Knickpoints in Fluvial Systems: Comparing Models of Basin-Wide Propagation and Initiation at Erosional Thresholds

Knickpoints, which we define morphologically as discrete negative steps in the long profiles of rivers, have been frequently associated with the dynamic adjustment of channels following a change in climate or tectonics. The danger in this process-based definition arises from the numerous circumstances, both static (e.g. substrate erodability contrasts) and dynamic (e.g. stream capture), that generate knickpoint morphologies. In addition, because changes in knickpoint form are often too slow to measure, their role as upstream propagating fronts of adjustment is most often inferred rather than observed. Most previous studies of knickpoint retreat have examined the response of a single channel to base level fall, but we propose that the timing and pattern of knickpoint distribution throughout entire fluvial networks must be characterized in order to ultimately understand landscape response times to external forcing and the history of sediment delivery to offshore basins. To explore this, we examined 236 knickpoints distributed within the fluvial network of the Waipaoa River on the North Island of New Zealand. A climatically triggered pulse of incision initiated 18,000 years ago lowered base level 50-100 m along the Waipaoa mainstem. Using field measurements, aerial photo analysis and digital elevation data, we studied the knickpoints' positions within the network. We found that ~70% of the knickpoints are located at drainage areas between 1 x 10⁵ m² and 1 x 10⁶ m² and more than half are < 1 km upstream of tributary junctions. This observed knickpoint distribution in the Waipaoa was compared to two end-member models for knickpoint behavior. In the first model, we examined the time-evolution of a knickpoint as it propagates upstream and is distributed throughout the network at a rate that is a power law function of drainage area. In the second, we examined if knickpoints form at threshold drainage areas where their fluvial erosive potential, as determined by water and sediment flux, is incapable of incising as rapidly as downstream reaches. Though neither model addressed along-stream variability in substrate or knickpoint form, surprisingly, both models provided highly accurate fits to the ~70% of knickpoints at drainage areas < 1 x 10⁶ m². Though the field and modeled results suggest that the present positions of the 236 observed knickpoints are determined by this threshold area behavior, explaining the basin-wide time evolution of a pulse of incision will require further model refinement and field observation.