SSN2: Spatial modeling on stream networks

Spatial statistical modeling and prediction for data on stream networks, including models based on in-stream distance. Models are created using moving average constructions. Spatial linear models, including explanatory variables, can be fit with (restricted) maximum likelihood. Mapping and other graphical functions are included.

Impact/Purpose

The SSN2 R package is a useful modeling tool for data on stream networks. It formally incorporates the idea of spatial covariance, which allows nearby observations to borrow information from one another. Distance is thought of in three components: upstream distance, downstream distance, and Euclidean distance. Models are fit and predictions are possible using familiar R functions. SSN2 is an update to the SSN R package, which has been used to support science in over a hundred peer-reviewed journal articles. 

Citation

Dumelle, M., E. Peterson, J. Ver Hoef, A. Pearse, AND D. Isaak. SSN2: Spatial modeling on stream networks. U.S. Environmental Protection Agency, Washington, DC, 2023.

• https://cran.r-project.org/web/packages/SSN2/index.html