Acoustic backscatter-Data and Python Code

These data were compiled for investigating the relationship between acoustic backscattering by riverbeds composed of various riverbed substrates (bed sediment), and for developing and testing a probabilistic model for substrate classification based on high-frequency multibeam acoustic backscatter. The model is described in Buscombe et al. (2017). The data consist of various quantities on coincident grids, from various sites along the Colorado River in Grand Canyon, including water depth, bed roughness, the area (or footprint) of the acoustic beam, unfiltered and filtered backscatter magnitude, sediment classification (for each location, 1 of 5 sediment classes in a categorical scheme), and the probabilities for each of 5 classes considered by the model. Files are organized by site, themselves denoted by river mile (RM) which is the linear distance downstream of Lees Ferry, Arizona. The so-called unfiltered backscatter has been corrected for various water, sediment and acoustic variables that might cause backscatter to vary independent of bed sediment, but has not been corrected for the effects of topography. It is shown by Buscombe et al. (2017) that topography has a major influence of the relationship between backscatter magnitude and substrate. Therefore, the topographic effects on backscatter are filtered out, resulting in filtered, or 'compositional' backscatter which is more strongly related to substrate type, and therefore serves as the basic and input to the probabilistic substrate classification model.