Floodplain Geomorphons for the Lake Champlain Basin of Vermont

This document provides supporting information for the floodplain geomorphons geospatial layers developed for the Lake Champlain Basin of Vermont in the northeastern United States.  In these layers, we used geomorphons 
to identify meso-scale topographic units in the floodplains of reaches in the Lake Champlain Basin of Vermont draining greater than 2 square miles.  Geomorphons were identified for the extent of the 500-year 
floodplain (Vermont Center for Geographic Information (VCGI), 2022).  The data are grouped into folders by HUC-8 subbasin: Missisquoi River (04300107), Lamoille River (04300105), Winooski River (04300103), Otter 
Creek (04300102), Mettawee River (04300101), and Lake Champlain (04300108).

Geomorphons are a classification system for landforms (Jasiewicz & Stepinski, 2013) based on generalized three-dimensional shapes of the ground surface.  Geomorphons can be identified at varying scales and are useful 
for visualizing spatial patterns in topography.  The geomorphons concept is based on a line-of-sight analysis for the eight topographic profiles in the cardinal directions surrounding a given cell in a digital elevation 
model (DEM).  While there are many different shapes that can be identified, many of the shapes are equivalent to each other via rotations and reflections.  In addition, some of the remaining shapes rarely occur in natural 
topography.  Jasiewicz & Stepinski (2013) identified 10 of the most common landform shapes (Table 1).

Table 1	
(Landform Shape, 10-class raster value, 3-class raster value, 3-class category)
Flat			1	1	flat
Peak		2	2	convex
Ridge		3	2	convex
Shoulder	4	2	convex
Spur		5	1	flat
Slope		6	1	flat
Hollow		7	1	flat
Footslope	8	3	concave
Valley		9	3	concave
Pit			10	3	concave

For this analysis, we used LiDAR-derived DEMs from the QL2 dataset available from the Vermont Center for Geographic Information (VCGI).  The DEMs were originally 0.7-meter resolution but were further hydroenforced and 
upscaled to 1-m resolution (Diehl et al., 2022).  DEM analysis to create the geomorphons was performed using Whitebox Tools (Wu & Brown, 2022) in R Studio.  To create geomorphons from DEMs, there are three 
hyperparameters within the tool: search radius, skip radius and relief threshold (Jasiewicz & Stepinski, 2013).  Search radius determines the maximum distance from a central point outward for which an individual landform 
can be identified.  Skip radius defines a minimum distance from a central point outward within which variations in topography are ignored.  Together, search radius and skip radius determine the maximum and minimum ranges, 
respectively, at which individual geomorphons can be identified.  Relief threshold (in degrees) sets a maximum limit for terrain roughness for individual landforms.  The hyperparameters can be adjusted to identify 
coarser or finer landforms as desired.  Different hyperparameter values were tested to find a scale of geomorphons that identified terrain features in a way that is consistent with our scale of interest, notably 
capturing meso-scale patterns in floodplain landforms.  A sensitivity analysis was performed for the study area to identify the optimal hyperparameter set that best captures topographic units of interest (i.e., meso-scale 
such as levees, floodplain channels, etc). Optimal values included a search radius = 34 meters, skip radius = 12 meters, and relief threshold = 1˚. Field surveys were used for verification. 

For each HUC-12 subbasin, we created a geomorphons layer using all 10 shape classes (Table 1); the file name for this layer begins with four digits corresponding to the last four digits of the HUC-12 number and ends with 
“geomorphons_clean.tif”.  We also created a second layer in which the 10 original classes were consolidated into three classes.  This second layer is intended to serve as an alternate layer with less complex topographic 
composition to simplify interpretability of the effects of topographic feature shape on flood routing.  The original 10 classes were grouped as follows: flats, slopes, spurs and hollows were grouped as ‘flat’ units; 
peaks, ridges and shoulders were grouped as ‘convex’ units; and pits, valleys and footslopes were grouped as ‘concave’ units.

In the alternate, 3-class layer the main channel of the reach was also masked out so that only topographic units for the floodplain were included.  The channel was masked out by removing geomorphons that overlapped with 
National Hydrography Dataset (NHD)  stream centerlines and were one of the following original 10 classes: flat, footslope, valley, or pit.  Prior to masking, the shapefile of NHD centerlines (U.S. Geological Survey, 2022) 
used for this step was edited to remove several first and second order streams that had little or no floodplains but flowed through floodplains of higher order reaches where they connected to these larger streams.  This 
was done so that the channels of these smaller tributaries were not masked out of the floodplain of the higher order reach with which they intersected.  The file name for the 3-class layer with the channel masked out 
begins with the four digits corresponding to the last four digits of the HUC-12 subwatershed code and ends with “geomorphons_mask3cls.tif”.

References

Diehl, R., K. Underwood, S. Lawson, S. Drago, J. Matt. Topographically defined Floodplains: Relative inundation for conservation and restoration planning in the Lake Champlain Basin, Vermont. May 25, 2022.
	https://www.arcgis.com/home/item.html?id=b05be7a01d56484593a2137c659bcb92
Jasiewicz, J., & Stepinski, T. F. (2013). Geomorphons—A pattern recognition approach to classification and mapping of landforms. Geomorphology, 182, 147–156. https://doi.org/10.1016/j.geomorph.2012.11.005
U.S. Geological Survey. (2022). USGS national hydrography dataset plus high resolution (NHDPlus HR) for 4-digit hydrologic units. [Dataset]. https://apps.nationalmap.gov/downloader/#/
Vermont Center for Geographic Information (VCGI). (2022). Topographically-defined Floodplains for the Lake Champlain Basin, 2013-2017 [Dataset].
Wu, Q., & Brown, A. (2022). Whitebox: “WhiteboxTools” R Frontend. (Version 2.2.0) [Computer software]. https://CRAN.R-project.org/package=whitebox