Skaitmeninio žemėlapio, sudaryto stereofotogrametriniu metodu, tikslumas

Abstract

The Geodesy Institute of VGTU is working on compiling the basic digital map for Vilnius city GIS by stereophotogrammetric approach according to common Vilnius municipality and Norway project “Geoinformation system of city” [1]. The map is constructed applying the stereodigitalisation of photogrammetric model using computerised analogue instruments - two Wild autographs A8 and Santoni-Gallileo stereosimplex [...]. The accuracy of the map is to be the same as in graphic plans at a scale of 1:1000. The aerophotography was taken at a scale of 1:6000 (ck 153,6 mm ). In 1996 and 1998, the map was constructed of area more than 80 km2. The aim of investigation of map accuracy was to define a real accuracy of the constructed map. Accuracy of the map was investigated according to the data of inner control of stereophotogrammetric processes repeating stereodigitalisation and geodetic field measurements by GPS. The data of stereomodels absolute orientation characterise the accuracy of stereodigitalisation quite well. There are absolute deviations vx, vy, vz of control (aerotriangulation) point ( about 10 points in the model) coordinates X, Y, Z as well as values of mean square Mp and Mh (Table 1). Deviations of control point coordinates and calculated by aerotriangulation which are not used for absolute orientation characterise effectively the accuracy of stereomodels. Mean square values of such deviations are shown in the Table 2. The map accuracy was investigated by repeating the process of stereophotogrammetric digitalisation, too. Aeropositives are inserted in the instrument anew and orientated and part of distinct features is constructed. Coordinates of points are compared with previous database. Repeated stereodigitalisation was carried out by the same or another operator using other instrument. Mean square coordinate difference is mΔxy= 0,29 m and maximal difference is ΔXY=0.65 m ( see Table 3). Objective accuracy of the map features was determined by geodetic measurements. Marked in the terrain and interrelated control points (32 points) were coordinated using GPS method. Distinct terrain features (comers of buildings, streets, pavements, poles, etc) were coordinated by polar method from points mentioned above (see the picture ). The number of such coordinated feature points are 448. Point coordinates determined by the geodetic method were compared with the identical point coordinates in the database. An accuracy of feature points position was also controlled by measuring lengths from GPS stations up to feature points or between features. The number of measured lengths are 59. These lengths were established in the database as well. Differences of controlled point coordinates and lengths ΔX, ΔY, ΔXY as well as frequencies ΔS and values of mean square error are shown in Tables 4, 5 and Fig 2. Mean square feature points position error of the constructed map is 0,26-0,34 m concerning GPS network of city and lengths accuracy is 0.29 m. Maximal difference of co-ordinates ΔXY=0.80 m and difference of lengths ΔS=0.78 m. Negative systemic errors were observed. Some identified in the terrain aerotriangulation and polygonometric point coordinates were controlled by GPS method, too (Tables 6 and 7). Also, covers of 15 underground utilities wells were coordinated using GPS. Coordinates of such covers were also determined in the scanned rastric plans at a scale of 1:500. It is observed that in this case the existing old topographical plans do not conform to the accuracy of plans at a scale of 1:500. An accuracy of digital map point elevations was investigated by comparing the same points elevations with the geodetic ones received from digitalisation in the existing plans at a scale of 1:500. Digitised relief points were included into database of constructed map and elevations of some points were measured by stereophotogrammetric approach (Table 9). An accuracy of elevations were investigated repeating stereodigitalisation in three stereomodels (see Table 9). Conclusions. Creating Vilnius city GIS basic digital map by stereophotogrammetric approach, distinct, clearly defined features are fixed approximately with an absolute accuracy of 0,30 - 0,35 m, and elevations with 10-15 cm. It satisfies the accuracy of graphic plans at a scale of 1:1000.