SSN & STARS: Tools for Spatial Statistical Modeling on Stream Networks

Spatial statistical models for streams provide a new set of analytical tools that can be used to improve predictions of physical, chemical, and biological characteristics on stream networks. These models are unique because they account for patterns of spatial autocorrelation among locations based on both Euclidean and in-stream distances. They also have practical applications for the design of monitoring strategies and the derivation of information from databases with non-random sample locations. Generating the spatial data needed to fit these statistical models requires practical skills in multiple disciplines including ecology, geospatial science, and statistics. This is the home page for two sets of tools that have been developed to make the methodology more accessible to users: the STARS ArcGIS toolset and the SSN package for R statistical software. These models were developed by researchers at NOAA and CSIRO.

STARS: ArcGIS Toolset The purpose of the Spatial Tools for the Analysis of River Systems (STARS) toolset is to generate and format the data needed to fit spatial statistical models in R software. The STARS toolset makes use of the Landscape Network, a data structure used to efficiently navigate throughout a stream network. Specific tools have been included to 1) Pre-process the Landscape Network; 2) Calculate the hydrologic distances (with flow-direction preserved), the spatial additive function used to weight converging stream segments, and the covariates for all observed and prediction locations in the stream network; and 3) Export the topological, spatial, and attribute information in a format that can be efficiently stored, accessed, and analyzed in R.

SSN: An R Package for Spatial Statistical Modeling on Stream Networks The Spatial Stream Network (SSN) package was developed for R statistical software, which is powerful open source statistical computing and graphics software. Once the streams data have been properly formatted using the STARS toolset, the SSN package allows users to: 1) import and store their spatial data in R, 2) calculate pair-wise distances and spatial weights based on the network topology, 3) fit spatial statistical models to streams data where autocorrelation is based on three spatial relationships (Euclidean, flow-connected, and flow-unconnected), 4) estimate relationships between stream variables (spatial regression), 5) make predictions at unobserved locations (prediction sites), 6) export spatial data

• Resource Title: SSN & STARS Software Downloads.

  File Name: Web Page, url: https://www.fs.fed.us/rm/boise/AWAE/projects/SSN_STARS/software_data.html#software

  Includes the following resources:

  • SSN Package – Current Release: Cran-R project
  • STARS Current Release (zip)
  • Example datasets for STARS Tutorial
  • SaltRiver_TroutDensity_BlockKrige (zip file of .ssn data folder)
  • Attribute Metadata (xlsx)
  • Figure 1 map (PDF)
  • NCEAS Presentation that steps through the predictors and response variables & then has maps showing each database at the end of the presentation.
  • Related links