Objective 1: Determine the requirements for identifying furrows from LiDAR data
The necessary requirements for identifying ridges and furrows from LiDAR data were seperated into three categories.
In order to identify furrows within a LiDAR dataset, the geometric properties of a typical furrow and ridge must be known. Furrows are created by plowing the land, creating parallel troughs with a uniform width that typically span the length of a field. From their study of surface roughness, Beaudoin et al. (1990) suggest that typical ridge-furrow spacing ranges from 0.4 to 1.20 meters in width. It was essential that the LiDAR data being used featured a data density high enough to capture these features. The Peterborough LiDAR Dataset has a pulse density listed at 8 pulses per square meter (OMAFRA, 2017). This density was deemed to be high enough that the furrows and ridges could be identified within the point cloud, regardless of their spacing.
It was also necessary for the LiDAR data to be collected at a time when the fields were barren or near barren, as crops would obscure the soil surface. Fortunately, it is common for aerial LiDAR scans to occur in the early spring, during the period when the winter’s snow has melted, but before plants are able to bloom (NOAA, 2009). The Peterborough dataset was collected in early May, which meant that the fields were barren or mostly barren.
Finally, it was necessary to use an algorithm capable of identifying furrows based upon their geometric properties. Turner et al. (2014) note that the surface roughness of tilled land is anisotropic due to the lack of variation in furrow orientation. The MaxAnisotropyDev tool in WhiteboxTools automatically identifies anisotropic landforms in a digital elevation model (DEM). If the LiDAR data is interpolated into a raster at a fine enough grid resolution, the MaxAnisotropyDev tool should be capable of distinguishing the furrows from the surrounding landscape.