Alexandra RÁŠOVÁ

Effect of digital elevation model quality on visibility analysis in geographical information systems

Rášová, A.: Effect of digital elevation model quality on visibility analysis in geographical information systems. Kartografické listy, 2014, 22 (1).

Abstract: Current geographical information systems (GIS) software packages usually offer a possibility to calculate viewshed from a given observation point and a digital elevation model (DEM). Although the inaccuracy of the DEM is acknowledged in its metadata, it is still not very common to consider it in analyses in GIS environment – although the DEM inaccuracy represents an important effect on the result. This article presents different methods of DEM error modelling and an approach to integrate the DEM inaccuracy in the process of visibility analysis. Probability of a cell being visible in consideration of the DEM error can be expressed using probable viewshed. In probable viewshed raster, each cell has assigned value from the range from 0 to 1 depending on its probability to be visible. An ArcGIS Model Builder Toolbox was created to calculate the probable viewshed; this toolbox contains two models for calculation with and without consideration of spatial correlation of the DEM error. Probable viewshed is computed using Monte Carlo simulation for a given number of random DEM realisations. Both models require an observation point, a DEM, and information about the DEM quality (root mean square error – RMSE and its distribution). The calculated probable viewshed – unlike the “common” viewshed – gives the information about the influence of DEM inaccuracy and therefore it can prevent misinterpreting the results of the visibility analysis and their incorrect application.

Keywords: visibility analysis, probable viewshed, digital elevation model, inaccuracy, uncertainty, error modelling, Monte Carlo simulation