Welcome to Brett Eaton’s GitHub directory
This page is designed to host various ongoing research projects. Most of the material here is in the form of R Notebooks, which present both the code for analysis and some interpretive text. The source code file can be downloaded (see the code button at the top of each page), and modified. The underlying idea is to contribute to Reproducible Research (RR) in the field of geomorphology. People can explore both the ideas and interpretation that I am working on, and explore the code itself. After all, the devil is in the details.
Introducing GSDtools
Recently, Dan Moore, Lucy MacKenzie and I developed a set of tools for working with pebble count data collected to characterize the bed surface texture in gravel bed rivers. The tools are freely available in an R Package called GSDtools, available from GitHub. In the package are functions to create cumulative frequency distributions from a set of b-axis diameter measurements, to estimate the confidence intervals for a grain size percentile of interest about a sample estimate that is likley to contain the true population grain size, and to compare two samples to see if the grain size estimates for a percentile of interest are statistically different from each other. We have created a demo that shows how to download and use the GSDtools package. The demo is an html file, but also provides an option to download the underlying R Markdown file used to generate it.
Rethinking channel stability
My group has recently been focussing on the threshold separating channels in what we might call “dynamic equilibrium” from those that are unstabile, and which experience rapid changes in channel pattern. Our experimental data suggests that the characteristics of sediment transport are differnt during these (brief) periods of instability, and that a negative feedback between mean boundary shear stress and bank erosion makes periods of instability very brief, bringing systems back into what M. Church would classify as a threshold state. To be clear, we are most interested in steep gravel bed rivers with alluvial boundaries, and many of these ideas probably do not apply to labile sand bed streams. Interestingly, the evidence that our collaborators at BGC Enginnering have collected in the field suggests that large floods on fans also behave in much the same way as our laboratory experiments.
The physics of channel stability
Anyway, the first idea in this theme is that we should probably refine our thinking about the physics of stream channel stability. See this R Notebook webpage for our initial ideas. Feel free to download the code, and play with the analysis we present. All you need is R Studio (which is free to download). You may also need the bibliography file if you want to compile the notebook and to include the references that I have placed in there.
Estimating uncertainty of bed surface grain size samples
One of the most common measurements taken of gravel bed rivers is the so-called Wolman sample of the bed surface texture. The data are typically analysed to estimate the various percentiles of the cumulative bed surface distribution. However, uncertainties in these estimates are almost never made, and most people using the data are only vaguely aware of the (limitted) statistical power of this conventional sampling methodology or of the uncertainty of the resulting grain size estimates. Based on some work by Dan Moore at UBC, we have developed a method for estimating uncertainty for this kind of sample, and we use data from our stream table to test the idea. We are working on developing an R package to conduct this kind of analysis, so that people can more easily distinguish statistically significant changes in grain size from those that are not significant, and to help people choose an appropriate sample size to achieve their desired level of statistical power. The write is in this R Notebook, and uses this data file.
Support or Contact
Having trouble with Pages? Check out our documentation or contact support and we’ll help you sort it out.