The StreamCat Dataset and Channel Geometry Models.
On this page:
EPA’s National Rivers and Streams Assessment (NRSA) is a probabilistic-based survey for the conterminous US (CONUS) that can provide national and regional information on the condition of the nation’s rivers and streams. By combining NRSA data with landscape characterizations, models can be developed that spatially predict condition over all 1.1 million perennial stream segments in the CONUS. The landscape characterizations that make this modeling possible were derived from the StreamCat dataset, which is an extensive geospatial dataset containing 2.6 million stream segments and their associated catchments within the CONUS. StreamCat currently contains 581 landscape metrics, including anthropogenic and natural features. Data are organized by catchment, which includes the local area contributing to the stream segment excluding upstream contributions, and by watershed, which includes the local catchment plus upstream catchments.
As an example of how StreamCat and NRSA data can be used in modeling, preliminary channel geometry models for wetted width, wetted depth, bankfull width, bankfull depth, and bankfull width to depth ratio were developed using 2008-09 and 2013-14 NRSA data as response variables and StreamCat data as independent variables, based on random forest modeling. Accuracy for the preliminary wetted width, wetted depth, bankfull width was high – around 80% in each case. The model for bankfull depth had poor accuracy. Results for the bankfull width to depth ratio depended on the approach: results were poor (16%) if the ratio was modeled directly, but good (58%) if the ratio was calculated after bankfull width and depth were each modeled separately.
StreamCat data and modeled channel geometry could be useful to the Stream Stability Advisory Committee, a committee organized by EPA Region 3 that is working to develop a Stream Stability Index that could be applied to the five mid-Atlantic states using GIS.