The first part of the model is the creation of a spatially
autocorrelated error term to be added to the original DEM, shown
in Figure 1. Starting at the beginning of the model a normal raster
is created – this is a raster with random cell values which follow
a normal distribution. Cell values will be both positive and
negative, with a mean of zero and a standard deviation of 1. This
normal raster is the starting point for the error model. However,
errors in DEMs are typically not random, but reveal strong signs
of spatial autocorrelation. Spatial autocorrelation is introduced
with the Focal Statistics tool. Within a circular neighborhood of 5
cells the random error is averaged