Random forest models to estimate bankfull and low flow channel widths and depths across the conterminous United States

Channel dimensions (width and depth) at varying flows influence a host of instream ecological processes, as well as habitat and biotic features; they are a major consideration in stream habitat restoration and instream flow assessments. Models of widths and depths are often used to assess climate change vulnerability, develop endangered species recovery plans, and model water quality. However, development and application of such models require specific skillsets and resources. To facilitate acquisition of such estimates, we created a dataset of modeled channel dimensions for perennial stream segments across the conterminous United States. We used random forest models to predict wetted width, thalweg depth, bankfull width, and bankfull depth from several thousand field measurements of the National Rivers and Streams Assessment. Observed channel widths varied from <5 to >2000 m and depths varied from <2 to >125 m. Metrics of watershed area, runoff, slope, land use, and more were used as model predictors. The models had high pseudo R2 values (0.70–0.91) and median absolute errors within ±6% to ±21% of the interquartile range of measured values across 10 stream orders. Predicted channel dimensions can be joined to 1.1 million stream segments of the 1:100 K resolution National Hydrography Dataset Plus (version 2.1). These predictions, combined with a rapidly growing body of nationally available data, will further enhance our ability to study and protect aquatic resources.

Impact/Purpose

The US Environmental Protection Agency (EPA) is working to calculate non-use values for streams and lakes across the conterminous United States. Focus groups conducted by the National Center for Environmental Economics determined that people understood water resources (lakes and streams) in terms of area. This perspective on water resources presented an interesting challenge as streams and rivers are often measured in the unit of length and not area. In this study, we used the USEPA StreamCat and the National Rivers and Stream Assessments datasets (2008-09 and 2013-14) to interpolate widths and depths of streams out to 1.1 million stream segments. These interpolated predictions will allow us to now calculate a stream area and to present lakes and streams in the same unit. In this paper, we describe this novel approach of interpolating channel width in rivers and streams. We discuss our use of a machine learning algorithm (random forest) as opposed to a commonly used power function. Additionally, we compare our interpolations to another well-known bankfull width model and show that it performs just as well, and often better in some regions. Outside of the EPA’s need for this research, state and local practitioners often require channel dimension (width and depth) estimates for fisheries management models. Estimates of channel width and depth could provide an important tool for fisheries and watershed managers to estimate the habitat amount, fisheries production models. Thus, we modeled channel depths in addition to channel widths. These estimates can help guide sampling designs for regional and national aquatic assessments.  Upon publication of this work we will make these datasets publicly available through the USEPA StreamCat database. This research supports work being conducted in collaboration with economist in the EPA’s National Center for Environmental Economics under SSWR 1.2.2 - Interpolation and stressor-response analyses that extend the use of NARS data to support regulatory program needs.

Citation

Doyle, J., R. Hill, S. Leibowitz, AND J. Ebersole. Random forest models to estimate bankfull and low flow channel widths and depths across the conterminous United States. American Water Resources Association, Middleburg, VA, 59(5):1099-1114, (2023). [DOI: 10.1111/1752-1688.13116]

DOI: Random forest models to estimate bankfull and low flow channel widths and depths across the conterminous United States