by the fact that these conditions are insufficient to unequivocally define the mutual location of the points.

In another attempt to improve the accuracy of the transformed coordinates, the use of the registered areas of the parcels as additional cadastral conditions in the adjustment process was considered. For the purposes of the examination, the differences were calculated between the registered area and the area calculated from the transformed coordinates. No correlation between the two type of differences was found, in other words, large differences in coordinates are not necessarily accompanied by large differences in the areas, and vice versa. Based on these findings and considering the relatively low accuracy of the registered areas, it was decided not to use the areas as conditions for adjusting coordinates. This finding of the relatively limited accuracy of the registered areas of cadastral parcels is also reflected in the current surveying regulations in Israel, which permit considerable tolerances between the measured area and the registered area (for example, the permitted tolerance for a parcel of 1000 m² is about 3%).

The alternative to the graphic digitizing of the cadastral maps is calculation of the field books. Since the measurements recorded in the field books have legal validity, it is possible at this stage to ignore their inherent accuracy and concentrate on other factors that affect the misfit between coordinates calculated from the field books and the reality in the field. One of these factors is the accuracy of the control points, particularly the method of calculating the traverse points on which the measurement lines are based. As noted, most of the traverses in Israel were formerly calculated by relying on the separate adjustment of each polygon and not on rigorous adjustment as a uniform network. In order to estimate the differences in location of the traverse points, the original measurements of the traverses in the examined areas were readjusted as a single uniform network by a least squares adjustment. The new coordinates obtained from the adjustment were compared with current coordinates in the database of the Survey of Israel (the Israeli Mapping Center). The comparison revealed significant differences (up to 0.20 m.) in the location of some of the points. The differences are not uniform and depend primarily on the geometric shapes of the original traverses (larger differences in zones of meandering/zigzagging shapes of the traverses). Thus calculating the field books cannot by itself ensure correct coordinate values for the cadastral boundary points (in relation to their true location in the field).

The use of geometric and cadastral conditions and the various adjustment techniques presented in the previous chapter is intended to bring the maximum number of points as close as possible to their "correct" location. The accuracy of the results depends primarily on the strength of the geometric shape (configuration) and the accuracy of the external information (Perelmuter, 1993). The results that were obtained produce a more accurate "graphic picture" than that of the maps. So far, this does not permit relating to the combined product as information that can constitute a proper legal and statutory alternative for the traditional cadastre (that is subjected to the conventional procedures and regulations). Thus, in conclusion, a processing procedure only of the existing data cannot ensure correspondence between the calculated coordinates and the cadastral reality. A way is needed to integrate a limited number of new measurements in the process of obtaining reliable coordinates for the turning points of the cadastral boundaries.

A substantial difference exists between calculating the coordinates of the boundary points from field books and between restoring them in the field by traditional methods. The conventional restoring process is local by its nature, relying only on findings in the near vicinity in the neighboring parcels of the restored boundary and the original measurements. The new proposed method, based on an analytical restoring process, which analytically imitates the traditional process, will rely on the precise location of several existing features in the field, features that are linked to the old measurements. The only way to obtain the location of these features is to identify them in the field and to measure them in relation to a new GPS based control network (a network free of the constraints, inaccuracies and errors inherent in old control networks). The existing coordinates of the control points are not to be used for calculations, but rather new coordinates in the new network are to be determined for a limited number of existing features that were located. The new method is thus complying with the following requirements:

Buildings were chosen to create the "link" between the current measurements and the historical measurements (between the new and the previous networks). The decision was based on the following reasons:

The simple geometric link between the building corners and the parcel boundary is shown schematically in Figure 2. As seen in the figure, a new measurement of two corners of a building, together with the previous measurements easily permits the calculation of the parcel corner coordinates.

The buildings (as secondary features that are not connected to the land boundaries) were not measured directly from the measurement lines, but were linked to the parcel boundaries. The level and accuracy of the building measurements is poor in comparison with the level and accuracy of measurement of the boundary points. As a result, differences are obtained between the coordinates of the same boundary points common to two adjacent parcels that were calculated separately according to the measurements of two neighboring buildings. Since each of the two adjacent buildings permits use of independent measurements for restoring the boundary points between them, it is possible to calculate common boundary points for the various parcels using adjustment procedures. Further strengthening of the shape and location of the boundaries is achieved by applying geometric and cadastral conditions.

The simulation is intended to estimate the accuracy attainable by employing the proposed method, and particularly to examine the effect of the new measurement configuration on the results. Simulated measurements are used to facilitate easy analysis of the examined results by comparing the adjusted coordinates and the "real" location of the cadastral boundary points. The discrepancies between the coordinate values obtained in the adjustment process and the original coordinates that formed the basis for the simulation, are a direct indication of the accuracy of the proposed system. In a conventional surveying process based on physical reconstruction of the boundary points in the field, the accuracy of the coordinate values obtained depends on the accuracy of the field measurements. As a result, the comparison between coordinates obtained by the new method and the coordinates obtained by measurement based on reconstruction in the field, is affected by errors and inaccuracies of the traditional reconstruction method, thus preventing the attainment of meaningful evaluations.

For purposes of the simulation, the coordinates of the boundary points of city block parcels and the buildings on each parcel of the block, were defined geometrically and calculated. The distance and length measurements were calculated in respect to these data (as a simulation of chain surveying – the conventional cadastral surveying procedures), as well as the geometric conditions defining the mutual location of buildings and parcels. In order to imitate the real surveying procedure, errors e were randomly introduced into these measurements according to the normal distribution with variance s (e ~ N(0, s ²)), where the magnitude of s is chosen in accordance with the surveying regulations and is to reflect the actual surveying accuracy. Assigning appropriate weighting to the geometric and cadastral conditions (as exist in the field) attained additional correspondence between the simulated measurements and the real surveying procedure. The simulation measurements were carried out on different sets of e and weighting, in order to estimate the effect of measurement accuracy on the accuracy of the results. The location accuracy as a function of distance accuracy can be seen in Table 2. The location accuracy is presented in the table by the average size of the halves of error ellipse axes (a and b). The values of "a" are in the range of the MSE of the distance measurements and "b" values are a little smaller.

Distance MSE	Mean a	Mean b
0.05	0.06	0.04
0.10	0.09	0.06

Table 2. Simulation accuracy (meters) as a function of distance accuracy (meters)

Besides estimating accuracy of coordinates, the simulation calculations were intended to clarify the extent of the effect of the geometric configuration of the buildings (number and distribution) on the accuracy of the adjusted coordinates. To this end, runs were performed with different numbers of buildings within the block, starting with a "full" configuration (a building on every parcel) and ending with the extreme case - only one building on the block. Since each building contributes two control points - the significance of this is the number and distribution of the control points on which the adjustment process of the city block is based. According to the experimental results presented in Table 3, it is easy to conclude that there is no significant difference in the accuracy of the adjusted coordinates between the calculation in which all the buildings are given and the case in which "control points" are located in the block corners only. On the other hand, as could be expected, accuracy falls sharply when using only one building in one of the block corners, and adjustment under these conditions is performed by full extrapolation. The explanation for the similar results under conditions of full interpolation (whether by employing all buildings or by employing four city block corner buildings only, stems from the rigid geometry created by the geometric conditions. In a similar manner, it is possible to see according to the data in Table 4, that the accuracy values of the calculated coordinates in the adjustment process behave in a manner similar to what was described above (in relation to the data in Table 3).

No. of buildings	Mean a	MSE a	Mean b	MSE b
All	0.07	0.02	0.05	0.01
Four corners	0.08	0.02	0.06	0.01
Two	0.10	0.02	0.08	0.01
One	0.14	0.04	0.10	0.03

The quality of the solution can be estimated according to the error ellipse of the point locations (see Figure 3). The maximum size of the large half axis of the error ellipses was obtained as 0.08m. This numerical value indicates that in the given configuration and the accuracy determined for the measurements, it is possible to obtain by the proposed method an accuracy level that complies with the surveying regulations in Israel.

No. of buildings	Mean Y	MSE Y	Mean X	MSE X
All	-0.01	0.04	-0.01	0.04
Four corners	0.00	0.06	0.02	0.04
Two	0.00	0.04	0.03	0.06
One	0.03	0.04	-0.03	0.04

Table 4. Effect of the configuration on the accuracy of the parcel corner coordinates

For a practical test of the method, typical city blocks were chosen (see Figure 4), most of whose buildings have been measured. Since more than 20 years have passed since the cadastral mapping, a field survey was made in order to identify the buildings and ascertain that no changes have taken place in their shape. Comparing new direct distance measurements with their recorded counterpart in the field books enabled this identification. For the purpose of measuring features in the field, a traverse network was established, based on GPS points. The corners of the buildings were measured from the traverse points by the polar method and their coordinates were calculated. These building corners are directly linked to the parcel boundaries (in terms of measurements in the old field books), thus enabling to calculate the coordinates of the boundary points. In order to take into account the redundancy of measurements and to integrate geometric and cadastral conditions, an adjustment procedure was implemented. The unified approach was to relate to all data (measurement and coordinate) as measurements of different weight, which permits taking into account the varying accuracies of the different types of measurement. Accuracy estimates for the adjusted coordinates (MSE of less than 0.10 m.) fulfill the requirements of the Israeli surveying regulations.

To obtain a basis for examining the quality of the new method and its results, some of the boundary points in the experiment area were restored by the traditional method. Restored for this purpose, according to original descriptions, were the old traverse points on which the previous measurement lines were based. The measurements recorded in the old field books were set out from the restored measurement lines. The results were compared and verified in the course of measurement by the existing reality in the field. In some cases, considerable discrepancies were discovered between the location of features in the field and their new marked location. In such cases, the final decision was made after using additional information (building plans) and the judgment of the surveyors making the measurements. In some cases it was impossible to restore the boundary points by the traditional method due to various obstacles. Among these obstacles are missing descriptions of some points; descriptions of points which could not be used due to changes that have taken place in the area; invisibility due to new buildings or growth, etc.

The restored boundary points were measured in relation to the same traverse and their coordinates were compared with the coordinates obtained from adjustment by the proposed new method. In addition to the new measurement of the restored boundary points, the coordinates of the original traverses were also recalculated. Comparison of the new and existing coordinates of the traverses revealed not inconsiderable differences (larger than 0.20 m.). As a result of these differences, differences were also obtained between the coordinates of boundary points calculated from the field books and the restored coordinates of these points. These facts of course intensify the conclusion of the essential need to integrate new field measurements as the basis for the analytical reconstruction of the boundary points. Based on the results obtained, it is possible to conclude that:

The article presents several possibilities for establishing an analytic cadastre, as well as the results of the various cadastral examinations made. It was demonstrated and presented that calculation of the coordinates of cadastral boundary points cannot rely only on the original measurements and on the coordinates of the current control network. It necessitates the integration of new measurements in the calculation process, in order to ensure the desired accuracy level and to maintain the judicial admissibility of the calculated coordinates. The proposed method permits the optimal integration of a limited amount of new measurements with the existing measurements and constitutes a possible solution for the establishment of an analytical cadastre in built-up areas.

The research upon which this paper is based was supported in part by the Israeli Survey Department under project 018-008. The authors wish to thank the Survey Department for their assistance.

Adler R., ,Shmutter B., Etrog U., Papo H., Stoch L., Sharni D., 1974. Optimization of Methods and Procedures in Cadastral Surveys. In Proceedings of the XIV FIG Congress (Commission 5), Washington D. C.

Berthon Jones P., 1972. A Comparison of the Precision of Traverses Adjusted by Bowditch Rule and by Least Squares. Survey Review, 164:253-273.

Doytsher Y., 1980. Numerical Processing of Graphical Cadastral Information. In Proceedings of the 16th FIG Congress, paper 304.5.

Doytsher Y., Shmutter B.,1991. Incorporating Known External Data in Information Digitized from Cadastral Maps. In Proceedings of ACSM/ASPRS Convention, pp. 61-68.

Morgenstern D., Prell K.M., Riemer H.G., 1989. Digitization and Geometrical Improvement of Inhomogeneous Cadastral Maps. Survey Review, 30(234): 149-159.

Perelmuter A., Steinberg G., 1993. LIS and Cadastre in Israel, In Proceedings of the Israeli GIS/LIS’92 Conference, pp. 73-78.

ESTABLISHING A DIGITAL CADASTRE IN ISRAEL

ETABLIER UN CADASTRE NUMERIQUE EN ISRAEL

EINRICHTUNG EINES DIGITALEN KATASTER IN ISRAEL