Geoscience Reference
In-Depth Information
Algorithm
Time (msec)
River width and re-classification
3053453
River flow
785187
Strahler order
140954
Data enrichment
299218
River course
1054672
Pruning
1427500
Total
6760984
Table 2:
Time elapsed by each step of the generalization process in msec on a dual Intel
Xeon 2.0 Ghz CPU with 3 Gb RAM; the total time amounts to 1 hour 52 minutes 40 sec.
Figure 8:
results of the network pruning: left source data, right generalized data.
The flow direction reconstruction algorithm worked very well but in one of
our tests it could not provide a complete reconstruction due to the high
number of flat and uphill sections: despite some little improvements that
can be made on the algorithm, this proves once again that to have a good
generalization it is important to have a good input data; in our case in the
absence of a directed graph (i.e. having the flow direction of the streams)
the z coordinate of the data must not contain errors. The reconstructed
river courses show in some cases sharp bends: it is under investigation the
use of “good continuity” as a parameter in the river course reconstruction
algorithm (e.g. see (Touya 2007)).
The pruning algorithm was found to perform well, even though the speed
of the density pruning could be probably improved. The pruning solved
also some braided streams sections removing the shorter branches. The
performances of the pruning algorithms are given in
Table 3
.
