Geoscience Reference
In-Depth Information
4- Methods: Algorithm Design
Several preprocessing steps extract the backbone of a centerline feature.
The first preprocessing step dissolves all flowlines on reach code; this re-
duces the number of records which must be searched during subsequent
steps. The original, non-dissolved version of data is retained, so that (sub-
sequent to centerline delineation) all segments for each reach in the center-
line will be individually attributed.
The second preprocessing step merges all polygonal water features con-
tained in the NHDWaterbody and NHDArea feature classes. Once merged,
a selection on minimum size is applied to select those polygons considered
large enough for the extraction of a centerline. The minimum size criterion
of 0.02 km
2
was chosen for this research to delineate a centerline from the
24k source data. The specific value might be expected to vary with source
scale and density of flowline channels.
In a third preprocessing step, dissolved reaches are intersected with the
merged and selected water polygons to form a “master centerline” list.
Comprehensive intersection of all reaches with all polygons is preferable
to beginning with the set of artificial paths: as
Figure 2
(upper panel)
shows. Selection of artificial paths alone misses many stream channels that
contribute to a continuous centerline, and includes many artificial paths
(flowing through isolated lakes and ponds) which are not viable centerline
features. Subsequent processing will identify these channels eventually,
but requires a lot of additional searching through the flowlines.
The lower panel shown in
Figure 2
demonstrates that a simple intersection
is sufficient to capture the majority of centerline features; but some gaps
persist that make the centerline non-continuous.
