«ACCURACY DETERMINATION FOR THE AUSTRIAN DIGITAL CADASTRAL MAP (DKM) Gerhard Navratil, Jeannine Hafner, Dmitri Jilin Vienna University of Technology, ...»
In order to provide a full picture of the DKM accuracy samples should be taken from all over Austria. This was not possible due to unavailable resources. We thus selected sample areas that were easily accessible and still provide insights. We decided that 176 Četvrti hrvatski kongres o katastru/Fourth Croatian Congress on Cadastre mountainous areas may contain the largest deviations but there may not be many plats since boundary changes are rare. The most frequent changes occur in rural areas due to reallocation processes and at the boundaries of cites when new parcels for building are
created. These two areas are different in two ways:
• Parcels in rural areas are typically much larger than in urban areas. Thus the dimensions are larger and deviations should be, too.
• Land in rural areas is typically cheaper than in urban areas because there is abundant land available. As an effect the survey of rural land should not cost too much and it may not be essential to describe the boundary as detailed as in urban areas.
Especially the second point conflicts with the original requirements for the Cadastre.
Urban land was not a tax object and thus the quality of cadastral maps in cities and villages was worse than the quality in agricultural areas. Thus plats in urban areas typically improved the quality of the may more than plats in rural areas did.
We selected a part of Vienna as an example for an urban area and an example from Carinthia for a rural area. In both areas we collected available cadastral surveys and checked that the boundaries in the DKM still show the contents of these plats. We then compared the dimensions shown in the survey with those computed from the coordinates. All plats were drawn originally in a scale of 1:2.880 and thus the results should be comparable.
5. TEST RESULTS
We could collect at total of approximately 250 pairs of distances. More than 200 pairs were collected in the Vienna area. The distances range from 5 m to 94 m. Figure 3 shows the distribution of the analyzed distances. It is evident that most of the distances are between 15 m and 35 m. These distances are typical for a parceling. Figure 4 shows the deviations of these values from the measures computed from the DKM. 176 of the 202 measures (87%) have deviations of less than 1 m, 149 measures (74%) have deviations of less that 50 cm and 88 measures (44%) have deviations of less than 20 cm. The average deviation is -7 cm with a standard deviation of 84 cm.
Četvrti hrvatski kongres o katastru/Fourth Croatian Congress on Cadastre 177 Length of measure [m]
Three deviations are larger than 3 m and may be gross errors. Gross errors may emerge
in various cases:
• We may have missed changes in the boundary lines. Then the plat would not show the same boundary as the DKM and the distances are incompatible.
• There may have been an error while incorporating the plat in the cadastral maps.
Any error in that process would go unnoticed until another surveyor compares the cadastral map with the plat. This is usually only done when changing parcel boundaries. Since these cases have been eliminated from our sample, there may be undetected errors.
• There may have been undocumented changes in the cadastral map.
The elimination of deviations larger than 3 m changes (3 values) the standard deviation to 40 cm.
39 pairs were collected in Carinthia. The distances range from 2 m to 84 m. Figure 5 shows the distribution of the distances. Half the distances are below 20 m but there in general the distribution of distances is more equal than in the case of Vienna. Figure 6 shows the deviations of these values from the measures computed from the DKM. 22 of the 39 measures (56%) have deviations of less than 1 m, 14 measures (36%) have deviations of less that 50 cm and 3 measures (8%) have deviations of less than 20 cm.
The average deviation is 50 cm with a standard deviation of 213 cm.
Četvrti hrvatski kongres o katastru/Fourth Croatian Congress on Cadastre 179 In this data set there is also a value that seems to be a gross error. The deviation is almost 6 m, which is 2.5 times the second largest deviation (2.3 m). Elimination of this value leads to an average of 35 cm and a standard deviation of 138 cm.
Deviation of the observations [m]
Some of the differences could originate from the different distribution of measurements. Thus we determined the same values for the distances up to 30 m. This gives standard deviations of 40 cm for Vienna and 99 cm for Carinthia. Thus even small distances are more distributed in Vienna than in Carinthia.
6. DISCUSSION OF THE RESULTS AND FUTURE WORK
The results show that the cadastral map in urban areas is more accurate than in rural areas. Even limiting the maximum distances did not change this picture. A difference between rural and urban areas could be that urban areas have a better distribution of plats. Thus there is a higher probability that recent surveys were performed in the vicinity of the measurement used in the sample. This could improve the overall quality of the cadastral map.
The analysis shows that areas surveyed in the first half of the 20th century may still have serious quality problems. Deviations of more than 1 m are not unlikely since 13% of the data in the urban sample and 45% of the sample in the rural sample exceed this value. Thus it is important to communicate to lay people that the DKM is not a very reliable source for distances even if signatures indicate that there have been surveys in the 20th century.
More work is necessary in two directions:
180 Četvrti hrvatski kongres o katastru/Fourth Croatian Congress on Cadastre
• The sample should be expanded to cover more of Austria to get better estimates.
The sample should then also cover other scales than just 1:2.880.
• Similar investigations are necessary for areas that have not yet been surveyed.
The problem with this endeavor is that the original are not documented. It is thus only possible to compare the derived result of the original measurements (the original cadastral map) with the DKM.
The goal should be finding simple quality parameters that can be communicated to lay people. This is necessary to avoid misinterpretation of the DKM, which leads to disputes and results in lawsuits. The large group in the target audience is lawyers and judges because they must advise their clients and interpret DKM data. It could be beneficial for them to have studies to rely upon.
 Austrian National Assembly (1968). (1968). Bundesgesetz vom 3. Juli 1968 über die Landesvermessung und den Grenzkataster (Vermessungsgesetz - VermG).
BGBl. Nr. 306/1968.
 Franz Josef I. (1869). Regelung der Grundsteuer. RGBl. 88.
 Franz Josef I. (1883). Gesetz über die Evidenthaltung des Grundsteuerkatasters (Law on updating the Cadastre). RGBl.Nr 1883/83: 249-268.
 Kugler, R. (1994). Der Aufbau aktueller und konsistenter Datenbanken des Katasters. AGIT'94, Salzburg, Austria, Institut für Geographie der Universität Salzburg.
 Lego, K. (1968). Geschichte des österreichischen Grundkatasters. Vienna, Bundesamt f. Eich- und Vermessungswesen.
 Twaroch, C. and G. Muggenhuber (1997). Evolution of Land Registration and Cadastre - Case Study: Austria. Joint European Conference on Geographical Information.
Četvrti hrvatski kongres o katastru/Fourth Croatian Congress on Cadastre 181