Abstract Most—if not all—alluvial rivers experience overbank floods. On some rivers, these floods form natural levees, whereas on others they do not. Existing theories for how levees form on riverbanks rely on the physics of sediment transport across the channel margin, but applying them to real rivers does not yield good predictions of why one river may have abundant levees, while another might have very few. To improve theories of levee formation, we adopted a statistical approach. We randomly sampled river reaches (∼10 km long) across the continental USA, evaluated whether they had no levees, sparse levees, or abundant levees using high resolution topography, and gathered a database of several readily available parameters that we hypothesized might influence levee formation. Then, we used a statistical model to deduce which of the variables were good predictors of the abundance of levees on each river reach. Discharge and discharge variability are poor predictors of levee abundance, whereas river slope and lateral migration rate are the strongest predictors. Levees abound on gently sloping and quickly migrating rivers. We hypothesize that the slope effect exists because strong flow velocities across the channel bank on steep floodplains smear out levee deposits. We think that levees are more common on rapidly migrating rivers because they tend to carry rich sediment loads, so suspended sediment is likely to escape and deposit during overbank floods. We found levees (either sparse or abundant) on about ¾ of all alluvial river reaches in our sample.
Barefoot et al. (Wed,) studied this question.