Knickpoint initiation and distribution within fluvial networks: 236 waterfalls in the Waipaoa River, North Island, New Zealand

If knickpoints transmit signals of base level fall in river networks, then improvements in our understanding of their retreat rate and basin wide distribution helps constrain the transient response following perturbation. Many studies of knickpoint retreat focus on the response of trunk streams to base level fall. Here we examine the response of an entire fluvial network, as recorded by 236 active knickpoints distributed within the Waipaoa River on the North Island of New Zealand. Base level fall within the Waipaoa catchment initiated 18,000 years ago in response to a climatically triggered and tectonically exacerbated pulse of incision. Using observations from field work, aerial photo analysis and a digital elevation model (DEM), we study the knickpoint positions within the network. We find that ∼ 70% of the knickpoints are located at drainage areas between 1 × 10⁵ m² and 1 × 10⁶ m² and more than half are < 1 km upstream of a large change in drainage area. For the knickpoints < 1 km upstream of large tributary junctions, we find that the retreat distances were well correlated with the tributaries' drainage areas. In order to determine how a pulse of incision distributes itself throughout a fluvial network, we develop two simple, end-member models and compare their behavior to the observed knickpoint distribution in the Waipaoa. In the first model, we propose that a knickpoint initiated at the basin outlet retreats upstream and distributes the signal throughout the network at a rate that is a power law function of drainage area. In the second model, we propose that knickpoints form near a threshold drainage area, below which channels cannot incise with the same efficiency as possible in downstream reaches. Though neither model addresses the along-stream variability in substrate or knickpoint form, the misfit between the modeled and the observed knickpoints' along-stream positions are surprisingly low (< 1 km; ∼ 1.25% of the total stream length) for knickpoints with drainage areas < 1 × 10⁶ m². Large misfits (up to 3.5 km) are observed for knickpoints with present-day drainage areas greater than 1 × 10⁶ m². The large, single step in channel elevation that characterizes knickpoints presently observed in tributaries of the Waipaoa River may not characterize the base level fall signal that propagated through the trunk streams. Evidence for progressive (rather than instantaneous) incision in the trunk streams, the knickpoints' vicinities to tributary junctions and the equivalent success of the two-end member models lead us to conclude that the present positions of the 236 observed knickpoints are largely a consequence of thresholds in channel incision at low drainage areas.