Integration of fuzzy logic and image analysis for the detection of gullies in the Calhoun critical zone observatory using airborne LiDAR data

Abstract:

The entire Piedmont of the Southeastern United States, where the Calhoun Critical Zone Observatory (CCZO) is located, experienced one of the most severe erosive events in the United States during last two centuries. Forested areas were cleared to cultivate cotton, tobacco and other crops during the nineteenth and early twentieth century and these land use change, together with intense rainfalls, initiated deep gullying. An accurate mapping of these landforms is important since, despite some gully stabilization and reforestation efforts, gullies are still major contributors of sediment to streams. Mapping gullies in the CCZO area is hindered by the presence of dense canopy which precludes the identification through aerial photogrammetry and other traditional remote sensing methods. Moreover, the wide spatial extent of the gullies makes detailed field surveys, for the identification and characterization of entire gullies, a very large and expensive proposition.

This work aims to develop and assess an automated set of algorithms to detect and map gullies using morphological characteristics retrieved by very high resolution imagery (VHRI). A one-meter resolution LiDAR DEM is used to derive different morphometric indices whose combination, carried out using spatial analysis methods and fuzzy logic rules, are a tool to identify gullies. This spatial model has been calibrated using the reference perimeters of two gullies that we measured during a recent field survey. The entire procedure attempts to provide estimates of gully erosion patterns, which characterize the entire Calhoun CZO area and to develop and evaluate a method to measure characteristic features of gullies (i.e. depth and volume).