Authors: Gaurav Sinha*, Ohio University, Samantha Arundel, Center of Excellence in Geographic Information Science - U.S. Geological Survey, Isabel Gregorek, Ohio University, Christian Slee, Ohio University, David Mark, University at Buffalo, SUNY
Topics: Geographic Information Science and Systems, Cartography, Cyberinfrastructure
Keywords: GNIS, elevation, terrain, landforms, geomorphometry, feature extraction, toponym, ontology
Session Type: Paper
Start / End Time: 9:55 AM / 11:35 AM
Room: Capitol Room, Omni, East
Presentation File: No File Uploaded
The U.S. Geological Survey (USGS) maintains the Geographic Names Information System (GNIS), which contains basic geographic information about 2.8 million geographic toponyms classified into 63 feature classes. The ability to automatically derive areal footprints of GNIS features is an important goal of the USGS National Geospatial Program (NGP). The authors are currently focused on creating methods for automated identification and delineation of approximate areal footprints for GNIS terrain (landform) features, which currently are represented simplistically as 2D points. Since terrain feature extraction and classification are intertwined, an important first step of this project is to define valid terrain feature categories. However, GNIS terrain feature classes vary widely in geomorphometric signatures. (e.g., summit covers topographic eminences of many different shapes, sizes and material composition). Thus, subcategories with narrower semantic scope are necessary for delineating GNIS terrain feature boundaries. The generic part of feature toponyms (e.g., mount, mountain, valley, canyon) are logical candidates to begin the search for intuitive, socio-linguistically defined subcategories. In this presentation, the authors will discuss their initial findings for the following interrelated research questions: i) Are there regional geographic patterns of toponym generics extracted from GNIS terrain feature classes? ii) Which geomorphometric parameters/features can be used to construct distinctive signatures for toponym generics? iii) How can the geomorphometric signatures of toponym generics be used to compare and derive a small set of GNIS subcategories? iv) How can such linguistic subcategories be ultimately used for designing feature extraction methods for GNIS features?