3DEP Feature Extraction and Conflation
Ongoing research in CEGIS has focused on creating semantically-accessible terrain features for the nation from the National Geospatial Program’s (NGP) pixel-based 3D Elevation Program (3DEP) data.
Due to the size and diversity of the United States, automated, or, at least semi-automated mapping techniques are essential for realizing this objective. Besides the methodological and technical challenges involved in automating terrain or landform mapping, another daunting obstacle to automating this mapping is verifying results due to the lack of validation data.
An important resource of validation can be the Board of Geographic Name’s Domestic Names Database, which is the “official repository of domestic geographic names data, the official vehicle for geographic names use by all departments of the Federal Government, and the source for applying geographic names to Federal electronic and printed products.” Domestic Names are delivered to the public by way of the Geographic Names Information System (GNIS), which is managed by the USGS. As of April 2018, the GNIS currently stores information about 2.28 million current and historical physical and cultural geographic features, located in the United States, associated areas, and Antarctica. The database does not include roads and highways. In the late 1970’s, GNIS developers introduced a classification system of 63 non-hierarchical, mutually-exclusive feature classes to encourage the standardization of feature type interpretation. This also increased information retrieval accuracy and efficiency. The names and descriptions of GNIS feature classes are simple to enable users’ understanding of the types of features each class represents. The GNIS database holds the federally-recognized name of each feature and defines the feature location by state, county, USGS topographic map, and geographic coordinates. For mapping purposes, all GNIS features are located by a primary point, while some also include secondary location points. Considered as a whole, the primary and any secondary points provide a general extent of the feature. Other attributes include names or spellings other than the official name, feature designations, feature classification, historical and descriptive information, although all fields are not always populated for every feature. Thus, this project seeks to i) create semantically-accessible terrain features from the pixel-based 3D Elevation Program (3DEP) data, and ii) enhance the usability of the USGS Geographic Names Information System (GNIS) by associating boundaries with GNIS features whose spatial representation is currently limited to 2D point locations.
