Visual impact studies often make use of Boolean viewshed maps created by st
andard functions implemented in most raster GIS software. The DEM is used i
n a deterministic way and possible inaccuracies in the DEM are neglected. M
onte Carlo simulations of the errors in the digital elevation data (DEM err
or) can be used to test the effect of the DEM error on the calculated views
hed. Both an explorational and a quantitative method based on binomial stat
istics are proposed to analyse the impact of the number of simulations on t
he accuracy of the probabilistic visibility of a cell. The probabilistic in
formation is used for the creation of a Boolean viewshed map with a known u
ser's and/or producer's accuracy. The importance of this methodology is ill
ustrated.