DCpts, DTpts, SLpts: Dune crest, dune toe, and mean high water shoreline positions: Cedar Island, VA, 2012

A customized MATLAB routine was used to extract beach and dune morphology. The routine determines the position of the mean high water (MHW) shoreline and the position and elevation of dune toe and dune crest. The routine is summarized below. Refer to Stockdon and others (2012) for details and figures. A version of these data are released in Doran and others (2017).
Processing was performed in MATLAB version 2017a. Source lidar point clouds were adjusted to MHW. The NAVD88 elevation of MHW is 0.34 m for the region encompassing Cedar Island (Weber and others, 2005). The input lidar dataset was the lidar point cloud (see Source Information section). The automated routine (see below) was executed iteratively to optimize the parameters based on visual comparison with high-resolution imagery. Results were considered low quality and required modifications to the parameters if numerous feature positions appeared to be erroneous or if there were large alongshore gaps in the output features. Once optimal parameters were determined, the routine was run for a final time.
The shoreline position extraction selects the MHW shoreline grid cell within a dynamic distance of a prior shoreline. Stockdon and others (2012, p. 18) used a distance of 3 standard deviations of all shoreline positions within the grid segment. Starting with 3 standard deviations, the search distance was optimized to produce both quality and quantity of shoreline points as determined by the operator through trial and error. This usually resulted in a search distance between 0.5 and 3 standard deviations of the shoreline points within the grid segment. The prior shoreline was a first approximation and was selected as the most recent shoreline in a shoreline database.
The following summarizes the feature extraction routine:
First, lidar point cloud data are interpolated to 10 m (alongshore) by 2.5 m (cross-shore) grids oriented parallel to a roughly shore-parallel reference line that is used for all such feature extractions (Doran and others, 2017). Interpolation uses a Hanning window that is twice as wide as the grid resolution to minimize noise. Automated routines then identify the position of the three features (dune crest, shoreline, and dune toe) for each elevation profile. Each step is performed by a separate MATLAB program.
To determine the dune crest position, the algorithm selects the peak of the most seaward sand dune. It does so by classifying inflection points in cross-shore slope as morphologic crests and troughs and identifying the highest crest elevation as the primary dune. In the absence of a dune, the beach berm or seaward edge of the bluff or hard structure (e.g., road, parking lot, seawall) was extracted. The selection is limited to areas within 200 m of the shoreline, seaward of built structures, less than 0.5 m vertical error, and higher than the MHW elevation.
For each profile, the position of the MHW shoreline is determined. The cross-shore location of the MHW shoreline is automatically extracted with the incorporation of the RMS error surface. The probability that a grid cell elevation value is equal to MHW elevation is estimated from a normal distribution of beach elevation. The grid cell with the highest probability within a defined dynamic distance (standard deviation multiplier of all shoreline points within the grid segment) of a prior shoreline is selected as the most likely shoreline. The dynamic search distance varies based on the generalized beach width of the region. A more precise location of MHW is interpolated from a linear regression of the selected point and adjacent grid cells. A confidence interval for the estimate is defined from the regression error and the error in the lidar data.
Once the dune crest position is established, the dune toe position is determined. First, if the dune crest position is understood to represent a beach berm because the elevation is less than the high water line (HWL), dune toe position is not determined for the given profile. In the remainder of cases, dune toe is identified as the point of maximum slope change between the shoreline and the dune crest unless the elevation of that point is within 0.5 m of the dune crest elevation or the curvature of the profile around the dune toe was negative.
The feature positions resulting from the automated routine are manually verified by overlaying the horizontal positions on high-resolution coastal imagery. Profiles with identified inconsistencies are viewed in a customized MATLAB Graphical User Interface (GUI) that displays the features with the lidar data (interpolated and point cloud) and allows the operator to relocate or delete feature positions. High density of built structures and complex topography often cause errors in the feature positions that require manual editing.
The positions are exported from Matlab with geographic coordinate values (WGS84) in a table of comma-separated values and converted to an Esri shapefile in NAD83 UTM Zone 18N (Make XY Event Layer tool in ArcGIS 10.5). Coordinates were transformed from WGS84 to NAD83 using the transformation WGS_1984_(ITRF00)_To_NAD_1983 (WKID: 108190, accuracy: 0.1 m).

Positions were converted from a table of comma-separated values to a shapefile using the function functions_warcpy.MorphologyCSV_to_FCsByFeature in the python package bi-transect-extractor (Sturdivant, 2018; version 1.0).
The dune crest, dune toe, and shoreline positions were overlaid on the DEM in ArcGIS 10.5 to check for incongruences between the morphology points and the elevation. Features that displayed logical inconsistency relative to the other features were flagged. For example, dune toe and crest points located seaward of shoreline points would be flagged and evaluated with reference to the DEM to determine which of the features were the more accurate.
Finally, the points were converted from shapefile format to comma-separated text using the ArcGIS tool Export Table (version 10.5). Both the shapefile and the CSV files are included in this data release.

Added keywords section with USGS persistent identifier as theme keyword.