Geoscience Reference
In-Depth Information
Without this piece of information it is impossible to say, at a branch, if two
sections are converging in one or one is branching in two. Flow direction
is usually embedded in the hydrography data models either implicitly (e.g.
the flow direction follows the order of the points of each edge) or explic-
itly (i.e. as an attribute). In the case the flow direction is not known, it can
be calculated from the z-coordinate: this was the case with our input data.
The algorithm to evaluate the flow direction in each section of the river
network iteratively analyzes every edge and for every current edge tries to
classify the edges touching it into fathers, children and siblings, comparing
the z-coordinate of the vertices of each edge.
For the current edge the algorithm finds the highest point
cpMax
and the
lowest point
cpMin
; the same is done for all the edges connected to it
(
pMax
and
pMin
respectively):
•
if an edge has pMax > cpMax and it is connected to c on the point
cpMax, it is a father,
•
if an edge has pMin < cpMin and it is connected to c on the point
cpMin, it is a child,
•
if an edge has pMax > cpMin and it is connected to c on the point
cpMin, it is a sibling,
•
if an edge has pMin < cpMax and it is connected to c on the point
cpMax, it is a sibling.
Furthermore,
•
if c doesn't have any father, it is a source,
•
if c doesn't have any children, it is a drain,
•
the flow direction of c is from cpMax to cpMin.
It is common that in cartographic datasets the values on the
z
plane have a
lower precision than the
x, y
data. For this reason the simple model above
had to be expanded to include two special cases:
•
flat edges
•
uphill edges
A flat edge happens when
cpMax = cpMin
. This may be caused when a
river flows on an almost flat surface (e.g. a big plain) or the edge is an arti-
ficial connector inside a water body (e.g. in a lake), or the edge is too short
to record a difference in the
z
values of its points. Flat edges are quite
