The distribution of surface distances (Figure 11) shows a concentration peak between 0.1 and 0.25, which corresponds to quite small differences. Bel Hadj Ali and Vauglin (1999) proposed a polygon object-matching process partly based on surface distance, and considered such a distribution as showing little difference.
They compared buildings from cadastral and topographic data (considered as quite close in position and shape) and noticed that 25% were under 0.05, which is not the case here, and 70% were between 0.15 and 0.45.