Molnár Gábor – Timár Gábor – Biszak Előd: Can the First Military Survey maps of the Habsburg Empire (1763-1790) be georeferenced by an accuracy of 200 meters? Conference: 9th International Workshop on Digital Approaches to Cartographic Heritage, Budapest, 2014. Volume: 9. 127–132.
9th International Workshop on Digital Approaches to Cartographic Heritage Budapest, 4-5 September 2014 tion center) crosses the Stephansdom tower in Vienna (with a few hundred meters of reliability), which may indicate that the survey has some cartographic background anyway. Figure 1. Correction vectors at the ground control points in Lower Austria, used for the correction grid compilation: The length of the longest vector is 80 arc seconds (cca. 2400 meters). In the second step, we use the GCP set again, to define a geodetic datum instead of the WGS84 used in the first step to reduce the horizontal errors. A five arc minute correction grid was compiled by simple bilinear interpolation. The latitude and longitude shifts at the GCPs were computed from their WGS84 coordinates (as starting points of the vectors) and the geographic coordinates computed from the mosaic pixel coordinates and the above listed parameters of the Cassini grid and its location (as endpoints of the vectors). The correction vectors for Lower Austria are shown in Fig. 1. It is obvious that these corrected error is not systematic, they cannot be modelled by simple mathematic methods available in GIS packages. Therefore the correction grid is the best and so far the only available method to handle these errors. Results The result for each province is (1) the coordinate system of the mosaic, characterized by the Cassini projection with its estimated parameters and the datum defined by the correction grid (GSB) and (2) the province image mosaic fit to (resampled into) this coordinate system. The horizontal control of the geo-referred mosaic is considerably better than it is without the correction grid. With the five arc minute grid, the error can be decreased to 100-200 meters (Fig. 2). However, the method has an option to increase the accuracy not in the all mapped region (by defining more GCPs and compiling a denser grid) but also in smaller parts of it (e.g. for the city of Vienna) by defining a high-resolution local grid for the more important parts of the area. The
