These data were not initially classified for vegetation removal. As such, we removed vegetation to create 
the raster datasets. The removal process will not match the removal from the classification performed on 
the data later. The later classification is uploaded to OpenTopography. To get the same rasters as the ones
we uploaded here, start with the complete unclassified datasets and follow our method below.

We filtered the lidar point clouds for returns from both trees and lower vegetation. For trees, we 
classified the points as trees/non-trees using the qCANUPO classifier [Brodu and Lague, 2012, 
http://dx.doi.org/10.1016/j.isprsjprs.2012.01.006] within  the software CloudCompare 
(GPL software version 2.6.2. (2016) at www.cloudcompare.org) and removed all of the tree points. 
To remove non-tree vegetation, we filtered the elevation data within 5 m grid cells. Points that 
were = 0.10 m above the minimum elevation for a grid cell were classified as low herbaceous vegetation 
and removed. While this filter is appropriate for the low-relief topography of the Wax Lake Delta, some 
points from steep channels or levees may have been incorrectly removed.

The 'bare-earth' elevations were gridded from the 2009 and 2013 point clouds using a 2-m pixel for ease 
of presentation and to remove spatial clustering bias in statistical calculations. The gridded bare-earth 
data from both surveys were adjusted from height above ellipsoid to NAVD88 using GEOID 12B. The 2009 lidar 
data were not initially classified  to remove water, so we used Normalized Difference Vegetation Index from 
a nearly concurrent (January 12, 2009) LANDSAT image as a stencil to remove water cells. As a second filter, 
we removed all points below NAVD -0.075 m from the 2009 survey.