A method to detect abrupt shifts in river channel position using a Landsat‐derived water occurrence record

Abstract

AbstractThe lateral migration of river channels is an important control on the evolution of alluvial fans, deltas, and floodplains. Lateral migration consists of both gradual riverbank migration and abrupt shifts in location due to avulsions or cutoffs. Methods exist to measure bank migration, but abrupt shifts in position are rarely considered or are not emphasized. Here we describe a new method using Landsat‐derived water occurrence images that primarily focuses on detecting when a channel has abruptly shifted position, either from avulsion or cutoff. The method does not assume any a priori model of channel geometry or evolution. Within a given area of interest, binary channel images created from the fluvial water occurrence record are stacked through time. Then a channel shift intensity, , is created by estimating the number of possible ending times for fluvial water voxels (a point in three‐dimensional space) in the stacked occurrence record. The number of possible end‐times for fluvial water voxels within a given region of the occurrence record reveals the likelihood that a reach of a river underwent an abrupt channel shift during the observation period. We present the results of this analysis for a 194 481 km2 region of the Amazonian basin. Follow‐up validation found three avulsions and 270 cutoffs within regions identified by this method. We show that areas above a threshold contain an avulsion or cutoff with high probability. This highlights the method's potential to detect and quantify abrupt channel shifts at the basin scale. The method also successfully distinguishes between abrupt channel movement and complex braiding behaviour. As this method is applicable to any binary water‐masked time series images, future applications of this method have the potential to provide insight into the controls and spatial variance of avulsions and cutoffs across a variety of scales.