#29: Part 3, Spatial statistical models for stream networks: Freeware tools for model implementation
Dan Isaak
Being given the right tools for free is best…
Hi Everyone,
It’s a new world today if you’re a stream ecologist, manager, or researcher that’s interested in understanding, accurately modeling, or managing stream ecosystems at spatial scales larger than a pool/riffle or an individual stream. As described previously, new types of spatial statistical models have been developed for stream networks (blog #27) that often provide fundamentally better inference (& new types of inference) from many commonly available datasets (blog #28). The GIS tools and R scripts for implementing these models have not previously been widely available, which has limited their broad application. Today, however, a new website was launched to provide these tools as freeware (http://www.fs.fed.us/rm/boise/AWAE/projects/SpatialStreamNetworks.shtml), thereby culminating almost a decade of research and development by Jay Ver Hoef and Erin Peterson.
On the website, you’ll find: 1) the STARS ArcGIS custom toolbox used to pre-process stream datasets and develop auxiliary information, 2) the SSN (SpatialStreamNetworks) package for the freeware R statistical program (downloadable here: http://www.r-project.org/), which is used to fit the spatial models, 3) example datasets, and 4) tutorials to develop familiarity with the tools and their functionality. If you’re planning to use the spatial models with datasets in the Northwest U.S., there’s also a set of NHD+ hydrocoverages that have already been reconditioned to work consistently with the spatial models (downloadable here: http://www.fs.fed.us/rm/boise/AWAE/projects/SSN_STARS/GIS_reconditioned_NHDstreams.shtml). Use of these hydrocoverages can save some time by eliminating a few of the pre-processing steps otherwise required in STARS.
Attached to this email are vignettes that step through several well annotated examples in R. Even if you don’t know how to operate in the R environment, it’s possible to get the gist from the thorough descriptions and interpretation of the statistical outputs. The datasets used in the vignette are also embedded within the SSN package to facilitate working directly through them in R after loading the SSN package onto your computer. With a basic level of familiarity, implementing the models with your stream databases just requires doing the preprocessing steps described in the STARS tutorial, then crunching through the statistical analysis much as one would with any other analysis.
Not much more to say than that, but have at it & let’s learn lots of new stuff about streams & their critters so we keep as many of them around this century as possible.
Until next time, best regards,
Dan