Stratified Database Pruning to Support Local Density Variations in Automated Generalization of the United States National Hydrography Dataset

Introduction: The U.S. Geological Survey’s (USGS) Center of Excellence in Geospatial Information Science (CEGIS) is conducting generalization research in cooperation with the University of Colorado—Boulder and Pennsylvania State University to support display and delivery of The National Map and other USGS geospatial data at multiple scales. This paper focuses on generalization of the National Hydrography Dataset (NHD). Objectives: Objectives of this research are to develop methods to sub-select, or prune, features from the multi-scale high-resolution (HR) NHD layer to automate generating a multiple representation database (MRDB) and simplify, or further generalize, remaining features for cartographic display. Methods should maintain hydrographic network connectivity and local density variations that typify physiographic or climate variations. Methodology: A four-subbasin region of HR NHD data in Iowa, having obvious natural network density variations, was pruned to four smaller scales—1:100,000 (100K); 1:500,000 (500K); 1:2,000,000 (2M); and 1:10,000,000 (10M). Methods employ a stratified database pruning approach that partitions HR NHD data based on localized network densities and prunes features to densities appropriate for smaller map scales based on reach codes and upstream drainage area (UDA) estimates. Database enrichment,