Cutoffs induce cutoffs on meandering rivers through persistent nonlocal effects

Abstract:

As a river meanders and increases in length, a bend may migrate into itself and initiate a cutoff. Cutoffs serve as important refugia for flora and fauna, create accommodation space for floodplains, and contribute to meander dynamics. As cutoffs occur, they locally increase water surface slope and nonlocally accelerate channel migration and widening upstream and downstream of their bend. These nonlocal effects suggest that cutoffs may accelerate the formation of more cutoffs in neighboring bends, resulting in clusters. Despite their significance, we do not know if cutoffs occur randomly along a river, or if they are clustered in time and space. Therefore, we set out to test if cutoffs can cluster by identifying cutoffs whose nonlocal effects overlapped both spatially and temporally with another cutoff. We mapped the locations of 599 cutoffs across the Amazon Basin and found that 48% of those cutoffs occurred within 50 channel widths of one another, suggesting some are clustered. Furthermore, we used yearly water masks derived from the Global Surface Water dataset of 10 reaches with potential clustering to identify the cause of clustering. First, we define the temporal signature of nonlocal effects induced by cutoffs, which can last for up to 7 years after occurrence. We track nonlocal effects by comparing background lateral migration rates to migration rates in a zone around the cutoff. We identify cutoff clusters where the migration rates in both regions are similar before the cutoff, but after the cutoff migration rate in the influenced region is at least 25% greater than in the background. We consider cutoffs clustered if a river meets these criteria and if at least one more cutoff occurred along a bend within the zone of influence of the initial cutoff. Our results show that cutoffs on meandering rivers can be clustered because of the persistent nature of nonlocal effects. Additionally, we find that cutoff clusters amplify nonlocal effects, as the accelerated migration rate resulting from cutoffs in a cluster is greater and lasts longer than would occur from an isolated event. Through illustrating that cutoffs can cluster due to nonlocal effects, our work suggests that the distribution of oxbow lakes and cutoff events may not be random in space.