Geoscience Reference
In-Depth Information
flagged as “not to be typified” otherwise all the segments in the same ditch
cluster will be typified together.
Once we have identified the ditches to generalize and grouped them in
ditch clusters, we need to typify them. That is to replace their representa-
tion with a new one, that is simpler (i.e. uses less geometries) but still con-
veys the same information of the former one.
Figure 4:
cluster of ditches and their cluster envelope.
The idea behind our algorithm is that the area interested by the presence of
ditches should be considered as a “canvas” where we can freely draw the
new ditches (as it will be explained later, this is not completely true and
this behavior is constrained). The new ditches will run in the same direc-
tion of those that are replaced and will be equally spaced, accordingly to a
distance parameter SP that is function of the target scale.
It is important to underline that the new ditches that will be drawn are
completely new features that are not present in the original dataset: this is
what makes this algorithm different from a selection algorithm.
The shape of the “canvas” where to draw the new ditches is obtained cal-
culating a hull or envelope around all the segments of each ditch cluster.
To obtain the shape of the “canvas” where to draw the new ditches we
chose not to use the convex hull: since the convex hull may not follow
closely the shape of the ditches, the resulting canvas could be too broad,
enclosing areas that in the original data were not interested by the presence
of ditches. Our choice was to be more conservative, in order to minimize
the possibility of topological errors while drawing the new ditches: the
