Error Propagation in the Production of Dem-dom-dlg by Digital Photogrammetry

This paper adopts the stochastic simulation method that pertains to geostatistics, on the basis of simulated DEM realizations with spatial autocorrelation errors, to explore the error propagation mechanism of DEM-DON-DLG under the digital photogrammetry environment. It follows this procedure: first, produce 100 error-disturbed DEMs by means of sequential Gaussian simulation; next, attain 100 realizations of DOM based on the digital photogrammetric system VirtuoZo; third, semi-automatically extract some boundaries of land cover to simulate 100 DLGs; last, proceed to perimeter and area error propagation of a number of polygons. Moreover, for comparison, error propagation in polygonal perimeters and areal extents was also performed using the law of variance-covariance propagation. The experiments show that the results of analytical-based error propagation are much lower than the standard deviations of perimeter or area calculated by the stochastic simulation, for the former overlooks the property of spatial autocorrelation and co-relation among variables which objectively exist in the spatial process (including DEM-DOM-DLG error propagation). The research will play a positive role in semi-automatically error processing in the process of digital photogrammetry spatial information acquisition. * [email protected]; phone:86-13971611505