October 2011 article (early view): “Digital Hydrologic Networks Supporting Applications Related to Spatially Referenced Regression Modeling,” by J.W. Brakebill, D.M. Wolock, and S.E. Terziotti. Part of Featured Collection on SPARROW.
Continuing along on Huge Dataset Week, this paper looks at the kind of stream network representation you need to model the conterminous United States.
Anyone familiar with the field knows there have been many representations of the U.S. stream network at various scales and levels of quality. The original “River Reach File” (RF1) schema was developed by Bob Horn of EPA way back in the 1970′s; with subsequent modifications, mergers, and updates, it forms the backbone of stream reach identification and attributes today in the form of the National Hydrography Dataset (NHD). It’s worth reading this paper for the historical background alone.
A key to successful stream modeling is to have a network with just the right amount of detail, and this paper shows how the authors achieved this for the SPARROW collection. Digital hydrologic networks modified from the RF1 and NHD geospatial datasets provided frameworks for SPARROW in six regions of the conterminous United States. In addition, characteristics of the modified RF1 were used to update estimates of mean annual streamflow. This produced more current flow estimates for use in SPARROW.
[Please note: I have quoted and paraphrased freely from the article, but the interpretation is my own!]