Beyond the case study: characterizing natural floodplain heterogeneity in the United States
Date
2023
Authors
Iskin, Emily Paige, author
Wohl, Ellen, advisor
Morrison, Ryan, committee member
McGrath, Daniel, committee member
Ronayne, Michael, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
With human degradation of natural river corridors, the number of natural, functional floodplains is rapidly decreasing due to dams, diversions, artificial levees, draining, development, agriculture, and invasive species. At the same time, small- to large-scale interest in and implementation of river restoration is expanding, with floodplain restoration soon to take a starring role. To properly manage and restore processes to floodplains, we first need a broad understanding of what they look like and why. A key component of natural river-floodplain systems is heterogeneity, defined as the spatial variation of geomorphic and vegetation classes and patches across a floodplain. Heterogeneity of floodplains both reflects and influences the fluvial processes acting on floodplains and can help shape our understanding of the form and function of floodplains. To begin characterizing floodplain spatial heterogeneity, I present in this dissertation: 1) the development of a method to combine field measurements and remote sensing data products to calculate integrative landscape-scale metrics of floodplain spatial heterogeneity, and the demonstration of which metrics from landscape ecology are likely to be useful for identifying qualities of natural floodplains at four case study sites; 2) a sensitivity analysis to determine whether and how the values of the heterogeneity metrics change when spatial and spectral resolution of the input data are increased, and the extraction of underlying data from the classification results to determine whether using higher resolution data allows identification of the resulting unsupervised classes in relation to field and remote data at four case study sites; and 3) quantification of floodplain spatial heterogeneity, evaluation of whether statistically significant patterns are present, and interpretation of the statistical analyses with respect to the influence of channel lateral mobility and valley-floor space available using a complete dataset of 15 sites representing diverse floodplains across the continental United States. I found that "stacking" Sentinel-2A multispectral satellite imagery and digital elevation model topographic data allows for unsupervised classification of floodplains, and that metrics from landscape ecology can differentiate between different floodplain types. I also found via a sensitivity analysis that increasing the spatial resolution of the topographic data to finer than 10 m and including band ratios related to vegetation improves the classification results. Comparison of the field classes with the remote sensing classes allows for general interpretation of the results, but it is the heterogeneity within the broad classes that I expect is most important to these ecosystems. Lastly, through classification of 15 diverse river corridors across the United States, calculation of five heterogeneity metrics, and completion of a comparative analysis, I found that these natural floodplains have moderate aggregation of classes (median aggregation index = 58.8%), high evenness (median Shannon's evenness index = 0.934) and intermixing of classes (median interspersion and juxtaposition index = 74.9%), and a wide range of patch densities (range of patch density = 491–1866 patches/100 ha). I also found that the river corridor characteristics of drainage area, floodplain width ratio (space available), and elevation, precipitation, total sinuosity, large wood volume, planform, and flow regime (channel mobility) emerge as important variables to understanding floodplain heterogeneity.
Description
Rights Access
Subject
fluvial geomorphology
remote sensing
spatial heterogeneity
landscape ecology
floodplain
river