In February 2015, the Czech Research Institute of Geodesy, Topography and Cartography (VÚGTK) and Upvision began a project to investigate the possible integration of innovative elements such as UAVs into the mapping process in the cadastre of the Czech Republic. Between 2015 and 2017, mapping was carried out in six locations, and the results compared to the various existing processes, in combination with geodetic methods, from the point of view of the accuracy achievable, as well as the effectiveness of the deployment of these technologies in cadastral mapping.
For this research, several types of UAV were used, most notably the MaVinci Sirius unmanned plane, which is ideal for detailed mapping of larger areas, as well as the DJI Matrice 600 Pro and Mikrocopter Hexa.
The first step was the creation and design of a geodetically measured and signalled base to determine the internal parameters of the non-measurable cameras in UAVs used for cadastral imaging. Several calibration bases have been established in the Czech Republic for UAVs, but none of them are optimal in all respects and needs. Following the testing of several bases, a new scheme for building a public base has been proposed and its location is now under discussion. The newly designed base will allows images to be captured with a ground sample distance (GSD) range of 1cm to 12cm. Base dimensions will be 400m x 400m, with the location of the default and control points designed to test cameras with different chip sizes and focal lengths.
Coordinates of the starting and control points will be measured at heights of very precise nivelation. Determination of the position of the signalled points will be done both by terrestrial determination in relation to the state point field (S-JTSK and dual measuring by GNSS technology (after two hours interval) with 15 minute observation.
Further experimental work in favour of mapping using UAVs has been carried out in the area of identification of signalled GCP, control and detailed measurement points. Large sets of temporary stabilisation points (land ownership markings) of different sizes and colours were shot from different heights. After performing the computational experiments, all signalling points were placed on flat terrain for clear and correct interpretation.
Another part of the experimental work was the study of the distribution of starting points in the localities with respect to the required precision of detailed cadastral mapping points. The layout and location of the ground control points (GCPs), control point (CP) and detailed points in the location must be planned with respect to the boundaries of the territory that is intended for cadastral mapping.
The position of these points is performed in the field so that these points are observable and measurable on at least nine shots (85% longitudinal overlap and 70% lateral overlap) and that the GCP points are evenly spaced at the mapping location and also at the boundary of the mapped location. Each GCP at the boundary of the location must be visible on the other four rows beyond the location boundary and each GCP at the end of the series must be on six more consecutive images. The total number of GCPs on the mapped location must be in proportion to the number of images taken from the UAV, with one GCP on 80-110 images. The number of control points should be at least two-thirds of the number of GCPs.
Using this procedure, the points are decomposed on a regular basis within the location, and the whole GCP system then exhibits higher stiffness, meaning there is a lower coordinate error in measurement and photogrammetric correlation calculations. The accuracy values in the GCP position and height and the control points entering the photogrammetric calculations shall have a relative positioning accuracy at the value of the coordinate standard deviation in Czech republic σx, y = 15mm.
Experimental cadastral locations
For testing, various experimental locations in the Pilsen Region of the Czech Republic were selected. For example, Tymakov was mapped from a UAV with a GSD of 2cm, longitudinal overlap 80% and transverse overlap 70%. The location itself was divided into three frame blocks, which in total represented 176ha. Mapping and orientation parameters were selected with respect to the scanning target, ie the maximum number of points generated on the surface of the object facades – the mapping rule of the Czech Republic’s Land Registry is that the location of the building is determined by its intersection with the terrain in the cadastre.
As a further experimental site to test UAV mapping in a poorly accessible area, a cadastre called Bohy was used. This contains the ruins of a castle and rock above the river. This showed that mapping from a UAV is ideal, fast and safe in these conditions.
Point clouds and orthophotos
For calculations and outputs, Agisoft PhotoScan Pro was used to process UAV mapping test locations. The resulting point clouds and orthophotos were divided into 1:500 map sheet maps. The assessment of the positional accuracy of the orthophoto and the digital surface model was done by land surveying methods. At the Tymakov test location, outputs from the UAV were compared with geodetic measurement of the position and the height of 80 control points, which were not used in photogrammetric processing.
Comparing the position values of 80 control points by land surveying methods in the state coordinate system and the coordinates of these control points interpreted from the orthophotomapy resulted in the determination of the mean absolute value of the differences of 2.52cm in position and 3.75cm in height. Furthermore, 1,351 detail points identical to land surveying measurements in the cadastral area were evaluated from the point clouds of DSM-forming points of the façade surface. All points for the experiment were measured by two experienced experts. The RMSExy value of these points was 13.54 cm and therefore complied with the requirements of the cadastral regulation of the accuracy of points entering the Land Register of the Czech Republic. Similar results were obtained at another test site.
Another of the results of the experiments was the economic evaluation of the use of UAV methods in favour of cadastral mapping. Comparison was made with conventional aircrafts with crew and use of standard UltraCamXP large-format cameras, as well as economical comparison with standard land surveying mapping procedures. In all the different experimental locations, the combination of the UAV mapping methods with the land surveying methods of measuring the detailed cadastral points is up to 22% more cost-effective as a separate land-surveying method.
The possible deployment of UAVs for mapping in the cadastre of the Czech Republic is now currently being considered by ÄÚZK. UAV mapping brings very interesting results and will have a future in Czech cadastre, we hope.
Jakub Karas is co-owner Upvision (www.upvision.cz). Václav ŠafáÅ is chief of department GIS and real estate cadastre at the Research Institute Geodesy, Cartography and Topography (www.vugtk.cz)