Geoscience Reference
In-Depth Information
1- Introduction
A frequent task for geographers is to distill geographic features and to
assemble these data in a map. Given the nature of this distillation process,
geographers modify data details for representation at a chosen scale. It is
this possibility to focus on a map region as a synopsis of given themes
such as hydrography, transportation, or landcover that enables spatial
analysis. However, the ability to construe geographic features in any given
map design is limited in part by feature extraction. Many in the Geo-
graphic Information Science (GIS) and Remote Sensing communities are
familiar with the difficulties of extracting features from a raster image such
as a satellite scene (Sowmya and Trinder 2000). In vector datasets, the
choice of which features provide important characteristic shapes for map-
ping can be a time-consuming task especially when selecting from a large
database (Peter 2001). Unfortunately, few vector databases contain feature
attributes which aid the geographer in the selection process. This paper
describes a database enrichment process which adds attribution delineating
a continuous channel centerline through a set of hydrographic flowlines.
2- Problem Context
One area of GIS research that has historically offered rich insights about
the synoptic link between maps and the real world is generalization. Carto-
graphers are frequently faced with the task of manipulating geographic
databases for the purpose of generalization, and in many cases feature
extraction and/or selection are necessary. Although many tools are currently
available through commercial GIS packages to analyze and generalize data
(Regnauld and McMaster 2007), the relationship between feature extrac-
tion and generalization is still an area with much research potential.
As Steiniger and Weibel (2005) note, feature-oriented generalization offers
a useful paradigm, which organizes generalization based on map objects
chosen to emulate certain processes, presumably within a certain theme.
More explicitly, the chosen theme is in reality a context which determines
what common features are expected from a cartographic product (Thomson
and Brooks 2007). Touya (2008) argued a similar point, that generaliza-
tion processes must collaborate based on the underlying geographic dis-
tinctions such as urban versus rural, residential versus commercial, etc.
